{
  "best_global_step": 48000,
  "best_metric": 4.58038854598999,
  "best_model_checkpoint": "./model_save/checkpoint-48000",
  "epoch": 1.0,
  "eval_steps": 1000,
  "global_step": 48833,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00010238977740462392,
      "grad_norm": 3.237281322479248,
      "learning_rate": 1.2e-06,
      "loss": 11.7826,
      "step": 5
    },
    {
      "epoch": 0.00020477955480924785,
      "grad_norm": 3.256563425064087,
      "learning_rate": 2.6999999999999996e-06,
      "loss": 11.7333,
      "step": 10
    },
    {
      "epoch": 0.00030716933221387174,
      "grad_norm": 3.1677451133728027,
      "learning_rate": 4.2e-06,
      "loss": 11.6291,
      "step": 15
    },
    {
      "epoch": 0.0004095591096184957,
      "grad_norm": 3.29661226272583,
      "learning_rate": 5.7e-06,
      "loss": 11.476,
      "step": 20
    },
    {
      "epoch": 0.0005119488870231196,
      "grad_norm": 3.558603048324585,
      "learning_rate": 7.2e-06,
      "loss": 11.2652,
      "step": 25
    },
    {
      "epoch": 0.0006143386644277435,
      "grad_norm": 2.8637077808380127,
      "learning_rate": 8.7e-06,
      "loss": 11.0138,
      "step": 30
    },
    {
      "epoch": 0.0007167284418323675,
      "grad_norm": 2.0960562229156494,
      "learning_rate": 1.02e-05,
      "loss": 10.7718,
      "step": 35
    },
    {
      "epoch": 0.0008191182192369914,
      "grad_norm": 1.7101972103118896,
      "learning_rate": 1.17e-05,
      "loss": 10.5829,
      "step": 40
    },
    {
      "epoch": 0.0009215079966416153,
      "grad_norm": 1.4555027484893799,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 10.3681,
      "step": 45
    },
    {
      "epoch": 0.0010238977740462392,
      "grad_norm": 1.3247417211532593,
      "learning_rate": 1.47e-05,
      "loss": 10.2536,
      "step": 50
    },
    {
      "epoch": 0.0011262875514508632,
      "grad_norm": 1.1643482446670532,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 10.1509,
      "step": 55
    },
    {
      "epoch": 0.001228677328855487,
      "grad_norm": 1.153939127922058,
      "learning_rate": 1.7699999999999997e-05,
      "loss": 10.0431,
      "step": 60
    },
    {
      "epoch": 0.001331067106260111,
      "grad_norm": 1.1378567218780518,
      "learning_rate": 1.92e-05,
      "loss": 9.9499,
      "step": 65
    },
    {
      "epoch": 0.001433456883664735,
      "grad_norm": 1.1022170782089233,
      "learning_rate": 2.07e-05,
      "loss": 9.8486,
      "step": 70
    },
    {
      "epoch": 0.0015358466610693588,
      "grad_norm": 1.0503519773483276,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 9.7697,
      "step": 75
    },
    {
      "epoch": 0.0016382364384739828,
      "grad_norm": 1.0037927627563477,
      "learning_rate": 2.3699999999999997e-05,
      "loss": 9.6688,
      "step": 80
    },
    {
      "epoch": 0.0017406262158786068,
      "grad_norm": 0.9526827335357666,
      "learning_rate": 2.52e-05,
      "loss": 9.5799,
      "step": 85
    },
    {
      "epoch": 0.0018430159932832306,
      "grad_norm": 0.8503618836402893,
      "learning_rate": 2.6699999999999995e-05,
      "loss": 9.5157,
      "step": 90
    },
    {
      "epoch": 0.0019454057706878546,
      "grad_norm": 0.8890340328216553,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 9.4274,
      "step": 95
    },
    {
      "epoch": 0.0020477955480924783,
      "grad_norm": 0.8748540878295898,
      "learning_rate": 2.97e-05,
      "loss": 9.3157,
      "step": 100
    },
    {
      "epoch": 0.0021501853254971026,
      "grad_norm": 0.8950842022895813,
      "learning_rate": 3.119999999999999e-05,
      "loss": 9.2445,
      "step": 105
    },
    {
      "epoch": 0.0022525751029017264,
      "grad_norm": 0.8296220898628235,
      "learning_rate": 3.2699999999999995e-05,
      "loss": 9.1356,
      "step": 110
    },
    {
      "epoch": 0.00235496488030635,
      "grad_norm": 0.8047326803207397,
      "learning_rate": 3.42e-05,
      "loss": 9.0434,
      "step": 115
    },
    {
      "epoch": 0.002457354657710974,
      "grad_norm": 0.8104546070098877,
      "learning_rate": 3.5699999999999994e-05,
      "loss": 8.9745,
      "step": 120
    },
    {
      "epoch": 0.002559744435115598,
      "grad_norm": 0.8774133324623108,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 8.9045,
      "step": 125
    },
    {
      "epoch": 0.002662134212520222,
      "grad_norm": 0.6810584664344788,
      "learning_rate": 3.87e-05,
      "loss": 8.8046,
      "step": 130
    },
    {
      "epoch": 0.0027645239899248457,
      "grad_norm": 0.6244851350784302,
      "learning_rate": 4.02e-05,
      "loss": 8.7529,
      "step": 135
    },
    {
      "epoch": 0.00286691376732947,
      "grad_norm": 0.7336015701293945,
      "learning_rate": 4.17e-05,
      "loss": 8.6713,
      "step": 140
    },
    {
      "epoch": 0.0029693035447340937,
      "grad_norm": 0.7508506774902344,
      "learning_rate": 4.319999999999999e-05,
      "loss": 8.6402,
      "step": 145
    },
    {
      "epoch": 0.0030716933221387175,
      "grad_norm": 0.7344472408294678,
      "learning_rate": 4.4699999999999996e-05,
      "loss": 8.5754,
      "step": 150
    },
    {
      "epoch": 0.0031740830995433417,
      "grad_norm": 0.5286904573440552,
      "learning_rate": 4.62e-05,
      "loss": 8.5107,
      "step": 155
    },
    {
      "epoch": 0.0032764728769479655,
      "grad_norm": 0.5070658326148987,
      "learning_rate": 4.7699999999999994e-05,
      "loss": 8.4627,
      "step": 160
    },
    {
      "epoch": 0.0033788626543525893,
      "grad_norm": 0.5503280162811279,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 8.4416,
      "step": 165
    },
    {
      "epoch": 0.0034812524317572135,
      "grad_norm": 0.5057103633880615,
      "learning_rate": 5.07e-05,
      "loss": 8.4661,
      "step": 170
    },
    {
      "epoch": 0.0035836422091618373,
      "grad_norm": 0.5336150527000427,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 8.3401,
      "step": 175
    },
    {
      "epoch": 0.003686031986566461,
      "grad_norm": 0.4095563590526581,
      "learning_rate": 5.369999999999999e-05,
      "loss": 8.2784,
      "step": 180
    },
    {
      "epoch": 0.0037884217639710853,
      "grad_norm": 0.37456509470939636,
      "learning_rate": 5.519999999999999e-05,
      "loss": 8.2367,
      "step": 185
    },
    {
      "epoch": 0.003890811541375709,
      "grad_norm": 0.4807591736316681,
      "learning_rate": 5.6699999999999996e-05,
      "loss": 8.2496,
      "step": 190
    },
    {
      "epoch": 0.003993201318780333,
      "grad_norm": 0.4516820013523102,
      "learning_rate": 5.82e-05,
      "loss": 8.1633,
      "step": 195
    },
    {
      "epoch": 0.004095591096184957,
      "grad_norm": 0.40326714515686035,
      "learning_rate": 5.97e-05,
      "loss": 8.1903,
      "step": 200
    },
    {
      "epoch": 0.0041979808735895805,
      "grad_norm": 0.4950360357761383,
      "learning_rate": 6.12e-05,
      "loss": 8.1316,
      "step": 205
    },
    {
      "epoch": 0.004300370650994205,
      "grad_norm": 0.6782643795013428,
      "learning_rate": 6.269999999999999e-05,
      "loss": 8.1855,
      "step": 210
    },
    {
      "epoch": 0.004402760428398829,
      "grad_norm": 0.5185160636901855,
      "learning_rate": 6.419999999999999e-05,
      "loss": 8.0867,
      "step": 215
    },
    {
      "epoch": 0.004505150205803453,
      "grad_norm": 0.5909910202026367,
      "learning_rate": 6.57e-05,
      "loss": 8.098,
      "step": 220
    },
    {
      "epoch": 0.0046075399832080765,
      "grad_norm": 0.5086753368377686,
      "learning_rate": 6.72e-05,
      "loss": 8.0583,
      "step": 225
    },
    {
      "epoch": 0.0047099297606127,
      "grad_norm": 0.5272602438926697,
      "learning_rate": 6.87e-05,
      "loss": 8.0315,
      "step": 230
    },
    {
      "epoch": 0.004812319538017324,
      "grad_norm": 0.9392020106315613,
      "learning_rate": 7.02e-05,
      "loss": 8.0517,
      "step": 235
    },
    {
      "epoch": 0.004914709315421948,
      "grad_norm": 0.696977436542511,
      "learning_rate": 7.17e-05,
      "loss": 8.0084,
      "step": 240
    },
    {
      "epoch": 0.0050170990928265725,
      "grad_norm": 0.6232326030731201,
      "learning_rate": 7.319999999999999e-05,
      "loss": 7.9619,
      "step": 245
    },
    {
      "epoch": 0.005119488870231196,
      "grad_norm": 1.3301762342453003,
      "learning_rate": 7.47e-05,
      "loss": 7.9713,
      "step": 250
    },
    {
      "epoch": 0.00522187864763582,
      "grad_norm": 3.402907609939575,
      "learning_rate": 7.62e-05,
      "loss": 8.0243,
      "step": 255
    },
    {
      "epoch": 0.005324268425040444,
      "grad_norm": 0.739121675491333,
      "learning_rate": 7.769999999999999e-05,
      "loss": 7.888,
      "step": 260
    },
    {
      "epoch": 0.005426658202445068,
      "grad_norm": 0.6031880974769592,
      "learning_rate": 7.92e-05,
      "loss": 7.8527,
      "step": 265
    },
    {
      "epoch": 0.0055290479798496914,
      "grad_norm": 0.7053850889205933,
      "learning_rate": 8.07e-05,
      "loss": 7.8751,
      "step": 270
    },
    {
      "epoch": 0.005631437757254316,
      "grad_norm": 0.564606249332428,
      "learning_rate": 8.22e-05,
      "loss": 7.7992,
      "step": 275
    },
    {
      "epoch": 0.00573382753465894,
      "grad_norm": 0.5167285203933716,
      "learning_rate": 8.37e-05,
      "loss": 7.7863,
      "step": 280
    },
    {
      "epoch": 0.005836217312063564,
      "grad_norm": 0.816970944404602,
      "learning_rate": 8.519999999999998e-05,
      "loss": 7.7767,
      "step": 285
    },
    {
      "epoch": 0.0059386070894681875,
      "grad_norm": 0.7622373700141907,
      "learning_rate": 8.669999999999998e-05,
      "loss": 7.7498,
      "step": 290
    },
    {
      "epoch": 0.006040996866872811,
      "grad_norm": 0.7365726828575134,
      "learning_rate": 8.819999999999999e-05,
      "loss": 7.769,
      "step": 295
    },
    {
      "epoch": 0.006143386644277435,
      "grad_norm": 0.5777131915092468,
      "learning_rate": 8.969999999999998e-05,
      "loss": 7.7428,
      "step": 300
    },
    {
      "epoch": 0.00624577642168206,
      "grad_norm": 0.7264725565910339,
      "learning_rate": 9.12e-05,
      "loss": 7.7056,
      "step": 305
    },
    {
      "epoch": 0.0063481661990866835,
      "grad_norm": 0.5708840489387512,
      "learning_rate": 9.269999999999999e-05,
      "loss": 7.661,
      "step": 310
    },
    {
      "epoch": 0.006450555976491307,
      "grad_norm": 0.6288994550704956,
      "learning_rate": 9.419999999999999e-05,
      "loss": 7.6555,
      "step": 315
    },
    {
      "epoch": 0.006552945753895931,
      "grad_norm": 1.0663909912109375,
      "learning_rate": 9.57e-05,
      "loss": 7.6393,
      "step": 320
    },
    {
      "epoch": 0.006655335531300555,
      "grad_norm": 0.7720368504524231,
      "learning_rate": 9.719999999999999e-05,
      "loss": 7.6684,
      "step": 325
    },
    {
      "epoch": 0.006757725308705179,
      "grad_norm": 0.983554482460022,
      "learning_rate": 9.87e-05,
      "loss": 7.633,
      "step": 330
    },
    {
      "epoch": 0.006860115086109802,
      "grad_norm": 0.7408352494239807,
      "learning_rate": 0.0001002,
      "loss": 7.6358,
      "step": 335
    },
    {
      "epoch": 0.006962504863514427,
      "grad_norm": 0.7378109097480774,
      "learning_rate": 0.00010169999999999999,
      "loss": 7.5782,
      "step": 340
    },
    {
      "epoch": 0.007064894640919051,
      "grad_norm": 0.6890093684196472,
      "learning_rate": 0.00010319999999999999,
      "loss": 7.5093,
      "step": 345
    },
    {
      "epoch": 0.007167284418323675,
      "grad_norm": 0.974371612071991,
      "learning_rate": 0.00010469999999999998,
      "loss": 7.4945,
      "step": 350
    },
    {
      "epoch": 0.007269674195728298,
      "grad_norm": 0.6680528521537781,
      "learning_rate": 0.00010619999999999998,
      "loss": 7.4819,
      "step": 355
    },
    {
      "epoch": 0.007372063973132922,
      "grad_norm": 1.2959836721420288,
      "learning_rate": 0.00010769999999999999,
      "loss": 7.5094,
      "step": 360
    },
    {
      "epoch": 0.007474453750537546,
      "grad_norm": 0.6721152663230896,
      "learning_rate": 0.00010919999999999998,
      "loss": 7.4758,
      "step": 365
    },
    {
      "epoch": 0.007576843527942171,
      "grad_norm": 0.6479594111442566,
      "learning_rate": 0.0001107,
      "loss": 7.4569,
      "step": 370
    },
    {
      "epoch": 0.0076792333053467944,
      "grad_norm": 1.128847599029541,
      "learning_rate": 0.00011219999999999999,
      "loss": 7.4654,
      "step": 375
    },
    {
      "epoch": 0.007781623082751418,
      "grad_norm": 0.9047718048095703,
      "learning_rate": 0.00011369999999999999,
      "loss": 7.4445,
      "step": 380
    },
    {
      "epoch": 0.007884012860156043,
      "grad_norm": 0.6440275311470032,
      "learning_rate": 0.0001152,
      "loss": 7.3781,
      "step": 385
    },
    {
      "epoch": 0.007986402637560666,
      "grad_norm": 0.9747110605239868,
      "learning_rate": 0.00011669999999999999,
      "loss": 7.378,
      "step": 390
    },
    {
      "epoch": 0.00808879241496529,
      "grad_norm": 0.8239765167236328,
      "learning_rate": 0.0001182,
      "loss": 7.3363,
      "step": 395
    },
    {
      "epoch": 0.008191182192369913,
      "grad_norm": 1.1421988010406494,
      "learning_rate": 0.0001197,
      "loss": 7.3996,
      "step": 400
    },
    {
      "epoch": 0.008293571969774538,
      "grad_norm": 0.9112151861190796,
      "learning_rate": 0.00012119999999999999,
      "loss": 7.4102,
      "step": 405
    },
    {
      "epoch": 0.008395961747179161,
      "grad_norm": 0.6386898756027222,
      "learning_rate": 0.00012269999999999997,
      "loss": 7.3811,
      "step": 410
    },
    {
      "epoch": 0.008498351524583786,
      "grad_norm": 0.7662557363510132,
      "learning_rate": 0.00012419999999999998,
      "loss": 7.3087,
      "step": 415
    },
    {
      "epoch": 0.00860074130198841,
      "grad_norm": 0.8661086559295654,
      "learning_rate": 0.0001257,
      "loss": 7.2616,
      "step": 420
    },
    {
      "epoch": 0.008703131079393033,
      "grad_norm": 0.7727162837982178,
      "learning_rate": 0.00012719999999999997,
      "loss": 7.2938,
      "step": 425
    },
    {
      "epoch": 0.008805520856797658,
      "grad_norm": 0.6405845284461975,
      "learning_rate": 0.00012869999999999998,
      "loss": 7.2836,
      "step": 430
    },
    {
      "epoch": 0.00890791063420228,
      "grad_norm": 1.2078088521957397,
      "learning_rate": 0.0001302,
      "loss": 7.2712,
      "step": 435
    },
    {
      "epoch": 0.009010300411606905,
      "grad_norm": 0.9001152515411377,
      "learning_rate": 0.00013169999999999998,
      "loss": 7.249,
      "step": 440
    },
    {
      "epoch": 0.009112690189011528,
      "grad_norm": 0.8163766264915466,
      "learning_rate": 0.00013319999999999999,
      "loss": 7.2852,
      "step": 445
    },
    {
      "epoch": 0.009215079966416153,
      "grad_norm": 0.9464946389198303,
      "learning_rate": 0.0001347,
      "loss": 7.1945,
      "step": 450
    },
    {
      "epoch": 0.009317469743820778,
      "grad_norm": 0.8073117733001709,
      "learning_rate": 0.0001362,
      "loss": 7.1885,
      "step": 455
    },
    {
      "epoch": 0.0094198595212254,
      "grad_norm": 0.7544310092926025,
      "learning_rate": 0.00013769999999999999,
      "loss": 7.219,
      "step": 460
    },
    {
      "epoch": 0.009522249298630025,
      "grad_norm": 1.2347111701965332,
      "learning_rate": 0.0001392,
      "loss": 7.1837,
      "step": 465
    },
    {
      "epoch": 0.009624639076034648,
      "grad_norm": 0.7268940210342407,
      "learning_rate": 0.00014069999999999998,
      "loss": 7.1897,
      "step": 470
    },
    {
      "epoch": 0.009727028853439273,
      "grad_norm": 0.9710267186164856,
      "learning_rate": 0.0001422,
      "loss": 7.1928,
      "step": 475
    },
    {
      "epoch": 0.009829418630843896,
      "grad_norm": 0.8327569365501404,
      "learning_rate": 0.00014369999999999997,
      "loss": 7.2138,
      "step": 480
    },
    {
      "epoch": 0.00993180840824852,
      "grad_norm": 0.9703325033187866,
      "learning_rate": 0.00014519999999999998,
      "loss": 7.1411,
      "step": 485
    },
    {
      "epoch": 0.010034198185653145,
      "grad_norm": 1.184667706489563,
      "learning_rate": 0.0001467,
      "loss": 7.1319,
      "step": 490
    },
    {
      "epoch": 0.010136587963057768,
      "grad_norm": 0.8813581466674805,
      "learning_rate": 0.0001482,
      "loss": 7.1491,
      "step": 495
    },
    {
      "epoch": 0.010238977740462393,
      "grad_norm": 0.7553069591522217,
      "learning_rate": 0.00014969999999999998,
      "loss": 7.1247,
      "step": 500
    },
    {
      "epoch": 0.010341367517867016,
      "grad_norm": 0.8524662256240845,
      "learning_rate": 0.0001512,
      "loss": 7.0651,
      "step": 505
    },
    {
      "epoch": 0.01044375729527164,
      "grad_norm": 0.9542851448059082,
      "learning_rate": 0.0001527,
      "loss": 7.0811,
      "step": 510
    },
    {
      "epoch": 0.010546147072676265,
      "grad_norm": 0.839388906955719,
      "learning_rate": 0.00015419999999999998,
      "loss": 7.1088,
      "step": 515
    },
    {
      "epoch": 0.010648536850080888,
      "grad_norm": 0.7219004034996033,
      "learning_rate": 0.0001557,
      "loss": 7.073,
      "step": 520
    },
    {
      "epoch": 0.010750926627485512,
      "grad_norm": 1.1495189666748047,
      "learning_rate": 0.0001572,
      "loss": 7.0617,
      "step": 525
    },
    {
      "epoch": 0.010853316404890135,
      "grad_norm": 0.9143120646476746,
      "learning_rate": 0.00015869999999999998,
      "loss": 7.095,
      "step": 530
    },
    {
      "epoch": 0.01095570618229476,
      "grad_norm": 0.6772273182868958,
      "learning_rate": 0.0001602,
      "loss": 7.0889,
      "step": 535
    },
    {
      "epoch": 0.011058095959699383,
      "grad_norm": 0.6970393657684326,
      "learning_rate": 0.0001617,
      "loss": 7.0662,
      "step": 540
    },
    {
      "epoch": 0.011160485737104008,
      "grad_norm": 0.7849879264831543,
      "learning_rate": 0.0001632,
      "loss": 7.0654,
      "step": 545
    },
    {
      "epoch": 0.011262875514508632,
      "grad_norm": 0.6196251511573792,
      "learning_rate": 0.0001647,
      "loss": 7.0607,
      "step": 550
    },
    {
      "epoch": 0.011365265291913255,
      "grad_norm": 0.9614342451095581,
      "learning_rate": 0.0001662,
      "loss": 7.0476,
      "step": 555
    },
    {
      "epoch": 0.01146765506931788,
      "grad_norm": 0.8006017208099365,
      "learning_rate": 0.0001677,
      "loss": 6.9906,
      "step": 560
    },
    {
      "epoch": 0.011570044846722503,
      "grad_norm": 0.7031441926956177,
      "learning_rate": 0.00016919999999999997,
      "loss": 6.9459,
      "step": 565
    },
    {
      "epoch": 0.011672434624127127,
      "grad_norm": 0.770129919052124,
      "learning_rate": 0.00017069999999999998,
      "loss": 6.997,
      "step": 570
    },
    {
      "epoch": 0.01177482440153175,
      "grad_norm": 0.7633779644966125,
      "learning_rate": 0.00017219999999999998,
      "loss": 7.0089,
      "step": 575
    },
    {
      "epoch": 0.011877214178936375,
      "grad_norm": 1.0391132831573486,
      "learning_rate": 0.00017369999999999997,
      "loss": 6.9777,
      "step": 580
    },
    {
      "epoch": 0.011979603956341,
      "grad_norm": 0.8871012330055237,
      "learning_rate": 0.00017519999999999998,
      "loss": 6.9827,
      "step": 585
    },
    {
      "epoch": 0.012081993733745622,
      "grad_norm": 0.6953468918800354,
      "learning_rate": 0.00017669999999999999,
      "loss": 6.9325,
      "step": 590
    },
    {
      "epoch": 0.012184383511150247,
      "grad_norm": 0.7784765362739563,
      "learning_rate": 0.00017819999999999997,
      "loss": 6.9506,
      "step": 595
    },
    {
      "epoch": 0.01228677328855487,
      "grad_norm": 0.8684126734733582,
      "learning_rate": 0.00017969999999999998,
      "loss": 6.9306,
      "step": 600
    },
    {
      "epoch": 0.012389163065959495,
      "grad_norm": 0.9275881052017212,
      "learning_rate": 0.00018119999999999999,
      "loss": 6.9174,
      "step": 605
    },
    {
      "epoch": 0.01249155284336412,
      "grad_norm": 0.6531779766082764,
      "learning_rate": 0.00018269999999999997,
      "loss": 6.9123,
      "step": 610
    },
    {
      "epoch": 0.012593942620768742,
      "grad_norm": 0.758029043674469,
      "learning_rate": 0.00018419999999999998,
      "loss": 6.9776,
      "step": 615
    },
    {
      "epoch": 0.012696332398173367,
      "grad_norm": 0.7189378142356873,
      "learning_rate": 0.0001857,
      "loss": 6.9247,
      "step": 620
    },
    {
      "epoch": 0.01279872217557799,
      "grad_norm": 0.8125119209289551,
      "learning_rate": 0.0001872,
      "loss": 6.8839,
      "step": 625
    },
    {
      "epoch": 0.012901111952982615,
      "grad_norm": 0.8374565243721008,
      "learning_rate": 0.00018869999999999998,
      "loss": 6.9138,
      "step": 630
    },
    {
      "epoch": 0.013003501730387237,
      "grad_norm": 0.7942729592323303,
      "learning_rate": 0.0001902,
      "loss": 6.866,
      "step": 635
    },
    {
      "epoch": 0.013105891507791862,
      "grad_norm": 0.9105733633041382,
      "learning_rate": 0.0001917,
      "loss": 6.876,
      "step": 640
    },
    {
      "epoch": 0.013208281285196487,
      "grad_norm": 0.873391330242157,
      "learning_rate": 0.00019319999999999998,
      "loss": 6.8388,
      "step": 645
    },
    {
      "epoch": 0.01331067106260111,
      "grad_norm": 0.7067049145698547,
      "learning_rate": 0.0001947,
      "loss": 6.8088,
      "step": 650
    },
    {
      "epoch": 0.013413060840005734,
      "grad_norm": 1.2663843631744385,
      "learning_rate": 0.0001962,
      "loss": 6.8106,
      "step": 655
    },
    {
      "epoch": 0.013515450617410357,
      "grad_norm": 1.0296598672866821,
      "learning_rate": 0.00019769999999999998,
      "loss": 6.8277,
      "step": 660
    },
    {
      "epoch": 0.013617840394814982,
      "grad_norm": 0.810079038143158,
      "learning_rate": 0.0001992,
      "loss": 6.8173,
      "step": 665
    },
    {
      "epoch": 0.013720230172219605,
      "grad_norm": 0.8228099942207336,
      "learning_rate": 0.0002007,
      "loss": 6.8861,
      "step": 670
    },
    {
      "epoch": 0.01382261994962423,
      "grad_norm": 0.7651252746582031,
      "learning_rate": 0.0002022,
      "loss": 6.8165,
      "step": 675
    },
    {
      "epoch": 0.013925009727028854,
      "grad_norm": 0.8029287457466125,
      "learning_rate": 0.0002037,
      "loss": 6.7644,
      "step": 680
    },
    {
      "epoch": 0.014027399504433477,
      "grad_norm": 0.852442741394043,
      "learning_rate": 0.0002052,
      "loss": 6.8568,
      "step": 685
    },
    {
      "epoch": 0.014129789281838102,
      "grad_norm": 0.8873968124389648,
      "learning_rate": 0.00020669999999999996,
      "loss": 6.7989,
      "step": 690
    },
    {
      "epoch": 0.014232179059242725,
      "grad_norm": 0.7122694849967957,
      "learning_rate": 0.00020819999999999996,
      "loss": 6.7896,
      "step": 695
    },
    {
      "epoch": 0.01433456883664735,
      "grad_norm": 0.668544590473175,
      "learning_rate": 0.00020969999999999997,
      "loss": 6.7068,
      "step": 700
    },
    {
      "epoch": 0.014436958614051972,
      "grad_norm": 1.1267056465148926,
      "learning_rate": 0.00021119999999999996,
      "loss": 6.7627,
      "step": 705
    },
    {
      "epoch": 0.014539348391456597,
      "grad_norm": 0.9785383939743042,
      "learning_rate": 0.00021269999999999997,
      "loss": 6.7088,
      "step": 710
    },
    {
      "epoch": 0.014641738168861222,
      "grad_norm": 0.6889818906784058,
      "learning_rate": 0.00021419999999999998,
      "loss": 6.7014,
      "step": 715
    },
    {
      "epoch": 0.014744127946265844,
      "grad_norm": 0.5881128907203674,
      "learning_rate": 0.00021569999999999998,
      "loss": 6.7047,
      "step": 720
    },
    {
      "epoch": 0.014846517723670469,
      "grad_norm": 0.49074748158454895,
      "learning_rate": 0.00021719999999999997,
      "loss": 6.7351,
      "step": 725
    },
    {
      "epoch": 0.014948907501075092,
      "grad_norm": 1.000606656074524,
      "learning_rate": 0.00021869999999999998,
      "loss": 6.6648,
      "step": 730
    },
    {
      "epoch": 0.015051297278479717,
      "grad_norm": 1.2626793384552002,
      "learning_rate": 0.00022019999999999999,
      "loss": 6.7403,
      "step": 735
    },
    {
      "epoch": 0.015153687055884341,
      "grad_norm": 1.0611534118652344,
      "learning_rate": 0.00022169999999999997,
      "loss": 6.6834,
      "step": 740
    },
    {
      "epoch": 0.015256076833288964,
      "grad_norm": 1.1611676216125488,
      "learning_rate": 0.00022319999999999998,
      "loss": 6.7988,
      "step": 745
    },
    {
      "epoch": 0.015358466610693589,
      "grad_norm": 0.6415315270423889,
      "learning_rate": 0.0002247,
      "loss": 6.6566,
      "step": 750
    },
    {
      "epoch": 0.015460856388098212,
      "grad_norm": 0.6816340684890747,
      "learning_rate": 0.00022619999999999997,
      "loss": 6.652,
      "step": 755
    },
    {
      "epoch": 0.015563246165502836,
      "grad_norm": 0.8404628038406372,
      "learning_rate": 0.00022769999999999998,
      "loss": 6.6516,
      "step": 760
    },
    {
      "epoch": 0.01566563594290746,
      "grad_norm": 1.1827441453933716,
      "learning_rate": 0.0002292,
      "loss": 6.6355,
      "step": 765
    },
    {
      "epoch": 0.015768025720312086,
      "grad_norm": 0.5615525841712952,
      "learning_rate": 0.0002307,
      "loss": 6.6248,
      "step": 770
    },
    {
      "epoch": 0.015870415497716707,
      "grad_norm": 0.7266453504562378,
      "learning_rate": 0.00023219999999999998,
      "loss": 6.6207,
      "step": 775
    },
    {
      "epoch": 0.01597280527512133,
      "grad_norm": 0.6322405338287354,
      "learning_rate": 0.0002337,
      "loss": 6.6276,
      "step": 780
    },
    {
      "epoch": 0.016075195052525956,
      "grad_norm": 0.80306077003479,
      "learning_rate": 0.0002352,
      "loss": 6.6492,
      "step": 785
    },
    {
      "epoch": 0.01617758482993058,
      "grad_norm": 0.6725737452507019,
      "learning_rate": 0.00023669999999999998,
      "loss": 6.6179,
      "step": 790
    },
    {
      "epoch": 0.016279974607335202,
      "grad_norm": 0.7415054440498352,
      "learning_rate": 0.0002382,
      "loss": 6.6516,
      "step": 795
    },
    {
      "epoch": 0.016382364384739827,
      "grad_norm": 0.5289933681488037,
      "learning_rate": 0.0002397,
      "loss": 6.584,
      "step": 800
    },
    {
      "epoch": 0.01648475416214445,
      "grad_norm": 0.6528019905090332,
      "learning_rate": 0.00024119999999999998,
      "loss": 6.6597,
      "step": 805
    },
    {
      "epoch": 0.016587143939549076,
      "grad_norm": 0.573921799659729,
      "learning_rate": 0.0002427,
      "loss": 6.581,
      "step": 810
    },
    {
      "epoch": 0.0166895337169537,
      "grad_norm": 0.7144637703895569,
      "learning_rate": 0.00024419999999999997,
      "loss": 6.5338,
      "step": 815
    },
    {
      "epoch": 0.016791923494358322,
      "grad_norm": 0.6318907141685486,
      "learning_rate": 0.00024569999999999995,
      "loss": 6.5926,
      "step": 820
    },
    {
      "epoch": 0.016894313271762947,
      "grad_norm": 0.665605902671814,
      "learning_rate": 0.0002472,
      "loss": 6.5585,
      "step": 825
    },
    {
      "epoch": 0.01699670304916757,
      "grad_norm": 0.5565012097358704,
      "learning_rate": 0.0002487,
      "loss": 6.5379,
      "step": 830
    },
    {
      "epoch": 0.017099092826572196,
      "grad_norm": 0.6355167031288147,
      "learning_rate": 0.00025019999999999996,
      "loss": 6.5098,
      "step": 835
    },
    {
      "epoch": 0.01720148260397682,
      "grad_norm": 0.6501747369766235,
      "learning_rate": 0.0002517,
      "loss": 6.5521,
      "step": 840
    },
    {
      "epoch": 0.01730387238138144,
      "grad_norm": 0.5432541370391846,
      "learning_rate": 0.0002532,
      "loss": 6.5465,
      "step": 845
    },
    {
      "epoch": 0.017406262158786066,
      "grad_norm": 0.561551570892334,
      "learning_rate": 0.00025469999999999996,
      "loss": 6.5426,
      "step": 850
    },
    {
      "epoch": 0.01750865193619069,
      "grad_norm": 0.865897536277771,
      "learning_rate": 0.0002562,
      "loss": 6.5195,
      "step": 855
    },
    {
      "epoch": 0.017611041713595316,
      "grad_norm": 0.7287527918815613,
      "learning_rate": 0.0002577,
      "loss": 6.5411,
      "step": 860
    },
    {
      "epoch": 0.017713431490999937,
      "grad_norm": 0.6758579015731812,
      "learning_rate": 0.00025919999999999996,
      "loss": 6.5528,
      "step": 865
    },
    {
      "epoch": 0.01781582126840456,
      "grad_norm": 0.7459971904754639,
      "learning_rate": 0.0002607,
      "loss": 6.5109,
      "step": 870
    },
    {
      "epoch": 0.017918211045809186,
      "grad_norm": 0.5554007291793823,
      "learning_rate": 0.0002622,
      "loss": 6.4838,
      "step": 875
    },
    {
      "epoch": 0.01802060082321381,
      "grad_norm": 0.7232329845428467,
      "learning_rate": 0.00026369999999999996,
      "loss": 6.5336,
      "step": 880
    },
    {
      "epoch": 0.018122990600618435,
      "grad_norm": 0.48811808228492737,
      "learning_rate": 0.0002652,
      "loss": 6.4909,
      "step": 885
    },
    {
      "epoch": 0.018225380378023057,
      "grad_norm": 0.673035204410553,
      "learning_rate": 0.0002667,
      "loss": 6.4889,
      "step": 890
    },
    {
      "epoch": 0.01832777015542768,
      "grad_norm": 0.5305101275444031,
      "learning_rate": 0.00026819999999999996,
      "loss": 6.4806,
      "step": 895
    },
    {
      "epoch": 0.018430159932832306,
      "grad_norm": 0.5400475263595581,
      "learning_rate": 0.0002697,
      "loss": 6.4376,
      "step": 900
    },
    {
      "epoch": 0.01853254971023693,
      "grad_norm": 0.5476040244102478,
      "learning_rate": 0.0002712,
      "loss": 6.4375,
      "step": 905
    },
    {
      "epoch": 0.018634939487641555,
      "grad_norm": 0.4433288276195526,
      "learning_rate": 0.00027269999999999996,
      "loss": 6.4463,
      "step": 910
    },
    {
      "epoch": 0.018737329265046176,
      "grad_norm": 0.640889048576355,
      "learning_rate": 0.0002742,
      "loss": 6.4159,
      "step": 915
    },
    {
      "epoch": 0.0188397190424508,
      "grad_norm": 0.5932295322418213,
      "learning_rate": 0.0002757,
      "loss": 6.4376,
      "step": 920
    },
    {
      "epoch": 0.018942108819855426,
      "grad_norm": 0.542849600315094,
      "learning_rate": 0.0002772,
      "loss": 6.4334,
      "step": 925
    },
    {
      "epoch": 0.01904449859726005,
      "grad_norm": 0.7035185694694519,
      "learning_rate": 0.0002787,
      "loss": 6.3826,
      "step": 930
    },
    {
      "epoch": 0.019146888374664675,
      "grad_norm": 0.5947180986404419,
      "learning_rate": 0.0002802,
      "loss": 6.4046,
      "step": 935
    },
    {
      "epoch": 0.019249278152069296,
      "grad_norm": 0.45393678545951843,
      "learning_rate": 0.00028169999999999996,
      "loss": 6.4006,
      "step": 940
    },
    {
      "epoch": 0.01935166792947392,
      "grad_norm": 0.5216274261474609,
      "learning_rate": 0.00028319999999999994,
      "loss": 6.4666,
      "step": 945
    },
    {
      "epoch": 0.019454057706878546,
      "grad_norm": 0.6625182032585144,
      "learning_rate": 0.0002847,
      "loss": 6.3587,
      "step": 950
    },
    {
      "epoch": 0.01955644748428317,
      "grad_norm": 0.6020975708961487,
      "learning_rate": 0.00028619999999999996,
      "loss": 6.3703,
      "step": 955
    },
    {
      "epoch": 0.01965883726168779,
      "grad_norm": 0.6037085056304932,
      "learning_rate": 0.00028769999999999995,
      "loss": 6.4085,
      "step": 960
    },
    {
      "epoch": 0.019761227039092416,
      "grad_norm": 0.5771493911743164,
      "learning_rate": 0.0002892,
      "loss": 6.5006,
      "step": 965
    },
    {
      "epoch": 0.01986361681649704,
      "grad_norm": 0.6366599798202515,
      "learning_rate": 0.00029069999999999996,
      "loss": 6.3859,
      "step": 970
    },
    {
      "epoch": 0.019966006593901665,
      "grad_norm": 0.48622769117355347,
      "learning_rate": 0.00029219999999999995,
      "loss": 6.3734,
      "step": 975
    },
    {
      "epoch": 0.02006839637130629,
      "grad_norm": 0.5495675206184387,
      "learning_rate": 0.0002937,
      "loss": 6.3456,
      "step": 980
    },
    {
      "epoch": 0.02017078614871091,
      "grad_norm": 0.48445188999176025,
      "learning_rate": 0.00029519999999999997,
      "loss": 6.3319,
      "step": 985
    },
    {
      "epoch": 0.020273175926115536,
      "grad_norm": 0.5041749477386475,
      "learning_rate": 0.00029669999999999995,
      "loss": 6.2705,
      "step": 990
    },
    {
      "epoch": 0.02037556570352016,
      "grad_norm": 0.7327763438224792,
      "learning_rate": 0.0002982,
      "loss": 6.3685,
      "step": 995
    },
    {
      "epoch": 0.020477955480924785,
      "grad_norm": 0.5152395367622375,
      "learning_rate": 0.00029969999999999997,
      "loss": 6.2687,
      "step": 1000
    },
    {
      "epoch": 0.020477955480924785,
      "eval_loss": 6.336370944976807,
      "eval_runtime": 128.3744,
      "eval_samples_per_second": 4.534,
      "eval_steps_per_second": 0.569,
      "step": 1000
    },
    {
      "epoch": 0.02058034525832941,
      "grad_norm": 0.6076420545578003,
      "learning_rate": 0.0003,
      "loss": 6.2989,
      "step": 1005
    },
    {
      "epoch": 0.02068273503573403,
      "grad_norm": 0.47126850485801697,
      "learning_rate": 0.0003,
      "loss": 6.2849,
      "step": 1010
    },
    {
      "epoch": 0.020785124813138656,
      "grad_norm": 0.5502199530601501,
      "learning_rate": 0.0003,
      "loss": 6.2745,
      "step": 1015
    },
    {
      "epoch": 0.02088751459054328,
      "grad_norm": 0.42955151200294495,
      "learning_rate": 0.0003,
      "loss": 6.2641,
      "step": 1020
    },
    {
      "epoch": 0.020989904367947905,
      "grad_norm": 0.48553207516670227,
      "learning_rate": 0.0003,
      "loss": 6.2998,
      "step": 1025
    },
    {
      "epoch": 0.02109229414535253,
      "grad_norm": 0.5554366707801819,
      "learning_rate": 0.0003,
      "loss": 6.1981,
      "step": 1030
    },
    {
      "epoch": 0.02119468392275715,
      "grad_norm": 0.41872456669807434,
      "learning_rate": 0.0003,
      "loss": 6.2481,
      "step": 1035
    },
    {
      "epoch": 0.021297073700161775,
      "grad_norm": 0.4277290403842926,
      "learning_rate": 0.0003,
      "loss": 6.2365,
      "step": 1040
    },
    {
      "epoch": 0.0213994634775664,
      "grad_norm": 0.543452799320221,
      "learning_rate": 0.0003,
      "loss": 6.2455,
      "step": 1045
    },
    {
      "epoch": 0.021501853254971025,
      "grad_norm": 0.5797136425971985,
      "learning_rate": 0.0003,
      "loss": 6.2531,
      "step": 1050
    },
    {
      "epoch": 0.021604243032375646,
      "grad_norm": 0.5384876132011414,
      "learning_rate": 0.0003,
      "loss": 6.2193,
      "step": 1055
    },
    {
      "epoch": 0.02170663280978027,
      "grad_norm": 0.7414798140525818,
      "learning_rate": 0.0003,
      "loss": 6.1985,
      "step": 1060
    },
    {
      "epoch": 0.021809022587184895,
      "grad_norm": 0.5251097083091736,
      "learning_rate": 0.0003,
      "loss": 6.1753,
      "step": 1065
    },
    {
      "epoch": 0.02191141236458952,
      "grad_norm": 0.4314242899417877,
      "learning_rate": 0.0003,
      "loss": 6.1949,
      "step": 1070
    },
    {
      "epoch": 0.022013802141994145,
      "grad_norm": 0.3870464861392975,
      "learning_rate": 0.0003,
      "loss": 6.1631,
      "step": 1075
    },
    {
      "epoch": 0.022116191919398766,
      "grad_norm": 0.5488783121109009,
      "learning_rate": 0.0003,
      "loss": 6.2236,
      "step": 1080
    },
    {
      "epoch": 0.02221858169680339,
      "grad_norm": 0.4516567587852478,
      "learning_rate": 0.0003,
      "loss": 6.1836,
      "step": 1085
    },
    {
      "epoch": 0.022320971474208015,
      "grad_norm": 0.44085627794265747,
      "learning_rate": 0.0003,
      "loss": 6.1345,
      "step": 1090
    },
    {
      "epoch": 0.02242336125161264,
      "grad_norm": 0.6064696907997131,
      "learning_rate": 0.0003,
      "loss": 6.1789,
      "step": 1095
    },
    {
      "epoch": 0.022525751029017264,
      "grad_norm": 0.4074086844921112,
      "learning_rate": 0.0003,
      "loss": 6.0935,
      "step": 1100
    },
    {
      "epoch": 0.022628140806421886,
      "grad_norm": 0.4455675482749939,
      "learning_rate": 0.0003,
      "loss": 6.163,
      "step": 1105
    },
    {
      "epoch": 0.02273053058382651,
      "grad_norm": 0.4718489944934845,
      "learning_rate": 0.0003,
      "loss": 6.139,
      "step": 1110
    },
    {
      "epoch": 0.022832920361231135,
      "grad_norm": 0.45494744181632996,
      "learning_rate": 0.0003,
      "loss": 6.1668,
      "step": 1115
    },
    {
      "epoch": 0.02293531013863576,
      "grad_norm": 0.5436424612998962,
      "learning_rate": 0.0003,
      "loss": 6.145,
      "step": 1120
    },
    {
      "epoch": 0.023037699916040384,
      "grad_norm": 0.42465776205062866,
      "learning_rate": 0.0003,
      "loss": 6.0842,
      "step": 1125
    },
    {
      "epoch": 0.023140089693445005,
      "grad_norm": 0.5106245279312134,
      "learning_rate": 0.0003,
      "loss": 6.076,
      "step": 1130
    },
    {
      "epoch": 0.02324247947084963,
      "grad_norm": 0.41347572207450867,
      "learning_rate": 0.0003,
      "loss": 6.0926,
      "step": 1135
    },
    {
      "epoch": 0.023344869248254255,
      "grad_norm": 0.48794978857040405,
      "learning_rate": 0.0003,
      "loss": 6.1093,
      "step": 1140
    },
    {
      "epoch": 0.02344725902565888,
      "grad_norm": 0.47482869029045105,
      "learning_rate": 0.0003,
      "loss": 6.087,
      "step": 1145
    },
    {
      "epoch": 0.0235496488030635,
      "grad_norm": 0.5310367941856384,
      "learning_rate": 0.0003,
      "loss": 6.0847,
      "step": 1150
    },
    {
      "epoch": 0.023652038580468125,
      "grad_norm": 0.4245409071445465,
      "learning_rate": 0.0003,
      "loss": 6.0588,
      "step": 1155
    },
    {
      "epoch": 0.02375442835787275,
      "grad_norm": 0.40255552530288696,
      "learning_rate": 0.0003,
      "loss": 6.0829,
      "step": 1160
    },
    {
      "epoch": 0.023856818135277374,
      "grad_norm": 0.4049079120159149,
      "learning_rate": 0.0003,
      "loss": 6.099,
      "step": 1165
    },
    {
      "epoch": 0.023959207912682,
      "grad_norm": 0.44209587574005127,
      "learning_rate": 0.0003,
      "loss": 6.0418,
      "step": 1170
    },
    {
      "epoch": 0.02406159769008662,
      "grad_norm": 0.4518246650695801,
      "learning_rate": 0.0003,
      "loss": 6.0572,
      "step": 1175
    },
    {
      "epoch": 0.024163987467491245,
      "grad_norm": 0.5068168640136719,
      "learning_rate": 0.0003,
      "loss": 5.985,
      "step": 1180
    },
    {
      "epoch": 0.02426637724489587,
      "grad_norm": 0.7139155268669128,
      "learning_rate": 0.0003,
      "loss": 6.0778,
      "step": 1185
    },
    {
      "epoch": 0.024368767022300494,
      "grad_norm": 0.5174667239189148,
      "learning_rate": 0.0003,
      "loss": 6.0872,
      "step": 1190
    },
    {
      "epoch": 0.02447115679970512,
      "grad_norm": 0.5178516507148743,
      "learning_rate": 0.0003,
      "loss": 6.0561,
      "step": 1195
    },
    {
      "epoch": 0.02457354657710974,
      "grad_norm": 0.4526061415672302,
      "learning_rate": 0.0003,
      "loss": 6.0428,
      "step": 1200
    },
    {
      "epoch": 0.024675936354514365,
      "grad_norm": 0.4358402490615845,
      "learning_rate": 0.0003,
      "loss": 6.0126,
      "step": 1205
    },
    {
      "epoch": 0.02477832613191899,
      "grad_norm": 0.4212406575679779,
      "learning_rate": 0.0003,
      "loss": 6.0066,
      "step": 1210
    },
    {
      "epoch": 0.024880715909323614,
      "grad_norm": 0.4510822296142578,
      "learning_rate": 0.0003,
      "loss": 5.9947,
      "step": 1215
    },
    {
      "epoch": 0.02498310568672824,
      "grad_norm": 0.4986138939857483,
      "learning_rate": 0.0003,
      "loss": 6.0421,
      "step": 1220
    },
    {
      "epoch": 0.02508549546413286,
      "grad_norm": 0.41960692405700684,
      "learning_rate": 0.0003,
      "loss": 5.9841,
      "step": 1225
    },
    {
      "epoch": 0.025187885241537485,
      "grad_norm": 0.3849349319934845,
      "learning_rate": 0.0003,
      "loss": 6.0143,
      "step": 1230
    },
    {
      "epoch": 0.02529027501894211,
      "grad_norm": 0.3687571883201599,
      "learning_rate": 0.0003,
      "loss": 5.9718,
      "step": 1235
    },
    {
      "epoch": 0.025392664796346734,
      "grad_norm": 0.40778475999832153,
      "learning_rate": 0.0003,
      "loss": 5.982,
      "step": 1240
    },
    {
      "epoch": 0.025495054573751355,
      "grad_norm": 0.4962833821773529,
      "learning_rate": 0.0003,
      "loss": 5.9226,
      "step": 1245
    },
    {
      "epoch": 0.02559744435115598,
      "grad_norm": 0.5068868398666382,
      "learning_rate": 0.0003,
      "loss": 5.976,
      "step": 1250
    },
    {
      "epoch": 0.025699834128560604,
      "grad_norm": 0.4068223237991333,
      "learning_rate": 0.0003,
      "loss": 5.954,
      "step": 1255
    },
    {
      "epoch": 0.02580222390596523,
      "grad_norm": 0.4254078269004822,
      "learning_rate": 0.0003,
      "loss": 5.95,
      "step": 1260
    },
    {
      "epoch": 0.025904613683369854,
      "grad_norm": 0.4287364184856415,
      "learning_rate": 0.0003,
      "loss": 5.91,
      "step": 1265
    },
    {
      "epoch": 0.026007003460774475,
      "grad_norm": 0.561913251876831,
      "learning_rate": 0.0003,
      "loss": 5.9931,
      "step": 1270
    },
    {
      "epoch": 0.0261093932381791,
      "grad_norm": 0.44675496220588684,
      "learning_rate": 0.0003,
      "loss": 5.9254,
      "step": 1275
    },
    {
      "epoch": 0.026211783015583724,
      "grad_norm": 0.3729199171066284,
      "learning_rate": 0.0003,
      "loss": 5.9828,
      "step": 1280
    },
    {
      "epoch": 0.02631417279298835,
      "grad_norm": 0.3478130102157593,
      "learning_rate": 0.0003,
      "loss": 5.917,
      "step": 1285
    },
    {
      "epoch": 0.026416562570392974,
      "grad_norm": 0.38890165090560913,
      "learning_rate": 0.0003,
      "loss": 5.933,
      "step": 1290
    },
    {
      "epoch": 0.026518952347797595,
      "grad_norm": 0.41554898023605347,
      "learning_rate": 0.0003,
      "loss": 5.9307,
      "step": 1295
    },
    {
      "epoch": 0.02662134212520222,
      "grad_norm": 0.42908337712287903,
      "learning_rate": 0.0003,
      "loss": 5.9184,
      "step": 1300
    },
    {
      "epoch": 0.026723731902606844,
      "grad_norm": 0.38750404119491577,
      "learning_rate": 0.0003,
      "loss": 5.9222,
      "step": 1305
    },
    {
      "epoch": 0.02682612168001147,
      "grad_norm": 0.4352513253688812,
      "learning_rate": 0.0003,
      "loss": 5.8937,
      "step": 1310
    },
    {
      "epoch": 0.026928511457416093,
      "grad_norm": 0.3794362545013428,
      "learning_rate": 0.0003,
      "loss": 5.8896,
      "step": 1315
    },
    {
      "epoch": 0.027030901234820714,
      "grad_norm": 0.4965048134326935,
      "learning_rate": 0.0003,
      "loss": 5.9118,
      "step": 1320
    },
    {
      "epoch": 0.02713329101222534,
      "grad_norm": 0.4754084050655365,
      "learning_rate": 0.0003,
      "loss": 5.9377,
      "step": 1325
    },
    {
      "epoch": 0.027235680789629964,
      "grad_norm": 0.48557162284851074,
      "learning_rate": 0.0003,
      "loss": 5.8814,
      "step": 1330
    },
    {
      "epoch": 0.02733807056703459,
      "grad_norm": 0.4073465168476105,
      "learning_rate": 0.0003,
      "loss": 5.878,
      "step": 1335
    },
    {
      "epoch": 0.02744046034443921,
      "grad_norm": 0.40075191855430603,
      "learning_rate": 0.0003,
      "loss": 5.8505,
      "step": 1340
    },
    {
      "epoch": 0.027542850121843834,
      "grad_norm": 0.35601678490638733,
      "learning_rate": 0.0003,
      "loss": 5.9291,
      "step": 1345
    },
    {
      "epoch": 0.02764523989924846,
      "grad_norm": 0.5753795504570007,
      "learning_rate": 0.0003,
      "loss": 5.8586,
      "step": 1350
    },
    {
      "epoch": 0.027747629676653084,
      "grad_norm": 0.3592877984046936,
      "learning_rate": 0.0003,
      "loss": 5.8507,
      "step": 1355
    },
    {
      "epoch": 0.027850019454057708,
      "grad_norm": 0.3982386291027069,
      "learning_rate": 0.0003,
      "loss": 5.8679,
      "step": 1360
    },
    {
      "epoch": 0.02795240923146233,
      "grad_norm": 0.4653341770172119,
      "learning_rate": 0.0003,
      "loss": 5.8429,
      "step": 1365
    },
    {
      "epoch": 0.028054799008866954,
      "grad_norm": 0.42373934388160706,
      "learning_rate": 0.0003,
      "loss": 5.8409,
      "step": 1370
    },
    {
      "epoch": 0.02815718878627158,
      "grad_norm": 0.4872317910194397,
      "learning_rate": 0.0003,
      "loss": 5.8394,
      "step": 1375
    },
    {
      "epoch": 0.028259578563676203,
      "grad_norm": 0.4176357388496399,
      "learning_rate": 0.0003,
      "loss": 5.869,
      "step": 1380
    },
    {
      "epoch": 0.028361968341080828,
      "grad_norm": 0.4286108911037445,
      "learning_rate": 0.0003,
      "loss": 5.8369,
      "step": 1385
    },
    {
      "epoch": 0.02846435811848545,
      "grad_norm": 0.39009803533554077,
      "learning_rate": 0.0003,
      "loss": 5.8566,
      "step": 1390
    },
    {
      "epoch": 0.028566747895890074,
      "grad_norm": 0.3626231253147125,
      "learning_rate": 0.0003,
      "loss": 5.8769,
      "step": 1395
    },
    {
      "epoch": 0.0286691376732947,
      "grad_norm": 0.39475512504577637,
      "learning_rate": 0.0003,
      "loss": 5.8343,
      "step": 1400
    },
    {
      "epoch": 0.028771527450699323,
      "grad_norm": 0.35049155354499817,
      "learning_rate": 0.0003,
      "loss": 5.8078,
      "step": 1405
    },
    {
      "epoch": 0.028873917228103944,
      "grad_norm": 0.4400930404663086,
      "learning_rate": 0.0003,
      "loss": 5.8022,
      "step": 1410
    },
    {
      "epoch": 0.02897630700550857,
      "grad_norm": 0.4743267595767975,
      "learning_rate": 0.0003,
      "loss": 5.8385,
      "step": 1415
    },
    {
      "epoch": 0.029078696782913194,
      "grad_norm": 0.3859167695045471,
      "learning_rate": 0.0003,
      "loss": 5.8517,
      "step": 1420
    },
    {
      "epoch": 0.02918108656031782,
      "grad_norm": 0.37883540987968445,
      "learning_rate": 0.0003,
      "loss": 5.8716,
      "step": 1425
    },
    {
      "epoch": 0.029283476337722443,
      "grad_norm": 0.383993923664093,
      "learning_rate": 0.0003,
      "loss": 5.7866,
      "step": 1430
    },
    {
      "epoch": 0.029385866115127064,
      "grad_norm": 0.42862668633461,
      "learning_rate": 0.0003,
      "loss": 5.8276,
      "step": 1435
    },
    {
      "epoch": 0.02948825589253169,
      "grad_norm": 0.36276373267173767,
      "learning_rate": 0.0003,
      "loss": 5.8164,
      "step": 1440
    },
    {
      "epoch": 0.029590645669936314,
      "grad_norm": 0.5727818012237549,
      "learning_rate": 0.0003,
      "loss": 5.8051,
      "step": 1445
    },
    {
      "epoch": 0.029693035447340938,
      "grad_norm": 0.35043594241142273,
      "learning_rate": 0.0003,
      "loss": 5.8202,
      "step": 1450
    },
    {
      "epoch": 0.029795425224745563,
      "grad_norm": 0.44809040427207947,
      "learning_rate": 0.0003,
      "loss": 5.8222,
      "step": 1455
    },
    {
      "epoch": 0.029897815002150184,
      "grad_norm": 0.4120275378227234,
      "learning_rate": 0.0003,
      "loss": 5.794,
      "step": 1460
    },
    {
      "epoch": 0.03000020477955481,
      "grad_norm": 0.3673813045024872,
      "learning_rate": 0.0003,
      "loss": 5.8181,
      "step": 1465
    },
    {
      "epoch": 0.030102594556959433,
      "grad_norm": 0.3413775563240051,
      "learning_rate": 0.0003,
      "loss": 5.7701,
      "step": 1470
    },
    {
      "epoch": 0.030204984334364058,
      "grad_norm": 0.35870635509490967,
      "learning_rate": 0.0003,
      "loss": 5.7869,
      "step": 1475
    },
    {
      "epoch": 0.030307374111768683,
      "grad_norm": 0.3878447711467743,
      "learning_rate": 0.0003,
      "loss": 5.7742,
      "step": 1480
    },
    {
      "epoch": 0.030409763889173304,
      "grad_norm": 0.4131092429161072,
      "learning_rate": 0.0003,
      "loss": 5.7552,
      "step": 1485
    },
    {
      "epoch": 0.03051215366657793,
      "grad_norm": 0.34513989090919495,
      "learning_rate": 0.0003,
      "loss": 5.7792,
      "step": 1490
    },
    {
      "epoch": 0.030614543443982553,
      "grad_norm": 0.38045361638069153,
      "learning_rate": 0.0003,
      "loss": 5.7677,
      "step": 1495
    },
    {
      "epoch": 0.030716933221387178,
      "grad_norm": 0.4234094023704529,
      "learning_rate": 0.0003,
      "loss": 5.7968,
      "step": 1500
    },
    {
      "epoch": 0.0308193229987918,
      "grad_norm": 0.3792508840560913,
      "learning_rate": 0.0003,
      "loss": 5.7779,
      "step": 1505
    },
    {
      "epoch": 0.030921712776196424,
      "grad_norm": 0.3751790225505829,
      "learning_rate": 0.0003,
      "loss": 5.8122,
      "step": 1510
    },
    {
      "epoch": 0.03102410255360105,
      "grad_norm": 0.38870561122894287,
      "learning_rate": 0.0003,
      "loss": 5.7776,
      "step": 1515
    },
    {
      "epoch": 0.031126492331005673,
      "grad_norm": 0.3328951299190521,
      "learning_rate": 0.0003,
      "loss": 5.7336,
      "step": 1520
    },
    {
      "epoch": 0.031228882108410298,
      "grad_norm": 0.8058025240898132,
      "learning_rate": 0.0003,
      "loss": 5.6826,
      "step": 1525
    },
    {
      "epoch": 0.03133127188581492,
      "grad_norm": 0.37183621525764465,
      "learning_rate": 0.0003,
      "loss": 5.7525,
      "step": 1530
    },
    {
      "epoch": 0.03143366166321954,
      "grad_norm": 0.33894869685173035,
      "learning_rate": 0.0003,
      "loss": 5.7739,
      "step": 1535
    },
    {
      "epoch": 0.03153605144062417,
      "grad_norm": 0.3959846794605255,
      "learning_rate": 0.0003,
      "loss": 5.7527,
      "step": 1540
    },
    {
      "epoch": 0.03163844121802879,
      "grad_norm": 0.41048547625541687,
      "learning_rate": 0.0003,
      "loss": 5.7598,
      "step": 1545
    },
    {
      "epoch": 0.031740830995433414,
      "grad_norm": 0.37840402126312256,
      "learning_rate": 0.0003,
      "loss": 5.7207,
      "step": 1550
    },
    {
      "epoch": 0.03184322077283804,
      "grad_norm": 0.37814024090766907,
      "learning_rate": 0.0003,
      "loss": 5.7232,
      "step": 1555
    },
    {
      "epoch": 0.03194561055024266,
      "grad_norm": 0.3821817636489868,
      "learning_rate": 0.0003,
      "loss": 5.7615,
      "step": 1560
    },
    {
      "epoch": 0.032048000327647284,
      "grad_norm": 0.42602258920669556,
      "learning_rate": 0.0003,
      "loss": 5.7825,
      "step": 1565
    },
    {
      "epoch": 0.03215039010505191,
      "grad_norm": 0.4071274995803833,
      "learning_rate": 0.0003,
      "loss": 5.7122,
      "step": 1570
    },
    {
      "epoch": 0.032252779882456534,
      "grad_norm": 0.3515342175960541,
      "learning_rate": 0.0003,
      "loss": 5.7415,
      "step": 1575
    },
    {
      "epoch": 0.03235516965986116,
      "grad_norm": 0.40821370482444763,
      "learning_rate": 0.0003,
      "loss": 5.6702,
      "step": 1580
    },
    {
      "epoch": 0.03245755943726578,
      "grad_norm": 0.4025024175643921,
      "learning_rate": 0.0003,
      "loss": 5.7376,
      "step": 1585
    },
    {
      "epoch": 0.032559949214670404,
      "grad_norm": 0.3244132995605469,
      "learning_rate": 0.0003,
      "loss": 5.7276,
      "step": 1590
    },
    {
      "epoch": 0.03266233899207503,
      "grad_norm": 0.3303980231285095,
      "learning_rate": 0.0003,
      "loss": 5.7456,
      "step": 1595
    },
    {
      "epoch": 0.032764728769479654,
      "grad_norm": 0.38295868039131165,
      "learning_rate": 0.0003,
      "loss": 5.7252,
      "step": 1600
    },
    {
      "epoch": 0.03286711854688428,
      "grad_norm": 0.3676038086414337,
      "learning_rate": 0.0003,
      "loss": 5.7262,
      "step": 1605
    },
    {
      "epoch": 0.0329695083242889,
      "grad_norm": 0.3270364999771118,
      "learning_rate": 0.0003,
      "loss": 5.6958,
      "step": 1610
    },
    {
      "epoch": 0.033071898101693524,
      "grad_norm": 0.38710907101631165,
      "learning_rate": 0.0003,
      "loss": 5.6863,
      "step": 1615
    },
    {
      "epoch": 0.03317428787909815,
      "grad_norm": 0.48394158482551575,
      "learning_rate": 0.0003,
      "loss": 5.7162,
      "step": 1620
    },
    {
      "epoch": 0.03327667765650277,
      "grad_norm": 0.3814449906349182,
      "learning_rate": 0.0003,
      "loss": 5.7125,
      "step": 1625
    },
    {
      "epoch": 0.0333790674339074,
      "grad_norm": 0.3561025857925415,
      "learning_rate": 0.0003,
      "loss": 5.7087,
      "step": 1630
    },
    {
      "epoch": 0.03348145721131202,
      "grad_norm": 0.3981541693210602,
      "learning_rate": 0.0003,
      "loss": 5.6742,
      "step": 1635
    },
    {
      "epoch": 0.033583846988716644,
      "grad_norm": 0.36963599920272827,
      "learning_rate": 0.0003,
      "loss": 5.6722,
      "step": 1640
    },
    {
      "epoch": 0.03368623676612127,
      "grad_norm": 0.3874456286430359,
      "learning_rate": 0.0003,
      "loss": 5.6887,
      "step": 1645
    },
    {
      "epoch": 0.03378862654352589,
      "grad_norm": 0.31456896662712097,
      "learning_rate": 0.0003,
      "loss": 5.6685,
      "step": 1650
    },
    {
      "epoch": 0.03389101632093052,
      "grad_norm": 0.3377586901187897,
      "learning_rate": 0.0003,
      "loss": 5.6969,
      "step": 1655
    },
    {
      "epoch": 0.03399340609833514,
      "grad_norm": 0.3572098910808563,
      "learning_rate": 0.0003,
      "loss": 5.676,
      "step": 1660
    },
    {
      "epoch": 0.034095795875739764,
      "grad_norm": 0.3694843351840973,
      "learning_rate": 0.0003,
      "loss": 5.6547,
      "step": 1665
    },
    {
      "epoch": 0.03419818565314439,
      "grad_norm": 0.4291844666004181,
      "learning_rate": 0.0003,
      "loss": 5.7127,
      "step": 1670
    },
    {
      "epoch": 0.03430057543054901,
      "grad_norm": 0.36142176389694214,
      "learning_rate": 0.0003,
      "loss": 5.7063,
      "step": 1675
    },
    {
      "epoch": 0.03440296520795364,
      "grad_norm": 0.3345441520214081,
      "learning_rate": 0.0003,
      "loss": 5.7012,
      "step": 1680
    },
    {
      "epoch": 0.03450535498535826,
      "grad_norm": 0.33553993701934814,
      "learning_rate": 0.0003,
      "loss": 5.6681,
      "step": 1685
    },
    {
      "epoch": 0.03460774476276288,
      "grad_norm": 0.3819684386253357,
      "learning_rate": 0.0003,
      "loss": 5.6353,
      "step": 1690
    },
    {
      "epoch": 0.03471013454016751,
      "grad_norm": 0.35187289118766785,
      "learning_rate": 0.0003,
      "loss": 5.6603,
      "step": 1695
    },
    {
      "epoch": 0.03481252431757213,
      "grad_norm": 0.3862445652484894,
      "learning_rate": 0.0003,
      "loss": 5.7001,
      "step": 1700
    },
    {
      "epoch": 0.03491491409497676,
      "grad_norm": 0.3471886217594147,
      "learning_rate": 0.0003,
      "loss": 5.6303,
      "step": 1705
    },
    {
      "epoch": 0.03501730387238138,
      "grad_norm": 0.330762654542923,
      "learning_rate": 0.0003,
      "loss": 5.7048,
      "step": 1710
    },
    {
      "epoch": 0.035119693649786,
      "grad_norm": 0.3590639531612396,
      "learning_rate": 0.0003,
      "loss": 5.6629,
      "step": 1715
    },
    {
      "epoch": 0.03522208342719063,
      "grad_norm": 0.4507967531681061,
      "learning_rate": 0.0003,
      "loss": 5.6756,
      "step": 1720
    },
    {
      "epoch": 0.03532447320459525,
      "grad_norm": 0.3698206841945648,
      "learning_rate": 0.0003,
      "loss": 5.727,
      "step": 1725
    },
    {
      "epoch": 0.035426862981999874,
      "grad_norm": 0.35770583152770996,
      "learning_rate": 0.0003,
      "loss": 5.5815,
      "step": 1730
    },
    {
      "epoch": 0.0355292527594045,
      "grad_norm": 0.33955705165863037,
      "learning_rate": 0.0003,
      "loss": 5.6646,
      "step": 1735
    },
    {
      "epoch": 0.03563164253680912,
      "grad_norm": 0.3695562481880188,
      "learning_rate": 0.0003,
      "loss": 5.6772,
      "step": 1740
    },
    {
      "epoch": 0.03573403231421375,
      "grad_norm": 0.3273179829120636,
      "learning_rate": 0.0003,
      "loss": 5.6608,
      "step": 1745
    },
    {
      "epoch": 0.03583642209161837,
      "grad_norm": 0.34312453866004944,
      "learning_rate": 0.0003,
      "loss": 5.6768,
      "step": 1750
    },
    {
      "epoch": 0.035938811869022994,
      "grad_norm": 0.34229740500450134,
      "learning_rate": 0.0003,
      "loss": 5.6108,
      "step": 1755
    },
    {
      "epoch": 0.03604120164642762,
      "grad_norm": 0.3227626383304596,
      "learning_rate": 0.0003,
      "loss": 5.6101,
      "step": 1760
    },
    {
      "epoch": 0.03614359142383224,
      "grad_norm": 0.3419533669948578,
      "learning_rate": 0.0003,
      "loss": 5.6117,
      "step": 1765
    },
    {
      "epoch": 0.03624598120123687,
      "grad_norm": 0.40896594524383545,
      "learning_rate": 0.0003,
      "loss": 5.6037,
      "step": 1770
    },
    {
      "epoch": 0.03634837097864149,
      "grad_norm": 0.32976123690605164,
      "learning_rate": 0.0003,
      "loss": 5.6104,
      "step": 1775
    },
    {
      "epoch": 0.03645076075604611,
      "grad_norm": 0.31900909543037415,
      "learning_rate": 0.0003,
      "loss": 5.6692,
      "step": 1780
    },
    {
      "epoch": 0.03655315053345074,
      "grad_norm": 0.30529239773750305,
      "learning_rate": 0.0003,
      "loss": 5.5969,
      "step": 1785
    },
    {
      "epoch": 0.03665554031085536,
      "grad_norm": 0.32662200927734375,
      "learning_rate": 0.0003,
      "loss": 5.6239,
      "step": 1790
    },
    {
      "epoch": 0.03675793008825999,
      "grad_norm": 0.35408344864845276,
      "learning_rate": 0.0003,
      "loss": 5.6779,
      "step": 1795
    },
    {
      "epoch": 0.03686031986566461,
      "grad_norm": 0.3842693269252777,
      "learning_rate": 0.0003,
      "loss": 5.6231,
      "step": 1800
    },
    {
      "epoch": 0.03696270964306923,
      "grad_norm": 0.33722659945487976,
      "learning_rate": 0.0003,
      "loss": 5.6586,
      "step": 1805
    },
    {
      "epoch": 0.03706509942047386,
      "grad_norm": 0.3564646542072296,
      "learning_rate": 0.0003,
      "loss": 5.5967,
      "step": 1810
    },
    {
      "epoch": 0.03716748919787848,
      "grad_norm": 0.32668283581733704,
      "learning_rate": 0.0003,
      "loss": 5.5891,
      "step": 1815
    },
    {
      "epoch": 0.03726987897528311,
      "grad_norm": 0.3343525230884552,
      "learning_rate": 0.0003,
      "loss": 5.6138,
      "step": 1820
    },
    {
      "epoch": 0.03737226875268773,
      "grad_norm": 0.4201762080192566,
      "learning_rate": 0.0003,
      "loss": 5.598,
      "step": 1825
    },
    {
      "epoch": 0.03747465853009235,
      "grad_norm": 0.5659616589546204,
      "learning_rate": 0.0003,
      "loss": 5.7533,
      "step": 1830
    },
    {
      "epoch": 0.03757704830749698,
      "grad_norm": 0.38928595185279846,
      "learning_rate": 0.0003,
      "loss": 5.6555,
      "step": 1835
    },
    {
      "epoch": 0.0376794380849016,
      "grad_norm": 0.45403486490249634,
      "learning_rate": 0.0003,
      "loss": 5.7067,
      "step": 1840
    },
    {
      "epoch": 0.03778182786230623,
      "grad_norm": 0.3449617624282837,
      "learning_rate": 0.0003,
      "loss": 5.6221,
      "step": 1845
    },
    {
      "epoch": 0.03788421763971085,
      "grad_norm": 0.353818416595459,
      "learning_rate": 0.0003,
      "loss": 5.614,
      "step": 1850
    },
    {
      "epoch": 0.03798660741711547,
      "grad_norm": 0.35148918628692627,
      "learning_rate": 0.0003,
      "loss": 5.6021,
      "step": 1855
    },
    {
      "epoch": 0.0380889971945201,
      "grad_norm": 0.2895936369895935,
      "learning_rate": 0.0003,
      "loss": 5.6375,
      "step": 1860
    },
    {
      "epoch": 0.03819138697192472,
      "grad_norm": 0.31613689661026,
      "learning_rate": 0.0003,
      "loss": 5.5729,
      "step": 1865
    },
    {
      "epoch": 0.03829377674932935,
      "grad_norm": 0.7037827372550964,
      "learning_rate": 0.0003,
      "loss": 5.5409,
      "step": 1870
    },
    {
      "epoch": 0.03839616652673397,
      "grad_norm": 0.40238696336746216,
      "learning_rate": 0.0003,
      "loss": 5.5955,
      "step": 1875
    },
    {
      "epoch": 0.03849855630413859,
      "grad_norm": 0.30799344182014465,
      "learning_rate": 0.0003,
      "loss": 5.6085,
      "step": 1880
    },
    {
      "epoch": 0.03860094608154322,
      "grad_norm": 0.3469482660293579,
      "learning_rate": 0.0003,
      "loss": 5.608,
      "step": 1885
    },
    {
      "epoch": 0.03870333585894784,
      "grad_norm": 0.3633992075920105,
      "learning_rate": 0.0003,
      "loss": 5.6317,
      "step": 1890
    },
    {
      "epoch": 0.03880572563635247,
      "grad_norm": 0.3335460126399994,
      "learning_rate": 0.0003,
      "loss": 5.577,
      "step": 1895
    },
    {
      "epoch": 0.03890811541375709,
      "grad_norm": 0.3882322907447815,
      "learning_rate": 0.0003,
      "loss": 5.6301,
      "step": 1900
    },
    {
      "epoch": 0.03901050519116171,
      "grad_norm": 0.33104488253593445,
      "learning_rate": 0.0003,
      "loss": 5.5612,
      "step": 1905
    },
    {
      "epoch": 0.03911289496856634,
      "grad_norm": 0.32825130224227905,
      "learning_rate": 0.0003,
      "loss": 5.5834,
      "step": 1910
    },
    {
      "epoch": 0.03921528474597096,
      "grad_norm": 0.3569926619529724,
      "learning_rate": 0.0003,
      "loss": 5.5913,
      "step": 1915
    },
    {
      "epoch": 0.03931767452337558,
      "grad_norm": 0.3393506109714508,
      "learning_rate": 0.0003,
      "loss": 5.6085,
      "step": 1920
    },
    {
      "epoch": 0.03942006430078021,
      "grad_norm": 0.3391747772693634,
      "learning_rate": 0.0003,
      "loss": 5.5552,
      "step": 1925
    },
    {
      "epoch": 0.03952245407818483,
      "grad_norm": 0.374737024307251,
      "learning_rate": 0.0003,
      "loss": 5.5973,
      "step": 1930
    },
    {
      "epoch": 0.03962484385558946,
      "grad_norm": 0.32326409220695496,
      "learning_rate": 0.0003,
      "loss": 5.5686,
      "step": 1935
    },
    {
      "epoch": 0.03972723363299408,
      "grad_norm": 0.33155369758605957,
      "learning_rate": 0.0003,
      "loss": 5.586,
      "step": 1940
    },
    {
      "epoch": 0.0398296234103987,
      "grad_norm": 0.30039525032043457,
      "learning_rate": 0.0003,
      "loss": 5.5708,
      "step": 1945
    },
    {
      "epoch": 0.03993201318780333,
      "grad_norm": 0.31259462237358093,
      "learning_rate": 0.0003,
      "loss": 5.5693,
      "step": 1950
    },
    {
      "epoch": 0.04003440296520795,
      "grad_norm": 0.32965245842933655,
      "learning_rate": 0.0003,
      "loss": 5.5784,
      "step": 1955
    },
    {
      "epoch": 0.04013679274261258,
      "grad_norm": 0.36182472109794617,
      "learning_rate": 0.0003,
      "loss": 5.5073,
      "step": 1960
    },
    {
      "epoch": 0.0402391825200172,
      "grad_norm": 0.3083086311817169,
      "learning_rate": 0.0003,
      "loss": 5.5432,
      "step": 1965
    },
    {
      "epoch": 0.04034157229742182,
      "grad_norm": 0.3532242178916931,
      "learning_rate": 0.0003,
      "loss": 5.5841,
      "step": 1970
    },
    {
      "epoch": 0.04044396207482645,
      "grad_norm": 0.3572655916213989,
      "learning_rate": 0.0003,
      "loss": 5.5665,
      "step": 1975
    },
    {
      "epoch": 0.04054635185223107,
      "grad_norm": 0.3412257432937622,
      "learning_rate": 0.0003,
      "loss": 5.5591,
      "step": 1980
    },
    {
      "epoch": 0.0406487416296357,
      "grad_norm": 0.3201288878917694,
      "learning_rate": 0.0003,
      "loss": 5.606,
      "step": 1985
    },
    {
      "epoch": 0.04075113140704032,
      "grad_norm": 0.38728052377700806,
      "learning_rate": 0.0003,
      "loss": 5.5509,
      "step": 1990
    },
    {
      "epoch": 0.04085352118444494,
      "grad_norm": 0.38100388646125793,
      "learning_rate": 0.0003,
      "loss": 5.6295,
      "step": 1995
    },
    {
      "epoch": 0.04095591096184957,
      "grad_norm": 0.34643521904945374,
      "learning_rate": 0.0003,
      "loss": 5.5531,
      "step": 2000
    },
    {
      "epoch": 0.04095591096184957,
      "eval_loss": 5.583315849304199,
      "eval_runtime": 17.2581,
      "eval_samples_per_second": 33.723,
      "eval_steps_per_second": 4.23,
      "step": 2000
    },
    {
      "epoch": 0.04105830073925419,
      "grad_norm": 0.3605150282382965,
      "learning_rate": 0.0003,
      "loss": 5.5599,
      "step": 2005
    },
    {
      "epoch": 0.04116069051665882,
      "grad_norm": 0.3214218318462372,
      "learning_rate": 0.0003,
      "loss": 5.5758,
      "step": 2010
    },
    {
      "epoch": 0.04126308029406344,
      "grad_norm": 0.33031126856803894,
      "learning_rate": 0.0003,
      "loss": 5.5487,
      "step": 2015
    },
    {
      "epoch": 0.04136547007146806,
      "grad_norm": 0.3428272604942322,
      "learning_rate": 0.0003,
      "loss": 5.5345,
      "step": 2020
    },
    {
      "epoch": 0.04146785984887269,
      "grad_norm": 0.32269689440727234,
      "learning_rate": 0.0003,
      "loss": 5.5702,
      "step": 2025
    },
    {
      "epoch": 0.04157024962627731,
      "grad_norm": 0.38060262799263,
      "learning_rate": 0.0003,
      "loss": 5.5379,
      "step": 2030
    },
    {
      "epoch": 0.04167263940368194,
      "grad_norm": 0.32364749908447266,
      "learning_rate": 0.0003,
      "loss": 5.5284,
      "step": 2035
    },
    {
      "epoch": 0.04177502918108656,
      "grad_norm": 0.3137708008289337,
      "learning_rate": 0.0003,
      "loss": 5.5021,
      "step": 2040
    },
    {
      "epoch": 0.04187741895849118,
      "grad_norm": 0.3555919826030731,
      "learning_rate": 0.0003,
      "loss": 5.5261,
      "step": 2045
    },
    {
      "epoch": 0.04197980873589581,
      "grad_norm": 0.3532278537750244,
      "learning_rate": 0.0003,
      "loss": 5.5472,
      "step": 2050
    },
    {
      "epoch": 0.04208219851330043,
      "grad_norm": 0.331758052110672,
      "learning_rate": 0.0003,
      "loss": 5.4677,
      "step": 2055
    },
    {
      "epoch": 0.04218458829070506,
      "grad_norm": 0.49990275502204895,
      "learning_rate": 0.0003,
      "loss": 5.5588,
      "step": 2060
    },
    {
      "epoch": 0.04228697806810968,
      "grad_norm": 0.30251890420913696,
      "learning_rate": 0.0003,
      "loss": 5.5241,
      "step": 2065
    },
    {
      "epoch": 0.0423893678455143,
      "grad_norm": 0.36536428332328796,
      "learning_rate": 0.0003,
      "loss": 5.5187,
      "step": 2070
    },
    {
      "epoch": 0.04249175762291893,
      "grad_norm": 0.29050710797309875,
      "learning_rate": 0.0003,
      "loss": 5.5183,
      "step": 2075
    },
    {
      "epoch": 0.04259414740032355,
      "grad_norm": 4.046236991882324,
      "learning_rate": 0.0003,
      "loss": 5.5574,
      "step": 2080
    },
    {
      "epoch": 0.04269653717772818,
      "grad_norm": 0.39397743344306946,
      "learning_rate": 0.0003,
      "loss": 5.576,
      "step": 2085
    },
    {
      "epoch": 0.0427989269551328,
      "grad_norm": 0.4010278880596161,
      "learning_rate": 0.0003,
      "loss": 5.513,
      "step": 2090
    },
    {
      "epoch": 0.04290131673253742,
      "grad_norm": 0.368369460105896,
      "learning_rate": 0.0003,
      "loss": 5.4895,
      "step": 2095
    },
    {
      "epoch": 0.04300370650994205,
      "grad_norm": 0.3162172734737396,
      "learning_rate": 0.0003,
      "loss": 5.6343,
      "step": 2100
    },
    {
      "epoch": 0.04310609628734667,
      "grad_norm": 0.32602453231811523,
      "learning_rate": 0.0003,
      "loss": 5.5658,
      "step": 2105
    },
    {
      "epoch": 0.04320848606475129,
      "grad_norm": 0.29531750082969666,
      "learning_rate": 0.0003,
      "loss": 5.5271,
      "step": 2110
    },
    {
      "epoch": 0.04331087584215592,
      "grad_norm": 0.31045177578926086,
      "learning_rate": 0.0003,
      "loss": 5.5363,
      "step": 2115
    },
    {
      "epoch": 0.04341326561956054,
      "grad_norm": 0.29968997836112976,
      "learning_rate": 0.0003,
      "loss": 5.4967,
      "step": 2120
    },
    {
      "epoch": 0.04351565539696517,
      "grad_norm": 0.32277289032936096,
      "learning_rate": 0.0003,
      "loss": 5.3919,
      "step": 2125
    },
    {
      "epoch": 0.04361804517436979,
      "grad_norm": 0.3237505555152893,
      "learning_rate": 0.0003,
      "loss": 5.5099,
      "step": 2130
    },
    {
      "epoch": 0.04372043495177441,
      "grad_norm": 0.3156304657459259,
      "learning_rate": 0.0003,
      "loss": 5.4859,
      "step": 2135
    },
    {
      "epoch": 0.04382282472917904,
      "grad_norm": 0.2915082573890686,
      "learning_rate": 0.0003,
      "loss": 5.5746,
      "step": 2140
    },
    {
      "epoch": 0.04392521450658366,
      "grad_norm": 0.3208426237106323,
      "learning_rate": 0.0003,
      "loss": 5.4987,
      "step": 2145
    },
    {
      "epoch": 0.04402760428398829,
      "grad_norm": 0.2956843674182892,
      "learning_rate": 0.0003,
      "loss": 5.5098,
      "step": 2150
    },
    {
      "epoch": 0.04412999406139291,
      "grad_norm": 0.294599711894989,
      "learning_rate": 0.0003,
      "loss": 5.5153,
      "step": 2155
    },
    {
      "epoch": 0.04423238383879753,
      "grad_norm": 0.3562420904636383,
      "learning_rate": 0.0003,
      "loss": 5.4741,
      "step": 2160
    },
    {
      "epoch": 0.04433477361620216,
      "grad_norm": 0.3201698362827301,
      "learning_rate": 0.0003,
      "loss": 5.532,
      "step": 2165
    },
    {
      "epoch": 0.04443716339360678,
      "grad_norm": 0.304730087518692,
      "learning_rate": 0.0003,
      "loss": 5.5303,
      "step": 2170
    },
    {
      "epoch": 0.04453955317101141,
      "grad_norm": 0.332582026720047,
      "learning_rate": 0.0003,
      "loss": 5.4733,
      "step": 2175
    },
    {
      "epoch": 0.04464194294841603,
      "grad_norm": 0.30200523138046265,
      "learning_rate": 0.0003,
      "loss": 5.4378,
      "step": 2180
    },
    {
      "epoch": 0.04474433272582065,
      "grad_norm": 0.3588789999485016,
      "learning_rate": 0.0003,
      "loss": 5.4876,
      "step": 2185
    },
    {
      "epoch": 0.04484672250322528,
      "grad_norm": 0.3039074242115021,
      "learning_rate": 0.0003,
      "loss": 5.4944,
      "step": 2190
    },
    {
      "epoch": 0.0449491122806299,
      "grad_norm": 0.48547083139419556,
      "learning_rate": 0.0003,
      "loss": 5.5256,
      "step": 2195
    },
    {
      "epoch": 0.04505150205803453,
      "grad_norm": 0.32456061244010925,
      "learning_rate": 0.0003,
      "loss": 5.4931,
      "step": 2200
    },
    {
      "epoch": 0.04515389183543915,
      "grad_norm": 0.36695829033851624,
      "learning_rate": 0.0003,
      "loss": 5.4575,
      "step": 2205
    },
    {
      "epoch": 0.04525628161284377,
      "grad_norm": 0.365884929895401,
      "learning_rate": 0.0003,
      "loss": 5.4911,
      "step": 2210
    },
    {
      "epoch": 0.0453586713902484,
      "grad_norm": 0.3172907829284668,
      "learning_rate": 0.0003,
      "loss": 5.5056,
      "step": 2215
    },
    {
      "epoch": 0.04546106116765302,
      "grad_norm": 0.31234583258628845,
      "learning_rate": 0.0003,
      "loss": 5.477,
      "step": 2220
    },
    {
      "epoch": 0.04556345094505765,
      "grad_norm": 0.3623846769332886,
      "learning_rate": 0.0003,
      "loss": 5.4839,
      "step": 2225
    },
    {
      "epoch": 0.04566584072246227,
      "grad_norm": 0.29917284846305847,
      "learning_rate": 0.0003,
      "loss": 5.533,
      "step": 2230
    },
    {
      "epoch": 0.04576823049986689,
      "grad_norm": 0.30479612946510315,
      "learning_rate": 0.0003,
      "loss": 5.467,
      "step": 2235
    },
    {
      "epoch": 0.04587062027727152,
      "grad_norm": 0.29524606466293335,
      "learning_rate": 0.0003,
      "loss": 5.4925,
      "step": 2240
    },
    {
      "epoch": 0.04597301005467614,
      "grad_norm": 0.37530916929244995,
      "learning_rate": 0.0003,
      "loss": 5.5117,
      "step": 2245
    },
    {
      "epoch": 0.04607539983208077,
      "grad_norm": 0.33626043796539307,
      "learning_rate": 0.0003,
      "loss": 5.4686,
      "step": 2250
    },
    {
      "epoch": 0.04617778960948539,
      "grad_norm": 0.30177098512649536,
      "learning_rate": 0.0003,
      "loss": 5.462,
      "step": 2255
    },
    {
      "epoch": 0.04628017938689001,
      "grad_norm": 0.30366501212120056,
      "learning_rate": 0.0003,
      "loss": 5.4903,
      "step": 2260
    },
    {
      "epoch": 0.04638256916429464,
      "grad_norm": 0.281887412071228,
      "learning_rate": 0.0003,
      "loss": 5.4703,
      "step": 2265
    },
    {
      "epoch": 0.04648495894169926,
      "grad_norm": 0.2937527298927307,
      "learning_rate": 0.0003,
      "loss": 5.4657,
      "step": 2270
    },
    {
      "epoch": 0.04658734871910388,
      "grad_norm": 0.34095871448516846,
      "learning_rate": 0.0003,
      "loss": 5.43,
      "step": 2275
    },
    {
      "epoch": 0.04668973849650851,
      "grad_norm": 0.29006507992744446,
      "learning_rate": 0.0003,
      "loss": 5.4479,
      "step": 2280
    },
    {
      "epoch": 0.04679212827391313,
      "grad_norm": 0.29680919647216797,
      "learning_rate": 0.0003,
      "loss": 5.4708,
      "step": 2285
    },
    {
      "epoch": 0.04689451805131776,
      "grad_norm": 0.3066211938858032,
      "learning_rate": 0.0003,
      "loss": 5.4698,
      "step": 2290
    },
    {
      "epoch": 0.04699690782872238,
      "grad_norm": 0.3496037423610687,
      "learning_rate": 0.0003,
      "loss": 5.4626,
      "step": 2295
    },
    {
      "epoch": 0.047099297606127,
      "grad_norm": 0.30650416016578674,
      "learning_rate": 0.0003,
      "loss": 5.503,
      "step": 2300
    },
    {
      "epoch": 0.04720168738353163,
      "grad_norm": 0.3337506353855133,
      "learning_rate": 0.0003,
      "loss": 5.4569,
      "step": 2305
    },
    {
      "epoch": 0.04730407716093625,
      "grad_norm": 0.3257908523082733,
      "learning_rate": 0.0003,
      "loss": 5.4877,
      "step": 2310
    },
    {
      "epoch": 0.04740646693834088,
      "grad_norm": 0.32778066396713257,
      "learning_rate": 0.0003,
      "loss": 5.4402,
      "step": 2315
    },
    {
      "epoch": 0.0475088567157455,
      "grad_norm": 0.3712349534034729,
      "learning_rate": 0.0003,
      "loss": 5.4845,
      "step": 2320
    },
    {
      "epoch": 0.04761124649315012,
      "grad_norm": 0.29005157947540283,
      "learning_rate": 0.0003,
      "loss": 5.4534,
      "step": 2325
    },
    {
      "epoch": 0.04771363627055475,
      "grad_norm": 0.2965439558029175,
      "learning_rate": 0.0003,
      "loss": 5.4702,
      "step": 2330
    },
    {
      "epoch": 0.04781602604795937,
      "grad_norm": 0.3257777988910675,
      "learning_rate": 0.0003,
      "loss": 5.4405,
      "step": 2335
    },
    {
      "epoch": 0.047918415825364,
      "grad_norm": 0.30828627943992615,
      "learning_rate": 0.0003,
      "loss": 5.5202,
      "step": 2340
    },
    {
      "epoch": 0.04802080560276862,
      "grad_norm": 0.3007051944732666,
      "learning_rate": 0.0003,
      "loss": 5.4896,
      "step": 2345
    },
    {
      "epoch": 0.04812319538017324,
      "grad_norm": 0.31143495440483093,
      "learning_rate": 0.0003,
      "loss": 5.5087,
      "step": 2350
    },
    {
      "epoch": 0.04822558515757787,
      "grad_norm": 0.3348305821418762,
      "learning_rate": 0.0003,
      "loss": 5.4619,
      "step": 2355
    },
    {
      "epoch": 0.04832797493498249,
      "grad_norm": 0.3939598798751831,
      "learning_rate": 0.0003,
      "loss": 5.4527,
      "step": 2360
    },
    {
      "epoch": 0.04843036471238712,
      "grad_norm": 0.418344110250473,
      "learning_rate": 0.0003,
      "loss": 5.457,
      "step": 2365
    },
    {
      "epoch": 0.04853275448979174,
      "grad_norm": 0.3089907765388489,
      "learning_rate": 0.0003,
      "loss": 5.4268,
      "step": 2370
    },
    {
      "epoch": 0.04863514426719636,
      "grad_norm": 0.30724790692329407,
      "learning_rate": 0.0003,
      "loss": 5.497,
      "step": 2375
    },
    {
      "epoch": 0.04873753404460099,
      "grad_norm": 0.2833951711654663,
      "learning_rate": 0.0003,
      "loss": 5.436,
      "step": 2380
    },
    {
      "epoch": 0.04883992382200561,
      "grad_norm": 0.26925790309906006,
      "learning_rate": 0.0003,
      "loss": 5.4077,
      "step": 2385
    },
    {
      "epoch": 0.04894231359941024,
      "grad_norm": 0.2891678810119629,
      "learning_rate": 0.0003,
      "loss": 5.4053,
      "step": 2390
    },
    {
      "epoch": 0.04904470337681486,
      "grad_norm": 0.37123313546180725,
      "learning_rate": 0.0003,
      "loss": 5.4378,
      "step": 2395
    },
    {
      "epoch": 0.04914709315421948,
      "grad_norm": 0.3456077575683594,
      "learning_rate": 0.0003,
      "loss": 5.4581,
      "step": 2400
    },
    {
      "epoch": 0.04924948293162411,
      "grad_norm": 0.3430868983268738,
      "learning_rate": 0.0003,
      "loss": 5.4521,
      "step": 2405
    },
    {
      "epoch": 0.04935187270902873,
      "grad_norm": 0.3325310945510864,
      "learning_rate": 0.0003,
      "loss": 5.4949,
      "step": 2410
    },
    {
      "epoch": 0.04945426248643336,
      "grad_norm": 0.6735697388648987,
      "learning_rate": 0.0003,
      "loss": 5.4431,
      "step": 2415
    },
    {
      "epoch": 0.04955665226383798,
      "grad_norm": 0.32990604639053345,
      "learning_rate": 0.0003,
      "loss": 5.4302,
      "step": 2420
    },
    {
      "epoch": 0.0496590420412426,
      "grad_norm": 0.8016563653945923,
      "learning_rate": 0.0003,
      "loss": 5.4676,
      "step": 2425
    },
    {
      "epoch": 0.04976143181864723,
      "grad_norm": 0.32512250542640686,
      "learning_rate": 0.0003,
      "loss": 5.4798,
      "step": 2430
    },
    {
      "epoch": 0.04986382159605185,
      "grad_norm": 0.31798192858695984,
      "learning_rate": 0.0003,
      "loss": 5.4696,
      "step": 2435
    },
    {
      "epoch": 0.04996621137345648,
      "grad_norm": 0.3064839541912079,
      "learning_rate": 0.0003,
      "loss": 5.4655,
      "step": 2440
    },
    {
      "epoch": 0.0500686011508611,
      "grad_norm": 0.3080918490886688,
      "learning_rate": 0.0003,
      "loss": 5.4392,
      "step": 2445
    },
    {
      "epoch": 0.05017099092826572,
      "grad_norm": 0.27605608105659485,
      "learning_rate": 0.0003,
      "loss": 5.4385,
      "step": 2450
    },
    {
      "epoch": 0.05027338070567035,
      "grad_norm": 0.2982933521270752,
      "learning_rate": 0.0003,
      "loss": 5.4325,
      "step": 2455
    },
    {
      "epoch": 0.05037577048307497,
      "grad_norm": 0.3271021544933319,
      "learning_rate": 0.0003,
      "loss": 5.4436,
      "step": 2460
    },
    {
      "epoch": 0.05047816026047959,
      "grad_norm": 0.3043646216392517,
      "learning_rate": 0.0003,
      "loss": 5.4242,
      "step": 2465
    },
    {
      "epoch": 0.05058055003788422,
      "grad_norm": 0.3109362721443176,
      "learning_rate": 0.0003,
      "loss": 5.4851,
      "step": 2470
    },
    {
      "epoch": 0.05068293981528884,
      "grad_norm": 0.2791377604007721,
      "learning_rate": 0.0003,
      "loss": 5.4403,
      "step": 2475
    },
    {
      "epoch": 0.05078532959269347,
      "grad_norm": 0.3052929639816284,
      "learning_rate": 0.0003,
      "loss": 5.4143,
      "step": 2480
    },
    {
      "epoch": 0.05088771937009809,
      "grad_norm": 0.2574477195739746,
      "learning_rate": 0.0003,
      "loss": 5.4014,
      "step": 2485
    },
    {
      "epoch": 0.05099010914750271,
      "grad_norm": 0.2741445302963257,
      "learning_rate": 0.0003,
      "loss": 5.4129,
      "step": 2490
    },
    {
      "epoch": 0.05109249892490734,
      "grad_norm": 0.32297956943511963,
      "learning_rate": 0.0003,
      "loss": 5.4256,
      "step": 2495
    },
    {
      "epoch": 0.05119488870231196,
      "grad_norm": 0.32383501529693604,
      "learning_rate": 0.0003,
      "loss": 5.4153,
      "step": 2500
    },
    {
      "epoch": 0.05129727847971659,
      "grad_norm": 0.33168911933898926,
      "learning_rate": 0.0003,
      "loss": 5.4351,
      "step": 2505
    },
    {
      "epoch": 0.05139966825712121,
      "grad_norm": 0.2909999191761017,
      "learning_rate": 0.0003,
      "loss": 5.4195,
      "step": 2510
    },
    {
      "epoch": 0.05150205803452583,
      "grad_norm": 0.29537051916122437,
      "learning_rate": 0.0003,
      "loss": 5.4515,
      "step": 2515
    },
    {
      "epoch": 0.05160444781193046,
      "grad_norm": 0.29898619651794434,
      "learning_rate": 0.0003,
      "loss": 5.3557,
      "step": 2520
    },
    {
      "epoch": 0.05170683758933508,
      "grad_norm": 0.2907998859882355,
      "learning_rate": 0.0003,
      "loss": 5.4229,
      "step": 2525
    },
    {
      "epoch": 0.05180922736673971,
      "grad_norm": 0.3177140951156616,
      "learning_rate": 0.0003,
      "loss": 5.4054,
      "step": 2530
    },
    {
      "epoch": 0.05191161714414433,
      "grad_norm": 0.28961414098739624,
      "learning_rate": 0.0003,
      "loss": 5.417,
      "step": 2535
    },
    {
      "epoch": 0.05201400692154895,
      "grad_norm": 0.3086441159248352,
      "learning_rate": 0.0003,
      "loss": 5.4408,
      "step": 2540
    },
    {
      "epoch": 0.05211639669895358,
      "grad_norm": 0.2974289655685425,
      "learning_rate": 0.0003,
      "loss": 5.4342,
      "step": 2545
    },
    {
      "epoch": 0.0522187864763582,
      "grad_norm": 0.3043321967124939,
      "learning_rate": 0.0003,
      "loss": 5.3939,
      "step": 2550
    },
    {
      "epoch": 0.05232117625376283,
      "grad_norm": 0.313244104385376,
      "learning_rate": 0.0003,
      "loss": 5.4296,
      "step": 2555
    },
    {
      "epoch": 0.05242356603116745,
      "grad_norm": 0.3215901851654053,
      "learning_rate": 0.0003,
      "loss": 5.3516,
      "step": 2560
    },
    {
      "epoch": 0.05252595580857207,
      "grad_norm": 0.287676602602005,
      "learning_rate": 0.0003,
      "loss": 5.3864,
      "step": 2565
    },
    {
      "epoch": 0.0526283455859767,
      "grad_norm": 0.29060837626457214,
      "learning_rate": 0.0003,
      "loss": 5.399,
      "step": 2570
    },
    {
      "epoch": 0.05273073536338132,
      "grad_norm": 0.31551116704940796,
      "learning_rate": 0.0003,
      "loss": 5.436,
      "step": 2575
    },
    {
      "epoch": 0.05283312514078595,
      "grad_norm": 0.3053077161312103,
      "learning_rate": 0.0003,
      "loss": 5.408,
      "step": 2580
    },
    {
      "epoch": 0.05293551491819057,
      "grad_norm": 0.2839934825897217,
      "learning_rate": 0.0003,
      "loss": 5.4055,
      "step": 2585
    },
    {
      "epoch": 0.05303790469559519,
      "grad_norm": 0.2997831106185913,
      "learning_rate": 0.0003,
      "loss": 5.4258,
      "step": 2590
    },
    {
      "epoch": 0.05314029447299982,
      "grad_norm": 0.27672094106674194,
      "learning_rate": 0.0003,
      "loss": 5.4314,
      "step": 2595
    },
    {
      "epoch": 0.05324268425040444,
      "grad_norm": 0.2853901982307434,
      "learning_rate": 0.0003,
      "loss": 5.4281,
      "step": 2600
    },
    {
      "epoch": 0.05334507402780907,
      "grad_norm": 0.432605117559433,
      "learning_rate": 0.0003,
      "loss": 5.2178,
      "step": 2605
    },
    {
      "epoch": 0.05344746380521369,
      "grad_norm": 0.28586146235466003,
      "learning_rate": 0.0003,
      "loss": 5.4123,
      "step": 2610
    },
    {
      "epoch": 0.05354985358261831,
      "grad_norm": 0.2829761803150177,
      "learning_rate": 0.0003,
      "loss": 5.4273,
      "step": 2615
    },
    {
      "epoch": 0.05365224336002294,
      "grad_norm": 0.33027493953704834,
      "learning_rate": 0.0003,
      "loss": 5.3922,
      "step": 2620
    },
    {
      "epoch": 0.05375463313742756,
      "grad_norm": 0.28707289695739746,
      "learning_rate": 0.0003,
      "loss": 5.396,
      "step": 2625
    },
    {
      "epoch": 0.05385702291483219,
      "grad_norm": 0.35139745473861694,
      "learning_rate": 0.0003,
      "loss": 5.405,
      "step": 2630
    },
    {
      "epoch": 0.05395941269223681,
      "grad_norm": 0.27695679664611816,
      "learning_rate": 0.0003,
      "loss": 5.4257,
      "step": 2635
    },
    {
      "epoch": 0.05406180246964143,
      "grad_norm": 0.30533459782600403,
      "learning_rate": 0.0003,
      "loss": 5.3721,
      "step": 2640
    },
    {
      "epoch": 0.05416419224704606,
      "grad_norm": 0.3125362694263458,
      "learning_rate": 0.0003,
      "loss": 5.4178,
      "step": 2645
    },
    {
      "epoch": 0.05426658202445068,
      "grad_norm": 0.26875168085098267,
      "learning_rate": 0.0003,
      "loss": 5.4108,
      "step": 2650
    },
    {
      "epoch": 0.0543689718018553,
      "grad_norm": 0.2825389802455902,
      "learning_rate": 0.0003,
      "loss": 5.4012,
      "step": 2655
    },
    {
      "epoch": 0.05447136157925993,
      "grad_norm": 0.29759857058525085,
      "learning_rate": 0.0003,
      "loss": 5.3857,
      "step": 2660
    },
    {
      "epoch": 0.05457375135666455,
      "grad_norm": 0.2873668670654297,
      "learning_rate": 0.0003,
      "loss": 5.3778,
      "step": 2665
    },
    {
      "epoch": 0.05467614113406918,
      "grad_norm": 0.30925366282463074,
      "learning_rate": 0.0003,
      "loss": 5.4058,
      "step": 2670
    },
    {
      "epoch": 0.0547785309114738,
      "grad_norm": 0.31262725591659546,
      "learning_rate": 0.0003,
      "loss": 5.3799,
      "step": 2675
    },
    {
      "epoch": 0.05488092068887842,
      "grad_norm": 0.3050954341888428,
      "learning_rate": 0.0003,
      "loss": 5.3993,
      "step": 2680
    },
    {
      "epoch": 0.05498331046628305,
      "grad_norm": 0.28599846363067627,
      "learning_rate": 0.0003,
      "loss": 5.4062,
      "step": 2685
    },
    {
      "epoch": 0.05508570024368767,
      "grad_norm": 0.3238472044467926,
      "learning_rate": 0.0003,
      "loss": 5.3648,
      "step": 2690
    },
    {
      "epoch": 0.0551880900210923,
      "grad_norm": 0.26982608437538147,
      "learning_rate": 0.0003,
      "loss": 5.3961,
      "step": 2695
    },
    {
      "epoch": 0.05529047979849692,
      "grad_norm": 0.2549554705619812,
      "learning_rate": 0.0003,
      "loss": 5.4047,
      "step": 2700
    },
    {
      "epoch": 0.05539286957590154,
      "grad_norm": 0.26843875646591187,
      "learning_rate": 0.0003,
      "loss": 5.4177,
      "step": 2705
    },
    {
      "epoch": 0.05549525935330617,
      "grad_norm": 0.3723680377006531,
      "learning_rate": 0.0003,
      "loss": 5.349,
      "step": 2710
    },
    {
      "epoch": 0.05559764913071079,
      "grad_norm": 0.2782374620437622,
      "learning_rate": 0.0003,
      "loss": 5.3586,
      "step": 2715
    },
    {
      "epoch": 0.055700038908115417,
      "grad_norm": 0.3722502589225769,
      "learning_rate": 0.0003,
      "loss": 5.3817,
      "step": 2720
    },
    {
      "epoch": 0.05580242868552004,
      "grad_norm": 0.3401469886302948,
      "learning_rate": 0.0003,
      "loss": 5.331,
      "step": 2725
    },
    {
      "epoch": 0.05590481846292466,
      "grad_norm": 0.33351531624794006,
      "learning_rate": 0.0003,
      "loss": 5.3896,
      "step": 2730
    },
    {
      "epoch": 0.05600720824032929,
      "grad_norm": 0.2746967077255249,
      "learning_rate": 0.0003,
      "loss": 5.3825,
      "step": 2735
    },
    {
      "epoch": 0.05610959801773391,
      "grad_norm": 0.3015638291835785,
      "learning_rate": 0.0003,
      "loss": 5.3828,
      "step": 2740
    },
    {
      "epoch": 0.056211987795138536,
      "grad_norm": 0.27876561880111694,
      "learning_rate": 0.0003,
      "loss": 5.4345,
      "step": 2745
    },
    {
      "epoch": 0.05631437757254316,
      "grad_norm": 0.31784892082214355,
      "learning_rate": 0.0003,
      "loss": 5.3858,
      "step": 2750
    },
    {
      "epoch": 0.05641676734994778,
      "grad_norm": 0.2963598072528839,
      "learning_rate": 0.0003,
      "loss": 5.3802,
      "step": 2755
    },
    {
      "epoch": 0.05651915712735241,
      "grad_norm": 0.31250786781311035,
      "learning_rate": 0.0003,
      "loss": 5.355,
      "step": 2760
    },
    {
      "epoch": 0.05662154690475703,
      "grad_norm": 0.327053964138031,
      "learning_rate": 0.0003,
      "loss": 5.4345,
      "step": 2765
    },
    {
      "epoch": 0.056723936682161656,
      "grad_norm": 0.2861989736557007,
      "learning_rate": 0.0003,
      "loss": 5.3666,
      "step": 2770
    },
    {
      "epoch": 0.05682632645956628,
      "grad_norm": 0.3195374011993408,
      "learning_rate": 0.0003,
      "loss": 5.4485,
      "step": 2775
    },
    {
      "epoch": 0.0569287162369709,
      "grad_norm": 0.30048274993896484,
      "learning_rate": 0.0003,
      "loss": 5.3655,
      "step": 2780
    },
    {
      "epoch": 0.05703110601437553,
      "grad_norm": 0.3026953339576721,
      "learning_rate": 0.0003,
      "loss": 5.3903,
      "step": 2785
    },
    {
      "epoch": 0.05713349579178015,
      "grad_norm": 0.2979716956615448,
      "learning_rate": 0.0003,
      "loss": 5.3727,
      "step": 2790
    },
    {
      "epoch": 0.057235885569184776,
      "grad_norm": 0.29879677295684814,
      "learning_rate": 0.0003,
      "loss": 5.3949,
      "step": 2795
    },
    {
      "epoch": 0.0573382753465894,
      "grad_norm": 0.2776120901107788,
      "learning_rate": 0.0003,
      "loss": 5.3619,
      "step": 2800
    },
    {
      "epoch": 0.05744066512399402,
      "grad_norm": 0.29277926683425903,
      "learning_rate": 0.0003,
      "loss": 5.3101,
      "step": 2805
    },
    {
      "epoch": 0.057543054901398646,
      "grad_norm": 0.35536375641822815,
      "learning_rate": 0.0003,
      "loss": 5.2366,
      "step": 2810
    },
    {
      "epoch": 0.05764544467880327,
      "grad_norm": 0.3723018765449524,
      "learning_rate": 0.0003,
      "loss": 5.3171,
      "step": 2815
    },
    {
      "epoch": 0.05774783445620789,
      "grad_norm": 0.28674349188804626,
      "learning_rate": 0.0003,
      "loss": 5.3853,
      "step": 2820
    },
    {
      "epoch": 0.05785022423361252,
      "grad_norm": 0.26566022634506226,
      "learning_rate": 0.0003,
      "loss": 5.3303,
      "step": 2825
    },
    {
      "epoch": 0.05795261401101714,
      "grad_norm": 0.30490708351135254,
      "learning_rate": 0.0003,
      "loss": 5.3849,
      "step": 2830
    },
    {
      "epoch": 0.058055003788421766,
      "grad_norm": 0.33971863985061646,
      "learning_rate": 0.0003,
      "loss": 5.3618,
      "step": 2835
    },
    {
      "epoch": 0.05815739356582639,
      "grad_norm": 0.2888649106025696,
      "learning_rate": 0.0003,
      "loss": 5.3214,
      "step": 2840
    },
    {
      "epoch": 0.05825978334323101,
      "grad_norm": 0.27611035108566284,
      "learning_rate": 0.0003,
      "loss": 5.3986,
      "step": 2845
    },
    {
      "epoch": 0.05836217312063564,
      "grad_norm": 0.2928638160228729,
      "learning_rate": 0.0003,
      "loss": 5.3831,
      "step": 2850
    },
    {
      "epoch": 0.05846456289804026,
      "grad_norm": 0.26954448223114014,
      "learning_rate": 0.0003,
      "loss": 5.3531,
      "step": 2855
    },
    {
      "epoch": 0.058566952675444886,
      "grad_norm": 0.27161797881126404,
      "learning_rate": 0.0003,
      "loss": 5.3587,
      "step": 2860
    },
    {
      "epoch": 0.05866934245284951,
      "grad_norm": 0.27939358353614807,
      "learning_rate": 0.0003,
      "loss": 5.3553,
      "step": 2865
    },
    {
      "epoch": 0.05877173223025413,
      "grad_norm": 0.2974206507205963,
      "learning_rate": 0.0003,
      "loss": 5.3483,
      "step": 2870
    },
    {
      "epoch": 0.058874122007658757,
      "grad_norm": 0.27986735105514526,
      "learning_rate": 0.0003,
      "loss": 5.3578,
      "step": 2875
    },
    {
      "epoch": 0.05897651178506338,
      "grad_norm": 0.303239643573761,
      "learning_rate": 0.0003,
      "loss": 5.3732,
      "step": 2880
    },
    {
      "epoch": 0.059078901562468006,
      "grad_norm": 0.2981972396373749,
      "learning_rate": 0.0003,
      "loss": 5.3519,
      "step": 2885
    },
    {
      "epoch": 0.05918129133987263,
      "grad_norm": 0.3639703094959259,
      "learning_rate": 0.0003,
      "loss": 5.3359,
      "step": 2890
    },
    {
      "epoch": 0.05928368111727725,
      "grad_norm": 0.29545578360557556,
      "learning_rate": 0.0003,
      "loss": 5.3495,
      "step": 2895
    },
    {
      "epoch": 0.059386070894681876,
      "grad_norm": 0.30039510130882263,
      "learning_rate": 0.0003,
      "loss": 5.3114,
      "step": 2900
    },
    {
      "epoch": 0.0594884606720865,
      "grad_norm": 0.4698617458343506,
      "learning_rate": 0.0003,
      "loss": 5.3936,
      "step": 2905
    },
    {
      "epoch": 0.059590850449491126,
      "grad_norm": 0.4134523868560791,
      "learning_rate": 0.0003,
      "loss": 5.4452,
      "step": 2910
    },
    {
      "epoch": 0.05969324022689575,
      "grad_norm": 0.3588584065437317,
      "learning_rate": 0.0003,
      "loss": 5.3283,
      "step": 2915
    },
    {
      "epoch": 0.05979563000430037,
      "grad_norm": 0.33248433470726013,
      "learning_rate": 0.0003,
      "loss": 5.3824,
      "step": 2920
    },
    {
      "epoch": 0.059898019781704996,
      "grad_norm": 0.2713482677936554,
      "learning_rate": 0.0003,
      "loss": 5.3401,
      "step": 2925
    },
    {
      "epoch": 0.06000040955910962,
      "grad_norm": 0.30715176463127136,
      "learning_rate": 0.0003,
      "loss": 5.3221,
      "step": 2930
    },
    {
      "epoch": 0.060102799336514245,
      "grad_norm": 0.3150102198123932,
      "learning_rate": 0.0003,
      "loss": 5.3757,
      "step": 2935
    },
    {
      "epoch": 0.06020518911391887,
      "grad_norm": 0.2869734466075897,
      "learning_rate": 0.0003,
      "loss": 5.3552,
      "step": 2940
    },
    {
      "epoch": 0.06030757889132349,
      "grad_norm": 0.2754596769809723,
      "learning_rate": 0.0003,
      "loss": 5.3078,
      "step": 2945
    },
    {
      "epoch": 0.060409968668728116,
      "grad_norm": 0.2720445692539215,
      "learning_rate": 0.0003,
      "loss": 5.341,
      "step": 2950
    },
    {
      "epoch": 0.06051235844613274,
      "grad_norm": 0.256799578666687,
      "learning_rate": 0.0003,
      "loss": 5.3398,
      "step": 2955
    },
    {
      "epoch": 0.060614748223537365,
      "grad_norm": 0.25818538665771484,
      "learning_rate": 0.0003,
      "loss": 5.3507,
      "step": 2960
    },
    {
      "epoch": 0.060717138000941986,
      "grad_norm": 0.27038925886154175,
      "learning_rate": 0.0003,
      "loss": 5.3145,
      "step": 2965
    },
    {
      "epoch": 0.06081952777834661,
      "grad_norm": 0.30888015031814575,
      "learning_rate": 0.0003,
      "loss": 5.299,
      "step": 2970
    },
    {
      "epoch": 0.060921917555751236,
      "grad_norm": 0.27706509828567505,
      "learning_rate": 0.0003,
      "loss": 5.316,
      "step": 2975
    },
    {
      "epoch": 0.06102430733315586,
      "grad_norm": 0.3341308832168579,
      "learning_rate": 0.0003,
      "loss": 5.3509,
      "step": 2980
    },
    {
      "epoch": 0.061126697110560485,
      "grad_norm": 0.258544385433197,
      "learning_rate": 0.0003,
      "loss": 5.3369,
      "step": 2985
    },
    {
      "epoch": 0.061229086887965106,
      "grad_norm": 0.28303077816963196,
      "learning_rate": 0.0003,
      "loss": 5.3017,
      "step": 2990
    },
    {
      "epoch": 0.06133147666536973,
      "grad_norm": 0.26469340920448303,
      "learning_rate": 0.0003,
      "loss": 5.2852,
      "step": 2995
    },
    {
      "epoch": 0.061433866442774356,
      "grad_norm": 0.42151668667793274,
      "learning_rate": 0.0003,
      "loss": 5.3421,
      "step": 3000
    },
    {
      "epoch": 0.061433866442774356,
      "eval_loss": 5.355992317199707,
      "eval_runtime": 17.2748,
      "eval_samples_per_second": 33.691,
      "eval_steps_per_second": 4.226,
      "step": 3000
    },
    {
      "epoch": 0.06153625622017898,
      "grad_norm": 0.28036433458328247,
      "learning_rate": 0.0003,
      "loss": 5.3528,
      "step": 3005
    },
    {
      "epoch": 0.0616386459975836,
      "grad_norm": 0.3133857548236847,
      "learning_rate": 0.0003,
      "loss": 5.3184,
      "step": 3010
    },
    {
      "epoch": 0.061741035774988226,
      "grad_norm": 0.25309228897094727,
      "learning_rate": 0.0003,
      "loss": 5.3449,
      "step": 3015
    },
    {
      "epoch": 0.06184342555239285,
      "grad_norm": 0.28472191095352173,
      "learning_rate": 0.0003,
      "loss": 5.3376,
      "step": 3020
    },
    {
      "epoch": 0.061945815329797475,
      "grad_norm": 0.28121504187583923,
      "learning_rate": 0.0003,
      "loss": 5.3259,
      "step": 3025
    },
    {
      "epoch": 0.0620482051072021,
      "grad_norm": 0.2599506378173828,
      "learning_rate": 0.0003,
      "loss": 5.3605,
      "step": 3030
    },
    {
      "epoch": 0.06215059488460672,
      "grad_norm": 0.2690852880477905,
      "learning_rate": 0.0003,
      "loss": 5.312,
      "step": 3035
    },
    {
      "epoch": 0.062252984662011346,
      "grad_norm": 0.26789405941963196,
      "learning_rate": 0.0003,
      "loss": 5.3435,
      "step": 3040
    },
    {
      "epoch": 0.06235537443941597,
      "grad_norm": 0.26487815380096436,
      "learning_rate": 0.0003,
      "loss": 5.327,
      "step": 3045
    },
    {
      "epoch": 0.062457764216820595,
      "grad_norm": 0.4305364787578583,
      "learning_rate": 0.0003,
      "loss": 5.2532,
      "step": 3050
    },
    {
      "epoch": 0.06256015399422522,
      "grad_norm": 0.3147276043891907,
      "learning_rate": 0.0003,
      "loss": 5.3129,
      "step": 3055
    },
    {
      "epoch": 0.06266254377162984,
      "grad_norm": 0.2905116677284241,
      "learning_rate": 0.0003,
      "loss": 5.3561,
      "step": 3060
    },
    {
      "epoch": 0.06276493354903447,
      "grad_norm": 0.27454766631126404,
      "learning_rate": 0.0003,
      "loss": 5.3661,
      "step": 3065
    },
    {
      "epoch": 0.06286732332643909,
      "grad_norm": 0.29055505990982056,
      "learning_rate": 0.0003,
      "loss": 5.3424,
      "step": 3070
    },
    {
      "epoch": 0.06296971310384371,
      "grad_norm": 0.2825135886669159,
      "learning_rate": 0.0003,
      "loss": 5.2881,
      "step": 3075
    },
    {
      "epoch": 0.06307210288124834,
      "grad_norm": 0.2770971953868866,
      "learning_rate": 0.0003,
      "loss": 5.3529,
      "step": 3080
    },
    {
      "epoch": 0.06317449265865296,
      "grad_norm": 0.2654070556163788,
      "learning_rate": 0.0003,
      "loss": 5.2791,
      "step": 3085
    },
    {
      "epoch": 0.06327688243605759,
      "grad_norm": 0.25536680221557617,
      "learning_rate": 0.0003,
      "loss": 5.2641,
      "step": 3090
    },
    {
      "epoch": 0.0633792722134622,
      "grad_norm": 0.26588839292526245,
      "learning_rate": 0.0003,
      "loss": 5.3355,
      "step": 3095
    },
    {
      "epoch": 0.06348166199086683,
      "grad_norm": 0.2800893783569336,
      "learning_rate": 0.0003,
      "loss": 5.3052,
      "step": 3100
    },
    {
      "epoch": 0.06358405176827145,
      "grad_norm": 0.3270575702190399,
      "learning_rate": 0.0003,
      "loss": 5.3568,
      "step": 3105
    },
    {
      "epoch": 0.06368644154567608,
      "grad_norm": 0.28551337122917175,
      "learning_rate": 0.0003,
      "loss": 5.2801,
      "step": 3110
    },
    {
      "epoch": 0.0637888313230807,
      "grad_norm": 0.2678089141845703,
      "learning_rate": 0.0003,
      "loss": 5.3226,
      "step": 3115
    },
    {
      "epoch": 0.06389122110048533,
      "grad_norm": 1.0088434219360352,
      "learning_rate": 0.0003,
      "loss": 5.3141,
      "step": 3120
    },
    {
      "epoch": 0.06399361087788995,
      "grad_norm": 0.3256169855594635,
      "learning_rate": 0.0003,
      "loss": 5.3208,
      "step": 3125
    },
    {
      "epoch": 0.06409600065529457,
      "grad_norm": 0.2671166956424713,
      "learning_rate": 0.0003,
      "loss": 5.3377,
      "step": 3130
    },
    {
      "epoch": 0.0641983904326992,
      "grad_norm": 0.24780243635177612,
      "learning_rate": 0.0003,
      "loss": 5.3314,
      "step": 3135
    },
    {
      "epoch": 0.06430078021010383,
      "grad_norm": 0.30526068806648254,
      "learning_rate": 0.0003,
      "loss": 5.3694,
      "step": 3140
    },
    {
      "epoch": 0.06440316998750845,
      "grad_norm": 0.3097380995750427,
      "learning_rate": 0.0003,
      "loss": 5.3049,
      "step": 3145
    },
    {
      "epoch": 0.06450555976491307,
      "grad_norm": 0.2833743691444397,
      "learning_rate": 0.0003,
      "loss": 5.316,
      "step": 3150
    },
    {
      "epoch": 0.06460794954231769,
      "grad_norm": 0.26801565289497375,
      "learning_rate": 0.0003,
      "loss": 5.3528,
      "step": 3155
    },
    {
      "epoch": 0.06471033931972232,
      "grad_norm": 0.33694198727607727,
      "learning_rate": 0.0003,
      "loss": 5.3172,
      "step": 3160
    },
    {
      "epoch": 0.06481272909712694,
      "grad_norm": 0.36817753314971924,
      "learning_rate": 0.0003,
      "loss": 5.294,
      "step": 3165
    },
    {
      "epoch": 0.06491511887453157,
      "grad_norm": 0.28540757298469543,
      "learning_rate": 0.0003,
      "loss": 5.3004,
      "step": 3170
    },
    {
      "epoch": 0.06501750865193619,
      "grad_norm": 0.39156943559646606,
      "learning_rate": 0.0003,
      "loss": 5.2591,
      "step": 3175
    },
    {
      "epoch": 0.06511989842934081,
      "grad_norm": 0.27260908484458923,
      "learning_rate": 0.0003,
      "loss": 5.2961,
      "step": 3180
    },
    {
      "epoch": 0.06522228820674544,
      "grad_norm": 0.35306307673454285,
      "learning_rate": 0.0003,
      "loss": 5.3259,
      "step": 3185
    },
    {
      "epoch": 0.06532467798415006,
      "grad_norm": 0.27560725808143616,
      "learning_rate": 0.0003,
      "loss": 5.3031,
      "step": 3190
    },
    {
      "epoch": 0.06542706776155469,
      "grad_norm": 0.30916744470596313,
      "learning_rate": 0.0003,
      "loss": 5.3213,
      "step": 3195
    },
    {
      "epoch": 0.06552945753895931,
      "grad_norm": 0.26047390699386597,
      "learning_rate": 0.0003,
      "loss": 5.3198,
      "step": 3200
    },
    {
      "epoch": 0.06563184731636393,
      "grad_norm": 0.2895469069480896,
      "learning_rate": 0.0003,
      "loss": 5.3226,
      "step": 3205
    },
    {
      "epoch": 0.06573423709376856,
      "grad_norm": 0.27080896496772766,
      "learning_rate": 0.0003,
      "loss": 5.2866,
      "step": 3210
    },
    {
      "epoch": 0.06583662687117318,
      "grad_norm": 0.2811758816242218,
      "learning_rate": 0.0003,
      "loss": 5.3196,
      "step": 3215
    },
    {
      "epoch": 0.0659390166485778,
      "grad_norm": 0.29708757996559143,
      "learning_rate": 0.0003,
      "loss": 5.2298,
      "step": 3220
    },
    {
      "epoch": 0.06604140642598243,
      "grad_norm": 0.29997655749320984,
      "learning_rate": 0.0003,
      "loss": 5.3038,
      "step": 3225
    },
    {
      "epoch": 0.06614379620338705,
      "grad_norm": 0.3130645751953125,
      "learning_rate": 0.0003,
      "loss": 5.2443,
      "step": 3230
    },
    {
      "epoch": 0.06624618598079168,
      "grad_norm": 0.3080459535121918,
      "learning_rate": 0.0003,
      "loss": 5.3079,
      "step": 3235
    },
    {
      "epoch": 0.0663485757581963,
      "grad_norm": 0.3791588842868805,
      "learning_rate": 0.0003,
      "loss": 5.2863,
      "step": 3240
    },
    {
      "epoch": 0.06645096553560093,
      "grad_norm": 0.3161385953426361,
      "learning_rate": 0.0003,
      "loss": 5.3139,
      "step": 3245
    },
    {
      "epoch": 0.06655335531300555,
      "grad_norm": 0.2740941345691681,
      "learning_rate": 0.0003,
      "loss": 5.3195,
      "step": 3250
    },
    {
      "epoch": 0.06665574509041017,
      "grad_norm": 0.26943519711494446,
      "learning_rate": 0.0003,
      "loss": 5.2867,
      "step": 3255
    },
    {
      "epoch": 0.0667581348678148,
      "grad_norm": 0.2833729088306427,
      "learning_rate": 0.0003,
      "loss": 5.2506,
      "step": 3260
    },
    {
      "epoch": 0.06686052464521942,
      "grad_norm": 0.3066137135028839,
      "learning_rate": 0.0003,
      "loss": 5.2694,
      "step": 3265
    },
    {
      "epoch": 0.06696291442262405,
      "grad_norm": 0.3070327043533325,
      "learning_rate": 0.0003,
      "loss": 5.2966,
      "step": 3270
    },
    {
      "epoch": 0.06706530420002867,
      "grad_norm": 0.27142801880836487,
      "learning_rate": 0.0003,
      "loss": 5.2903,
      "step": 3275
    },
    {
      "epoch": 0.06716769397743329,
      "grad_norm": 0.26440364122390747,
      "learning_rate": 0.0003,
      "loss": 5.296,
      "step": 3280
    },
    {
      "epoch": 0.06727008375483792,
      "grad_norm": 0.27129313349723816,
      "learning_rate": 0.0003,
      "loss": 5.2924,
      "step": 3285
    },
    {
      "epoch": 0.06737247353224254,
      "grad_norm": 0.38910025358200073,
      "learning_rate": 0.0003,
      "loss": 5.339,
      "step": 3290
    },
    {
      "epoch": 0.06747486330964717,
      "grad_norm": 0.30716171860694885,
      "learning_rate": 0.0003,
      "loss": 5.2944,
      "step": 3295
    },
    {
      "epoch": 0.06757725308705179,
      "grad_norm": 0.3139477074146271,
      "learning_rate": 0.0003,
      "loss": 5.3103,
      "step": 3300
    },
    {
      "epoch": 0.06767964286445641,
      "grad_norm": 0.2722444534301758,
      "learning_rate": 0.0003,
      "loss": 5.2683,
      "step": 3305
    },
    {
      "epoch": 0.06778203264186104,
      "grad_norm": 0.2985815107822418,
      "learning_rate": 0.0003,
      "loss": 5.3044,
      "step": 3310
    },
    {
      "epoch": 0.06788442241926566,
      "grad_norm": 0.2560136318206787,
      "learning_rate": 0.0003,
      "loss": 5.3083,
      "step": 3315
    },
    {
      "epoch": 0.06798681219667028,
      "grad_norm": 0.25087544322013855,
      "learning_rate": 0.0003,
      "loss": 5.2957,
      "step": 3320
    },
    {
      "epoch": 0.0680892019740749,
      "grad_norm": 0.26948386430740356,
      "learning_rate": 0.0003,
      "loss": 5.2761,
      "step": 3325
    },
    {
      "epoch": 0.06819159175147953,
      "grad_norm": 0.2510646879673004,
      "learning_rate": 0.0003,
      "loss": 5.2919,
      "step": 3330
    },
    {
      "epoch": 0.06829398152888416,
      "grad_norm": 0.26186272501945496,
      "learning_rate": 0.0003,
      "loss": 5.3016,
      "step": 3335
    },
    {
      "epoch": 0.06839637130628878,
      "grad_norm": 0.3214239776134491,
      "learning_rate": 0.0003,
      "loss": 5.3502,
      "step": 3340
    },
    {
      "epoch": 0.0684987610836934,
      "grad_norm": 0.2626754939556122,
      "learning_rate": 0.0003,
      "loss": 5.2684,
      "step": 3345
    },
    {
      "epoch": 0.06860115086109803,
      "grad_norm": 0.2504951059818268,
      "learning_rate": 0.0003,
      "loss": 5.2904,
      "step": 3350
    },
    {
      "epoch": 0.06870354063850265,
      "grad_norm": 0.3028484284877777,
      "learning_rate": 0.0003,
      "loss": 5.3044,
      "step": 3355
    },
    {
      "epoch": 0.06880593041590728,
      "grad_norm": 0.2650691270828247,
      "learning_rate": 0.0003,
      "loss": 5.2882,
      "step": 3360
    },
    {
      "epoch": 0.0689083201933119,
      "grad_norm": 0.289113312959671,
      "learning_rate": 0.0003,
      "loss": 5.2831,
      "step": 3365
    },
    {
      "epoch": 0.06901070997071652,
      "grad_norm": 0.2897968888282776,
      "learning_rate": 0.0003,
      "loss": 5.2817,
      "step": 3370
    },
    {
      "epoch": 0.06911309974812115,
      "grad_norm": 0.27607211470603943,
      "learning_rate": 0.0003,
      "loss": 5.2752,
      "step": 3375
    },
    {
      "epoch": 0.06921548952552577,
      "grad_norm": 0.292235791683197,
      "learning_rate": 0.0003,
      "loss": 5.2336,
      "step": 3380
    },
    {
      "epoch": 0.0693178793029304,
      "grad_norm": 0.26884227991104126,
      "learning_rate": 0.0003,
      "loss": 5.2912,
      "step": 3385
    },
    {
      "epoch": 0.06942026908033502,
      "grad_norm": 0.2577894926071167,
      "learning_rate": 0.0003,
      "loss": 5.2823,
      "step": 3390
    },
    {
      "epoch": 0.06952265885773964,
      "grad_norm": 0.2647188603878021,
      "learning_rate": 0.0003,
      "loss": 5.2007,
      "step": 3395
    },
    {
      "epoch": 0.06962504863514427,
      "grad_norm": 0.2736581563949585,
      "learning_rate": 0.0003,
      "loss": 5.3041,
      "step": 3400
    },
    {
      "epoch": 0.06972743841254889,
      "grad_norm": 0.32651710510253906,
      "learning_rate": 0.0003,
      "loss": 5.3108,
      "step": 3405
    },
    {
      "epoch": 0.06982982818995352,
      "grad_norm": 0.2615613639354706,
      "learning_rate": 0.0003,
      "loss": 5.2654,
      "step": 3410
    },
    {
      "epoch": 0.06993221796735814,
      "grad_norm": 0.25817254185676575,
      "learning_rate": 0.0003,
      "loss": 5.2547,
      "step": 3415
    },
    {
      "epoch": 0.07003460774476276,
      "grad_norm": 0.2617282271385193,
      "learning_rate": 0.0003,
      "loss": 5.2543,
      "step": 3420
    },
    {
      "epoch": 0.07013699752216739,
      "grad_norm": 0.2925521433353424,
      "learning_rate": 0.0003,
      "loss": 5.2743,
      "step": 3425
    },
    {
      "epoch": 0.070239387299572,
      "grad_norm": 0.28042152523994446,
      "learning_rate": 0.0003,
      "loss": 5.2751,
      "step": 3430
    },
    {
      "epoch": 0.07034177707697664,
      "grad_norm": 0.30079346895217896,
      "learning_rate": 0.0003,
      "loss": 5.258,
      "step": 3435
    },
    {
      "epoch": 0.07044416685438126,
      "grad_norm": 0.2699873447418213,
      "learning_rate": 0.0003,
      "loss": 5.3195,
      "step": 3440
    },
    {
      "epoch": 0.07054655663178588,
      "grad_norm": 0.2844352126121521,
      "learning_rate": 0.0003,
      "loss": 5.2584,
      "step": 3445
    },
    {
      "epoch": 0.0706489464091905,
      "grad_norm": 0.42051392793655396,
      "learning_rate": 0.0003,
      "loss": 5.3133,
      "step": 3450
    },
    {
      "epoch": 0.07075133618659513,
      "grad_norm": 0.2783925533294678,
      "learning_rate": 0.0003,
      "loss": 5.2723,
      "step": 3455
    },
    {
      "epoch": 0.07085372596399975,
      "grad_norm": 0.3405534327030182,
      "learning_rate": 0.0003,
      "loss": 5.2908,
      "step": 3460
    },
    {
      "epoch": 0.07095611574140438,
      "grad_norm": 0.28797343373298645,
      "learning_rate": 0.0003,
      "loss": 5.2353,
      "step": 3465
    },
    {
      "epoch": 0.071058505518809,
      "grad_norm": 0.28846192359924316,
      "learning_rate": 0.0003,
      "loss": 5.2495,
      "step": 3470
    },
    {
      "epoch": 0.07116089529621362,
      "grad_norm": 0.25400757789611816,
      "learning_rate": 0.0003,
      "loss": 5.2741,
      "step": 3475
    },
    {
      "epoch": 0.07126328507361825,
      "grad_norm": 0.24486589431762695,
      "learning_rate": 0.0003,
      "loss": 5.2918,
      "step": 3480
    },
    {
      "epoch": 0.07136567485102287,
      "grad_norm": 0.3392948508262634,
      "learning_rate": 0.0003,
      "loss": 5.2564,
      "step": 3485
    },
    {
      "epoch": 0.0714680646284275,
      "grad_norm": 0.27733278274536133,
      "learning_rate": 0.0003,
      "loss": 5.2967,
      "step": 3490
    },
    {
      "epoch": 0.07157045440583212,
      "grad_norm": 0.28385838866233826,
      "learning_rate": 0.0003,
      "loss": 5.284,
      "step": 3495
    },
    {
      "epoch": 0.07167284418323674,
      "grad_norm": 0.3224107325077057,
      "learning_rate": 0.0003,
      "loss": 5.1633,
      "step": 3500
    },
    {
      "epoch": 0.07177523396064137,
      "grad_norm": 0.2476559430360794,
      "learning_rate": 0.0003,
      "loss": 5.2906,
      "step": 3505
    },
    {
      "epoch": 0.07187762373804599,
      "grad_norm": 0.2646051347255707,
      "learning_rate": 0.0003,
      "loss": 5.2459,
      "step": 3510
    },
    {
      "epoch": 0.07198001351545062,
      "grad_norm": 0.2576812207698822,
      "learning_rate": 0.0003,
      "loss": 5.3111,
      "step": 3515
    },
    {
      "epoch": 0.07208240329285524,
      "grad_norm": 0.260553240776062,
      "learning_rate": 0.0003,
      "loss": 5.2321,
      "step": 3520
    },
    {
      "epoch": 0.07218479307025986,
      "grad_norm": 0.261356383562088,
      "learning_rate": 0.0003,
      "loss": 5.2137,
      "step": 3525
    },
    {
      "epoch": 0.07228718284766449,
      "grad_norm": 0.24664048850536346,
      "learning_rate": 0.0003,
      "loss": 5.2492,
      "step": 3530
    },
    {
      "epoch": 0.0723895726250691,
      "grad_norm": 0.26263877749443054,
      "learning_rate": 0.0003,
      "loss": 5.229,
      "step": 3535
    },
    {
      "epoch": 0.07249196240247374,
      "grad_norm": 0.28026822209358215,
      "learning_rate": 0.0003,
      "loss": 5.2291,
      "step": 3540
    },
    {
      "epoch": 0.07259435217987836,
      "grad_norm": 0.2834453582763672,
      "learning_rate": 0.0003,
      "loss": 5.231,
      "step": 3545
    },
    {
      "epoch": 0.07269674195728298,
      "grad_norm": 0.28786909580230713,
      "learning_rate": 0.0003,
      "loss": 5.2743,
      "step": 3550
    },
    {
      "epoch": 0.0727991317346876,
      "grad_norm": 0.260613352060318,
      "learning_rate": 0.0003,
      "loss": 5.2542,
      "step": 3555
    },
    {
      "epoch": 0.07290152151209223,
      "grad_norm": 0.32520389556884766,
      "learning_rate": 0.0003,
      "loss": 5.2166,
      "step": 3560
    },
    {
      "epoch": 0.07300391128949686,
      "grad_norm": 0.2567857503890991,
      "learning_rate": 0.0003,
      "loss": 5.3022,
      "step": 3565
    },
    {
      "epoch": 0.07310630106690148,
      "grad_norm": 0.2632225453853607,
      "learning_rate": 0.0003,
      "loss": 5.2454,
      "step": 3570
    },
    {
      "epoch": 0.0732086908443061,
      "grad_norm": 0.26637014746665955,
      "learning_rate": 0.0003,
      "loss": 5.2041,
      "step": 3575
    },
    {
      "epoch": 0.07331108062171073,
      "grad_norm": 0.28755322098731995,
      "learning_rate": 0.0003,
      "loss": 5.2866,
      "step": 3580
    },
    {
      "epoch": 0.07341347039911535,
      "grad_norm": 0.32612115144729614,
      "learning_rate": 0.0003,
      "loss": 5.3024,
      "step": 3585
    },
    {
      "epoch": 0.07351586017651998,
      "grad_norm": 0.30339866876602173,
      "learning_rate": 0.0003,
      "loss": 5.2562,
      "step": 3590
    },
    {
      "epoch": 0.0736182499539246,
      "grad_norm": 0.2615393400192261,
      "learning_rate": 0.0003,
      "loss": 5.2801,
      "step": 3595
    },
    {
      "epoch": 0.07372063973132922,
      "grad_norm": 0.28131192922592163,
      "learning_rate": 0.0003,
      "loss": 5.2344,
      "step": 3600
    },
    {
      "epoch": 0.07382302950873385,
      "grad_norm": 0.2984217703342438,
      "learning_rate": 0.0003,
      "loss": 5.2558,
      "step": 3605
    },
    {
      "epoch": 0.07392541928613847,
      "grad_norm": 0.3154425323009491,
      "learning_rate": 0.0003,
      "loss": 5.2796,
      "step": 3610
    },
    {
      "epoch": 0.0740278090635431,
      "grad_norm": 0.27574992179870605,
      "learning_rate": 0.0003,
      "loss": 5.2575,
      "step": 3615
    },
    {
      "epoch": 0.07413019884094772,
      "grad_norm": 0.2771117687225342,
      "learning_rate": 0.0003,
      "loss": 5.2447,
      "step": 3620
    },
    {
      "epoch": 0.07423258861835234,
      "grad_norm": 0.2700023949146271,
      "learning_rate": 0.0003,
      "loss": 5.2851,
      "step": 3625
    },
    {
      "epoch": 0.07433497839575696,
      "grad_norm": 0.24821822345256805,
      "learning_rate": 0.0003,
      "loss": 5.2321,
      "step": 3630
    },
    {
      "epoch": 0.07443736817316159,
      "grad_norm": 0.2916540801525116,
      "learning_rate": 0.0003,
      "loss": 5.2355,
      "step": 3635
    },
    {
      "epoch": 0.07453975795056622,
      "grad_norm": 0.24802926182746887,
      "learning_rate": 0.0003,
      "loss": 5.2403,
      "step": 3640
    },
    {
      "epoch": 0.07464214772797084,
      "grad_norm": 0.24772369861602783,
      "learning_rate": 0.0003,
      "loss": 5.2455,
      "step": 3645
    },
    {
      "epoch": 0.07474453750537546,
      "grad_norm": 0.25520002841949463,
      "learning_rate": 0.0003,
      "loss": 5.2628,
      "step": 3650
    },
    {
      "epoch": 0.07484692728278008,
      "grad_norm": 0.2472926676273346,
      "learning_rate": 0.0003,
      "loss": 5.2709,
      "step": 3655
    },
    {
      "epoch": 0.0749493170601847,
      "grad_norm": 0.2765505015850067,
      "learning_rate": 0.0003,
      "loss": 5.2843,
      "step": 3660
    },
    {
      "epoch": 0.07505170683758934,
      "grad_norm": 0.26966553926467896,
      "learning_rate": 0.0003,
      "loss": 5.2948,
      "step": 3665
    },
    {
      "epoch": 0.07515409661499396,
      "grad_norm": 0.2616786062717438,
      "learning_rate": 0.0003,
      "loss": 5.2205,
      "step": 3670
    },
    {
      "epoch": 0.07525648639239858,
      "grad_norm": 0.27883464097976685,
      "learning_rate": 0.0003,
      "loss": 5.2227,
      "step": 3675
    },
    {
      "epoch": 0.0753588761698032,
      "grad_norm": 0.2898852825164795,
      "learning_rate": 0.0003,
      "loss": 5.2056,
      "step": 3680
    },
    {
      "epoch": 0.07546126594720783,
      "grad_norm": 0.2799520194530487,
      "learning_rate": 0.0003,
      "loss": 5.2248,
      "step": 3685
    },
    {
      "epoch": 0.07556365572461246,
      "grad_norm": 0.24998842179775238,
      "learning_rate": 0.0003,
      "loss": 5.2606,
      "step": 3690
    },
    {
      "epoch": 0.07566604550201708,
      "grad_norm": 0.32577699422836304,
      "learning_rate": 0.0003,
      "loss": 5.2199,
      "step": 3695
    },
    {
      "epoch": 0.0757684352794217,
      "grad_norm": 0.3090455234050751,
      "learning_rate": 0.0003,
      "loss": 5.2008,
      "step": 3700
    },
    {
      "epoch": 0.07587082505682632,
      "grad_norm": 0.29049402475357056,
      "learning_rate": 0.0003,
      "loss": 5.2657,
      "step": 3705
    },
    {
      "epoch": 0.07597321483423095,
      "grad_norm": 0.2764868140220642,
      "learning_rate": 0.0003,
      "loss": 5.2243,
      "step": 3710
    },
    {
      "epoch": 0.07607560461163558,
      "grad_norm": 0.2757784426212311,
      "learning_rate": 0.0003,
      "loss": 5.2708,
      "step": 3715
    },
    {
      "epoch": 0.0761779943890402,
      "grad_norm": 0.2754042446613312,
      "learning_rate": 0.0003,
      "loss": 5.2574,
      "step": 3720
    },
    {
      "epoch": 0.07628038416644482,
      "grad_norm": 0.2590422034263611,
      "learning_rate": 0.0003,
      "loss": 5.2617,
      "step": 3725
    },
    {
      "epoch": 0.07638277394384944,
      "grad_norm": 0.2684026062488556,
      "learning_rate": 0.0003,
      "loss": 5.201,
      "step": 3730
    },
    {
      "epoch": 0.07648516372125407,
      "grad_norm": 0.28633198142051697,
      "learning_rate": 0.0003,
      "loss": 5.2652,
      "step": 3735
    },
    {
      "epoch": 0.0765875534986587,
      "grad_norm": 0.2651969790458679,
      "learning_rate": 0.0003,
      "loss": 5.2194,
      "step": 3740
    },
    {
      "epoch": 0.07668994327606332,
      "grad_norm": 0.25093379616737366,
      "learning_rate": 0.0003,
      "loss": 5.2727,
      "step": 3745
    },
    {
      "epoch": 0.07679233305346794,
      "grad_norm": 0.2667597234249115,
      "learning_rate": 0.0003,
      "loss": 5.2158,
      "step": 3750
    },
    {
      "epoch": 0.07689472283087256,
      "grad_norm": 0.26323240995407104,
      "learning_rate": 0.0003,
      "loss": 5.2185,
      "step": 3755
    },
    {
      "epoch": 0.07699711260827719,
      "grad_norm": 0.24883873760700226,
      "learning_rate": 0.0003,
      "loss": 5.2437,
      "step": 3760
    },
    {
      "epoch": 0.07709950238568182,
      "grad_norm": 0.24001909792423248,
      "learning_rate": 0.0003,
      "loss": 5.2574,
      "step": 3765
    },
    {
      "epoch": 0.07720189216308644,
      "grad_norm": 0.2598496377468109,
      "learning_rate": 0.0003,
      "loss": 5.1732,
      "step": 3770
    },
    {
      "epoch": 0.07730428194049106,
      "grad_norm": 0.26076823472976685,
      "learning_rate": 0.0003,
      "loss": 5.2255,
      "step": 3775
    },
    {
      "epoch": 0.07740667171789568,
      "grad_norm": 0.24912242591381073,
      "learning_rate": 0.0003,
      "loss": 5.2146,
      "step": 3780
    },
    {
      "epoch": 0.0775090614953003,
      "grad_norm": 0.2977564036846161,
      "learning_rate": 0.0003,
      "loss": 5.1693,
      "step": 3785
    },
    {
      "epoch": 0.07761145127270494,
      "grad_norm": 0.27076271176338196,
      "learning_rate": 0.0003,
      "loss": 5.2399,
      "step": 3790
    },
    {
      "epoch": 0.07771384105010956,
      "grad_norm": 0.2570875883102417,
      "learning_rate": 0.0003,
      "loss": 5.2905,
      "step": 3795
    },
    {
      "epoch": 0.07781623082751418,
      "grad_norm": 0.25177836418151855,
      "learning_rate": 0.0003,
      "loss": 5.2525,
      "step": 3800
    },
    {
      "epoch": 0.0779186206049188,
      "grad_norm": 0.32862451672554016,
      "learning_rate": 0.0003,
      "loss": 5.1975,
      "step": 3805
    },
    {
      "epoch": 0.07802101038232342,
      "grad_norm": 0.274938702583313,
      "learning_rate": 0.0003,
      "loss": 5.2303,
      "step": 3810
    },
    {
      "epoch": 0.07812340015972805,
      "grad_norm": 0.27677834033966064,
      "learning_rate": 0.0003,
      "loss": 5.2273,
      "step": 3815
    },
    {
      "epoch": 0.07822578993713268,
      "grad_norm": 0.2708391845226288,
      "learning_rate": 0.0003,
      "loss": 5.242,
      "step": 3820
    },
    {
      "epoch": 0.0783281797145373,
      "grad_norm": 0.25845611095428467,
      "learning_rate": 0.0003,
      "loss": 5.2319,
      "step": 3825
    },
    {
      "epoch": 0.07843056949194192,
      "grad_norm": 0.2609349489212036,
      "learning_rate": 0.0003,
      "loss": 5.2428,
      "step": 3830
    },
    {
      "epoch": 0.07853295926934654,
      "grad_norm": 0.2555552124977112,
      "learning_rate": 0.0003,
      "loss": 5.2292,
      "step": 3835
    },
    {
      "epoch": 0.07863534904675117,
      "grad_norm": 0.2581983506679535,
      "learning_rate": 0.0003,
      "loss": 5.1964,
      "step": 3840
    },
    {
      "epoch": 0.0787377388241558,
      "grad_norm": 0.3047882318496704,
      "learning_rate": 0.0003,
      "loss": 5.3112,
      "step": 3845
    },
    {
      "epoch": 0.07884012860156042,
      "grad_norm": 0.7512697577476501,
      "learning_rate": 0.0003,
      "loss": 5.2443,
      "step": 3850
    },
    {
      "epoch": 0.07894251837896504,
      "grad_norm": 0.321704238653183,
      "learning_rate": 0.0003,
      "loss": 5.2639,
      "step": 3855
    },
    {
      "epoch": 0.07904490815636966,
      "grad_norm": 0.27286747097969055,
      "learning_rate": 0.0003,
      "loss": 5.1557,
      "step": 3860
    },
    {
      "epoch": 0.07914729793377429,
      "grad_norm": 0.2746713161468506,
      "learning_rate": 0.0003,
      "loss": 5.2666,
      "step": 3865
    },
    {
      "epoch": 0.07924968771117892,
      "grad_norm": 0.3512607216835022,
      "learning_rate": 0.0003,
      "loss": 5.2191,
      "step": 3870
    },
    {
      "epoch": 0.07935207748858354,
      "grad_norm": 0.2750149965286255,
      "learning_rate": 0.0003,
      "loss": 5.2028,
      "step": 3875
    },
    {
      "epoch": 0.07945446726598816,
      "grad_norm": 0.2558034658432007,
      "learning_rate": 0.0003,
      "loss": 5.2566,
      "step": 3880
    },
    {
      "epoch": 0.07955685704339278,
      "grad_norm": 0.2888765037059784,
      "learning_rate": 0.0003,
      "loss": 5.2306,
      "step": 3885
    },
    {
      "epoch": 0.0796592468207974,
      "grad_norm": 0.3252604305744171,
      "learning_rate": 0.0003,
      "loss": 5.173,
      "step": 3890
    },
    {
      "epoch": 0.07976163659820204,
      "grad_norm": 0.2762053906917572,
      "learning_rate": 0.0003,
      "loss": 5.1917,
      "step": 3895
    },
    {
      "epoch": 0.07986402637560666,
      "grad_norm": 0.270160973072052,
      "learning_rate": 0.0003,
      "loss": 5.2439,
      "step": 3900
    },
    {
      "epoch": 0.07996641615301128,
      "grad_norm": 0.2815668284893036,
      "learning_rate": 0.0003,
      "loss": 5.2011,
      "step": 3905
    },
    {
      "epoch": 0.0800688059304159,
      "grad_norm": 0.24999023973941803,
      "learning_rate": 0.0003,
      "loss": 5.2585,
      "step": 3910
    },
    {
      "epoch": 0.08017119570782053,
      "grad_norm": 0.2633795142173767,
      "learning_rate": 0.0003,
      "loss": 5.1888,
      "step": 3915
    },
    {
      "epoch": 0.08027358548522516,
      "grad_norm": 0.25494393706321716,
      "learning_rate": 0.0003,
      "loss": 5.2547,
      "step": 3920
    },
    {
      "epoch": 0.08037597526262978,
      "grad_norm": 0.2790360450744629,
      "learning_rate": 0.0003,
      "loss": 5.1955,
      "step": 3925
    },
    {
      "epoch": 0.0804783650400344,
      "grad_norm": 0.2529444396495819,
      "learning_rate": 0.0003,
      "loss": 5.2178,
      "step": 3930
    },
    {
      "epoch": 0.08058075481743902,
      "grad_norm": 0.2555560767650604,
      "learning_rate": 0.0003,
      "loss": 5.2064,
      "step": 3935
    },
    {
      "epoch": 0.08068314459484364,
      "grad_norm": 0.2914392352104187,
      "learning_rate": 0.0003,
      "loss": 5.2048,
      "step": 3940
    },
    {
      "epoch": 0.08078553437224828,
      "grad_norm": 0.2684366703033447,
      "learning_rate": 0.0003,
      "loss": 5.1977,
      "step": 3945
    },
    {
      "epoch": 0.0808879241496529,
      "grad_norm": 0.2719501256942749,
      "learning_rate": 0.0003,
      "loss": 5.1895,
      "step": 3950
    },
    {
      "epoch": 0.08099031392705752,
      "grad_norm": 0.27822640538215637,
      "learning_rate": 0.0003,
      "loss": 5.1856,
      "step": 3955
    },
    {
      "epoch": 0.08109270370446214,
      "grad_norm": 0.30609002709388733,
      "learning_rate": 0.0003,
      "loss": 5.1872,
      "step": 3960
    },
    {
      "epoch": 0.08119509348186676,
      "grad_norm": 0.26783978939056396,
      "learning_rate": 0.0003,
      "loss": 5.2106,
      "step": 3965
    },
    {
      "epoch": 0.0812974832592714,
      "grad_norm": 0.2516503930091858,
      "learning_rate": 0.0003,
      "loss": 5.3007,
      "step": 3970
    },
    {
      "epoch": 0.08139987303667602,
      "grad_norm": 0.26788803935050964,
      "learning_rate": 0.0003,
      "loss": 5.2178,
      "step": 3975
    },
    {
      "epoch": 0.08150226281408064,
      "grad_norm": 0.25448089838027954,
      "learning_rate": 0.0003,
      "loss": 5.2223,
      "step": 3980
    },
    {
      "epoch": 0.08160465259148526,
      "grad_norm": 0.26501232385635376,
      "learning_rate": 0.0003,
      "loss": 5.2898,
      "step": 3985
    },
    {
      "epoch": 0.08170704236888988,
      "grad_norm": 0.26919659972190857,
      "learning_rate": 0.0003,
      "loss": 5.1992,
      "step": 3990
    },
    {
      "epoch": 0.08180943214629452,
      "grad_norm": 0.29884156584739685,
      "learning_rate": 0.0003,
      "loss": 5.19,
      "step": 3995
    },
    {
      "epoch": 0.08191182192369914,
      "grad_norm": 0.26546797156333923,
      "learning_rate": 0.0003,
      "loss": 5.207,
      "step": 4000
    },
    {
      "epoch": 0.08191182192369914,
      "eval_loss": 5.231751441955566,
      "eval_runtime": 17.2576,
      "eval_samples_per_second": 33.724,
      "eval_steps_per_second": 4.23,
      "step": 4000
    },
    {
      "epoch": 0.08201421170110376,
      "grad_norm": 0.28061461448669434,
      "learning_rate": 0.0003,
      "loss": 5.1794,
      "step": 4005
    },
    {
      "epoch": 0.08211660147850838,
      "grad_norm": 0.3089412748813629,
      "learning_rate": 0.0003,
      "loss": 5.2058,
      "step": 4010
    },
    {
      "epoch": 0.082218991255913,
      "grad_norm": 0.24918992817401886,
      "learning_rate": 0.0003,
      "loss": 5.1594,
      "step": 4015
    },
    {
      "epoch": 0.08232138103331764,
      "grad_norm": 0.2777044475078583,
      "learning_rate": 0.0003,
      "loss": 5.209,
      "step": 4020
    },
    {
      "epoch": 0.08242377081072226,
      "grad_norm": 0.262815922498703,
      "learning_rate": 0.0003,
      "loss": 5.2152,
      "step": 4025
    },
    {
      "epoch": 0.08252616058812688,
      "grad_norm": 0.29473477602005005,
      "learning_rate": 0.0003,
      "loss": 5.1525,
      "step": 4030
    },
    {
      "epoch": 0.0826285503655315,
      "grad_norm": 0.2584522068500519,
      "learning_rate": 0.0003,
      "loss": 5.2217,
      "step": 4035
    },
    {
      "epoch": 0.08273094014293612,
      "grad_norm": 0.22683992981910706,
      "learning_rate": 0.0003,
      "loss": 5.2038,
      "step": 4040
    },
    {
      "epoch": 0.08283332992034076,
      "grad_norm": 0.23628821969032288,
      "learning_rate": 0.0003,
      "loss": 5.2474,
      "step": 4045
    },
    {
      "epoch": 0.08293571969774538,
      "grad_norm": 0.2681495249271393,
      "learning_rate": 0.0003,
      "loss": 5.2409,
      "step": 4050
    },
    {
      "epoch": 0.08303810947515,
      "grad_norm": 0.29070958495140076,
      "learning_rate": 0.0003,
      "loss": 5.1777,
      "step": 4055
    },
    {
      "epoch": 0.08314049925255462,
      "grad_norm": 0.3845941126346588,
      "learning_rate": 0.0003,
      "loss": 5.2047,
      "step": 4060
    },
    {
      "epoch": 0.08324288902995924,
      "grad_norm": 0.2862512171268463,
      "learning_rate": 0.0003,
      "loss": 5.1744,
      "step": 4065
    },
    {
      "epoch": 0.08334527880736388,
      "grad_norm": 0.2699524760246277,
      "learning_rate": 0.0003,
      "loss": 5.1751,
      "step": 4070
    },
    {
      "epoch": 0.0834476685847685,
      "grad_norm": 0.277503103017807,
      "learning_rate": 0.0003,
      "loss": 5.2415,
      "step": 4075
    },
    {
      "epoch": 0.08355005836217312,
      "grad_norm": 0.26220452785491943,
      "learning_rate": 0.0003,
      "loss": 5.2355,
      "step": 4080
    },
    {
      "epoch": 0.08365244813957774,
      "grad_norm": 0.25356751680374146,
      "learning_rate": 0.0003,
      "loss": 5.2204,
      "step": 4085
    },
    {
      "epoch": 0.08375483791698236,
      "grad_norm": 0.24540142714977264,
      "learning_rate": 0.0003,
      "loss": 5.181,
      "step": 4090
    },
    {
      "epoch": 0.083857227694387,
      "grad_norm": 0.24948079884052277,
      "learning_rate": 0.0003,
      "loss": 5.1546,
      "step": 4095
    },
    {
      "epoch": 0.08395961747179162,
      "grad_norm": 0.2514345645904541,
      "learning_rate": 0.0003,
      "loss": 5.2463,
      "step": 4100
    },
    {
      "epoch": 0.08406200724919624,
      "grad_norm": 0.22873541712760925,
      "learning_rate": 0.0003,
      "loss": 5.1944,
      "step": 4105
    },
    {
      "epoch": 0.08416439702660086,
      "grad_norm": 0.29980912804603577,
      "learning_rate": 0.0003,
      "loss": 5.2013,
      "step": 4110
    },
    {
      "epoch": 0.08426678680400548,
      "grad_norm": 0.37752649188041687,
      "learning_rate": 0.0003,
      "loss": 5.1881,
      "step": 4115
    },
    {
      "epoch": 0.08436917658141012,
      "grad_norm": 0.27054810523986816,
      "learning_rate": 0.0003,
      "loss": 5.1772,
      "step": 4120
    },
    {
      "epoch": 0.08447156635881474,
      "grad_norm": 0.2550160586833954,
      "learning_rate": 0.0003,
      "loss": 5.2239,
      "step": 4125
    },
    {
      "epoch": 0.08457395613621936,
      "grad_norm": 0.2638941705226898,
      "learning_rate": 0.0003,
      "loss": 5.221,
      "step": 4130
    },
    {
      "epoch": 0.08467634591362398,
      "grad_norm": 0.2580711245536804,
      "learning_rate": 0.0003,
      "loss": 5.1863,
      "step": 4135
    },
    {
      "epoch": 0.0847787356910286,
      "grad_norm": 0.26525193452835083,
      "learning_rate": 0.0003,
      "loss": 5.1725,
      "step": 4140
    },
    {
      "epoch": 0.08488112546843324,
      "grad_norm": 0.2745465040206909,
      "learning_rate": 0.0003,
      "loss": 5.2119,
      "step": 4145
    },
    {
      "epoch": 0.08498351524583786,
      "grad_norm": 0.2355852574110031,
      "learning_rate": 0.0003,
      "loss": 5.1977,
      "step": 4150
    },
    {
      "epoch": 0.08508590502324248,
      "grad_norm": 0.2682251036167145,
      "learning_rate": 0.0003,
      "loss": 5.2275,
      "step": 4155
    },
    {
      "epoch": 0.0851882948006471,
      "grad_norm": 0.26811787486076355,
      "learning_rate": 0.0003,
      "loss": 5.1731,
      "step": 4160
    },
    {
      "epoch": 0.08529068457805172,
      "grad_norm": 0.26039308309555054,
      "learning_rate": 0.0003,
      "loss": 5.1438,
      "step": 4165
    },
    {
      "epoch": 0.08539307435545636,
      "grad_norm": 0.31183162331581116,
      "learning_rate": 0.0003,
      "loss": 5.2215,
      "step": 4170
    },
    {
      "epoch": 0.08549546413286098,
      "grad_norm": 0.30162638425827026,
      "learning_rate": 0.0003,
      "loss": 5.1634,
      "step": 4175
    },
    {
      "epoch": 0.0855978539102656,
      "grad_norm": 0.27771133184432983,
      "learning_rate": 0.0003,
      "loss": 5.1333,
      "step": 4180
    },
    {
      "epoch": 0.08570024368767022,
      "grad_norm": 0.30736786127090454,
      "learning_rate": 0.0003,
      "loss": 5.2121,
      "step": 4185
    },
    {
      "epoch": 0.08580263346507484,
      "grad_norm": 0.29298821091651917,
      "learning_rate": 0.0003,
      "loss": 5.224,
      "step": 4190
    },
    {
      "epoch": 0.08590502324247946,
      "grad_norm": 0.2820582091808319,
      "learning_rate": 0.0003,
      "loss": 5.1978,
      "step": 4195
    },
    {
      "epoch": 0.0860074130198841,
      "grad_norm": 0.26031070947647095,
      "learning_rate": 0.0003,
      "loss": 5.2011,
      "step": 4200
    },
    {
      "epoch": 0.08610980279728872,
      "grad_norm": 0.321780264377594,
      "learning_rate": 0.0003,
      "loss": 5.1574,
      "step": 4205
    },
    {
      "epoch": 0.08621219257469334,
      "grad_norm": 0.31359395384788513,
      "learning_rate": 0.0003,
      "loss": 5.1827,
      "step": 4210
    },
    {
      "epoch": 0.08631458235209796,
      "grad_norm": 0.7416604161262512,
      "learning_rate": 0.0003,
      "loss": 5.0659,
      "step": 4215
    },
    {
      "epoch": 0.08641697212950258,
      "grad_norm": 0.2704704701900482,
      "learning_rate": 0.0003,
      "loss": 5.1829,
      "step": 4220
    },
    {
      "epoch": 0.08651936190690722,
      "grad_norm": 0.2521301209926605,
      "learning_rate": 0.0003,
      "loss": 5.2136,
      "step": 4225
    },
    {
      "epoch": 0.08662175168431184,
      "grad_norm": 0.25025126338005066,
      "learning_rate": 0.0003,
      "loss": 5.1878,
      "step": 4230
    },
    {
      "epoch": 0.08672414146171646,
      "grad_norm": 0.3045847415924072,
      "learning_rate": 0.0003,
      "loss": 5.1402,
      "step": 4235
    },
    {
      "epoch": 0.08682653123912108,
      "grad_norm": 0.5664308071136475,
      "learning_rate": 0.0003,
      "loss": 5.1917,
      "step": 4240
    },
    {
      "epoch": 0.0869289210165257,
      "grad_norm": 0.2421114146709442,
      "learning_rate": 0.0003,
      "loss": 5.1617,
      "step": 4245
    },
    {
      "epoch": 0.08703131079393034,
      "grad_norm": 0.26926347613334656,
      "learning_rate": 0.0003,
      "loss": 5.2445,
      "step": 4250
    },
    {
      "epoch": 0.08713370057133496,
      "grad_norm": 0.2814469337463379,
      "learning_rate": 0.0003,
      "loss": 5.2284,
      "step": 4255
    },
    {
      "epoch": 0.08723609034873958,
      "grad_norm": 0.2786811590194702,
      "learning_rate": 0.0003,
      "loss": 5.2273,
      "step": 4260
    },
    {
      "epoch": 0.0873384801261442,
      "grad_norm": 0.30829358100891113,
      "learning_rate": 0.0003,
      "loss": 5.1942,
      "step": 4265
    },
    {
      "epoch": 0.08744086990354882,
      "grad_norm": 0.2992086708545685,
      "learning_rate": 0.0003,
      "loss": 5.1899,
      "step": 4270
    },
    {
      "epoch": 0.08754325968095346,
      "grad_norm": 0.285158634185791,
      "learning_rate": 0.0003,
      "loss": 5.218,
      "step": 4275
    },
    {
      "epoch": 0.08764564945835808,
      "grad_norm": 0.2520645558834076,
      "learning_rate": 0.0003,
      "loss": 5.2074,
      "step": 4280
    },
    {
      "epoch": 0.0877480392357627,
      "grad_norm": 0.23772001266479492,
      "learning_rate": 0.0003,
      "loss": 5.21,
      "step": 4285
    },
    {
      "epoch": 0.08785042901316732,
      "grad_norm": 0.2620167136192322,
      "learning_rate": 0.0003,
      "loss": 5.1579,
      "step": 4290
    },
    {
      "epoch": 0.08795281879057194,
      "grad_norm": 0.24005870521068573,
      "learning_rate": 0.0003,
      "loss": 5.1406,
      "step": 4295
    },
    {
      "epoch": 0.08805520856797658,
      "grad_norm": 0.27313220500946045,
      "learning_rate": 0.0003,
      "loss": 5.2065,
      "step": 4300
    },
    {
      "epoch": 0.0881575983453812,
      "grad_norm": 0.2567211091518402,
      "learning_rate": 0.0003,
      "loss": 5.1661,
      "step": 4305
    },
    {
      "epoch": 0.08825998812278582,
      "grad_norm": 0.2909543216228485,
      "learning_rate": 0.0003,
      "loss": 5.1768,
      "step": 4310
    },
    {
      "epoch": 0.08836237790019044,
      "grad_norm": 0.2807057201862335,
      "learning_rate": 0.0003,
      "loss": 5.2037,
      "step": 4315
    },
    {
      "epoch": 0.08846476767759506,
      "grad_norm": 0.24894212186336517,
      "learning_rate": 0.0003,
      "loss": 5.1616,
      "step": 4320
    },
    {
      "epoch": 0.0885671574549997,
      "grad_norm": 0.2790505290031433,
      "learning_rate": 0.0003,
      "loss": 5.1889,
      "step": 4325
    },
    {
      "epoch": 0.08866954723240432,
      "grad_norm": 0.24801431596279144,
      "learning_rate": 0.0003,
      "loss": 5.1444,
      "step": 4330
    },
    {
      "epoch": 0.08877193700980894,
      "grad_norm": 0.2559351921081543,
      "learning_rate": 0.0003,
      "loss": 5.1971,
      "step": 4335
    },
    {
      "epoch": 0.08887432678721356,
      "grad_norm": 0.2676590383052826,
      "learning_rate": 0.0003,
      "loss": 5.1643,
      "step": 4340
    },
    {
      "epoch": 0.08897671656461818,
      "grad_norm": 0.26642993092536926,
      "learning_rate": 0.0003,
      "loss": 5.1413,
      "step": 4345
    },
    {
      "epoch": 0.08907910634202282,
      "grad_norm": 0.386616587638855,
      "learning_rate": 0.0003,
      "loss": 5.1662,
      "step": 4350
    },
    {
      "epoch": 0.08918149611942744,
      "grad_norm": 0.26433825492858887,
      "learning_rate": 0.0003,
      "loss": 5.1705,
      "step": 4355
    },
    {
      "epoch": 0.08928388589683206,
      "grad_norm": 0.2568153142929077,
      "learning_rate": 0.0003,
      "loss": 5.1989,
      "step": 4360
    },
    {
      "epoch": 0.08938627567423668,
      "grad_norm": 0.24500320851802826,
      "learning_rate": 0.0003,
      "loss": 5.1974,
      "step": 4365
    },
    {
      "epoch": 0.0894886654516413,
      "grad_norm": 0.2564222514629364,
      "learning_rate": 0.0003,
      "loss": 5.2068,
      "step": 4370
    },
    {
      "epoch": 0.08959105522904594,
      "grad_norm": 0.3021552562713623,
      "learning_rate": 0.0003,
      "loss": 5.166,
      "step": 4375
    },
    {
      "epoch": 0.08969344500645056,
      "grad_norm": 0.509596586227417,
      "learning_rate": 0.0003,
      "loss": 5.2846,
      "step": 4380
    },
    {
      "epoch": 0.08979583478385518,
      "grad_norm": 0.30619922280311584,
      "learning_rate": 0.0003,
      "loss": 5.1868,
      "step": 4385
    },
    {
      "epoch": 0.0898982245612598,
      "grad_norm": 0.3148152530193329,
      "learning_rate": 0.0003,
      "loss": 5.1385,
      "step": 4390
    },
    {
      "epoch": 0.09000061433866442,
      "grad_norm": 0.29068124294281006,
      "learning_rate": 0.0003,
      "loss": 5.1808,
      "step": 4395
    },
    {
      "epoch": 0.09010300411606906,
      "grad_norm": 0.25925976037979126,
      "learning_rate": 0.0003,
      "loss": 5.188,
      "step": 4400
    },
    {
      "epoch": 0.09020539389347368,
      "grad_norm": 0.2427985966205597,
      "learning_rate": 0.0003,
      "loss": 5.1415,
      "step": 4405
    },
    {
      "epoch": 0.0903077836708783,
      "grad_norm": 0.25115054845809937,
      "learning_rate": 0.0003,
      "loss": 5.1682,
      "step": 4410
    },
    {
      "epoch": 0.09041017344828292,
      "grad_norm": 0.33867621421813965,
      "learning_rate": 0.0003,
      "loss": 5.1595,
      "step": 4415
    },
    {
      "epoch": 0.09051256322568754,
      "grad_norm": 0.27806276082992554,
      "learning_rate": 0.0003,
      "loss": 5.1811,
      "step": 4420
    },
    {
      "epoch": 0.09061495300309218,
      "grad_norm": 0.2543964684009552,
      "learning_rate": 0.0003,
      "loss": 5.174,
      "step": 4425
    },
    {
      "epoch": 0.0907173427804968,
      "grad_norm": 0.26773425936698914,
      "learning_rate": 0.0003,
      "loss": 5.1857,
      "step": 4430
    },
    {
      "epoch": 0.09081973255790142,
      "grad_norm": 0.257776141166687,
      "learning_rate": 0.0003,
      "loss": 5.157,
      "step": 4435
    },
    {
      "epoch": 0.09092212233530604,
      "grad_norm": 0.25122201442718506,
      "learning_rate": 0.0003,
      "loss": 5.2171,
      "step": 4440
    },
    {
      "epoch": 0.09102451211271066,
      "grad_norm": 0.2553665041923523,
      "learning_rate": 0.0003,
      "loss": 5.1882,
      "step": 4445
    },
    {
      "epoch": 0.0911269018901153,
      "grad_norm": 0.25656768679618835,
      "learning_rate": 0.0003,
      "loss": 5.154,
      "step": 4450
    },
    {
      "epoch": 0.09122929166751992,
      "grad_norm": 0.24263906478881836,
      "learning_rate": 0.0003,
      "loss": 5.1328,
      "step": 4455
    },
    {
      "epoch": 0.09133168144492454,
      "grad_norm": 0.270973801612854,
      "learning_rate": 0.0003,
      "loss": 5.2115,
      "step": 4460
    },
    {
      "epoch": 0.09143407122232916,
      "grad_norm": 0.26029303669929504,
      "learning_rate": 0.0003,
      "loss": 5.0833,
      "step": 4465
    },
    {
      "epoch": 0.09153646099973378,
      "grad_norm": 0.260354220867157,
      "learning_rate": 0.0003,
      "loss": 5.1212,
      "step": 4470
    },
    {
      "epoch": 0.09163885077713842,
      "grad_norm": 0.2506909966468811,
      "learning_rate": 0.0003,
      "loss": 5.1503,
      "step": 4475
    },
    {
      "epoch": 0.09174124055454304,
      "grad_norm": 0.2777837812900543,
      "learning_rate": 0.0003,
      "loss": 5.1238,
      "step": 4480
    },
    {
      "epoch": 0.09184363033194766,
      "grad_norm": 0.300954133272171,
      "learning_rate": 0.0003,
      "loss": 5.1712,
      "step": 4485
    },
    {
      "epoch": 0.09194602010935228,
      "grad_norm": 0.2487037479877472,
      "learning_rate": 0.0003,
      "loss": 5.1237,
      "step": 4490
    },
    {
      "epoch": 0.0920484098867569,
      "grad_norm": 0.2557465732097626,
      "learning_rate": 0.0003,
      "loss": 5.1414,
      "step": 4495
    },
    {
      "epoch": 0.09215079966416154,
      "grad_norm": 0.27927854657173157,
      "learning_rate": 0.0003,
      "loss": 5.1912,
      "step": 4500
    },
    {
      "epoch": 0.09225318944156616,
      "grad_norm": 0.2784734070301056,
      "learning_rate": 0.0003,
      "loss": 5.1419,
      "step": 4505
    },
    {
      "epoch": 0.09235557921897078,
      "grad_norm": 0.2585296928882599,
      "learning_rate": 0.0003,
      "loss": 5.1454,
      "step": 4510
    },
    {
      "epoch": 0.0924579689963754,
      "grad_norm": 0.24738222360610962,
      "learning_rate": 0.0003,
      "loss": 5.1355,
      "step": 4515
    },
    {
      "epoch": 0.09256035877378002,
      "grad_norm": 0.2655542492866516,
      "learning_rate": 0.0003,
      "loss": 5.1845,
      "step": 4520
    },
    {
      "epoch": 0.09266274855118466,
      "grad_norm": 0.24730607867240906,
      "learning_rate": 0.0003,
      "loss": 5.1778,
      "step": 4525
    },
    {
      "epoch": 0.09276513832858928,
      "grad_norm": 0.24619928002357483,
      "learning_rate": 0.0003,
      "loss": 5.1195,
      "step": 4530
    },
    {
      "epoch": 0.0928675281059939,
      "grad_norm": 0.23726853728294373,
      "learning_rate": 0.0003,
      "loss": 5.1595,
      "step": 4535
    },
    {
      "epoch": 0.09296991788339852,
      "grad_norm": 0.24438048899173737,
      "learning_rate": 0.0003,
      "loss": 5.2443,
      "step": 4540
    },
    {
      "epoch": 0.09307230766080314,
      "grad_norm": 0.26595500111579895,
      "learning_rate": 0.0003,
      "loss": 5.1405,
      "step": 4545
    },
    {
      "epoch": 0.09317469743820776,
      "grad_norm": 0.2615746855735779,
      "learning_rate": 0.0003,
      "loss": 5.1826,
      "step": 4550
    },
    {
      "epoch": 0.0932770872156124,
      "grad_norm": 0.282305508852005,
      "learning_rate": 0.0003,
      "loss": 5.1594,
      "step": 4555
    },
    {
      "epoch": 0.09337947699301702,
      "grad_norm": 0.24737706780433655,
      "learning_rate": 0.0003,
      "loss": 5.1662,
      "step": 4560
    },
    {
      "epoch": 0.09348186677042164,
      "grad_norm": 0.2563910186290741,
      "learning_rate": 0.0003,
      "loss": 5.1875,
      "step": 4565
    },
    {
      "epoch": 0.09358425654782626,
      "grad_norm": 0.26837265491485596,
      "learning_rate": 0.0003,
      "loss": 5.1543,
      "step": 4570
    },
    {
      "epoch": 0.09368664632523088,
      "grad_norm": 0.26752883195877075,
      "learning_rate": 0.0003,
      "loss": 5.198,
      "step": 4575
    },
    {
      "epoch": 0.09378903610263552,
      "grad_norm": 0.33696436882019043,
      "learning_rate": 0.0003,
      "loss": 5.1321,
      "step": 4580
    },
    {
      "epoch": 0.09389142588004014,
      "grad_norm": 0.27908802032470703,
      "learning_rate": 0.0003,
      "loss": 5.1815,
      "step": 4585
    },
    {
      "epoch": 0.09399381565744476,
      "grad_norm": 0.2778615951538086,
      "learning_rate": 0.0003,
      "loss": 5.156,
      "step": 4590
    },
    {
      "epoch": 0.09409620543484938,
      "grad_norm": 0.27270758152008057,
      "learning_rate": 0.0003,
      "loss": 5.1562,
      "step": 4595
    },
    {
      "epoch": 0.094198595212254,
      "grad_norm": 0.2752048969268799,
      "learning_rate": 0.0003,
      "loss": 5.152,
      "step": 4600
    },
    {
      "epoch": 0.09430098498965864,
      "grad_norm": 0.2512056827545166,
      "learning_rate": 0.0003,
      "loss": 5.1777,
      "step": 4605
    },
    {
      "epoch": 0.09440337476706326,
      "grad_norm": 0.28345462679862976,
      "learning_rate": 0.0003,
      "loss": 5.1545,
      "step": 4610
    },
    {
      "epoch": 0.09450576454446788,
      "grad_norm": 0.26428791880607605,
      "learning_rate": 0.0003,
      "loss": 5.162,
      "step": 4615
    },
    {
      "epoch": 0.0946081543218725,
      "grad_norm": 0.26608046889305115,
      "learning_rate": 0.0003,
      "loss": 5.1649,
      "step": 4620
    },
    {
      "epoch": 0.09471054409927712,
      "grad_norm": 0.2517186403274536,
      "learning_rate": 0.0003,
      "loss": 5.1542,
      "step": 4625
    },
    {
      "epoch": 0.09481293387668176,
      "grad_norm": 0.27337828278541565,
      "learning_rate": 0.0003,
      "loss": 5.1467,
      "step": 4630
    },
    {
      "epoch": 0.09491532365408638,
      "grad_norm": 0.26084741950035095,
      "learning_rate": 0.0003,
      "loss": 5.201,
      "step": 4635
    },
    {
      "epoch": 0.095017713431491,
      "grad_norm": 0.26555636525154114,
      "learning_rate": 0.0003,
      "loss": 5.1691,
      "step": 4640
    },
    {
      "epoch": 0.09512010320889562,
      "grad_norm": 0.25768858194351196,
      "learning_rate": 0.0003,
      "loss": 5.1592,
      "step": 4645
    },
    {
      "epoch": 0.09522249298630024,
      "grad_norm": 0.2555982172489166,
      "learning_rate": 0.0003,
      "loss": 5.153,
      "step": 4650
    },
    {
      "epoch": 0.09532488276370488,
      "grad_norm": 0.26049119234085083,
      "learning_rate": 0.0003,
      "loss": 5.0963,
      "step": 4655
    },
    {
      "epoch": 0.0954272725411095,
      "grad_norm": 0.266277015209198,
      "learning_rate": 0.0003,
      "loss": 5.1567,
      "step": 4660
    },
    {
      "epoch": 0.09552966231851412,
      "grad_norm": 0.2679003179073334,
      "learning_rate": 0.0003,
      "loss": 5.1182,
      "step": 4665
    },
    {
      "epoch": 0.09563205209591874,
      "grad_norm": 0.23862691223621368,
      "learning_rate": 0.0003,
      "loss": 5.1466,
      "step": 4670
    },
    {
      "epoch": 0.09573444187332336,
      "grad_norm": 0.24389131367206573,
      "learning_rate": 0.0003,
      "loss": 5.1213,
      "step": 4675
    },
    {
      "epoch": 0.095836831650728,
      "grad_norm": 0.24601227045059204,
      "learning_rate": 0.0003,
      "loss": 5.1277,
      "step": 4680
    },
    {
      "epoch": 0.09593922142813262,
      "grad_norm": 0.2755354344844818,
      "learning_rate": 0.0003,
      "loss": 5.0533,
      "step": 4685
    },
    {
      "epoch": 0.09604161120553724,
      "grad_norm": 0.3208519518375397,
      "learning_rate": 0.0003,
      "loss": 5.0945,
      "step": 4690
    },
    {
      "epoch": 0.09614400098294186,
      "grad_norm": 0.23996461927890778,
      "learning_rate": 0.0003,
      "loss": 5.2057,
      "step": 4695
    },
    {
      "epoch": 0.09624639076034648,
      "grad_norm": 0.2781566083431244,
      "learning_rate": 0.0003,
      "loss": 5.1222,
      "step": 4700
    },
    {
      "epoch": 0.09634878053775112,
      "grad_norm": 0.243170365691185,
      "learning_rate": 0.0003,
      "loss": 5.1748,
      "step": 4705
    },
    {
      "epoch": 0.09645117031515574,
      "grad_norm": 0.2466488778591156,
      "learning_rate": 0.0003,
      "loss": 5.1225,
      "step": 4710
    },
    {
      "epoch": 0.09655356009256036,
      "grad_norm": 0.2508876621723175,
      "learning_rate": 0.0003,
      "loss": 5.1061,
      "step": 4715
    },
    {
      "epoch": 0.09665594986996498,
      "grad_norm": 0.27872124314308167,
      "learning_rate": 0.0003,
      "loss": 5.1281,
      "step": 4720
    },
    {
      "epoch": 0.0967583396473696,
      "grad_norm": 0.2469724863767624,
      "learning_rate": 0.0003,
      "loss": 5.1669,
      "step": 4725
    },
    {
      "epoch": 0.09686072942477424,
      "grad_norm": 0.2937515377998352,
      "learning_rate": 0.0003,
      "loss": 5.1424,
      "step": 4730
    },
    {
      "epoch": 0.09696311920217886,
      "grad_norm": 0.23722520470619202,
      "learning_rate": 0.0003,
      "loss": 5.1555,
      "step": 4735
    },
    {
      "epoch": 0.09706550897958348,
      "grad_norm": 0.24938583374023438,
      "learning_rate": 0.0003,
      "loss": 5.1483,
      "step": 4740
    },
    {
      "epoch": 0.0971678987569881,
      "grad_norm": 0.23783664405345917,
      "learning_rate": 0.0003,
      "loss": 5.1784,
      "step": 4745
    },
    {
      "epoch": 0.09727028853439272,
      "grad_norm": 0.24777854979038239,
      "learning_rate": 0.0003,
      "loss": 5.1539,
      "step": 4750
    },
    {
      "epoch": 0.09737267831179736,
      "grad_norm": 0.2501198947429657,
      "learning_rate": 0.0003,
      "loss": 5.1269,
      "step": 4755
    },
    {
      "epoch": 0.09747506808920198,
      "grad_norm": 0.2623465359210968,
      "learning_rate": 0.0003,
      "loss": 5.1248,
      "step": 4760
    },
    {
      "epoch": 0.0975774578666066,
      "grad_norm": 0.2732715904712677,
      "learning_rate": 0.0003,
      "loss": 5.1493,
      "step": 4765
    },
    {
      "epoch": 0.09767984764401122,
      "grad_norm": 0.22563378512859344,
      "learning_rate": 0.0003,
      "loss": 5.1442,
      "step": 4770
    },
    {
      "epoch": 0.09778223742141584,
      "grad_norm": 0.2790745198726654,
      "learning_rate": 0.0003,
      "loss": 5.1472,
      "step": 4775
    },
    {
      "epoch": 0.09788462719882048,
      "grad_norm": 0.2794435918331146,
      "learning_rate": 0.0003,
      "loss": 5.1209,
      "step": 4780
    },
    {
      "epoch": 0.0979870169762251,
      "grad_norm": 0.26980283856391907,
      "learning_rate": 0.0003,
      "loss": 5.1331,
      "step": 4785
    },
    {
      "epoch": 0.09808940675362972,
      "grad_norm": 0.3303547203540802,
      "learning_rate": 0.0003,
      "loss": 5.0662,
      "step": 4790
    },
    {
      "epoch": 0.09819179653103434,
      "grad_norm": 0.24784614145755768,
      "learning_rate": 0.0003,
      "loss": 5.1229,
      "step": 4795
    },
    {
      "epoch": 0.09829418630843896,
      "grad_norm": 0.25418931245803833,
      "learning_rate": 0.0003,
      "loss": 5.1011,
      "step": 4800
    },
    {
      "epoch": 0.0983965760858436,
      "grad_norm": 0.27420976758003235,
      "learning_rate": 0.0003,
      "loss": 5.1764,
      "step": 4805
    },
    {
      "epoch": 0.09849896586324822,
      "grad_norm": 0.2599352300167084,
      "learning_rate": 0.0003,
      "loss": 5.1369,
      "step": 4810
    },
    {
      "epoch": 0.09860135564065284,
      "grad_norm": 0.26636365056037903,
      "learning_rate": 0.0003,
      "loss": 5.1434,
      "step": 4815
    },
    {
      "epoch": 0.09870374541805746,
      "grad_norm": 0.2413569837808609,
      "learning_rate": 0.0003,
      "loss": 5.2023,
      "step": 4820
    },
    {
      "epoch": 0.09880613519546208,
      "grad_norm": 0.3400757312774658,
      "learning_rate": 0.0003,
      "loss": 5.1625,
      "step": 4825
    },
    {
      "epoch": 0.09890852497286672,
      "grad_norm": 0.26301613450050354,
      "learning_rate": 0.0003,
      "loss": 5.12,
      "step": 4830
    },
    {
      "epoch": 0.09901091475027134,
      "grad_norm": 0.2481679618358612,
      "learning_rate": 0.0003,
      "loss": 5.1281,
      "step": 4835
    },
    {
      "epoch": 0.09911330452767596,
      "grad_norm": 0.2625468075275421,
      "learning_rate": 0.0003,
      "loss": 5.1418,
      "step": 4840
    },
    {
      "epoch": 0.09921569430508058,
      "grad_norm": 0.2667504549026489,
      "learning_rate": 0.0003,
      "loss": 5.1343,
      "step": 4845
    },
    {
      "epoch": 0.0993180840824852,
      "grad_norm": 0.244068443775177,
      "learning_rate": 0.0003,
      "loss": 5.1611,
      "step": 4850
    },
    {
      "epoch": 0.09942047385988984,
      "grad_norm": 0.27124324440956116,
      "learning_rate": 0.0003,
      "loss": 5.1413,
      "step": 4855
    },
    {
      "epoch": 0.09952286363729446,
      "grad_norm": 0.2569420635700226,
      "learning_rate": 0.0003,
      "loss": 5.1242,
      "step": 4860
    },
    {
      "epoch": 0.09962525341469908,
      "grad_norm": 0.2730373442173004,
      "learning_rate": 0.0003,
      "loss": 5.1154,
      "step": 4865
    },
    {
      "epoch": 0.0997276431921037,
      "grad_norm": 0.2586059868335724,
      "learning_rate": 0.0003,
      "loss": 5.1066,
      "step": 4870
    },
    {
      "epoch": 0.09983003296950832,
      "grad_norm": 0.24679683148860931,
      "learning_rate": 0.0003,
      "loss": 5.1419,
      "step": 4875
    },
    {
      "epoch": 0.09993242274691296,
      "grad_norm": 0.24525699019432068,
      "learning_rate": 0.0003,
      "loss": 5.1382,
      "step": 4880
    },
    {
      "epoch": 0.10003481252431758,
      "grad_norm": 0.2788681089878082,
      "learning_rate": 0.0003,
      "loss": 5.1226,
      "step": 4885
    },
    {
      "epoch": 0.1001372023017222,
      "grad_norm": 0.2719569504261017,
      "learning_rate": 0.0003,
      "loss": 5.1617,
      "step": 4890
    },
    {
      "epoch": 0.10023959207912682,
      "grad_norm": 0.2520180642604828,
      "learning_rate": 0.0003,
      "loss": 5.201,
      "step": 4895
    },
    {
      "epoch": 0.10034198185653144,
      "grad_norm": 0.2797073423862457,
      "learning_rate": 0.0003,
      "loss": 5.1073,
      "step": 4900
    },
    {
      "epoch": 0.10044437163393606,
      "grad_norm": 0.23679588735103607,
      "learning_rate": 0.0003,
      "loss": 5.0595,
      "step": 4905
    },
    {
      "epoch": 0.1005467614113407,
      "grad_norm": 0.2508576512336731,
      "learning_rate": 0.0003,
      "loss": 5.1936,
      "step": 4910
    },
    {
      "epoch": 0.10064915118874532,
      "grad_norm": 0.24073272943496704,
      "learning_rate": 0.0003,
      "loss": 5.1966,
      "step": 4915
    },
    {
      "epoch": 0.10075154096614994,
      "grad_norm": 0.30414456129074097,
      "learning_rate": 0.0003,
      "loss": 5.1499,
      "step": 4920
    },
    {
      "epoch": 0.10085393074355456,
      "grad_norm": 0.26562175154685974,
      "learning_rate": 0.0003,
      "loss": 5.1271,
      "step": 4925
    },
    {
      "epoch": 0.10095632052095918,
      "grad_norm": 0.28138473629951477,
      "learning_rate": 0.0003,
      "loss": 5.1332,
      "step": 4930
    },
    {
      "epoch": 0.10105871029836382,
      "grad_norm": 0.2860601246356964,
      "learning_rate": 0.0003,
      "loss": 5.1772,
      "step": 4935
    },
    {
      "epoch": 0.10116110007576844,
      "grad_norm": 0.35466572642326355,
      "learning_rate": 0.0003,
      "loss": 5.1402,
      "step": 4940
    },
    {
      "epoch": 0.10126348985317306,
      "grad_norm": 0.3055334687232971,
      "learning_rate": 0.0003,
      "loss": 5.109,
      "step": 4945
    },
    {
      "epoch": 0.10136587963057768,
      "grad_norm": 0.2625313401222229,
      "learning_rate": 0.0003,
      "loss": 5.1465,
      "step": 4950
    },
    {
      "epoch": 0.1014682694079823,
      "grad_norm": 0.3070855438709259,
      "learning_rate": 0.0003,
      "loss": 5.1775,
      "step": 4955
    },
    {
      "epoch": 0.10157065918538694,
      "grad_norm": 0.252029687166214,
      "learning_rate": 0.0003,
      "loss": 5.1638,
      "step": 4960
    },
    {
      "epoch": 0.10167304896279156,
      "grad_norm": 0.2720160186290741,
      "learning_rate": 0.0003,
      "loss": 5.1241,
      "step": 4965
    },
    {
      "epoch": 0.10177543874019618,
      "grad_norm": 0.29157376289367676,
      "learning_rate": 0.0003,
      "loss": 5.1356,
      "step": 4970
    },
    {
      "epoch": 0.1018778285176008,
      "grad_norm": 0.2681252062320709,
      "learning_rate": 0.0003,
      "loss": 5.1856,
      "step": 4975
    },
    {
      "epoch": 0.10198021829500542,
      "grad_norm": 0.25204727053642273,
      "learning_rate": 0.0003,
      "loss": 5.094,
      "step": 4980
    },
    {
      "epoch": 0.10208260807241006,
      "grad_norm": 0.2691825032234192,
      "learning_rate": 0.0003,
      "loss": 5.105,
      "step": 4985
    },
    {
      "epoch": 0.10218499784981468,
      "grad_norm": 0.27777227759361267,
      "learning_rate": 0.0003,
      "loss": 5.1761,
      "step": 4990
    },
    {
      "epoch": 0.1022873876272193,
      "grad_norm": 0.28451526165008545,
      "learning_rate": 0.0003,
      "loss": 5.1084,
      "step": 4995
    },
    {
      "epoch": 0.10238977740462392,
      "grad_norm": 0.25006020069122314,
      "learning_rate": 0.0003,
      "loss": 5.1251,
      "step": 5000
    },
    {
      "epoch": 0.10238977740462392,
      "eval_loss": 5.143056869506836,
      "eval_runtime": 17.2601,
      "eval_samples_per_second": 33.719,
      "eval_steps_per_second": 4.229,
      "step": 5000
    },
    {
      "epoch": 0.10249216718202854,
      "grad_norm": 0.23722681403160095,
      "learning_rate": 0.0003,
      "loss": 5.1111,
      "step": 5005
    },
    {
      "epoch": 0.10259455695943318,
      "grad_norm": 0.25522348284721375,
      "learning_rate": 0.0003,
      "loss": 5.0909,
      "step": 5010
    },
    {
      "epoch": 0.1026969467368378,
      "grad_norm": 0.27011430263519287,
      "learning_rate": 0.0003,
      "loss": 5.1469,
      "step": 5015
    },
    {
      "epoch": 0.10279933651424242,
      "grad_norm": 0.28230318427085876,
      "learning_rate": 0.0003,
      "loss": 5.1279,
      "step": 5020
    },
    {
      "epoch": 0.10290172629164704,
      "grad_norm": 0.2487412542104721,
      "learning_rate": 0.0003,
      "loss": 5.1361,
      "step": 5025
    },
    {
      "epoch": 0.10300411606905166,
      "grad_norm": 0.24795986711978912,
      "learning_rate": 0.0003,
      "loss": 5.1209,
      "step": 5030
    },
    {
      "epoch": 0.1031065058464563,
      "grad_norm": 0.2390604168176651,
      "learning_rate": 0.0003,
      "loss": 5.1062,
      "step": 5035
    },
    {
      "epoch": 0.10320889562386092,
      "grad_norm": 0.28911134600639343,
      "learning_rate": 0.0003,
      "loss": 5.112,
      "step": 5040
    },
    {
      "epoch": 0.10331128540126554,
      "grad_norm": 0.27236247062683105,
      "learning_rate": 0.0003,
      "loss": 5.0818,
      "step": 5045
    },
    {
      "epoch": 0.10341367517867016,
      "grad_norm": 0.28922855854034424,
      "learning_rate": 0.0003,
      "loss": 5.1046,
      "step": 5050
    },
    {
      "epoch": 0.10351606495607478,
      "grad_norm": 0.31324613094329834,
      "learning_rate": 0.0003,
      "loss": 5.1404,
      "step": 5055
    },
    {
      "epoch": 0.10361845473347941,
      "grad_norm": 0.26908165216445923,
      "learning_rate": 0.0003,
      "loss": 5.0967,
      "step": 5060
    },
    {
      "epoch": 0.10372084451088404,
      "grad_norm": 0.2754386365413666,
      "learning_rate": 0.0003,
      "loss": 5.1459,
      "step": 5065
    },
    {
      "epoch": 0.10382323428828866,
      "grad_norm": 0.29012352228164673,
      "learning_rate": 0.0003,
      "loss": 5.1556,
      "step": 5070
    },
    {
      "epoch": 0.10392562406569328,
      "grad_norm": 0.31764522194862366,
      "learning_rate": 0.0003,
      "loss": 5.1194,
      "step": 5075
    },
    {
      "epoch": 0.1040280138430979,
      "grad_norm": 0.27008572220802307,
      "learning_rate": 0.0003,
      "loss": 5.1184,
      "step": 5080
    },
    {
      "epoch": 0.10413040362050253,
      "grad_norm": 0.25800609588623047,
      "learning_rate": 0.0003,
      "loss": 5.0866,
      "step": 5085
    },
    {
      "epoch": 0.10423279339790716,
      "grad_norm": 0.6253742575645447,
      "learning_rate": 0.0003,
      "loss": 5.1341,
      "step": 5090
    },
    {
      "epoch": 0.10433518317531178,
      "grad_norm": 0.26275634765625,
      "learning_rate": 0.0003,
      "loss": 5.1114,
      "step": 5095
    },
    {
      "epoch": 0.1044375729527164,
      "grad_norm": 0.25894594192504883,
      "learning_rate": 0.0003,
      "loss": 5.0714,
      "step": 5100
    },
    {
      "epoch": 0.10453996273012102,
      "grad_norm": 0.2545381784439087,
      "learning_rate": 0.0003,
      "loss": 5.1005,
      "step": 5105
    },
    {
      "epoch": 0.10464235250752565,
      "grad_norm": 0.2790254056453705,
      "learning_rate": 0.0003,
      "loss": 5.1025,
      "step": 5110
    },
    {
      "epoch": 0.10474474228493028,
      "grad_norm": 0.25671303272247314,
      "learning_rate": 0.0003,
      "loss": 5.1659,
      "step": 5115
    },
    {
      "epoch": 0.1048471320623349,
      "grad_norm": 0.2371091991662979,
      "learning_rate": 0.0003,
      "loss": 5.1035,
      "step": 5120
    },
    {
      "epoch": 0.10494952183973952,
      "grad_norm": 0.2600834369659424,
      "learning_rate": 0.0003,
      "loss": 5.0701,
      "step": 5125
    },
    {
      "epoch": 0.10505191161714414,
      "grad_norm": 0.3032551109790802,
      "learning_rate": 0.0003,
      "loss": 5.094,
      "step": 5130
    },
    {
      "epoch": 0.10515430139454877,
      "grad_norm": 0.24528567492961884,
      "learning_rate": 0.0003,
      "loss": 5.1415,
      "step": 5135
    },
    {
      "epoch": 0.1052566911719534,
      "grad_norm": 0.291016161441803,
      "learning_rate": 0.0003,
      "loss": 5.0553,
      "step": 5140
    },
    {
      "epoch": 0.10535908094935802,
      "grad_norm": 0.23490361869335175,
      "learning_rate": 0.0003,
      "loss": 5.1263,
      "step": 5145
    },
    {
      "epoch": 0.10546147072676264,
      "grad_norm": 0.42114129662513733,
      "learning_rate": 0.0003,
      "loss": 5.1286,
      "step": 5150
    },
    {
      "epoch": 0.10556386050416726,
      "grad_norm": 0.25405389070510864,
      "learning_rate": 0.0003,
      "loss": 5.1484,
      "step": 5155
    },
    {
      "epoch": 0.1056662502815719,
      "grad_norm": 0.29397106170654297,
      "learning_rate": 0.0003,
      "loss": 5.1388,
      "step": 5160
    },
    {
      "epoch": 0.10576864005897652,
      "grad_norm": 0.2647407054901123,
      "learning_rate": 0.0003,
      "loss": 5.1146,
      "step": 5165
    },
    {
      "epoch": 0.10587102983638114,
      "grad_norm": 0.2777150273323059,
      "learning_rate": 0.0003,
      "loss": 5.1521,
      "step": 5170
    },
    {
      "epoch": 0.10597341961378576,
      "grad_norm": 0.25496071577072144,
      "learning_rate": 0.0003,
      "loss": 5.1233,
      "step": 5175
    },
    {
      "epoch": 0.10607580939119038,
      "grad_norm": 0.2745848298072815,
      "learning_rate": 0.0003,
      "loss": 5.1353,
      "step": 5180
    },
    {
      "epoch": 0.10617819916859501,
      "grad_norm": 0.2660798132419586,
      "learning_rate": 0.0003,
      "loss": 5.133,
      "step": 5185
    },
    {
      "epoch": 0.10628058894599964,
      "grad_norm": 0.2540626525878906,
      "learning_rate": 0.0003,
      "loss": 5.0843,
      "step": 5190
    },
    {
      "epoch": 0.10638297872340426,
      "grad_norm": 0.24836969375610352,
      "learning_rate": 0.0003,
      "loss": 5.1156,
      "step": 5195
    },
    {
      "epoch": 0.10648536850080888,
      "grad_norm": 0.28216132521629333,
      "learning_rate": 0.0003,
      "loss": 5.1307,
      "step": 5200
    },
    {
      "epoch": 0.1065877582782135,
      "grad_norm": 0.23361997306346893,
      "learning_rate": 0.0003,
      "loss": 5.128,
      "step": 5205
    },
    {
      "epoch": 0.10669014805561813,
      "grad_norm": 0.2421647608280182,
      "learning_rate": 0.0003,
      "loss": 5.1002,
      "step": 5210
    },
    {
      "epoch": 0.10679253783302275,
      "grad_norm": 0.24135902523994446,
      "learning_rate": 0.0003,
      "loss": 5.0917,
      "step": 5215
    },
    {
      "epoch": 0.10689492761042738,
      "grad_norm": 0.26153501868247986,
      "learning_rate": 0.0003,
      "loss": 5.0929,
      "step": 5220
    },
    {
      "epoch": 0.106997317387832,
      "grad_norm": 0.2569580078125,
      "learning_rate": 0.0003,
      "loss": 5.1,
      "step": 5225
    },
    {
      "epoch": 0.10709970716523662,
      "grad_norm": 0.34273838996887207,
      "learning_rate": 0.0003,
      "loss": 5.1024,
      "step": 5230
    },
    {
      "epoch": 0.10720209694264125,
      "grad_norm": 0.25896260142326355,
      "learning_rate": 0.0003,
      "loss": 5.116,
      "step": 5235
    },
    {
      "epoch": 0.10730448672004587,
      "grad_norm": 0.25686803460121155,
      "learning_rate": 0.0003,
      "loss": 5.117,
      "step": 5240
    },
    {
      "epoch": 0.1074068764974505,
      "grad_norm": 0.2619083821773529,
      "learning_rate": 0.0003,
      "loss": 5.0904,
      "step": 5245
    },
    {
      "epoch": 0.10750926627485512,
      "grad_norm": 0.26761579513549805,
      "learning_rate": 0.0003,
      "loss": 5.1233,
      "step": 5250
    },
    {
      "epoch": 0.10761165605225974,
      "grad_norm": 0.2648669481277466,
      "learning_rate": 0.0003,
      "loss": 5.1176,
      "step": 5255
    },
    {
      "epoch": 0.10771404582966437,
      "grad_norm": 0.2606008052825928,
      "learning_rate": 0.0003,
      "loss": 5.1285,
      "step": 5260
    },
    {
      "epoch": 0.107816435607069,
      "grad_norm": 0.2694513499736786,
      "learning_rate": 0.0003,
      "loss": 5.0945,
      "step": 5265
    },
    {
      "epoch": 0.10791882538447362,
      "grad_norm": 0.267244815826416,
      "learning_rate": 0.0003,
      "loss": 5.1243,
      "step": 5270
    },
    {
      "epoch": 0.10802121516187824,
      "grad_norm": 0.24808046221733093,
      "learning_rate": 0.0003,
      "loss": 5.0959,
      "step": 5275
    },
    {
      "epoch": 0.10812360493928286,
      "grad_norm": 0.3454941213130951,
      "learning_rate": 0.0003,
      "loss": 5.083,
      "step": 5280
    },
    {
      "epoch": 0.10822599471668748,
      "grad_norm": 0.2596804201602936,
      "learning_rate": 0.0003,
      "loss": 5.1347,
      "step": 5285
    },
    {
      "epoch": 0.10832838449409211,
      "grad_norm": 0.2772788107395172,
      "learning_rate": 0.0003,
      "loss": 5.1316,
      "step": 5290
    },
    {
      "epoch": 0.10843077427149674,
      "grad_norm": 0.31930217146873474,
      "learning_rate": 0.0003,
      "loss": 5.0711,
      "step": 5295
    },
    {
      "epoch": 0.10853316404890136,
      "grad_norm": 0.2737719416618347,
      "learning_rate": 0.0003,
      "loss": 5.1095,
      "step": 5300
    },
    {
      "epoch": 0.10863555382630598,
      "grad_norm": 0.2516084313392639,
      "learning_rate": 0.0003,
      "loss": 5.0975,
      "step": 5305
    },
    {
      "epoch": 0.1087379436037106,
      "grad_norm": 0.24778488278388977,
      "learning_rate": 0.0003,
      "loss": 5.0538,
      "step": 5310
    },
    {
      "epoch": 0.10884033338111523,
      "grad_norm": 0.33298465609550476,
      "learning_rate": 0.0003,
      "loss": 5.0902,
      "step": 5315
    },
    {
      "epoch": 0.10894272315851986,
      "grad_norm": 0.30207422375679016,
      "learning_rate": 0.0003,
      "loss": 5.0882,
      "step": 5320
    },
    {
      "epoch": 0.10904511293592448,
      "grad_norm": 0.26138100028038025,
      "learning_rate": 0.0003,
      "loss": 5.0834,
      "step": 5325
    },
    {
      "epoch": 0.1091475027133291,
      "grad_norm": 0.24217452108860016,
      "learning_rate": 0.0003,
      "loss": 5.1091,
      "step": 5330
    },
    {
      "epoch": 0.10924989249073372,
      "grad_norm": 0.2715062201023102,
      "learning_rate": 0.0003,
      "loss": 5.0695,
      "step": 5335
    },
    {
      "epoch": 0.10935228226813835,
      "grad_norm": 0.24975372850894928,
      "learning_rate": 0.0003,
      "loss": 5.1061,
      "step": 5340
    },
    {
      "epoch": 0.10945467204554298,
      "grad_norm": 0.24967104196548462,
      "learning_rate": 0.0003,
      "loss": 5.1191,
      "step": 5345
    },
    {
      "epoch": 0.1095570618229476,
      "grad_norm": 0.23948703706264496,
      "learning_rate": 0.0003,
      "loss": 5.1065,
      "step": 5350
    },
    {
      "epoch": 0.10965945160035222,
      "grad_norm": 0.2402109056711197,
      "learning_rate": 0.0003,
      "loss": 5.0674,
      "step": 5355
    },
    {
      "epoch": 0.10976184137775684,
      "grad_norm": 0.252261221408844,
      "learning_rate": 0.0003,
      "loss": 5.0853,
      "step": 5360
    },
    {
      "epoch": 0.10986423115516147,
      "grad_norm": 0.2358742654323578,
      "learning_rate": 0.0003,
      "loss": 5.1215,
      "step": 5365
    },
    {
      "epoch": 0.1099666209325661,
      "grad_norm": 0.2711077332496643,
      "learning_rate": 0.0003,
      "loss": 5.1049,
      "step": 5370
    },
    {
      "epoch": 0.11006901070997072,
      "grad_norm": 0.2335502803325653,
      "learning_rate": 0.0003,
      "loss": 5.0961,
      "step": 5375
    },
    {
      "epoch": 0.11017140048737534,
      "grad_norm": 0.28193357586860657,
      "learning_rate": 0.0003,
      "loss": 4.9784,
      "step": 5380
    },
    {
      "epoch": 0.11027379026477996,
      "grad_norm": 0.27377572655677795,
      "learning_rate": 0.0003,
      "loss": 5.1219,
      "step": 5385
    },
    {
      "epoch": 0.1103761800421846,
      "grad_norm": 0.25776010751724243,
      "learning_rate": 0.0003,
      "loss": 5.1087,
      "step": 5390
    },
    {
      "epoch": 0.11047856981958921,
      "grad_norm": 0.23744498193264008,
      "learning_rate": 0.0003,
      "loss": 5.1089,
      "step": 5395
    },
    {
      "epoch": 0.11058095959699384,
      "grad_norm": 0.24810472130775452,
      "learning_rate": 0.0003,
      "loss": 5.1097,
      "step": 5400
    },
    {
      "epoch": 0.11068334937439846,
      "grad_norm": 0.2660318613052368,
      "learning_rate": 0.0003,
      "loss": 5.0666,
      "step": 5405
    },
    {
      "epoch": 0.11078573915180308,
      "grad_norm": 0.2684977650642395,
      "learning_rate": 0.0003,
      "loss": 5.1226,
      "step": 5410
    },
    {
      "epoch": 0.11088812892920771,
      "grad_norm": 0.23833255469799042,
      "learning_rate": 0.0003,
      "loss": 5.059,
      "step": 5415
    },
    {
      "epoch": 0.11099051870661233,
      "grad_norm": 0.22599267959594727,
      "learning_rate": 0.0003,
      "loss": 5.0964,
      "step": 5420
    },
    {
      "epoch": 0.11109290848401696,
      "grad_norm": 0.2854502499103546,
      "learning_rate": 0.0003,
      "loss": 5.0965,
      "step": 5425
    },
    {
      "epoch": 0.11119529826142158,
      "grad_norm": 0.25768956542015076,
      "learning_rate": 0.0003,
      "loss": 5.0795,
      "step": 5430
    },
    {
      "epoch": 0.1112976880388262,
      "grad_norm": 0.25362953543663025,
      "learning_rate": 0.0003,
      "loss": 5.1051,
      "step": 5435
    },
    {
      "epoch": 0.11140007781623083,
      "grad_norm": 0.25006183981895447,
      "learning_rate": 0.0003,
      "loss": 5.1367,
      "step": 5440
    },
    {
      "epoch": 0.11150246759363545,
      "grad_norm": 0.26781198382377625,
      "learning_rate": 0.0003,
      "loss": 5.1088,
      "step": 5445
    },
    {
      "epoch": 0.11160485737104008,
      "grad_norm": 0.24142861366271973,
      "learning_rate": 0.0003,
      "loss": 5.0972,
      "step": 5450
    },
    {
      "epoch": 0.1117072471484447,
      "grad_norm": 0.5353167057037354,
      "learning_rate": 0.0003,
      "loss": 5.0888,
      "step": 5455
    },
    {
      "epoch": 0.11180963692584932,
      "grad_norm": 0.2552117705345154,
      "learning_rate": 0.0003,
      "loss": 5.1282,
      "step": 5460
    },
    {
      "epoch": 0.11191202670325395,
      "grad_norm": 0.2614104747772217,
      "learning_rate": 0.0003,
      "loss": 5.0819,
      "step": 5465
    },
    {
      "epoch": 0.11201441648065857,
      "grad_norm": 0.2724562883377075,
      "learning_rate": 0.0003,
      "loss": 5.0925,
      "step": 5470
    },
    {
      "epoch": 0.1121168062580632,
      "grad_norm": 0.25823545455932617,
      "learning_rate": 0.0003,
      "loss": 5.0824,
      "step": 5475
    },
    {
      "epoch": 0.11221919603546782,
      "grad_norm": 0.2416844666004181,
      "learning_rate": 0.0003,
      "loss": 5.086,
      "step": 5480
    },
    {
      "epoch": 0.11232158581287244,
      "grad_norm": 0.24802665412425995,
      "learning_rate": 0.0003,
      "loss": 5.116,
      "step": 5485
    },
    {
      "epoch": 0.11242397559027707,
      "grad_norm": 0.26273369789123535,
      "learning_rate": 0.0003,
      "loss": 5.0873,
      "step": 5490
    },
    {
      "epoch": 0.1125263653676817,
      "grad_norm": 0.25046268105506897,
      "learning_rate": 0.0003,
      "loss": 5.0717,
      "step": 5495
    },
    {
      "epoch": 0.11262875514508632,
      "grad_norm": 0.2463729828596115,
      "learning_rate": 0.0003,
      "loss": 5.0657,
      "step": 5500
    },
    {
      "epoch": 0.11273114492249094,
      "grad_norm": 0.2476728856563568,
      "learning_rate": 0.0003,
      "loss": 5.1009,
      "step": 5505
    },
    {
      "epoch": 0.11283353469989556,
      "grad_norm": 0.23306916654109955,
      "learning_rate": 0.0003,
      "loss": 5.1019,
      "step": 5510
    },
    {
      "epoch": 0.11293592447730019,
      "grad_norm": 0.2420760691165924,
      "learning_rate": 0.0003,
      "loss": 5.1188,
      "step": 5515
    },
    {
      "epoch": 0.11303831425470481,
      "grad_norm": 0.2833799421787262,
      "learning_rate": 0.0003,
      "loss": 5.1152,
      "step": 5520
    },
    {
      "epoch": 0.11314070403210943,
      "grad_norm": 0.27480828762054443,
      "learning_rate": 0.0003,
      "loss": 5.0933,
      "step": 5525
    },
    {
      "epoch": 0.11324309380951406,
      "grad_norm": 0.32995694875717163,
      "learning_rate": 0.0003,
      "loss": 5.0861,
      "step": 5530
    },
    {
      "epoch": 0.11334548358691868,
      "grad_norm": 0.2790719270706177,
      "learning_rate": 0.0003,
      "loss": 5.0864,
      "step": 5535
    },
    {
      "epoch": 0.11344787336432331,
      "grad_norm": 0.260960191488266,
      "learning_rate": 0.0003,
      "loss": 5.0609,
      "step": 5540
    },
    {
      "epoch": 0.11355026314172793,
      "grad_norm": 0.24379654228687286,
      "learning_rate": 0.0003,
      "loss": 5.1043,
      "step": 5545
    },
    {
      "epoch": 0.11365265291913255,
      "grad_norm": 0.5207554697990417,
      "learning_rate": 0.0003,
      "loss": 5.1387,
      "step": 5550
    },
    {
      "epoch": 0.11375504269653718,
      "grad_norm": 0.24510209262371063,
      "learning_rate": 0.0003,
      "loss": 5.0969,
      "step": 5555
    },
    {
      "epoch": 0.1138574324739418,
      "grad_norm": 0.2376794070005417,
      "learning_rate": 0.0003,
      "loss": 5.1169,
      "step": 5560
    },
    {
      "epoch": 0.11395982225134643,
      "grad_norm": 0.27716517448425293,
      "learning_rate": 0.0003,
      "loss": 5.1212,
      "step": 5565
    },
    {
      "epoch": 0.11406221202875105,
      "grad_norm": 0.2802460193634033,
      "learning_rate": 0.0003,
      "loss": 5.0595,
      "step": 5570
    },
    {
      "epoch": 0.11416460180615567,
      "grad_norm": 0.2466181516647339,
      "learning_rate": 0.0003,
      "loss": 5.1075,
      "step": 5575
    },
    {
      "epoch": 0.1142669915835603,
      "grad_norm": 0.23184147477149963,
      "learning_rate": 0.0003,
      "loss": 5.0972,
      "step": 5580
    },
    {
      "epoch": 0.11436938136096492,
      "grad_norm": 0.25597503781318665,
      "learning_rate": 0.0003,
      "loss": 5.0806,
      "step": 5585
    },
    {
      "epoch": 0.11447177113836955,
      "grad_norm": 0.2580608129501343,
      "learning_rate": 0.0003,
      "loss": 5.0831,
      "step": 5590
    },
    {
      "epoch": 0.11457416091577417,
      "grad_norm": 0.26163652539253235,
      "learning_rate": 0.0003,
      "loss": 5.085,
      "step": 5595
    },
    {
      "epoch": 0.1146765506931788,
      "grad_norm": 0.25621700286865234,
      "learning_rate": 0.0003,
      "loss": 5.0736,
      "step": 5600
    },
    {
      "epoch": 0.11477894047058342,
      "grad_norm": 0.25294628739356995,
      "learning_rate": 0.0003,
      "loss": 5.0473,
      "step": 5605
    },
    {
      "epoch": 0.11488133024798804,
      "grad_norm": 0.23033612966537476,
      "learning_rate": 0.0003,
      "loss": 5.1719,
      "step": 5610
    },
    {
      "epoch": 0.11498372002539267,
      "grad_norm": 1.2823646068572998,
      "learning_rate": 0.0003,
      "loss": 5.1058,
      "step": 5615
    },
    {
      "epoch": 0.11508610980279729,
      "grad_norm": 0.2869229018688202,
      "learning_rate": 0.0003,
      "loss": 5.1134,
      "step": 5620
    },
    {
      "epoch": 0.11518849958020191,
      "grad_norm": 0.2845302224159241,
      "learning_rate": 0.0003,
      "loss": 5.099,
      "step": 5625
    },
    {
      "epoch": 0.11529088935760654,
      "grad_norm": 0.28526628017425537,
      "learning_rate": 0.0003,
      "loss": 5.0788,
      "step": 5630
    },
    {
      "epoch": 0.11539327913501116,
      "grad_norm": 0.2521724998950958,
      "learning_rate": 0.0003,
      "loss": 5.0675,
      "step": 5635
    },
    {
      "epoch": 0.11549566891241578,
      "grad_norm": 0.24957355856895447,
      "learning_rate": 0.0003,
      "loss": 5.1247,
      "step": 5640
    },
    {
      "epoch": 0.11559805868982041,
      "grad_norm": 0.3723267912864685,
      "learning_rate": 0.0003,
      "loss": 5.0761,
      "step": 5645
    },
    {
      "epoch": 0.11570044846722503,
      "grad_norm": 0.2532084286212921,
      "learning_rate": 0.0003,
      "loss": 5.0858,
      "step": 5650
    },
    {
      "epoch": 0.11580283824462966,
      "grad_norm": 0.23800311982631683,
      "learning_rate": 0.0003,
      "loss": 5.0817,
      "step": 5655
    },
    {
      "epoch": 0.11590522802203428,
      "grad_norm": 0.27352309226989746,
      "learning_rate": 0.0003,
      "loss": 5.045,
      "step": 5660
    },
    {
      "epoch": 0.1160076177994389,
      "grad_norm": 0.2616291046142578,
      "learning_rate": 0.0003,
      "loss": 5.0713,
      "step": 5665
    },
    {
      "epoch": 0.11611000757684353,
      "grad_norm": 0.24883075058460236,
      "learning_rate": 0.0003,
      "loss": 5.0622,
      "step": 5670
    },
    {
      "epoch": 0.11621239735424815,
      "grad_norm": 0.30054086446762085,
      "learning_rate": 0.0003,
      "loss": 5.0688,
      "step": 5675
    },
    {
      "epoch": 0.11631478713165277,
      "grad_norm": 0.2814779281616211,
      "learning_rate": 0.0003,
      "loss": 5.0949,
      "step": 5680
    },
    {
      "epoch": 0.1164171769090574,
      "grad_norm": 0.23773646354675293,
      "learning_rate": 0.0003,
      "loss": 5.0514,
      "step": 5685
    },
    {
      "epoch": 0.11651956668646202,
      "grad_norm": 0.2319362759590149,
      "learning_rate": 0.0003,
      "loss": 5.0607,
      "step": 5690
    },
    {
      "epoch": 0.11662195646386665,
      "grad_norm": 0.28966546058654785,
      "learning_rate": 0.0003,
      "loss": 5.0548,
      "step": 5695
    },
    {
      "epoch": 0.11672434624127127,
      "grad_norm": 0.26125651597976685,
      "learning_rate": 0.0003,
      "loss": 5.072,
      "step": 5700
    },
    {
      "epoch": 0.1168267360186759,
      "grad_norm": 0.2489207535982132,
      "learning_rate": 0.0003,
      "loss": 5.091,
      "step": 5705
    },
    {
      "epoch": 0.11692912579608052,
      "grad_norm": 0.2681567966938019,
      "learning_rate": 0.0003,
      "loss": 5.0954,
      "step": 5710
    },
    {
      "epoch": 0.11703151557348514,
      "grad_norm": 0.3284648060798645,
      "learning_rate": 0.0003,
      "loss": 5.0812,
      "step": 5715
    },
    {
      "epoch": 0.11713390535088977,
      "grad_norm": 0.803409993648529,
      "learning_rate": 0.0003,
      "loss": 5.0538,
      "step": 5720
    },
    {
      "epoch": 0.1172362951282944,
      "grad_norm": 0.26291313767433167,
      "learning_rate": 0.0003,
      "loss": 5.0987,
      "step": 5725
    },
    {
      "epoch": 0.11733868490569901,
      "grad_norm": 0.2675195634365082,
      "learning_rate": 0.0003,
      "loss": 5.0876,
      "step": 5730
    },
    {
      "epoch": 0.11744107468310364,
      "grad_norm": 0.267531156539917,
      "learning_rate": 0.0003,
      "loss": 5.0532,
      "step": 5735
    },
    {
      "epoch": 0.11754346446050826,
      "grad_norm": 0.23051545023918152,
      "learning_rate": 0.0003,
      "loss": 5.0877,
      "step": 5740
    },
    {
      "epoch": 0.11764585423791289,
      "grad_norm": 0.24739649891853333,
      "learning_rate": 0.0003,
      "loss": 5.0688,
      "step": 5745
    },
    {
      "epoch": 0.11774824401531751,
      "grad_norm": 0.23418428003787994,
      "learning_rate": 0.0003,
      "loss": 5.1165,
      "step": 5750
    },
    {
      "epoch": 0.11785063379272213,
      "grad_norm": 0.24748854339122772,
      "learning_rate": 0.0003,
      "loss": 5.0548,
      "step": 5755
    },
    {
      "epoch": 0.11795302357012676,
      "grad_norm": 0.24577189981937408,
      "learning_rate": 0.0003,
      "loss": 5.0686,
      "step": 5760
    },
    {
      "epoch": 0.11805541334753138,
      "grad_norm": 0.2512575089931488,
      "learning_rate": 0.0003,
      "loss": 5.0668,
      "step": 5765
    },
    {
      "epoch": 0.11815780312493601,
      "grad_norm": 0.2299077957868576,
      "learning_rate": 0.0003,
      "loss": 5.1238,
      "step": 5770
    },
    {
      "epoch": 0.11826019290234063,
      "grad_norm": 0.24888236820697784,
      "learning_rate": 0.0003,
      "loss": 5.0951,
      "step": 5775
    },
    {
      "epoch": 0.11836258267974525,
      "grad_norm": 0.24673014879226685,
      "learning_rate": 0.0003,
      "loss": 5.0171,
      "step": 5780
    },
    {
      "epoch": 0.11846497245714988,
      "grad_norm": 0.24857375025749207,
      "learning_rate": 0.0003,
      "loss": 5.0761,
      "step": 5785
    },
    {
      "epoch": 0.1185673622345545,
      "grad_norm": 0.28456902503967285,
      "learning_rate": 0.0003,
      "loss": 5.0392,
      "step": 5790
    },
    {
      "epoch": 0.11866975201195913,
      "grad_norm": 0.24286267161369324,
      "learning_rate": 0.0003,
      "loss": 5.0783,
      "step": 5795
    },
    {
      "epoch": 0.11877214178936375,
      "grad_norm": 0.30398255586624146,
      "learning_rate": 0.0003,
      "loss": 5.0754,
      "step": 5800
    },
    {
      "epoch": 0.11887453156676837,
      "grad_norm": 0.2428668886423111,
      "learning_rate": 0.0003,
      "loss": 5.073,
      "step": 5805
    },
    {
      "epoch": 0.118976921344173,
      "grad_norm": 0.2462380826473236,
      "learning_rate": 0.0003,
      "loss": 5.0761,
      "step": 5810
    },
    {
      "epoch": 0.11907931112157762,
      "grad_norm": 0.2458244115114212,
      "learning_rate": 0.0003,
      "loss": 5.094,
      "step": 5815
    },
    {
      "epoch": 0.11918170089898225,
      "grad_norm": 0.23394592106342316,
      "learning_rate": 0.0003,
      "loss": 5.0877,
      "step": 5820
    },
    {
      "epoch": 0.11928409067638687,
      "grad_norm": 2.4448344707489014,
      "learning_rate": 0.0003,
      "loss": 5.1191,
      "step": 5825
    },
    {
      "epoch": 0.1193864804537915,
      "grad_norm": 0.285965234041214,
      "learning_rate": 0.0003,
      "loss": 5.0547,
      "step": 5830
    },
    {
      "epoch": 0.11948887023119611,
      "grad_norm": 0.3072868585586548,
      "learning_rate": 0.0003,
      "loss": 5.0573,
      "step": 5835
    },
    {
      "epoch": 0.11959126000860074,
      "grad_norm": 0.23148755729198456,
      "learning_rate": 0.0003,
      "loss": 5.1279,
      "step": 5840
    },
    {
      "epoch": 0.11969364978600537,
      "grad_norm": 0.2666797339916229,
      "learning_rate": 0.0003,
      "loss": 5.0577,
      "step": 5845
    },
    {
      "epoch": 0.11979603956340999,
      "grad_norm": 0.2828826308250427,
      "learning_rate": 0.0003,
      "loss": 5.0758,
      "step": 5850
    },
    {
      "epoch": 0.11989842934081461,
      "grad_norm": 0.26379990577697754,
      "learning_rate": 0.0003,
      "loss": 5.0427,
      "step": 5855
    },
    {
      "epoch": 0.12000081911821923,
      "grad_norm": 0.2764631509780884,
      "learning_rate": 0.0003,
      "loss": 5.0696,
      "step": 5860
    },
    {
      "epoch": 0.12010320889562386,
      "grad_norm": 0.358840674161911,
      "learning_rate": 0.0003,
      "loss": 5.0659,
      "step": 5865
    },
    {
      "epoch": 0.12020559867302849,
      "grad_norm": 0.2416507452726364,
      "learning_rate": 0.0003,
      "loss": 5.1139,
      "step": 5870
    },
    {
      "epoch": 0.12030798845043311,
      "grad_norm": 0.268307089805603,
      "learning_rate": 0.0003,
      "loss": 5.083,
      "step": 5875
    },
    {
      "epoch": 0.12041037822783773,
      "grad_norm": 0.2705295979976654,
      "learning_rate": 0.0003,
      "loss": 5.0614,
      "step": 5880
    },
    {
      "epoch": 0.12051276800524235,
      "grad_norm": 0.2383246272802353,
      "learning_rate": 0.0003,
      "loss": 5.0237,
      "step": 5885
    },
    {
      "epoch": 0.12061515778264698,
      "grad_norm": 0.24550370872020721,
      "learning_rate": 0.0003,
      "loss": 5.043,
      "step": 5890
    },
    {
      "epoch": 0.12071754756005161,
      "grad_norm": 0.30784371495246887,
      "learning_rate": 0.0003,
      "loss": 5.0565,
      "step": 5895
    },
    {
      "epoch": 0.12081993733745623,
      "grad_norm": 0.3065190315246582,
      "learning_rate": 0.0003,
      "loss": 5.0672,
      "step": 5900
    },
    {
      "epoch": 0.12092232711486085,
      "grad_norm": 0.2554434537887573,
      "learning_rate": 0.0003,
      "loss": 5.0605,
      "step": 5905
    },
    {
      "epoch": 0.12102471689226547,
      "grad_norm": 0.3871344327926636,
      "learning_rate": 0.0003,
      "loss": 5.0641,
      "step": 5910
    },
    {
      "epoch": 0.1211271066696701,
      "grad_norm": 0.2830437123775482,
      "learning_rate": 0.0003,
      "loss": 5.0544,
      "step": 5915
    },
    {
      "epoch": 0.12122949644707473,
      "grad_norm": 0.24620437622070312,
      "learning_rate": 0.0003,
      "loss": 5.0572,
      "step": 5920
    },
    {
      "epoch": 0.12133188622447935,
      "grad_norm": 0.225809246301651,
      "learning_rate": 0.0003,
      "loss": 5.033,
      "step": 5925
    },
    {
      "epoch": 0.12143427600188397,
      "grad_norm": 0.4193023145198822,
      "learning_rate": 0.0003,
      "loss": 5.0471,
      "step": 5930
    },
    {
      "epoch": 0.1215366657792886,
      "grad_norm": 0.2928660809993744,
      "learning_rate": 0.0003,
      "loss": 5.0839,
      "step": 5935
    },
    {
      "epoch": 0.12163905555669322,
      "grad_norm": 0.2889542877674103,
      "learning_rate": 0.0003,
      "loss": 5.0712,
      "step": 5940
    },
    {
      "epoch": 0.12174144533409785,
      "grad_norm": 0.25429001450538635,
      "learning_rate": 0.0003,
      "loss": 5.0725,
      "step": 5945
    },
    {
      "epoch": 0.12184383511150247,
      "grad_norm": 0.2578778564929962,
      "learning_rate": 0.0003,
      "loss": 5.087,
      "step": 5950
    },
    {
      "epoch": 0.12194622488890709,
      "grad_norm": 0.2804707884788513,
      "learning_rate": 0.0003,
      "loss": 5.0361,
      "step": 5955
    },
    {
      "epoch": 0.12204861466631171,
      "grad_norm": 0.2534593939781189,
      "learning_rate": 0.0003,
      "loss": 5.0565,
      "step": 5960
    },
    {
      "epoch": 0.12215100444371634,
      "grad_norm": 0.26346203684806824,
      "learning_rate": 0.0003,
      "loss": 5.0886,
      "step": 5965
    },
    {
      "epoch": 0.12225339422112097,
      "grad_norm": 0.2457103580236435,
      "learning_rate": 0.0003,
      "loss": 5.0454,
      "step": 5970
    },
    {
      "epoch": 0.12235578399852559,
      "grad_norm": 0.28149500489234924,
      "learning_rate": 0.0003,
      "loss": 5.083,
      "step": 5975
    },
    {
      "epoch": 0.12245817377593021,
      "grad_norm": 0.24643804132938385,
      "learning_rate": 0.0003,
      "loss": 5.0571,
      "step": 5980
    },
    {
      "epoch": 0.12256056355333483,
      "grad_norm": 0.24789750576019287,
      "learning_rate": 0.0003,
      "loss": 5.0151,
      "step": 5985
    },
    {
      "epoch": 0.12266295333073945,
      "grad_norm": 0.24370989203453064,
      "learning_rate": 0.0003,
      "loss": 5.0348,
      "step": 5990
    },
    {
      "epoch": 0.12276534310814408,
      "grad_norm": 0.26576387882232666,
      "learning_rate": 0.0003,
      "loss": 5.0047,
      "step": 5995
    },
    {
      "epoch": 0.12286773288554871,
      "grad_norm": 0.28600040078163147,
      "learning_rate": 0.0003,
      "loss": 5.0614,
      "step": 6000
    },
    {
      "epoch": 0.12286773288554871,
      "eval_loss": 5.087298393249512,
      "eval_runtime": 26.3694,
      "eval_samples_per_second": 22.071,
      "eval_steps_per_second": 2.768,
      "step": 6000
    },
    {
      "epoch": 0.12297012266295333,
      "grad_norm": 0.2305298149585724,
      "learning_rate": 0.0003,
      "loss": 5.0592,
      "step": 6005
    },
    {
      "epoch": 0.12307251244035795,
      "grad_norm": 0.25346919894218445,
      "learning_rate": 0.0003,
      "loss": 5.0096,
      "step": 6010
    },
    {
      "epoch": 0.12317490221776257,
      "grad_norm": 0.26450151205062866,
      "learning_rate": 0.0003,
      "loss": 5.0459,
      "step": 6015
    },
    {
      "epoch": 0.1232772919951672,
      "grad_norm": 0.24275422096252441,
      "learning_rate": 0.0003,
      "loss": 5.0712,
      "step": 6020
    },
    {
      "epoch": 0.12337968177257183,
      "grad_norm": 0.25413811206817627,
      "learning_rate": 0.0003,
      "loss": 5.0723,
      "step": 6025
    },
    {
      "epoch": 0.12348207154997645,
      "grad_norm": 0.25593388080596924,
      "learning_rate": 0.0003,
      "loss": 5.0804,
      "step": 6030
    },
    {
      "epoch": 0.12358446132738107,
      "grad_norm": 0.2864149510860443,
      "learning_rate": 0.0003,
      "loss": 5.0175,
      "step": 6035
    },
    {
      "epoch": 0.1236868511047857,
      "grad_norm": 0.2595575451850891,
      "learning_rate": 0.0003,
      "loss": 5.0374,
      "step": 6040
    },
    {
      "epoch": 0.12378924088219032,
      "grad_norm": 0.23770998418331146,
      "learning_rate": 0.0003,
      "loss": 5.0571,
      "step": 6045
    },
    {
      "epoch": 0.12389163065959495,
      "grad_norm": 0.2449321448802948,
      "learning_rate": 0.0003,
      "loss": 5.0374,
      "step": 6050
    },
    {
      "epoch": 0.12399402043699957,
      "grad_norm": 0.26707953214645386,
      "learning_rate": 0.0003,
      "loss": 5.0781,
      "step": 6055
    },
    {
      "epoch": 0.1240964102144042,
      "grad_norm": 0.22687005996704102,
      "learning_rate": 0.0003,
      "loss": 5.0732,
      "step": 6060
    },
    {
      "epoch": 0.12419879999180881,
      "grad_norm": 0.22488383948802948,
      "learning_rate": 0.0003,
      "loss": 5.0528,
      "step": 6065
    },
    {
      "epoch": 0.12430118976921344,
      "grad_norm": 0.2513805031776428,
      "learning_rate": 0.0003,
      "loss": 5.0521,
      "step": 6070
    },
    {
      "epoch": 0.12440357954661807,
      "grad_norm": 0.23436325788497925,
      "learning_rate": 0.0003,
      "loss": 5.048,
      "step": 6075
    },
    {
      "epoch": 0.12450596932402269,
      "grad_norm": 0.24925151467323303,
      "learning_rate": 0.0003,
      "loss": 5.0982,
      "step": 6080
    },
    {
      "epoch": 0.12460835910142731,
      "grad_norm": 0.28738659620285034,
      "learning_rate": 0.0003,
      "loss": 5.043,
      "step": 6085
    },
    {
      "epoch": 0.12471074887883193,
      "grad_norm": 0.2321338951587677,
      "learning_rate": 0.0003,
      "loss": 5.0053,
      "step": 6090
    },
    {
      "epoch": 0.12481313865623656,
      "grad_norm": 0.24323894083499908,
      "learning_rate": 0.0003,
      "loss": 5.021,
      "step": 6095
    },
    {
      "epoch": 0.12491552843364119,
      "grad_norm": 0.23402823507785797,
      "learning_rate": 0.0003,
      "loss": 5.0816,
      "step": 6100
    },
    {
      "epoch": 0.1250179182110458,
      "grad_norm": 0.26620617508888245,
      "learning_rate": 0.0003,
      "loss": 5.0764,
      "step": 6105
    },
    {
      "epoch": 0.12512030798845045,
      "grad_norm": 0.4839159846305847,
      "learning_rate": 0.0003,
      "loss": 5.0616,
      "step": 6110
    },
    {
      "epoch": 0.12522269776585507,
      "grad_norm": 0.3115023970603943,
      "learning_rate": 0.0003,
      "loss": 5.0959,
      "step": 6115
    },
    {
      "epoch": 0.1253250875432597,
      "grad_norm": 0.24713709950447083,
      "learning_rate": 0.0003,
      "loss": 5.0652,
      "step": 6120
    },
    {
      "epoch": 0.1254274773206643,
      "grad_norm": 0.2884705066680908,
      "learning_rate": 0.0003,
      "loss": 5.0471,
      "step": 6125
    },
    {
      "epoch": 0.12552986709806893,
      "grad_norm": 0.29362165927886963,
      "learning_rate": 0.0003,
      "loss": 5.1331,
      "step": 6130
    },
    {
      "epoch": 0.12563225687547355,
      "grad_norm": 0.2612384557723999,
      "learning_rate": 0.0003,
      "loss": 5.0427,
      "step": 6135
    },
    {
      "epoch": 0.12573464665287817,
      "grad_norm": 0.27509012818336487,
      "learning_rate": 0.0003,
      "loss": 5.0732,
      "step": 6140
    },
    {
      "epoch": 0.1258370364302828,
      "grad_norm": 0.29035264253616333,
      "learning_rate": 0.0003,
      "loss": 5.0591,
      "step": 6145
    },
    {
      "epoch": 0.12593942620768742,
      "grad_norm": 0.27949240803718567,
      "learning_rate": 0.0003,
      "loss": 5.0833,
      "step": 6150
    },
    {
      "epoch": 0.12604181598509204,
      "grad_norm": 0.2851013243198395,
      "learning_rate": 0.0003,
      "loss": 5.0396,
      "step": 6155
    },
    {
      "epoch": 0.1261442057624967,
      "grad_norm": 0.2831968367099762,
      "learning_rate": 0.0003,
      "loss": 5.0329,
      "step": 6160
    },
    {
      "epoch": 0.1262465955399013,
      "grad_norm": 0.2665991187095642,
      "learning_rate": 0.0003,
      "loss": 5.1208,
      "step": 6165
    },
    {
      "epoch": 0.12634898531730593,
      "grad_norm": 0.24800831079483032,
      "learning_rate": 0.0003,
      "loss": 5.041,
      "step": 6170
    },
    {
      "epoch": 0.12645137509471055,
      "grad_norm": 0.23127220571041107,
      "learning_rate": 0.0003,
      "loss": 5.0198,
      "step": 6175
    },
    {
      "epoch": 0.12655376487211517,
      "grad_norm": 0.28454628586769104,
      "learning_rate": 0.0003,
      "loss": 5.0206,
      "step": 6180
    },
    {
      "epoch": 0.1266561546495198,
      "grad_norm": 0.237688809633255,
      "learning_rate": 0.0003,
      "loss": 5.0353,
      "step": 6185
    },
    {
      "epoch": 0.1267585444269244,
      "grad_norm": 0.2456773966550827,
      "learning_rate": 0.0003,
      "loss": 5.0516,
      "step": 6190
    },
    {
      "epoch": 0.12686093420432903,
      "grad_norm": 0.256892591714859,
      "learning_rate": 0.0003,
      "loss": 5.0707,
      "step": 6195
    },
    {
      "epoch": 0.12696332398173366,
      "grad_norm": 0.23865951597690582,
      "learning_rate": 0.0003,
      "loss": 5.0908,
      "step": 6200
    },
    {
      "epoch": 0.12706571375913828,
      "grad_norm": 0.2427162379026413,
      "learning_rate": 0.0003,
      "loss": 5.0418,
      "step": 6205
    },
    {
      "epoch": 0.1271681035365429,
      "grad_norm": 0.2523399591445923,
      "learning_rate": 0.0003,
      "loss": 5.0393,
      "step": 6210
    },
    {
      "epoch": 0.12727049331394755,
      "grad_norm": 0.2777412235736847,
      "learning_rate": 0.0003,
      "loss": 5.0497,
      "step": 6215
    },
    {
      "epoch": 0.12737288309135217,
      "grad_norm": 0.26289626955986023,
      "learning_rate": 0.0003,
      "loss": 5.0905,
      "step": 6220
    },
    {
      "epoch": 0.1274752728687568,
      "grad_norm": 0.2896578013896942,
      "learning_rate": 0.0003,
      "loss": 5.1153,
      "step": 6225
    },
    {
      "epoch": 0.1275776626461614,
      "grad_norm": 0.26890280842781067,
      "learning_rate": 0.0003,
      "loss": 4.965,
      "step": 6230
    },
    {
      "epoch": 0.12768005242356603,
      "grad_norm": 0.24488884210586548,
      "learning_rate": 0.0003,
      "loss": 5.0857,
      "step": 6235
    },
    {
      "epoch": 0.12778244220097065,
      "grad_norm": 0.3304186761379242,
      "learning_rate": 0.0003,
      "loss": 5.0644,
      "step": 6240
    },
    {
      "epoch": 0.12788483197837527,
      "grad_norm": 0.2628065347671509,
      "learning_rate": 0.0003,
      "loss": 5.0786,
      "step": 6245
    },
    {
      "epoch": 0.1279872217557799,
      "grad_norm": 1.1583470106124878,
      "learning_rate": 0.0003,
      "loss": 5.0756,
      "step": 6250
    },
    {
      "epoch": 0.12808961153318452,
      "grad_norm": 0.26013973355293274,
      "learning_rate": 0.0003,
      "loss": 5.074,
      "step": 6255
    },
    {
      "epoch": 0.12819200131058914,
      "grad_norm": 0.24703949689865112,
      "learning_rate": 0.0003,
      "loss": 5.037,
      "step": 6260
    },
    {
      "epoch": 0.1282943910879938,
      "grad_norm": 0.30810126662254333,
      "learning_rate": 0.0003,
      "loss": 5.028,
      "step": 6265
    },
    {
      "epoch": 0.1283967808653984,
      "grad_norm": 0.34699803590774536,
      "learning_rate": 0.0003,
      "loss": 5.0482,
      "step": 6270
    },
    {
      "epoch": 0.12849917064280303,
      "grad_norm": 0.2741762399673462,
      "learning_rate": 0.0003,
      "loss": 5.0463,
      "step": 6275
    },
    {
      "epoch": 0.12860156042020765,
      "grad_norm": 0.2544890344142914,
      "learning_rate": 0.0003,
      "loss": 5.0228,
      "step": 6280
    },
    {
      "epoch": 0.12870395019761227,
      "grad_norm": 0.23670463263988495,
      "learning_rate": 0.0003,
      "loss": 5.0601,
      "step": 6285
    },
    {
      "epoch": 0.1288063399750169,
      "grad_norm": 0.23813705146312714,
      "learning_rate": 0.0003,
      "loss": 5.029,
      "step": 6290
    },
    {
      "epoch": 0.1289087297524215,
      "grad_norm": 0.23131109774112701,
      "learning_rate": 0.0003,
      "loss": 5.0519,
      "step": 6295
    },
    {
      "epoch": 0.12901111952982613,
      "grad_norm": 0.24065515398979187,
      "learning_rate": 0.0003,
      "loss": 5.0494,
      "step": 6300
    },
    {
      "epoch": 0.12911350930723076,
      "grad_norm": 0.23854351043701172,
      "learning_rate": 0.0003,
      "loss": 5.0359,
      "step": 6305
    },
    {
      "epoch": 0.12921589908463538,
      "grad_norm": 0.2923884391784668,
      "learning_rate": 0.0003,
      "loss": 4.9889,
      "step": 6310
    },
    {
      "epoch": 0.12931828886204003,
      "grad_norm": 0.2597198486328125,
      "learning_rate": 0.0003,
      "loss": 5.0757,
      "step": 6315
    },
    {
      "epoch": 0.12942067863944465,
      "grad_norm": 0.3665262460708618,
      "learning_rate": 0.0003,
      "loss": 5.0151,
      "step": 6320
    },
    {
      "epoch": 0.12952306841684927,
      "grad_norm": 0.24946120381355286,
      "learning_rate": 0.0003,
      "loss": 5.0674,
      "step": 6325
    },
    {
      "epoch": 0.1296254581942539,
      "grad_norm": 0.25710663199424744,
      "learning_rate": 0.0003,
      "loss": 5.0506,
      "step": 6330
    },
    {
      "epoch": 0.1297278479716585,
      "grad_norm": 0.25336015224456787,
      "learning_rate": 0.0003,
      "loss": 5.0538,
      "step": 6335
    },
    {
      "epoch": 0.12983023774906313,
      "grad_norm": 0.26922979950904846,
      "learning_rate": 0.0003,
      "loss": 5.006,
      "step": 6340
    },
    {
      "epoch": 0.12993262752646775,
      "grad_norm": 0.24607473611831665,
      "learning_rate": 0.0003,
      "loss": 5.0584,
      "step": 6345
    },
    {
      "epoch": 0.13003501730387237,
      "grad_norm": 0.28263700008392334,
      "learning_rate": 0.0003,
      "loss": 5.0866,
      "step": 6350
    },
    {
      "epoch": 0.130137407081277,
      "grad_norm": 0.24423837661743164,
      "learning_rate": 0.0003,
      "loss": 5.0601,
      "step": 6355
    },
    {
      "epoch": 0.13023979685868162,
      "grad_norm": 0.24322588741779327,
      "learning_rate": 0.0003,
      "loss": 5.0181,
      "step": 6360
    },
    {
      "epoch": 0.13034218663608627,
      "grad_norm": 0.2543542981147766,
      "learning_rate": 0.0003,
      "loss": 5.0332,
      "step": 6365
    },
    {
      "epoch": 0.1304445764134909,
      "grad_norm": 0.25317901372909546,
      "learning_rate": 0.0003,
      "loss": 5.0258,
      "step": 6370
    },
    {
      "epoch": 0.1305469661908955,
      "grad_norm": 0.262848436832428,
      "learning_rate": 0.0003,
      "loss": 5.0362,
      "step": 6375
    },
    {
      "epoch": 0.13064935596830013,
      "grad_norm": 0.26350048184394836,
      "learning_rate": 0.0003,
      "loss": 4.9923,
      "step": 6380
    },
    {
      "epoch": 0.13075174574570475,
      "grad_norm": 0.23977550864219666,
      "learning_rate": 0.0003,
      "loss": 5.0658,
      "step": 6385
    },
    {
      "epoch": 0.13085413552310937,
      "grad_norm": 0.2451464980840683,
      "learning_rate": 0.0003,
      "loss": 5.0401,
      "step": 6390
    },
    {
      "epoch": 0.130956525300514,
      "grad_norm": 0.24401719868183136,
      "learning_rate": 0.0003,
      "loss": 5.0332,
      "step": 6395
    },
    {
      "epoch": 0.13105891507791861,
      "grad_norm": 0.25680965185165405,
      "learning_rate": 0.0003,
      "loss": 4.9413,
      "step": 6400
    },
    {
      "epoch": 0.13116130485532324,
      "grad_norm": 0.2555413544178009,
      "learning_rate": 0.0003,
      "loss": 5.023,
      "step": 6405
    },
    {
      "epoch": 0.13126369463272786,
      "grad_norm": 0.23671738803386688,
      "learning_rate": 0.0003,
      "loss": 5.041,
      "step": 6410
    },
    {
      "epoch": 0.1313660844101325,
      "grad_norm": 0.25377485156059265,
      "learning_rate": 0.0003,
      "loss": 5.0157,
      "step": 6415
    },
    {
      "epoch": 0.13146847418753713,
      "grad_norm": 0.28589197993278503,
      "learning_rate": 0.0003,
      "loss": 4.9992,
      "step": 6420
    },
    {
      "epoch": 0.13157086396494175,
      "grad_norm": 0.3326317369937897,
      "learning_rate": 0.0003,
      "loss": 5.0644,
      "step": 6425
    },
    {
      "epoch": 0.13167325374234637,
      "grad_norm": 0.2430926114320755,
      "learning_rate": 0.0003,
      "loss": 5.0062,
      "step": 6430
    },
    {
      "epoch": 0.131775643519751,
      "grad_norm": 0.24135854840278625,
      "learning_rate": 0.0003,
      "loss": 5.0695,
      "step": 6435
    },
    {
      "epoch": 0.1318780332971556,
      "grad_norm": 0.2510199546813965,
      "learning_rate": 0.0003,
      "loss": 5.0286,
      "step": 6440
    },
    {
      "epoch": 0.13198042307456023,
      "grad_norm": 0.25724342465400696,
      "learning_rate": 0.0003,
      "loss": 5.0512,
      "step": 6445
    },
    {
      "epoch": 0.13208281285196485,
      "grad_norm": 0.2570081055164337,
      "learning_rate": 0.0003,
      "loss": 5.0157,
      "step": 6450
    },
    {
      "epoch": 0.13218520262936947,
      "grad_norm": 0.3087227940559387,
      "learning_rate": 0.0003,
      "loss": 5.053,
      "step": 6455
    },
    {
      "epoch": 0.1322875924067741,
      "grad_norm": 0.2415827363729477,
      "learning_rate": 0.0003,
      "loss": 5.0362,
      "step": 6460
    },
    {
      "epoch": 0.13238998218417874,
      "grad_norm": 1.4355356693267822,
      "learning_rate": 0.0003,
      "loss": 5.0126,
      "step": 6465
    },
    {
      "epoch": 0.13249237196158337,
      "grad_norm": 0.2502768635749817,
      "learning_rate": 0.0003,
      "loss": 5.0455,
      "step": 6470
    },
    {
      "epoch": 0.132594761738988,
      "grad_norm": 0.2415204793214798,
      "learning_rate": 0.0003,
      "loss": 5.0338,
      "step": 6475
    },
    {
      "epoch": 0.1326971515163926,
      "grad_norm": 0.2749311327934265,
      "learning_rate": 0.0003,
      "loss": 4.9887,
      "step": 6480
    },
    {
      "epoch": 0.13279954129379723,
      "grad_norm": 0.2325315922498703,
      "learning_rate": 0.0003,
      "loss": 5.0488,
      "step": 6485
    },
    {
      "epoch": 0.13290193107120185,
      "grad_norm": 0.23839184641838074,
      "learning_rate": 0.0003,
      "loss": 5.0283,
      "step": 6490
    },
    {
      "epoch": 0.13300432084860647,
      "grad_norm": 0.27810049057006836,
      "learning_rate": 0.0003,
      "loss": 5.0375,
      "step": 6495
    },
    {
      "epoch": 0.1331067106260111,
      "grad_norm": 0.24558596312999725,
      "learning_rate": 0.0003,
      "loss": 5.0604,
      "step": 6500
    },
    {
      "epoch": 0.13320910040341571,
      "grad_norm": 0.2379176914691925,
      "learning_rate": 0.0003,
      "loss": 5.0146,
      "step": 6505
    },
    {
      "epoch": 0.13331149018082034,
      "grad_norm": 0.24951337277889252,
      "learning_rate": 0.0003,
      "loss": 5.0139,
      "step": 6510
    },
    {
      "epoch": 0.13341387995822498,
      "grad_norm": 0.2671154737472534,
      "learning_rate": 0.0003,
      "loss": 5.0479,
      "step": 6515
    },
    {
      "epoch": 0.1335162697356296,
      "grad_norm": 0.27417638897895813,
      "learning_rate": 0.0003,
      "loss": 5.0236,
      "step": 6520
    },
    {
      "epoch": 0.13361865951303423,
      "grad_norm": 0.2691974937915802,
      "learning_rate": 0.0003,
      "loss": 5.0675,
      "step": 6525
    },
    {
      "epoch": 0.13372104929043885,
      "grad_norm": 0.2938447892665863,
      "learning_rate": 0.0003,
      "loss": 5.0354,
      "step": 6530
    },
    {
      "epoch": 0.13382343906784347,
      "grad_norm": 0.260647714138031,
      "learning_rate": 0.0003,
      "loss": 5.1266,
      "step": 6535
    },
    {
      "epoch": 0.1339258288452481,
      "grad_norm": 0.2850337028503418,
      "learning_rate": 0.0003,
      "loss": 5.036,
      "step": 6540
    },
    {
      "epoch": 0.1340282186226527,
      "grad_norm": 0.256000280380249,
      "learning_rate": 0.0003,
      "loss": 5.044,
      "step": 6545
    },
    {
      "epoch": 0.13413060840005733,
      "grad_norm": 0.24956929683685303,
      "learning_rate": 0.0003,
      "loss": 5.0733,
      "step": 6550
    },
    {
      "epoch": 0.13423299817746195,
      "grad_norm": 0.24049870669841766,
      "learning_rate": 0.0003,
      "loss": 5.0482,
      "step": 6555
    },
    {
      "epoch": 0.13433538795486658,
      "grad_norm": 0.257454514503479,
      "learning_rate": 0.0003,
      "loss": 5.0521,
      "step": 6560
    },
    {
      "epoch": 0.1344377777322712,
      "grad_norm": 0.26156890392303467,
      "learning_rate": 0.0003,
      "loss": 5.0397,
      "step": 6565
    },
    {
      "epoch": 0.13454016750967585,
      "grad_norm": 0.2320571094751358,
      "learning_rate": 0.0003,
      "loss": 5.0159,
      "step": 6570
    },
    {
      "epoch": 0.13464255728708047,
      "grad_norm": 0.24003830552101135,
      "learning_rate": 0.0003,
      "loss": 5.003,
      "step": 6575
    },
    {
      "epoch": 0.1347449470644851,
      "grad_norm": 0.25751402974128723,
      "learning_rate": 0.0003,
      "loss": 5.0338,
      "step": 6580
    },
    {
      "epoch": 0.1348473368418897,
      "grad_norm": 0.2699262499809265,
      "learning_rate": 0.0003,
      "loss": 5.0267,
      "step": 6585
    },
    {
      "epoch": 0.13494972661929433,
      "grad_norm": 0.2557588517665863,
      "learning_rate": 0.0003,
      "loss": 5.0298,
      "step": 6590
    },
    {
      "epoch": 0.13505211639669895,
      "grad_norm": 0.24363237619400024,
      "learning_rate": 0.0003,
      "loss": 5.0677,
      "step": 6595
    },
    {
      "epoch": 0.13515450617410357,
      "grad_norm": 0.2713419497013092,
      "learning_rate": 0.0003,
      "loss": 5.0305,
      "step": 6600
    },
    {
      "epoch": 0.1352568959515082,
      "grad_norm": 0.24032361805438995,
      "learning_rate": 0.0003,
      "loss": 5.0273,
      "step": 6605
    },
    {
      "epoch": 0.13535928572891281,
      "grad_norm": 0.2483200877904892,
      "learning_rate": 0.0003,
      "loss": 4.9871,
      "step": 6610
    },
    {
      "epoch": 0.13546167550631744,
      "grad_norm": 0.260397732257843,
      "learning_rate": 0.0003,
      "loss": 5.031,
      "step": 6615
    },
    {
      "epoch": 0.13556406528372208,
      "grad_norm": 0.26650041341781616,
      "learning_rate": 0.0003,
      "loss": 5.0102,
      "step": 6620
    },
    {
      "epoch": 0.1356664550611267,
      "grad_norm": 0.23946025967597961,
      "learning_rate": 0.0003,
      "loss": 5.0234,
      "step": 6625
    },
    {
      "epoch": 0.13576884483853133,
      "grad_norm": 0.2972355782985687,
      "learning_rate": 0.0003,
      "loss": 5.0263,
      "step": 6630
    },
    {
      "epoch": 0.13587123461593595,
      "grad_norm": 0.2401089370250702,
      "learning_rate": 0.0003,
      "loss": 5.0458,
      "step": 6635
    },
    {
      "epoch": 0.13597362439334057,
      "grad_norm": 0.24230457842350006,
      "learning_rate": 0.0003,
      "loss": 4.9748,
      "step": 6640
    },
    {
      "epoch": 0.1360760141707452,
      "grad_norm": 0.24211932718753815,
      "learning_rate": 0.0003,
      "loss": 5.0433,
      "step": 6645
    },
    {
      "epoch": 0.1361784039481498,
      "grad_norm": 0.25131285190582275,
      "learning_rate": 0.0003,
      "loss": 5.0782,
      "step": 6650
    },
    {
      "epoch": 0.13628079372555443,
      "grad_norm": 0.28355124592781067,
      "learning_rate": 0.0003,
      "loss": 5.0281,
      "step": 6655
    },
    {
      "epoch": 0.13638318350295905,
      "grad_norm": 0.2314610481262207,
      "learning_rate": 0.0003,
      "loss": 5.0297,
      "step": 6660
    },
    {
      "epoch": 0.13648557328036368,
      "grad_norm": 0.26233312487602234,
      "learning_rate": 0.0003,
      "loss": 5.0942,
      "step": 6665
    },
    {
      "epoch": 0.13658796305776832,
      "grad_norm": 0.24610739946365356,
      "learning_rate": 0.0003,
      "loss": 5.0764,
      "step": 6670
    },
    {
      "epoch": 0.13669035283517295,
      "grad_norm": 0.28637972474098206,
      "learning_rate": 0.0003,
      "loss": 5.0386,
      "step": 6675
    },
    {
      "epoch": 0.13679274261257757,
      "grad_norm": 0.2488769143819809,
      "learning_rate": 0.0003,
      "loss": 5.0485,
      "step": 6680
    },
    {
      "epoch": 0.1368951323899822,
      "grad_norm": 0.2477581650018692,
      "learning_rate": 0.0003,
      "loss": 5.038,
      "step": 6685
    },
    {
      "epoch": 0.1369975221673868,
      "grad_norm": 0.25378379225730896,
      "learning_rate": 0.0003,
      "loss": 5.0652,
      "step": 6690
    },
    {
      "epoch": 0.13709991194479143,
      "grad_norm": 0.25242990255355835,
      "learning_rate": 0.0003,
      "loss": 4.9996,
      "step": 6695
    },
    {
      "epoch": 0.13720230172219605,
      "grad_norm": 0.26495328545570374,
      "learning_rate": 0.0003,
      "loss": 5.0127,
      "step": 6700
    },
    {
      "epoch": 0.13730469149960067,
      "grad_norm": 0.24489976465702057,
      "learning_rate": 0.0003,
      "loss": 5.0007,
      "step": 6705
    },
    {
      "epoch": 0.1374070812770053,
      "grad_norm": 0.2487313151359558,
      "learning_rate": 0.0003,
      "loss": 5.0113,
      "step": 6710
    },
    {
      "epoch": 0.13750947105440992,
      "grad_norm": 0.23390714824199677,
      "learning_rate": 0.0003,
      "loss": 5.0327,
      "step": 6715
    },
    {
      "epoch": 0.13761186083181456,
      "grad_norm": 0.2524976134300232,
      "learning_rate": 0.0003,
      "loss": 4.9991,
      "step": 6720
    },
    {
      "epoch": 0.13771425060921919,
      "grad_norm": 0.25584086775779724,
      "learning_rate": 0.0003,
      "loss": 5.0311,
      "step": 6725
    },
    {
      "epoch": 0.1378166403866238,
      "grad_norm": 0.25065338611602783,
      "learning_rate": 0.0003,
      "loss": 5.0097,
      "step": 6730
    },
    {
      "epoch": 0.13791903016402843,
      "grad_norm": 0.22397209703922272,
      "learning_rate": 0.0003,
      "loss": 4.9872,
      "step": 6735
    },
    {
      "epoch": 0.13802141994143305,
      "grad_norm": 0.233913391828537,
      "learning_rate": 0.0003,
      "loss": 5.0608,
      "step": 6740
    },
    {
      "epoch": 0.13812380971883767,
      "grad_norm": 0.2484273463487625,
      "learning_rate": 0.0003,
      "loss": 4.9967,
      "step": 6745
    },
    {
      "epoch": 0.1382261994962423,
      "grad_norm": 0.2793211340904236,
      "learning_rate": 0.0003,
      "loss": 4.9444,
      "step": 6750
    },
    {
      "epoch": 0.1383285892736469,
      "grad_norm": 0.25882551074028015,
      "learning_rate": 0.0003,
      "loss": 5.0358,
      "step": 6755
    },
    {
      "epoch": 0.13843097905105153,
      "grad_norm": 0.2440735548734665,
      "learning_rate": 0.0003,
      "loss": 5.03,
      "step": 6760
    },
    {
      "epoch": 0.13853336882845615,
      "grad_norm": 0.25386354327201843,
      "learning_rate": 0.0003,
      "loss": 5.0468,
      "step": 6765
    },
    {
      "epoch": 0.1386357586058608,
      "grad_norm": 0.24036642909049988,
      "learning_rate": 0.0003,
      "loss": 5.0044,
      "step": 6770
    },
    {
      "epoch": 0.13873814838326542,
      "grad_norm": 0.255259245634079,
      "learning_rate": 0.0003,
      "loss": 5.0651,
      "step": 6775
    },
    {
      "epoch": 0.13884053816067005,
      "grad_norm": 0.3121148347854614,
      "learning_rate": 0.0003,
      "loss": 4.9908,
      "step": 6780
    },
    {
      "epoch": 0.13894292793807467,
      "grad_norm": 0.24176989495754242,
      "learning_rate": 0.0003,
      "loss": 4.9985,
      "step": 6785
    },
    {
      "epoch": 0.1390453177154793,
      "grad_norm": 0.23679430782794952,
      "learning_rate": 0.0003,
      "loss": 5.0802,
      "step": 6790
    },
    {
      "epoch": 0.1391477074928839,
      "grad_norm": 0.24225348234176636,
      "learning_rate": 0.0003,
      "loss": 5.0759,
      "step": 6795
    },
    {
      "epoch": 0.13925009727028853,
      "grad_norm": 0.23297856748104095,
      "learning_rate": 0.0003,
      "loss": 4.9883,
      "step": 6800
    },
    {
      "epoch": 0.13935248704769315,
      "grad_norm": 0.26487743854522705,
      "learning_rate": 0.0003,
      "loss": 5.0093,
      "step": 6805
    },
    {
      "epoch": 0.13945487682509777,
      "grad_norm": 0.2507880628108978,
      "learning_rate": 0.0003,
      "loss": 5.0326,
      "step": 6810
    },
    {
      "epoch": 0.1395572666025024,
      "grad_norm": 0.23700927197933197,
      "learning_rate": 0.0003,
      "loss": 5.0265,
      "step": 6815
    },
    {
      "epoch": 0.13965965637990704,
      "grad_norm": 0.24869215488433838,
      "learning_rate": 0.0003,
      "loss": 5.0231,
      "step": 6820
    },
    {
      "epoch": 0.13976204615731166,
      "grad_norm": 0.3493526577949524,
      "learning_rate": 0.0003,
      "loss": 5.0427,
      "step": 6825
    },
    {
      "epoch": 0.13986443593471629,
      "grad_norm": 0.24178656935691833,
      "learning_rate": 0.0003,
      "loss": 5.0023,
      "step": 6830
    },
    {
      "epoch": 0.1399668257121209,
      "grad_norm": 0.26609402894973755,
      "learning_rate": 0.0003,
      "loss": 4.9957,
      "step": 6835
    },
    {
      "epoch": 0.14006921548952553,
      "grad_norm": 0.24732144176959991,
      "learning_rate": 0.0003,
      "loss": 5.0224,
      "step": 6840
    },
    {
      "epoch": 0.14017160526693015,
      "grad_norm": 0.23194563388824463,
      "learning_rate": 0.0003,
      "loss": 4.9699,
      "step": 6845
    },
    {
      "epoch": 0.14027399504433477,
      "grad_norm": 0.23709094524383545,
      "learning_rate": 0.0003,
      "loss": 5.0184,
      "step": 6850
    },
    {
      "epoch": 0.1403763848217394,
      "grad_norm": 0.22520099580287933,
      "learning_rate": 0.0003,
      "loss": 5.0448,
      "step": 6855
    },
    {
      "epoch": 0.140478774599144,
      "grad_norm": 0.25094732642173767,
      "learning_rate": 0.0003,
      "loss": 5.0009,
      "step": 6860
    },
    {
      "epoch": 0.14058116437654863,
      "grad_norm": 0.25067663192749023,
      "learning_rate": 0.0003,
      "loss": 5.0565,
      "step": 6865
    },
    {
      "epoch": 0.14068355415395328,
      "grad_norm": 0.22902968525886536,
      "learning_rate": 0.0003,
      "loss": 4.9944,
      "step": 6870
    },
    {
      "epoch": 0.1407859439313579,
      "grad_norm": 0.294519305229187,
      "learning_rate": 0.0003,
      "loss": 5.0484,
      "step": 6875
    },
    {
      "epoch": 0.14088833370876253,
      "grad_norm": 0.2572459876537323,
      "learning_rate": 0.0003,
      "loss": 5.0189,
      "step": 6880
    },
    {
      "epoch": 0.14099072348616715,
      "grad_norm": 0.271657794713974,
      "learning_rate": 0.0003,
      "loss": 4.9645,
      "step": 6885
    },
    {
      "epoch": 0.14109311326357177,
      "grad_norm": 0.25202035903930664,
      "learning_rate": 0.0003,
      "loss": 5.0502,
      "step": 6890
    },
    {
      "epoch": 0.1411955030409764,
      "grad_norm": 0.2642892301082611,
      "learning_rate": 0.0003,
      "loss": 5.0016,
      "step": 6895
    },
    {
      "epoch": 0.141297892818381,
      "grad_norm": 0.26839500665664673,
      "learning_rate": 0.0003,
      "loss": 4.9992,
      "step": 6900
    },
    {
      "epoch": 0.14140028259578563,
      "grad_norm": 0.2610272467136383,
      "learning_rate": 0.0003,
      "loss": 5.0438,
      "step": 6905
    },
    {
      "epoch": 0.14150267237319025,
      "grad_norm": 0.22692175209522247,
      "learning_rate": 0.0003,
      "loss": 4.9622,
      "step": 6910
    },
    {
      "epoch": 0.14160506215059487,
      "grad_norm": 0.22383558750152588,
      "learning_rate": 0.0003,
      "loss": 5.0094,
      "step": 6915
    },
    {
      "epoch": 0.1417074519279995,
      "grad_norm": 0.24731644988059998,
      "learning_rate": 0.0003,
      "loss": 5.0331,
      "step": 6920
    },
    {
      "epoch": 0.14180984170540414,
      "grad_norm": 0.2571280896663666,
      "learning_rate": 0.0003,
      "loss": 5.0036,
      "step": 6925
    },
    {
      "epoch": 0.14191223148280876,
      "grad_norm": 0.23507055640220642,
      "learning_rate": 0.0003,
      "loss": 5.0071,
      "step": 6930
    },
    {
      "epoch": 0.1420146212602134,
      "grad_norm": 0.41560256481170654,
      "learning_rate": 0.0003,
      "loss": 4.9902,
      "step": 6935
    },
    {
      "epoch": 0.142117011037618,
      "grad_norm": 0.2307703197002411,
      "learning_rate": 0.0003,
      "loss": 4.9913,
      "step": 6940
    },
    {
      "epoch": 0.14221940081502263,
      "grad_norm": 0.2455960363149643,
      "learning_rate": 0.0003,
      "loss": 5.0613,
      "step": 6945
    },
    {
      "epoch": 0.14232179059242725,
      "grad_norm": 0.2564148008823395,
      "learning_rate": 0.0003,
      "loss": 5.0081,
      "step": 6950
    },
    {
      "epoch": 0.14242418036983187,
      "grad_norm": 0.3252364993095398,
      "learning_rate": 0.0003,
      "loss": 5.0755,
      "step": 6955
    },
    {
      "epoch": 0.1425265701472365,
      "grad_norm": 0.2593346834182739,
      "learning_rate": 0.0003,
      "loss": 5.0672,
      "step": 6960
    },
    {
      "epoch": 0.1426289599246411,
      "grad_norm": 0.27190688252449036,
      "learning_rate": 0.0003,
      "loss": 4.9919,
      "step": 6965
    },
    {
      "epoch": 0.14273134970204573,
      "grad_norm": 0.3587278425693512,
      "learning_rate": 0.0003,
      "loss": 5.0529,
      "step": 6970
    },
    {
      "epoch": 0.14283373947945038,
      "grad_norm": 0.2731194794178009,
      "learning_rate": 0.0003,
      "loss": 5.0135,
      "step": 6975
    },
    {
      "epoch": 0.142936129256855,
      "grad_norm": 0.35815128684043884,
      "learning_rate": 0.0003,
      "loss": 5.0411,
      "step": 6980
    },
    {
      "epoch": 0.14303851903425963,
      "grad_norm": 0.2469884753227234,
      "learning_rate": 0.0003,
      "loss": 4.9931,
      "step": 6985
    },
    {
      "epoch": 0.14314090881166425,
      "grad_norm": 0.24399548768997192,
      "learning_rate": 0.0003,
      "loss": 4.9633,
      "step": 6990
    },
    {
      "epoch": 0.14324329858906887,
      "grad_norm": 0.2295876443386078,
      "learning_rate": 0.0003,
      "loss": 4.9609,
      "step": 6995
    },
    {
      "epoch": 0.1433456883664735,
      "grad_norm": 0.24459685385227203,
      "learning_rate": 0.0003,
      "loss": 5.0562,
      "step": 7000
    },
    {
      "epoch": 0.1433456883664735,
      "eval_loss": 5.037739276885986,
      "eval_runtime": 17.2414,
      "eval_samples_per_second": 33.756,
      "eval_steps_per_second": 4.234,
      "step": 7000
    },
    {
      "epoch": 0.1434480781438781,
      "grad_norm": 0.24191655218601227,
      "learning_rate": 0.0003,
      "loss": 5.012,
      "step": 7005
    },
    {
      "epoch": 0.14355046792128273,
      "grad_norm": 0.3429293930530548,
      "learning_rate": 0.0003,
      "loss": 4.9455,
      "step": 7010
    },
    {
      "epoch": 0.14365285769868735,
      "grad_norm": 0.2853206396102905,
      "learning_rate": 0.0003,
      "loss": 5.0329,
      "step": 7015
    },
    {
      "epoch": 0.14375524747609197,
      "grad_norm": 0.23877167701721191,
      "learning_rate": 0.0003,
      "loss": 4.9954,
      "step": 7020
    },
    {
      "epoch": 0.14385763725349662,
      "grad_norm": 0.2498999685049057,
      "learning_rate": 0.0003,
      "loss": 5.0355,
      "step": 7025
    },
    {
      "epoch": 0.14396002703090124,
      "grad_norm": 0.23083969950675964,
      "learning_rate": 0.0003,
      "loss": 5.0247,
      "step": 7030
    },
    {
      "epoch": 0.14406241680830587,
      "grad_norm": 0.267577201128006,
      "learning_rate": 0.0003,
      "loss": 4.9991,
      "step": 7035
    },
    {
      "epoch": 0.1441648065857105,
      "grad_norm": 0.2475428432226181,
      "learning_rate": 0.0003,
      "loss": 5.0372,
      "step": 7040
    },
    {
      "epoch": 0.1442671963631151,
      "grad_norm": 0.2713354527950287,
      "learning_rate": 0.0003,
      "loss": 4.9865,
      "step": 7045
    },
    {
      "epoch": 0.14436958614051973,
      "grad_norm": 0.30416226387023926,
      "learning_rate": 0.0003,
      "loss": 4.9769,
      "step": 7050
    },
    {
      "epoch": 0.14447197591792435,
      "grad_norm": 0.26562967896461487,
      "learning_rate": 0.0003,
      "loss": 5.0089,
      "step": 7055
    },
    {
      "epoch": 0.14457436569532897,
      "grad_norm": 0.230251744389534,
      "learning_rate": 0.0003,
      "loss": 5.0115,
      "step": 7060
    },
    {
      "epoch": 0.1446767554727336,
      "grad_norm": 0.30862435698509216,
      "learning_rate": 0.0003,
      "loss": 5.029,
      "step": 7065
    },
    {
      "epoch": 0.1447791452501382,
      "grad_norm": 0.2607644498348236,
      "learning_rate": 0.0003,
      "loss": 5.0915,
      "step": 7070
    },
    {
      "epoch": 0.14488153502754286,
      "grad_norm": 0.24924525618553162,
      "learning_rate": 0.0003,
      "loss": 5.0379,
      "step": 7075
    },
    {
      "epoch": 0.14498392480494748,
      "grad_norm": 0.23883071541786194,
      "learning_rate": 0.0003,
      "loss": 5.0237,
      "step": 7080
    },
    {
      "epoch": 0.1450863145823521,
      "grad_norm": 0.23020990192890167,
      "learning_rate": 0.0003,
      "loss": 5.006,
      "step": 7085
    },
    {
      "epoch": 0.14518870435975673,
      "grad_norm": 0.25107336044311523,
      "learning_rate": 0.0003,
      "loss": 4.9507,
      "step": 7090
    },
    {
      "epoch": 0.14529109413716135,
      "grad_norm": 0.245454341173172,
      "learning_rate": 0.0003,
      "loss": 4.9903,
      "step": 7095
    },
    {
      "epoch": 0.14539348391456597,
      "grad_norm": 0.3312162160873413,
      "learning_rate": 0.0003,
      "loss": 5.0121,
      "step": 7100
    },
    {
      "epoch": 0.1454958736919706,
      "grad_norm": 0.2554459869861603,
      "learning_rate": 0.0003,
      "loss": 4.9821,
      "step": 7105
    },
    {
      "epoch": 0.1455982634693752,
      "grad_norm": 0.2653450667858124,
      "learning_rate": 0.0003,
      "loss": 5.0476,
      "step": 7110
    },
    {
      "epoch": 0.14570065324677983,
      "grad_norm": 0.2704363763332367,
      "learning_rate": 0.0003,
      "loss": 5.0029,
      "step": 7115
    },
    {
      "epoch": 0.14580304302418445,
      "grad_norm": 0.23047906160354614,
      "learning_rate": 0.0003,
      "loss": 5.013,
      "step": 7120
    },
    {
      "epoch": 0.1459054328015891,
      "grad_norm": 0.23883530497550964,
      "learning_rate": 0.0003,
      "loss": 4.9518,
      "step": 7125
    },
    {
      "epoch": 0.14600782257899372,
      "grad_norm": 0.28106656670570374,
      "learning_rate": 0.0003,
      "loss": 4.9999,
      "step": 7130
    },
    {
      "epoch": 0.14611021235639834,
      "grad_norm": 0.25772610306739807,
      "learning_rate": 0.0003,
      "loss": 5.0161,
      "step": 7135
    },
    {
      "epoch": 0.14621260213380297,
      "grad_norm": 0.22792254388332367,
      "learning_rate": 0.0003,
      "loss": 5.0558,
      "step": 7140
    },
    {
      "epoch": 0.1463149919112076,
      "grad_norm": 0.24308115243911743,
      "learning_rate": 0.0003,
      "loss": 5.0187,
      "step": 7145
    },
    {
      "epoch": 0.1464173816886122,
      "grad_norm": 0.23834307491779327,
      "learning_rate": 0.0003,
      "loss": 4.9911,
      "step": 7150
    },
    {
      "epoch": 0.14651977146601683,
      "grad_norm": 0.2690189778804779,
      "learning_rate": 0.0003,
      "loss": 4.9441,
      "step": 7155
    },
    {
      "epoch": 0.14662216124342145,
      "grad_norm": 0.22987601161003113,
      "learning_rate": 0.0003,
      "loss": 5.0239,
      "step": 7160
    },
    {
      "epoch": 0.14672455102082607,
      "grad_norm": 0.22981633245944977,
      "learning_rate": 0.0003,
      "loss": 5.033,
      "step": 7165
    },
    {
      "epoch": 0.1468269407982307,
      "grad_norm": 0.2626011371612549,
      "learning_rate": 0.0003,
      "loss": 5.0435,
      "step": 7170
    },
    {
      "epoch": 0.14692933057563534,
      "grad_norm": 0.26072123646736145,
      "learning_rate": 0.0003,
      "loss": 5.0708,
      "step": 7175
    },
    {
      "epoch": 0.14703172035303996,
      "grad_norm": 0.2979556918144226,
      "learning_rate": 0.0003,
      "loss": 5.0462,
      "step": 7180
    },
    {
      "epoch": 0.14713411013044458,
      "grad_norm": 0.2395155280828476,
      "learning_rate": 0.0003,
      "loss": 4.9404,
      "step": 7185
    },
    {
      "epoch": 0.1472364999078492,
      "grad_norm": 0.2690223455429077,
      "learning_rate": 0.0003,
      "loss": 5.0182,
      "step": 7190
    },
    {
      "epoch": 0.14733888968525383,
      "grad_norm": 0.2435748279094696,
      "learning_rate": 0.0003,
      "loss": 4.9982,
      "step": 7195
    },
    {
      "epoch": 0.14744127946265845,
      "grad_norm": 0.25203970074653625,
      "learning_rate": 0.0003,
      "loss": 4.9533,
      "step": 7200
    },
    {
      "epoch": 0.14754366924006307,
      "grad_norm": 0.2593334913253784,
      "learning_rate": 0.0003,
      "loss": 5.0437,
      "step": 7205
    },
    {
      "epoch": 0.1476460590174677,
      "grad_norm": 0.22257547080516815,
      "learning_rate": 0.0003,
      "loss": 5.0518,
      "step": 7210
    },
    {
      "epoch": 0.1477484487948723,
      "grad_norm": 0.24585096538066864,
      "learning_rate": 0.0003,
      "loss": 4.9979,
      "step": 7215
    },
    {
      "epoch": 0.14785083857227693,
      "grad_norm": 0.26046517491340637,
      "learning_rate": 0.0003,
      "loss": 5.0929,
      "step": 7220
    },
    {
      "epoch": 0.14795322834968158,
      "grad_norm": 0.26173135638237,
      "learning_rate": 0.0003,
      "loss": 5.0058,
      "step": 7225
    },
    {
      "epoch": 0.1480556181270862,
      "grad_norm": 0.2430068701505661,
      "learning_rate": 0.0003,
      "loss": 5.0021,
      "step": 7230
    },
    {
      "epoch": 0.14815800790449082,
      "grad_norm": 0.23499150574207306,
      "learning_rate": 0.0003,
      "loss": 4.9641,
      "step": 7235
    },
    {
      "epoch": 0.14826039768189544,
      "grad_norm": 0.24758745729923248,
      "learning_rate": 0.0003,
      "loss": 5.0309,
      "step": 7240
    },
    {
      "epoch": 0.14836278745930007,
      "grad_norm": 0.24381928145885468,
      "learning_rate": 0.0003,
      "loss": 5.0278,
      "step": 7245
    },
    {
      "epoch": 0.1484651772367047,
      "grad_norm": 0.27376264333724976,
      "learning_rate": 0.0003,
      "loss": 4.9957,
      "step": 7250
    },
    {
      "epoch": 0.1485675670141093,
      "grad_norm": 0.26902592182159424,
      "learning_rate": 0.0003,
      "loss": 5.0226,
      "step": 7255
    },
    {
      "epoch": 0.14866995679151393,
      "grad_norm": 0.2667185366153717,
      "learning_rate": 0.0003,
      "loss": 5.0353,
      "step": 7260
    },
    {
      "epoch": 0.14877234656891855,
      "grad_norm": 0.25455304980278015,
      "learning_rate": 0.0003,
      "loss": 5.0368,
      "step": 7265
    },
    {
      "epoch": 0.14887473634632317,
      "grad_norm": 0.26712462306022644,
      "learning_rate": 0.0003,
      "loss": 4.9974,
      "step": 7270
    },
    {
      "epoch": 0.1489771261237278,
      "grad_norm": 0.2454889863729477,
      "learning_rate": 0.0003,
      "loss": 5.0114,
      "step": 7275
    },
    {
      "epoch": 0.14907951590113244,
      "grad_norm": 0.22794292867183685,
      "learning_rate": 0.0003,
      "loss": 4.996,
      "step": 7280
    },
    {
      "epoch": 0.14918190567853706,
      "grad_norm": 0.26641157269477844,
      "learning_rate": 0.0003,
      "loss": 5.0429,
      "step": 7285
    },
    {
      "epoch": 0.14928429545594168,
      "grad_norm": 0.24585548043251038,
      "learning_rate": 0.0003,
      "loss": 5.0082,
      "step": 7290
    },
    {
      "epoch": 0.1493866852333463,
      "grad_norm": 0.2460235208272934,
      "learning_rate": 0.0003,
      "loss": 5.013,
      "step": 7295
    },
    {
      "epoch": 0.14948907501075093,
      "grad_norm": 0.24668200314044952,
      "learning_rate": 0.0003,
      "loss": 5.0051,
      "step": 7300
    },
    {
      "epoch": 0.14959146478815555,
      "grad_norm": 0.2402060478925705,
      "learning_rate": 0.0003,
      "loss": 4.9951,
      "step": 7305
    },
    {
      "epoch": 0.14969385456556017,
      "grad_norm": 0.27260613441467285,
      "learning_rate": 0.0003,
      "loss": 5.0488,
      "step": 7310
    },
    {
      "epoch": 0.1497962443429648,
      "grad_norm": 0.23753002285957336,
      "learning_rate": 0.0003,
      "loss": 4.9958,
      "step": 7315
    },
    {
      "epoch": 0.1498986341203694,
      "grad_norm": 0.24035422503948212,
      "learning_rate": 0.0003,
      "loss": 5.0024,
      "step": 7320
    },
    {
      "epoch": 0.15000102389777403,
      "grad_norm": 0.25172603130340576,
      "learning_rate": 0.0003,
      "loss": 4.9639,
      "step": 7325
    },
    {
      "epoch": 0.15010341367517868,
      "grad_norm": 0.2577626407146454,
      "learning_rate": 0.0003,
      "loss": 4.9987,
      "step": 7330
    },
    {
      "epoch": 0.1502058034525833,
      "grad_norm": 0.3757743239402771,
      "learning_rate": 0.0003,
      "loss": 4.9497,
      "step": 7335
    },
    {
      "epoch": 0.15030819322998792,
      "grad_norm": 0.26163744926452637,
      "learning_rate": 0.0003,
      "loss": 5.0107,
      "step": 7340
    },
    {
      "epoch": 0.15041058300739255,
      "grad_norm": 0.2646372318267822,
      "learning_rate": 0.0003,
      "loss": 4.9945,
      "step": 7345
    },
    {
      "epoch": 0.15051297278479717,
      "grad_norm": 0.2732544243335724,
      "learning_rate": 0.0003,
      "loss": 4.9914,
      "step": 7350
    },
    {
      "epoch": 0.1506153625622018,
      "grad_norm": 0.23641030490398407,
      "learning_rate": 0.0003,
      "loss": 4.9703,
      "step": 7355
    },
    {
      "epoch": 0.1507177523396064,
      "grad_norm": 0.24585741758346558,
      "learning_rate": 0.0003,
      "loss": 5.0141,
      "step": 7360
    },
    {
      "epoch": 0.15082014211701103,
      "grad_norm": 0.25379955768585205,
      "learning_rate": 0.0003,
      "loss": 5.056,
      "step": 7365
    },
    {
      "epoch": 0.15092253189441565,
      "grad_norm": 0.2581508755683899,
      "learning_rate": 0.0003,
      "loss": 5.0128,
      "step": 7370
    },
    {
      "epoch": 0.15102492167182027,
      "grad_norm": 0.24063491821289062,
      "learning_rate": 0.0003,
      "loss": 5.0242,
      "step": 7375
    },
    {
      "epoch": 0.15112731144922492,
      "grad_norm": 0.26179948449134827,
      "learning_rate": 0.0003,
      "loss": 5.0013,
      "step": 7380
    },
    {
      "epoch": 0.15122970122662954,
      "grad_norm": 0.3701592981815338,
      "learning_rate": 0.0003,
      "loss": 5.0073,
      "step": 7385
    },
    {
      "epoch": 0.15133209100403416,
      "grad_norm": 0.27745503187179565,
      "learning_rate": 0.0003,
      "loss": 5.0051,
      "step": 7390
    },
    {
      "epoch": 0.15143448078143878,
      "grad_norm": 0.2534632086753845,
      "learning_rate": 0.0003,
      "loss": 4.9536,
      "step": 7395
    },
    {
      "epoch": 0.1515368705588434,
      "grad_norm": 0.2302110195159912,
      "learning_rate": 0.0003,
      "loss": 5.0014,
      "step": 7400
    },
    {
      "epoch": 0.15163926033624803,
      "grad_norm": 0.2508939504623413,
      "learning_rate": 0.0003,
      "loss": 5.0082,
      "step": 7405
    },
    {
      "epoch": 0.15174165011365265,
      "grad_norm": 0.256247878074646,
      "learning_rate": 0.0003,
      "loss": 4.9864,
      "step": 7410
    },
    {
      "epoch": 0.15184403989105727,
      "grad_norm": 0.2760912775993347,
      "learning_rate": 0.0003,
      "loss": 4.9779,
      "step": 7415
    },
    {
      "epoch": 0.1519464296684619,
      "grad_norm": 0.25915077328681946,
      "learning_rate": 0.0003,
      "loss": 4.9908,
      "step": 7420
    },
    {
      "epoch": 0.1520488194458665,
      "grad_norm": 0.41883066296577454,
      "learning_rate": 0.0003,
      "loss": 5.0017,
      "step": 7425
    },
    {
      "epoch": 0.15215120922327116,
      "grad_norm": 0.32636791467666626,
      "learning_rate": 0.0003,
      "loss": 4.9843,
      "step": 7430
    },
    {
      "epoch": 0.15225359900067578,
      "grad_norm": 0.24279625713825226,
      "learning_rate": 0.0003,
      "loss": 5.0171,
      "step": 7435
    },
    {
      "epoch": 0.1523559887780804,
      "grad_norm": 0.2643495202064514,
      "learning_rate": 0.0003,
      "loss": 5.0126,
      "step": 7440
    },
    {
      "epoch": 0.15245837855548502,
      "grad_norm": 0.23208200931549072,
      "learning_rate": 0.0003,
      "loss": 5.0198,
      "step": 7445
    },
    {
      "epoch": 0.15256076833288965,
      "grad_norm": 0.2417708933353424,
      "learning_rate": 0.0003,
      "loss": 4.9749,
      "step": 7450
    },
    {
      "epoch": 0.15266315811029427,
      "grad_norm": 0.25086939334869385,
      "learning_rate": 0.0003,
      "loss": 5.0015,
      "step": 7455
    },
    {
      "epoch": 0.1527655478876989,
      "grad_norm": 0.26524269580841064,
      "learning_rate": 0.0003,
      "loss": 4.9887,
      "step": 7460
    },
    {
      "epoch": 0.1528679376651035,
      "grad_norm": 0.45219752192497253,
      "learning_rate": 0.0003,
      "loss": 5.0129,
      "step": 7465
    },
    {
      "epoch": 0.15297032744250813,
      "grad_norm": 0.2360047549009323,
      "learning_rate": 0.0003,
      "loss": 5.0366,
      "step": 7470
    },
    {
      "epoch": 0.15307271721991275,
      "grad_norm": 0.23776350915431976,
      "learning_rate": 0.0003,
      "loss": 4.9592,
      "step": 7475
    },
    {
      "epoch": 0.1531751069973174,
      "grad_norm": 0.23889972269535065,
      "learning_rate": 0.0003,
      "loss": 5.0274,
      "step": 7480
    },
    {
      "epoch": 0.15327749677472202,
      "grad_norm": 0.2513630986213684,
      "learning_rate": 0.0003,
      "loss": 4.9837,
      "step": 7485
    },
    {
      "epoch": 0.15337988655212664,
      "grad_norm": 0.2382023185491562,
      "learning_rate": 0.0003,
      "loss": 5.0247,
      "step": 7490
    },
    {
      "epoch": 0.15348227632953126,
      "grad_norm": 0.23475874960422516,
      "learning_rate": 0.0003,
      "loss": 5.0181,
      "step": 7495
    },
    {
      "epoch": 0.15358466610693589,
      "grad_norm": 0.2395312786102295,
      "learning_rate": 0.0003,
      "loss": 5.0178,
      "step": 7500
    },
    {
      "epoch": 0.1536870558843405,
      "grad_norm": 0.2524147629737854,
      "learning_rate": 0.0003,
      "loss": 4.9548,
      "step": 7505
    },
    {
      "epoch": 0.15378944566174513,
      "grad_norm": 0.28112271428108215,
      "learning_rate": 0.0003,
      "loss": 5.0397,
      "step": 7510
    },
    {
      "epoch": 0.15389183543914975,
      "grad_norm": 0.23278845846652985,
      "learning_rate": 0.0003,
      "loss": 5.0145,
      "step": 7515
    },
    {
      "epoch": 0.15399422521655437,
      "grad_norm": 0.27111172676086426,
      "learning_rate": 0.0003,
      "loss": 4.9722,
      "step": 7520
    },
    {
      "epoch": 0.154096614993959,
      "grad_norm": 0.2546147406101227,
      "learning_rate": 0.0003,
      "loss": 4.9905,
      "step": 7525
    },
    {
      "epoch": 0.15419900477136364,
      "grad_norm": 0.25940370559692383,
      "learning_rate": 0.0003,
      "loss": 4.9608,
      "step": 7530
    },
    {
      "epoch": 0.15430139454876826,
      "grad_norm": 0.23555074632167816,
      "learning_rate": 0.0003,
      "loss": 4.9884,
      "step": 7535
    },
    {
      "epoch": 0.15440378432617288,
      "grad_norm": 0.24311836063861847,
      "learning_rate": 0.0003,
      "loss": 4.9787,
      "step": 7540
    },
    {
      "epoch": 0.1545061741035775,
      "grad_norm": 0.24595271050930023,
      "learning_rate": 0.0003,
      "loss": 5.0135,
      "step": 7545
    },
    {
      "epoch": 0.15460856388098212,
      "grad_norm": 0.2362990379333496,
      "learning_rate": 0.0003,
      "loss": 4.9453,
      "step": 7550
    },
    {
      "epoch": 0.15471095365838675,
      "grad_norm": 0.24405013024806976,
      "learning_rate": 0.0003,
      "loss": 5.0109,
      "step": 7555
    },
    {
      "epoch": 0.15481334343579137,
      "grad_norm": 0.2482171505689621,
      "learning_rate": 0.0003,
      "loss": 5.0172,
      "step": 7560
    },
    {
      "epoch": 0.154915733213196,
      "grad_norm": 0.245453879237175,
      "learning_rate": 0.0003,
      "loss": 5.003,
      "step": 7565
    },
    {
      "epoch": 0.1550181229906006,
      "grad_norm": 0.32339438796043396,
      "learning_rate": 0.0003,
      "loss": 4.9981,
      "step": 7570
    },
    {
      "epoch": 0.15512051276800523,
      "grad_norm": 0.2471870481967926,
      "learning_rate": 0.0003,
      "loss": 5.0272,
      "step": 7575
    },
    {
      "epoch": 0.15522290254540988,
      "grad_norm": 0.25140976905822754,
      "learning_rate": 0.0003,
      "loss": 4.9293,
      "step": 7580
    },
    {
      "epoch": 0.1553252923228145,
      "grad_norm": 0.2848415970802307,
      "learning_rate": 0.0003,
      "loss": 4.9867,
      "step": 7585
    },
    {
      "epoch": 0.15542768210021912,
      "grad_norm": 0.2239174246788025,
      "learning_rate": 0.0003,
      "loss": 4.9804,
      "step": 7590
    },
    {
      "epoch": 0.15553007187762374,
      "grad_norm": 0.2289915233850479,
      "learning_rate": 0.0003,
      "loss": 4.9677,
      "step": 7595
    },
    {
      "epoch": 0.15563246165502836,
      "grad_norm": 0.25757190585136414,
      "learning_rate": 0.0003,
      "loss": 4.9976,
      "step": 7600
    },
    {
      "epoch": 0.15573485143243299,
      "grad_norm": 0.2640559673309326,
      "learning_rate": 0.0003,
      "loss": 4.9737,
      "step": 7605
    },
    {
      "epoch": 0.1558372412098376,
      "grad_norm": 0.2947389483451843,
      "learning_rate": 0.0003,
      "loss": 4.9666,
      "step": 7610
    },
    {
      "epoch": 0.15593963098724223,
      "grad_norm": 0.2592456638813019,
      "learning_rate": 0.0003,
      "loss": 4.9817,
      "step": 7615
    },
    {
      "epoch": 0.15604202076464685,
      "grad_norm": 0.25153815746307373,
      "learning_rate": 0.0003,
      "loss": 4.9565,
      "step": 7620
    },
    {
      "epoch": 0.15614441054205147,
      "grad_norm": 0.24976173043251038,
      "learning_rate": 0.0003,
      "loss": 4.988,
      "step": 7625
    },
    {
      "epoch": 0.1562468003194561,
      "grad_norm": 0.2521374821662903,
      "learning_rate": 0.0003,
      "loss": 4.963,
      "step": 7630
    },
    {
      "epoch": 0.15634919009686074,
      "grad_norm": 0.24638235569000244,
      "learning_rate": 0.0003,
      "loss": 4.9822,
      "step": 7635
    },
    {
      "epoch": 0.15645157987426536,
      "grad_norm": 0.2727237045764923,
      "learning_rate": 0.0003,
      "loss": 5.0165,
      "step": 7640
    },
    {
      "epoch": 0.15655396965166998,
      "grad_norm": 0.24292902648448944,
      "learning_rate": 0.0003,
      "loss": 4.9786,
      "step": 7645
    },
    {
      "epoch": 0.1566563594290746,
      "grad_norm": 0.29140084981918335,
      "learning_rate": 0.0003,
      "loss": 4.9874,
      "step": 7650
    },
    {
      "epoch": 0.15675874920647923,
      "grad_norm": 0.23380549252033234,
      "learning_rate": 0.0003,
      "loss": 5.0012,
      "step": 7655
    },
    {
      "epoch": 0.15686113898388385,
      "grad_norm": 0.24549448490142822,
      "learning_rate": 0.0003,
      "loss": 5.0244,
      "step": 7660
    },
    {
      "epoch": 0.15696352876128847,
      "grad_norm": 0.24343416094779968,
      "learning_rate": 0.0003,
      "loss": 5.0487,
      "step": 7665
    },
    {
      "epoch": 0.1570659185386931,
      "grad_norm": 0.252719521522522,
      "learning_rate": 0.0003,
      "loss": 4.9932,
      "step": 7670
    },
    {
      "epoch": 0.1571683083160977,
      "grad_norm": 0.245554581284523,
      "learning_rate": 0.0003,
      "loss": 5.0099,
      "step": 7675
    },
    {
      "epoch": 0.15727069809350233,
      "grad_norm": 0.24932564795017242,
      "learning_rate": 0.0003,
      "loss": 5.0463,
      "step": 7680
    },
    {
      "epoch": 0.15737308787090698,
      "grad_norm": 0.29451170563697815,
      "learning_rate": 0.0003,
      "loss": 5.005,
      "step": 7685
    },
    {
      "epoch": 0.1574754776483116,
      "grad_norm": 0.23839223384857178,
      "learning_rate": 0.0003,
      "loss": 4.977,
      "step": 7690
    },
    {
      "epoch": 0.15757786742571622,
      "grad_norm": 0.24138836562633514,
      "learning_rate": 0.0003,
      "loss": 4.9934,
      "step": 7695
    },
    {
      "epoch": 0.15768025720312084,
      "grad_norm": 0.36683306097984314,
      "learning_rate": 0.0003,
      "loss": 5.024,
      "step": 7700
    },
    {
      "epoch": 0.15778264698052546,
      "grad_norm": 0.26034170389175415,
      "learning_rate": 0.0003,
      "loss": 5.0189,
      "step": 7705
    },
    {
      "epoch": 0.1578850367579301,
      "grad_norm": 0.23770155012607574,
      "learning_rate": 0.0003,
      "loss": 4.9818,
      "step": 7710
    },
    {
      "epoch": 0.1579874265353347,
      "grad_norm": 0.25007253885269165,
      "learning_rate": 0.0003,
      "loss": 5.0418,
      "step": 7715
    },
    {
      "epoch": 0.15808981631273933,
      "grad_norm": 0.25594475865364075,
      "learning_rate": 0.0003,
      "loss": 4.9985,
      "step": 7720
    },
    {
      "epoch": 0.15819220609014395,
      "grad_norm": 0.26413026452064514,
      "learning_rate": 0.0003,
      "loss": 5.019,
      "step": 7725
    },
    {
      "epoch": 0.15829459586754857,
      "grad_norm": 0.25864240527153015,
      "learning_rate": 0.0003,
      "loss": 4.9674,
      "step": 7730
    },
    {
      "epoch": 0.15839698564495322,
      "grad_norm": 0.24335207045078278,
      "learning_rate": 0.0003,
      "loss": 5.0241,
      "step": 7735
    },
    {
      "epoch": 0.15849937542235784,
      "grad_norm": 0.2253551185131073,
      "learning_rate": 0.0003,
      "loss": 5.0276,
      "step": 7740
    },
    {
      "epoch": 0.15860176519976246,
      "grad_norm": 0.2327730655670166,
      "learning_rate": 0.0003,
      "loss": 4.9949,
      "step": 7745
    },
    {
      "epoch": 0.15870415497716708,
      "grad_norm": 0.26976391673088074,
      "learning_rate": 0.0003,
      "loss": 5.0058,
      "step": 7750
    },
    {
      "epoch": 0.1588065447545717,
      "grad_norm": 0.24418939650058746,
      "learning_rate": 0.0003,
      "loss": 5.0076,
      "step": 7755
    },
    {
      "epoch": 0.15890893453197633,
      "grad_norm": 0.3299313485622406,
      "learning_rate": 0.0003,
      "loss": 4.8469,
      "step": 7760
    },
    {
      "epoch": 0.15901132430938095,
      "grad_norm": 0.2607487142086029,
      "learning_rate": 0.0003,
      "loss": 4.9981,
      "step": 7765
    },
    {
      "epoch": 0.15911371408678557,
      "grad_norm": 0.29214009642601013,
      "learning_rate": 0.0003,
      "loss": 4.9962,
      "step": 7770
    },
    {
      "epoch": 0.1592161038641902,
      "grad_norm": 0.2634567320346832,
      "learning_rate": 0.0003,
      "loss": 4.9739,
      "step": 7775
    },
    {
      "epoch": 0.1593184936415948,
      "grad_norm": 0.2418191134929657,
      "learning_rate": 0.0003,
      "loss": 5.0173,
      "step": 7780
    },
    {
      "epoch": 0.15942088341899946,
      "grad_norm": 0.24370892345905304,
      "learning_rate": 0.0003,
      "loss": 4.9085,
      "step": 7785
    },
    {
      "epoch": 0.15952327319640408,
      "grad_norm": 0.2694243788719177,
      "learning_rate": 0.0003,
      "loss": 4.9661,
      "step": 7790
    },
    {
      "epoch": 0.1596256629738087,
      "grad_norm": 0.309671014547348,
      "learning_rate": 0.0003,
      "loss": 4.9758,
      "step": 7795
    },
    {
      "epoch": 0.15972805275121332,
      "grad_norm": 0.28414806723594666,
      "learning_rate": 0.0003,
      "loss": 5.0043,
      "step": 7800
    },
    {
      "epoch": 0.15983044252861794,
      "grad_norm": 0.23653919994831085,
      "learning_rate": 0.0003,
      "loss": 4.9264,
      "step": 7805
    },
    {
      "epoch": 0.15993283230602257,
      "grad_norm": 0.5142612457275391,
      "learning_rate": 0.0003,
      "loss": 4.9463,
      "step": 7810
    },
    {
      "epoch": 0.1600352220834272,
      "grad_norm": 0.2367003709077835,
      "learning_rate": 0.0003,
      "loss": 5.0234,
      "step": 7815
    },
    {
      "epoch": 0.1601376118608318,
      "grad_norm": 0.2358447164297104,
      "learning_rate": 0.0003,
      "loss": 4.9898,
      "step": 7820
    },
    {
      "epoch": 0.16024000163823643,
      "grad_norm": 0.22502097487449646,
      "learning_rate": 0.0003,
      "loss": 4.973,
      "step": 7825
    },
    {
      "epoch": 0.16034239141564105,
      "grad_norm": 0.25816860795021057,
      "learning_rate": 0.0003,
      "loss": 4.9557,
      "step": 7830
    },
    {
      "epoch": 0.1604447811930457,
      "grad_norm": 0.2492964267730713,
      "learning_rate": 0.0003,
      "loss": 4.9745,
      "step": 7835
    },
    {
      "epoch": 0.16054717097045032,
      "grad_norm": 0.25804537534713745,
      "learning_rate": 0.0003,
      "loss": 4.9777,
      "step": 7840
    },
    {
      "epoch": 0.16064956074785494,
      "grad_norm": 0.3088281750679016,
      "learning_rate": 0.0003,
      "loss": 4.9348,
      "step": 7845
    },
    {
      "epoch": 0.16075195052525956,
      "grad_norm": 0.2266150712966919,
      "learning_rate": 0.0003,
      "loss": 4.981,
      "step": 7850
    },
    {
      "epoch": 0.16085434030266418,
      "grad_norm": 0.23817439377307892,
      "learning_rate": 0.0003,
      "loss": 4.9638,
      "step": 7855
    },
    {
      "epoch": 0.1609567300800688,
      "grad_norm": 0.2314368039369583,
      "learning_rate": 0.0003,
      "loss": 4.9474,
      "step": 7860
    },
    {
      "epoch": 0.16105911985747343,
      "grad_norm": 0.22157548367977142,
      "learning_rate": 0.0003,
      "loss": 4.9844,
      "step": 7865
    },
    {
      "epoch": 0.16116150963487805,
      "grad_norm": 0.2731040418148041,
      "learning_rate": 0.0003,
      "loss": 4.9799,
      "step": 7870
    },
    {
      "epoch": 0.16126389941228267,
      "grad_norm": 0.2672834098339081,
      "learning_rate": 0.0003,
      "loss": 4.9617,
      "step": 7875
    },
    {
      "epoch": 0.1613662891896873,
      "grad_norm": 0.24145518243312836,
      "learning_rate": 0.0003,
      "loss": 4.9869,
      "step": 7880
    },
    {
      "epoch": 0.16146867896709194,
      "grad_norm": 0.33095917105674744,
      "learning_rate": 0.0003,
      "loss": 4.9788,
      "step": 7885
    },
    {
      "epoch": 0.16157106874449656,
      "grad_norm": 0.23778679966926575,
      "learning_rate": 0.0003,
      "loss": 4.9911,
      "step": 7890
    },
    {
      "epoch": 0.16167345852190118,
      "grad_norm": 0.2597041130065918,
      "learning_rate": 0.0003,
      "loss": 4.9643,
      "step": 7895
    },
    {
      "epoch": 0.1617758482993058,
      "grad_norm": 0.32607340812683105,
      "learning_rate": 0.0003,
      "loss": 5.0795,
      "step": 7900
    },
    {
      "epoch": 0.16187823807671042,
      "grad_norm": 0.29756247997283936,
      "learning_rate": 0.0003,
      "loss": 5.0274,
      "step": 7905
    },
    {
      "epoch": 0.16198062785411504,
      "grad_norm": 0.2715311646461487,
      "learning_rate": 0.0003,
      "loss": 4.9335,
      "step": 7910
    },
    {
      "epoch": 0.16208301763151967,
      "grad_norm": 0.27370917797088623,
      "learning_rate": 0.0003,
      "loss": 4.9757,
      "step": 7915
    },
    {
      "epoch": 0.1621854074089243,
      "grad_norm": 0.2792314887046814,
      "learning_rate": 0.0003,
      "loss": 4.9968,
      "step": 7920
    },
    {
      "epoch": 0.1622877971863289,
      "grad_norm": 0.23655901849269867,
      "learning_rate": 0.0003,
      "loss": 5.0076,
      "step": 7925
    },
    {
      "epoch": 0.16239018696373353,
      "grad_norm": 0.315487802028656,
      "learning_rate": 0.0003,
      "loss": 4.9748,
      "step": 7930
    },
    {
      "epoch": 0.16249257674113818,
      "grad_norm": 0.2521093189716339,
      "learning_rate": 0.0003,
      "loss": 4.9814,
      "step": 7935
    },
    {
      "epoch": 0.1625949665185428,
      "grad_norm": 0.2644865810871124,
      "learning_rate": 0.0003,
      "loss": 4.9929,
      "step": 7940
    },
    {
      "epoch": 0.16269735629594742,
      "grad_norm": 0.2822543680667877,
      "learning_rate": 0.0003,
      "loss": 4.9796,
      "step": 7945
    },
    {
      "epoch": 0.16279974607335204,
      "grad_norm": 0.2337653636932373,
      "learning_rate": 0.0003,
      "loss": 4.9272,
      "step": 7950
    },
    {
      "epoch": 0.16290213585075666,
      "grad_norm": 0.2425181120634079,
      "learning_rate": 0.0003,
      "loss": 5.0105,
      "step": 7955
    },
    {
      "epoch": 0.16300452562816128,
      "grad_norm": 0.24469076097011566,
      "learning_rate": 0.0003,
      "loss": 4.9679,
      "step": 7960
    },
    {
      "epoch": 0.1631069154055659,
      "grad_norm": 0.2601161301136017,
      "learning_rate": 0.0003,
      "loss": 4.8934,
      "step": 7965
    },
    {
      "epoch": 0.16320930518297053,
      "grad_norm": 0.3003526031970978,
      "learning_rate": 0.0003,
      "loss": 4.9514,
      "step": 7970
    },
    {
      "epoch": 0.16331169496037515,
      "grad_norm": 0.2524234652519226,
      "learning_rate": 0.0003,
      "loss": 4.9787,
      "step": 7975
    },
    {
      "epoch": 0.16341408473777977,
      "grad_norm": 0.2413032203912735,
      "learning_rate": 0.0003,
      "loss": 5.0099,
      "step": 7980
    },
    {
      "epoch": 0.1635164745151844,
      "grad_norm": 0.23179535567760468,
      "learning_rate": 0.0003,
      "loss": 5.0039,
      "step": 7985
    },
    {
      "epoch": 0.16361886429258904,
      "grad_norm": 0.24087181687355042,
      "learning_rate": 0.0003,
      "loss": 4.9662,
      "step": 7990
    },
    {
      "epoch": 0.16372125406999366,
      "grad_norm": 0.23221777379512787,
      "learning_rate": 0.0003,
      "loss": 4.9369,
      "step": 7995
    },
    {
      "epoch": 0.16382364384739828,
      "grad_norm": 0.21853451430797577,
      "learning_rate": 0.0003,
      "loss": 4.8567,
      "step": 8000
    },
    {
      "epoch": 0.16382364384739828,
      "eval_loss": 5.000431060791016,
      "eval_runtime": 17.2744,
      "eval_samples_per_second": 33.692,
      "eval_steps_per_second": 4.226,
      "step": 8000
    },
    {
      "epoch": 0.1639260336248029,
      "grad_norm": 0.23773954808712006,
      "learning_rate": 0.0003,
      "loss": 4.9538,
      "step": 8005
    },
    {
      "epoch": 0.16402842340220752,
      "grad_norm": 0.21447262167930603,
      "learning_rate": 0.0003,
      "loss": 4.9967,
      "step": 8010
    },
    {
      "epoch": 0.16413081317961215,
      "grad_norm": 0.23721234500408173,
      "learning_rate": 0.0003,
      "loss": 4.9341,
      "step": 8015
    },
    {
      "epoch": 0.16423320295701677,
      "grad_norm": 0.24699951708316803,
      "learning_rate": 0.0003,
      "loss": 4.9569,
      "step": 8020
    },
    {
      "epoch": 0.1643355927344214,
      "grad_norm": 0.2360193431377411,
      "learning_rate": 0.0003,
      "loss": 4.9537,
      "step": 8025
    },
    {
      "epoch": 0.164437982511826,
      "grad_norm": 0.24413712322711945,
      "learning_rate": 0.0003,
      "loss": 5.0083,
      "step": 8030
    },
    {
      "epoch": 0.16454037228923063,
      "grad_norm": 0.2634340822696686,
      "learning_rate": 0.0003,
      "loss": 5.0273,
      "step": 8035
    },
    {
      "epoch": 0.16464276206663528,
      "grad_norm": 0.2583164870738983,
      "learning_rate": 0.0003,
      "loss": 4.9409,
      "step": 8040
    },
    {
      "epoch": 0.1647451518440399,
      "grad_norm": 0.23808503150939941,
      "learning_rate": 0.0003,
      "loss": 4.9729,
      "step": 8045
    },
    {
      "epoch": 0.16484754162144452,
      "grad_norm": 0.2662254869937897,
      "learning_rate": 0.0003,
      "loss": 4.9806,
      "step": 8050
    },
    {
      "epoch": 0.16494993139884914,
      "grad_norm": 0.25939038395881653,
      "learning_rate": 0.0003,
      "loss": 4.9699,
      "step": 8055
    },
    {
      "epoch": 0.16505232117625376,
      "grad_norm": 0.23665112257003784,
      "learning_rate": 0.0003,
      "loss": 4.9592,
      "step": 8060
    },
    {
      "epoch": 0.16515471095365838,
      "grad_norm": 0.2312067747116089,
      "learning_rate": 0.0003,
      "loss": 4.9986,
      "step": 8065
    },
    {
      "epoch": 0.165257100731063,
      "grad_norm": 0.22978922724723816,
      "learning_rate": 0.0003,
      "loss": 4.9237,
      "step": 8070
    },
    {
      "epoch": 0.16535949050846763,
      "grad_norm": 0.22837577760219574,
      "learning_rate": 0.0003,
      "loss": 4.9564,
      "step": 8075
    },
    {
      "epoch": 0.16546188028587225,
      "grad_norm": 0.2294594943523407,
      "learning_rate": 0.0003,
      "loss": 4.9235,
      "step": 8080
    },
    {
      "epoch": 0.16556427006327687,
      "grad_norm": 0.25017088651657104,
      "learning_rate": 0.0003,
      "loss": 4.9978,
      "step": 8085
    },
    {
      "epoch": 0.16566665984068152,
      "grad_norm": 0.23632113635540009,
      "learning_rate": 0.0003,
      "loss": 4.9601,
      "step": 8090
    },
    {
      "epoch": 0.16576904961808614,
      "grad_norm": 0.24469229578971863,
      "learning_rate": 0.0003,
      "loss": 4.9547,
      "step": 8095
    },
    {
      "epoch": 0.16587143939549076,
      "grad_norm": 0.23987941443920135,
      "learning_rate": 0.0003,
      "loss": 4.9769,
      "step": 8100
    },
    {
      "epoch": 0.16597382917289538,
      "grad_norm": 0.2555577754974365,
      "learning_rate": 0.0003,
      "loss": 4.99,
      "step": 8105
    },
    {
      "epoch": 0.1660762189503,
      "grad_norm": 0.22785189747810364,
      "learning_rate": 0.0003,
      "loss": 4.9571,
      "step": 8110
    },
    {
      "epoch": 0.16617860872770462,
      "grad_norm": 0.22487638890743256,
      "learning_rate": 0.0003,
      "loss": 4.9281,
      "step": 8115
    },
    {
      "epoch": 0.16628099850510925,
      "grad_norm": 0.2585775554180145,
      "learning_rate": 0.0003,
      "loss": 4.9635,
      "step": 8120
    },
    {
      "epoch": 0.16638338828251387,
      "grad_norm": 0.24215564131736755,
      "learning_rate": 0.0003,
      "loss": 4.9886,
      "step": 8125
    },
    {
      "epoch": 0.1664857780599185,
      "grad_norm": 0.2455897480249405,
      "learning_rate": 0.0003,
      "loss": 5.0244,
      "step": 8130
    },
    {
      "epoch": 0.1665881678373231,
      "grad_norm": 0.2518976330757141,
      "learning_rate": 0.0003,
      "loss": 4.9591,
      "step": 8135
    },
    {
      "epoch": 0.16669055761472776,
      "grad_norm": 0.2537906765937805,
      "learning_rate": 0.0003,
      "loss": 5.0466,
      "step": 8140
    },
    {
      "epoch": 0.16679294739213238,
      "grad_norm": 0.24115991592407227,
      "learning_rate": 0.0003,
      "loss": 4.9882,
      "step": 8145
    },
    {
      "epoch": 0.166895337169537,
      "grad_norm": 0.243105947971344,
      "learning_rate": 0.0003,
      "loss": 4.9919,
      "step": 8150
    },
    {
      "epoch": 0.16699772694694162,
      "grad_norm": 0.24019856750965118,
      "learning_rate": 0.0003,
      "loss": 5.0229,
      "step": 8155
    },
    {
      "epoch": 0.16710011672434624,
      "grad_norm": 0.238954558968544,
      "learning_rate": 0.0003,
      "loss": 4.9798,
      "step": 8160
    },
    {
      "epoch": 0.16720250650175086,
      "grad_norm": 0.2641277611255646,
      "learning_rate": 0.0003,
      "loss": 4.9897,
      "step": 8165
    },
    {
      "epoch": 0.16730489627915549,
      "grad_norm": 0.24453116953372955,
      "learning_rate": 0.0003,
      "loss": 4.9824,
      "step": 8170
    },
    {
      "epoch": 0.1674072860565601,
      "grad_norm": 0.34915784001350403,
      "learning_rate": 0.0003,
      "loss": 4.928,
      "step": 8175
    },
    {
      "epoch": 0.16750967583396473,
      "grad_norm": 0.22643156349658966,
      "learning_rate": 0.0003,
      "loss": 4.985,
      "step": 8180
    },
    {
      "epoch": 0.16761206561136935,
      "grad_norm": 0.23364658653736115,
      "learning_rate": 0.0003,
      "loss": 4.9712,
      "step": 8185
    },
    {
      "epoch": 0.167714455388774,
      "grad_norm": 0.26778602600097656,
      "learning_rate": 0.0003,
      "loss": 4.9234,
      "step": 8190
    },
    {
      "epoch": 0.16781684516617862,
      "grad_norm": 0.2409345656633377,
      "learning_rate": 0.0003,
      "loss": 4.9569,
      "step": 8195
    },
    {
      "epoch": 0.16791923494358324,
      "grad_norm": 0.26513752341270447,
      "learning_rate": 0.0003,
      "loss": 4.9827,
      "step": 8200
    },
    {
      "epoch": 0.16802162472098786,
      "grad_norm": 0.24178951978683472,
      "learning_rate": 0.0003,
      "loss": 4.988,
      "step": 8205
    },
    {
      "epoch": 0.16812401449839248,
      "grad_norm": 0.26739561557769775,
      "learning_rate": 0.0003,
      "loss": 4.965,
      "step": 8210
    },
    {
      "epoch": 0.1682264042757971,
      "grad_norm": 0.2487669438123703,
      "learning_rate": 0.0003,
      "loss": 4.9987,
      "step": 8215
    },
    {
      "epoch": 0.16832879405320172,
      "grad_norm": 0.24459333717823029,
      "learning_rate": 0.0003,
      "loss": 4.9298,
      "step": 8220
    },
    {
      "epoch": 0.16843118383060635,
      "grad_norm": 0.24657560884952545,
      "learning_rate": 0.0003,
      "loss": 4.9931,
      "step": 8225
    },
    {
      "epoch": 0.16853357360801097,
      "grad_norm": 0.2773890495300293,
      "learning_rate": 0.0003,
      "loss": 4.9824,
      "step": 8230
    },
    {
      "epoch": 0.1686359633854156,
      "grad_norm": 0.24502816796302795,
      "learning_rate": 0.0003,
      "loss": 4.94,
      "step": 8235
    },
    {
      "epoch": 0.16873835316282024,
      "grad_norm": 0.25656858086586,
      "learning_rate": 0.0003,
      "loss": 4.973,
      "step": 8240
    },
    {
      "epoch": 0.16884074294022486,
      "grad_norm": 0.2743070125579834,
      "learning_rate": 0.0003,
      "loss": 4.9453,
      "step": 8245
    },
    {
      "epoch": 0.16894313271762948,
      "grad_norm": 0.232648566365242,
      "learning_rate": 0.0003,
      "loss": 4.9835,
      "step": 8250
    },
    {
      "epoch": 0.1690455224950341,
      "grad_norm": 0.2495628297328949,
      "learning_rate": 0.0003,
      "loss": 4.9835,
      "step": 8255
    },
    {
      "epoch": 0.16914791227243872,
      "grad_norm": 0.24052630364894867,
      "learning_rate": 0.0003,
      "loss": 4.9369,
      "step": 8260
    },
    {
      "epoch": 0.16925030204984334,
      "grad_norm": 0.23074476420879364,
      "learning_rate": 0.0003,
      "loss": 4.9835,
      "step": 8265
    },
    {
      "epoch": 0.16935269182724796,
      "grad_norm": 0.21705974638462067,
      "learning_rate": 0.0003,
      "loss": 4.9878,
      "step": 8270
    },
    {
      "epoch": 0.16945508160465259,
      "grad_norm": 0.25947219133377075,
      "learning_rate": 0.0003,
      "loss": 4.9692,
      "step": 8275
    },
    {
      "epoch": 0.1695574713820572,
      "grad_norm": 0.23706823587417603,
      "learning_rate": 0.0003,
      "loss": 4.9927,
      "step": 8280
    },
    {
      "epoch": 0.16965986115946183,
      "grad_norm": 0.23366497457027435,
      "learning_rate": 0.0003,
      "loss": 4.9778,
      "step": 8285
    },
    {
      "epoch": 0.16976225093686648,
      "grad_norm": 0.23043937981128693,
      "learning_rate": 0.0003,
      "loss": 4.9375,
      "step": 8290
    },
    {
      "epoch": 0.1698646407142711,
      "grad_norm": 0.22917410731315613,
      "learning_rate": 0.0003,
      "loss": 4.9518,
      "step": 8295
    },
    {
      "epoch": 0.16996703049167572,
      "grad_norm": 0.23943403363227844,
      "learning_rate": 0.0003,
      "loss": 4.9337,
      "step": 8300
    },
    {
      "epoch": 0.17006942026908034,
      "grad_norm": 0.24286477267742157,
      "learning_rate": 0.0003,
      "loss": 5.0025,
      "step": 8305
    },
    {
      "epoch": 0.17017181004648496,
      "grad_norm": 0.23036329448223114,
      "learning_rate": 0.0003,
      "loss": 4.9614,
      "step": 8310
    },
    {
      "epoch": 0.17027419982388958,
      "grad_norm": 0.2531217038631439,
      "learning_rate": 0.0003,
      "loss": 4.9789,
      "step": 8315
    },
    {
      "epoch": 0.1703765896012942,
      "grad_norm": 0.2279614508152008,
      "learning_rate": 0.0003,
      "loss": 4.9902,
      "step": 8320
    },
    {
      "epoch": 0.17047897937869883,
      "grad_norm": 0.2787529528141022,
      "learning_rate": 0.0003,
      "loss": 4.9625,
      "step": 8325
    },
    {
      "epoch": 0.17058136915610345,
      "grad_norm": 0.23746182024478912,
      "learning_rate": 0.0003,
      "loss": 4.9784,
      "step": 8330
    },
    {
      "epoch": 0.17068375893350807,
      "grad_norm": 0.24724778532981873,
      "learning_rate": 0.0003,
      "loss": 4.9431,
      "step": 8335
    },
    {
      "epoch": 0.17078614871091272,
      "grad_norm": 0.23509597778320312,
      "learning_rate": 0.0003,
      "loss": 4.9764,
      "step": 8340
    },
    {
      "epoch": 0.17088853848831734,
      "grad_norm": 0.23609371483325958,
      "learning_rate": 0.0003,
      "loss": 4.9762,
      "step": 8345
    },
    {
      "epoch": 0.17099092826572196,
      "grad_norm": 0.23974719643592834,
      "learning_rate": 0.0003,
      "loss": 4.9617,
      "step": 8350
    },
    {
      "epoch": 0.17109331804312658,
      "grad_norm": 0.22124601900577545,
      "learning_rate": 0.0003,
      "loss": 4.9675,
      "step": 8355
    },
    {
      "epoch": 0.1711957078205312,
      "grad_norm": 0.2362033724784851,
      "learning_rate": 0.0003,
      "loss": 4.999,
      "step": 8360
    },
    {
      "epoch": 0.17129809759793582,
      "grad_norm": 0.2428901493549347,
      "learning_rate": 0.0003,
      "loss": 5.0014,
      "step": 8365
    },
    {
      "epoch": 0.17140048737534044,
      "grad_norm": 0.24708890914916992,
      "learning_rate": 0.0003,
      "loss": 4.9538,
      "step": 8370
    },
    {
      "epoch": 0.17150287715274506,
      "grad_norm": 0.24944816529750824,
      "learning_rate": 0.0003,
      "loss": 4.9225,
      "step": 8375
    },
    {
      "epoch": 0.17160526693014969,
      "grad_norm": 0.2560562789440155,
      "learning_rate": 0.0003,
      "loss": 4.9858,
      "step": 8380
    },
    {
      "epoch": 0.1717076567075543,
      "grad_norm": 0.26690977811813354,
      "learning_rate": 0.0003,
      "loss": 4.9503,
      "step": 8385
    },
    {
      "epoch": 0.17181004648495893,
      "grad_norm": 0.24184927344322205,
      "learning_rate": 0.0003,
      "loss": 4.9364,
      "step": 8390
    },
    {
      "epoch": 0.17191243626236358,
      "grad_norm": 0.24029432237148285,
      "learning_rate": 0.0003,
      "loss": 4.9838,
      "step": 8395
    },
    {
      "epoch": 0.1720148260397682,
      "grad_norm": 0.2464255541563034,
      "learning_rate": 0.0003,
      "loss": 4.979,
      "step": 8400
    },
    {
      "epoch": 0.17211721581717282,
      "grad_norm": 0.22707878053188324,
      "learning_rate": 0.0003,
      "loss": 4.9349,
      "step": 8405
    },
    {
      "epoch": 0.17221960559457744,
      "grad_norm": 0.23702223598957062,
      "learning_rate": 0.0003,
      "loss": 4.9087,
      "step": 8410
    },
    {
      "epoch": 0.17232199537198206,
      "grad_norm": 0.2341252565383911,
      "learning_rate": 0.0003,
      "loss": 4.9247,
      "step": 8415
    },
    {
      "epoch": 0.17242438514938668,
      "grad_norm": 0.2504211366176605,
      "learning_rate": 0.0003,
      "loss": 4.9337,
      "step": 8420
    },
    {
      "epoch": 0.1725267749267913,
      "grad_norm": 0.25000959634780884,
      "learning_rate": 0.0003,
      "loss": 4.9552,
      "step": 8425
    },
    {
      "epoch": 0.17262916470419593,
      "grad_norm": 0.2468976229429245,
      "learning_rate": 0.0003,
      "loss": 4.9799,
      "step": 8430
    },
    {
      "epoch": 0.17273155448160055,
      "grad_norm": 0.2432153970003128,
      "learning_rate": 0.0003,
      "loss": 4.9608,
      "step": 8435
    },
    {
      "epoch": 0.17283394425900517,
      "grad_norm": 0.24488693475723267,
      "learning_rate": 0.0003,
      "loss": 4.9715,
      "step": 8440
    },
    {
      "epoch": 0.17293633403640982,
      "grad_norm": 0.26678532361984253,
      "learning_rate": 0.0003,
      "loss": 4.9634,
      "step": 8445
    },
    {
      "epoch": 0.17303872381381444,
      "grad_norm": 0.3612540066242218,
      "learning_rate": 0.0003,
      "loss": 4.9252,
      "step": 8450
    },
    {
      "epoch": 0.17314111359121906,
      "grad_norm": 0.258616179227829,
      "learning_rate": 0.0003,
      "loss": 4.951,
      "step": 8455
    },
    {
      "epoch": 0.17324350336862368,
      "grad_norm": 0.2403879016637802,
      "learning_rate": 0.0003,
      "loss": 4.9513,
      "step": 8460
    },
    {
      "epoch": 0.1733458931460283,
      "grad_norm": 0.2637866139411926,
      "learning_rate": 0.0003,
      "loss": 4.9427,
      "step": 8465
    },
    {
      "epoch": 0.17344828292343292,
      "grad_norm": 0.4013460576534271,
      "learning_rate": 0.0003,
      "loss": 4.9076,
      "step": 8470
    },
    {
      "epoch": 0.17355067270083754,
      "grad_norm": 0.2604331970214844,
      "learning_rate": 0.0003,
      "loss": 4.968,
      "step": 8475
    },
    {
      "epoch": 0.17365306247824217,
      "grad_norm": 0.2690982222557068,
      "learning_rate": 0.0003,
      "loss": 4.9317,
      "step": 8480
    },
    {
      "epoch": 0.1737554522556468,
      "grad_norm": 0.23837333917617798,
      "learning_rate": 0.0003,
      "loss": 5.0024,
      "step": 8485
    },
    {
      "epoch": 0.1738578420330514,
      "grad_norm": 0.24582943320274353,
      "learning_rate": 0.0003,
      "loss": 4.9811,
      "step": 8490
    },
    {
      "epoch": 0.17396023181045606,
      "grad_norm": 0.2452264428138733,
      "learning_rate": 0.0003,
      "loss": 4.961,
      "step": 8495
    },
    {
      "epoch": 0.17406262158786068,
      "grad_norm": 0.28215664625167847,
      "learning_rate": 0.0003,
      "loss": 5.0268,
      "step": 8500
    },
    {
      "epoch": 0.1741650113652653,
      "grad_norm": 0.2856769859790802,
      "learning_rate": 0.0003,
      "loss": 4.9394,
      "step": 8505
    },
    {
      "epoch": 0.17426740114266992,
      "grad_norm": 1.280678391456604,
      "learning_rate": 0.0003,
      "loss": 4.9656,
      "step": 8510
    },
    {
      "epoch": 0.17436979092007454,
      "grad_norm": 0.30562543869018555,
      "learning_rate": 0.0003,
      "loss": 5.0055,
      "step": 8515
    },
    {
      "epoch": 0.17447218069747916,
      "grad_norm": 0.29330888390541077,
      "learning_rate": 0.0003,
      "loss": 4.9697,
      "step": 8520
    },
    {
      "epoch": 0.17457457047488378,
      "grad_norm": 0.2480739951133728,
      "learning_rate": 0.0003,
      "loss": 4.9607,
      "step": 8525
    },
    {
      "epoch": 0.1746769602522884,
      "grad_norm": 0.24799346923828125,
      "learning_rate": 0.0003,
      "loss": 5.0038,
      "step": 8530
    },
    {
      "epoch": 0.17477935002969303,
      "grad_norm": 0.2737793028354645,
      "learning_rate": 0.0003,
      "loss": 4.9091,
      "step": 8535
    },
    {
      "epoch": 0.17488173980709765,
      "grad_norm": 0.25553327798843384,
      "learning_rate": 0.0003,
      "loss": 5.0308,
      "step": 8540
    },
    {
      "epoch": 0.1749841295845023,
      "grad_norm": 0.23692303895950317,
      "learning_rate": 0.0003,
      "loss": 4.9652,
      "step": 8545
    },
    {
      "epoch": 0.17508651936190692,
      "grad_norm": 0.30589890480041504,
      "learning_rate": 0.0003,
      "loss": 4.9889,
      "step": 8550
    },
    {
      "epoch": 0.17518890913931154,
      "grad_norm": 0.243758425116539,
      "learning_rate": 0.0003,
      "loss": 5.0105,
      "step": 8555
    },
    {
      "epoch": 0.17529129891671616,
      "grad_norm": 0.22573265433311462,
      "learning_rate": 0.0003,
      "loss": 4.9811,
      "step": 8560
    },
    {
      "epoch": 0.17539368869412078,
      "grad_norm": 0.23498474061489105,
      "learning_rate": 0.0003,
      "loss": 5.009,
      "step": 8565
    },
    {
      "epoch": 0.1754960784715254,
      "grad_norm": 0.2366688996553421,
      "learning_rate": 0.0003,
      "loss": 4.952,
      "step": 8570
    },
    {
      "epoch": 0.17559846824893002,
      "grad_norm": 0.2420990914106369,
      "learning_rate": 0.0003,
      "loss": 4.9078,
      "step": 8575
    },
    {
      "epoch": 0.17570085802633464,
      "grad_norm": 0.25633060932159424,
      "learning_rate": 0.0003,
      "loss": 4.9264,
      "step": 8580
    },
    {
      "epoch": 0.17580324780373927,
      "grad_norm": 0.630459189414978,
      "learning_rate": 0.0003,
      "loss": 4.987,
      "step": 8585
    },
    {
      "epoch": 0.1759056375811439,
      "grad_norm": 0.2781355381011963,
      "learning_rate": 0.0003,
      "loss": 4.9808,
      "step": 8590
    },
    {
      "epoch": 0.17600802735854854,
      "grad_norm": 0.2405020296573639,
      "learning_rate": 0.0003,
      "loss": 4.9342,
      "step": 8595
    },
    {
      "epoch": 0.17611041713595316,
      "grad_norm": 0.2748335003852844,
      "learning_rate": 0.0003,
      "loss": 4.99,
      "step": 8600
    },
    {
      "epoch": 0.17621280691335778,
      "grad_norm": 0.2357555776834488,
      "learning_rate": 0.0003,
      "loss": 4.9803,
      "step": 8605
    },
    {
      "epoch": 0.1763151966907624,
      "grad_norm": 0.2591405212879181,
      "learning_rate": 0.0003,
      "loss": 4.9526,
      "step": 8610
    },
    {
      "epoch": 0.17641758646816702,
      "grad_norm": 0.2714153826236725,
      "learning_rate": 0.0003,
      "loss": 4.9511,
      "step": 8615
    },
    {
      "epoch": 0.17651997624557164,
      "grad_norm": 0.2358967512845993,
      "learning_rate": 0.0003,
      "loss": 4.9982,
      "step": 8620
    },
    {
      "epoch": 0.17662236602297626,
      "grad_norm": 0.25634580850601196,
      "learning_rate": 0.0003,
      "loss": 4.9297,
      "step": 8625
    },
    {
      "epoch": 0.17672475580038088,
      "grad_norm": 0.24031628668308258,
      "learning_rate": 0.0003,
      "loss": 4.9554,
      "step": 8630
    },
    {
      "epoch": 0.1768271455777855,
      "grad_norm": 0.22725112736225128,
      "learning_rate": 0.0003,
      "loss": 4.9684,
      "step": 8635
    },
    {
      "epoch": 0.17692953535519013,
      "grad_norm": 0.23442846536636353,
      "learning_rate": 0.0003,
      "loss": 4.9531,
      "step": 8640
    },
    {
      "epoch": 0.17703192513259478,
      "grad_norm": 0.23823332786560059,
      "learning_rate": 0.0003,
      "loss": 4.9727,
      "step": 8645
    },
    {
      "epoch": 0.1771343149099994,
      "grad_norm": 0.2390361726284027,
      "learning_rate": 0.0003,
      "loss": 4.9173,
      "step": 8650
    },
    {
      "epoch": 0.17723670468740402,
      "grad_norm": 0.25551703572273254,
      "learning_rate": 0.0003,
      "loss": 5.0018,
      "step": 8655
    },
    {
      "epoch": 0.17733909446480864,
      "grad_norm": 0.2616801857948303,
      "learning_rate": 0.0003,
      "loss": 4.916,
      "step": 8660
    },
    {
      "epoch": 0.17744148424221326,
      "grad_norm": 0.2571796178817749,
      "learning_rate": 0.0003,
      "loss": 4.9279,
      "step": 8665
    },
    {
      "epoch": 0.17754387401961788,
      "grad_norm": 0.231767475605011,
      "learning_rate": 0.0003,
      "loss": 4.9712,
      "step": 8670
    },
    {
      "epoch": 0.1776462637970225,
      "grad_norm": 0.27216964960098267,
      "learning_rate": 0.0003,
      "loss": 4.9273,
      "step": 8675
    },
    {
      "epoch": 0.17774865357442712,
      "grad_norm": 0.22776266932487488,
      "learning_rate": 0.0003,
      "loss": 4.9075,
      "step": 8680
    },
    {
      "epoch": 0.17785104335183174,
      "grad_norm": 0.27528294920921326,
      "learning_rate": 0.0003,
      "loss": 4.8684,
      "step": 8685
    },
    {
      "epoch": 0.17795343312923637,
      "grad_norm": 0.24167364835739136,
      "learning_rate": 0.0003,
      "loss": 4.9285,
      "step": 8690
    },
    {
      "epoch": 0.17805582290664101,
      "grad_norm": 0.28756558895111084,
      "learning_rate": 0.0003,
      "loss": 4.9294,
      "step": 8695
    },
    {
      "epoch": 0.17815821268404564,
      "grad_norm": 0.27697524428367615,
      "learning_rate": 0.0003,
      "loss": 4.9078,
      "step": 8700
    },
    {
      "epoch": 0.17826060246145026,
      "grad_norm": 0.4069802165031433,
      "learning_rate": 0.0003,
      "loss": 4.9637,
      "step": 8705
    },
    {
      "epoch": 0.17836299223885488,
      "grad_norm": 0.25043371319770813,
      "learning_rate": 0.0003,
      "loss": 4.9357,
      "step": 8710
    },
    {
      "epoch": 0.1784653820162595,
      "grad_norm": 0.2558344304561615,
      "learning_rate": 0.0003,
      "loss": 5.004,
      "step": 8715
    },
    {
      "epoch": 0.17856777179366412,
      "grad_norm": 0.26395145058631897,
      "learning_rate": 0.0003,
      "loss": 4.9534,
      "step": 8720
    },
    {
      "epoch": 0.17867016157106874,
      "grad_norm": 0.23396216332912445,
      "learning_rate": 0.0003,
      "loss": 4.9541,
      "step": 8725
    },
    {
      "epoch": 0.17877255134847336,
      "grad_norm": 0.22810359299182892,
      "learning_rate": 0.0003,
      "loss": 4.9797,
      "step": 8730
    },
    {
      "epoch": 0.17887494112587798,
      "grad_norm": 0.24070174992084503,
      "learning_rate": 0.0003,
      "loss": 4.9558,
      "step": 8735
    },
    {
      "epoch": 0.1789773309032826,
      "grad_norm": 0.2460024207830429,
      "learning_rate": 0.0003,
      "loss": 4.9305,
      "step": 8740
    },
    {
      "epoch": 0.17907972068068723,
      "grad_norm": 0.2506413459777832,
      "learning_rate": 0.0003,
      "loss": 4.9893,
      "step": 8745
    },
    {
      "epoch": 0.17918211045809188,
      "grad_norm": 0.2459816187620163,
      "learning_rate": 0.0003,
      "loss": 5.0012,
      "step": 8750
    },
    {
      "epoch": 0.1792845002354965,
      "grad_norm": 0.2408568561077118,
      "learning_rate": 0.0003,
      "loss": 4.9595,
      "step": 8755
    },
    {
      "epoch": 0.17938689001290112,
      "grad_norm": 0.23039047420024872,
      "learning_rate": 0.0003,
      "loss": 4.976,
      "step": 8760
    },
    {
      "epoch": 0.17948927979030574,
      "grad_norm": 0.23503048717975616,
      "learning_rate": 0.0003,
      "loss": 4.9216,
      "step": 8765
    },
    {
      "epoch": 0.17959166956771036,
      "grad_norm": 0.23038825392723083,
      "learning_rate": 0.0003,
      "loss": 4.9127,
      "step": 8770
    },
    {
      "epoch": 0.17969405934511498,
      "grad_norm": 0.22313566505908966,
      "learning_rate": 0.0003,
      "loss": 4.9082,
      "step": 8775
    },
    {
      "epoch": 0.1797964491225196,
      "grad_norm": 0.23721888661384583,
      "learning_rate": 0.0003,
      "loss": 4.9445,
      "step": 8780
    },
    {
      "epoch": 0.17989883889992422,
      "grad_norm": 0.2366582602262497,
      "learning_rate": 0.0003,
      "loss": 4.9145,
      "step": 8785
    },
    {
      "epoch": 0.18000122867732885,
      "grad_norm": 0.24238401651382446,
      "learning_rate": 0.0003,
      "loss": 4.9726,
      "step": 8790
    },
    {
      "epoch": 0.18010361845473347,
      "grad_norm": 0.27836301922798157,
      "learning_rate": 0.0003,
      "loss": 4.9378,
      "step": 8795
    },
    {
      "epoch": 0.18020600823213812,
      "grad_norm": 0.3087904453277588,
      "learning_rate": 0.0003,
      "loss": 4.947,
      "step": 8800
    },
    {
      "epoch": 0.18030839800954274,
      "grad_norm": 0.2912931740283966,
      "learning_rate": 0.0003,
      "loss": 4.9846,
      "step": 8805
    },
    {
      "epoch": 0.18041078778694736,
      "grad_norm": 0.2829636037349701,
      "learning_rate": 0.0003,
      "loss": 4.9376,
      "step": 8810
    },
    {
      "epoch": 0.18051317756435198,
      "grad_norm": 0.24942532181739807,
      "learning_rate": 0.0003,
      "loss": 4.9389,
      "step": 8815
    },
    {
      "epoch": 0.1806155673417566,
      "grad_norm": 0.24345135688781738,
      "learning_rate": 0.0003,
      "loss": 4.9813,
      "step": 8820
    },
    {
      "epoch": 0.18071795711916122,
      "grad_norm": 0.24752293527126312,
      "learning_rate": 0.0003,
      "loss": 4.9714,
      "step": 8825
    },
    {
      "epoch": 0.18082034689656584,
      "grad_norm": 0.22899101674556732,
      "learning_rate": 0.0003,
      "loss": 4.9175,
      "step": 8830
    },
    {
      "epoch": 0.18092273667397046,
      "grad_norm": 0.23931820690631866,
      "learning_rate": 0.0003,
      "loss": 4.9516,
      "step": 8835
    },
    {
      "epoch": 0.18102512645137508,
      "grad_norm": 0.2398419827222824,
      "learning_rate": 0.0003,
      "loss": 4.9473,
      "step": 8840
    },
    {
      "epoch": 0.1811275162287797,
      "grad_norm": 0.31384849548339844,
      "learning_rate": 0.0003,
      "loss": 4.9467,
      "step": 8845
    },
    {
      "epoch": 0.18122990600618435,
      "grad_norm": 0.2399289309978485,
      "learning_rate": 0.0003,
      "loss": 4.9555,
      "step": 8850
    },
    {
      "epoch": 0.18133229578358898,
      "grad_norm": 0.23261117935180664,
      "learning_rate": 0.0003,
      "loss": 4.9576,
      "step": 8855
    },
    {
      "epoch": 0.1814346855609936,
      "grad_norm": 0.23390693962574005,
      "learning_rate": 0.0003,
      "loss": 4.95,
      "step": 8860
    },
    {
      "epoch": 0.18153707533839822,
      "grad_norm": 0.2741168439388275,
      "learning_rate": 0.0003,
      "loss": 4.9439,
      "step": 8865
    },
    {
      "epoch": 0.18163946511580284,
      "grad_norm": 0.2826792895793915,
      "learning_rate": 0.0003,
      "loss": 4.9285,
      "step": 8870
    },
    {
      "epoch": 0.18174185489320746,
      "grad_norm": 0.24303020536899567,
      "learning_rate": 0.0003,
      "loss": 4.9537,
      "step": 8875
    },
    {
      "epoch": 0.18184424467061208,
      "grad_norm": 0.23413772881031036,
      "learning_rate": 0.0003,
      "loss": 4.9463,
      "step": 8880
    },
    {
      "epoch": 0.1819466344480167,
      "grad_norm": 0.25310632586479187,
      "learning_rate": 0.0003,
      "loss": 4.9643,
      "step": 8885
    },
    {
      "epoch": 0.18204902422542132,
      "grad_norm": 0.25220412015914917,
      "learning_rate": 0.0003,
      "loss": 4.9889,
      "step": 8890
    },
    {
      "epoch": 0.18215141400282595,
      "grad_norm": 0.23816744983196259,
      "learning_rate": 0.0003,
      "loss": 4.9433,
      "step": 8895
    },
    {
      "epoch": 0.1822538037802306,
      "grad_norm": 0.24224130809307098,
      "learning_rate": 0.0003,
      "loss": 5.0065,
      "step": 8900
    },
    {
      "epoch": 0.18235619355763522,
      "grad_norm": 0.23748779296875,
      "learning_rate": 0.0003,
      "loss": 4.9892,
      "step": 8905
    },
    {
      "epoch": 0.18245858333503984,
      "grad_norm": 0.23334605991840363,
      "learning_rate": 0.0003,
      "loss": 4.9289,
      "step": 8910
    },
    {
      "epoch": 0.18256097311244446,
      "grad_norm": 0.2519617974758148,
      "learning_rate": 0.0003,
      "loss": 4.8981,
      "step": 8915
    },
    {
      "epoch": 0.18266336288984908,
      "grad_norm": 0.2552023231983185,
      "learning_rate": 0.0003,
      "loss": 4.974,
      "step": 8920
    },
    {
      "epoch": 0.1827657526672537,
      "grad_norm": 0.24951639771461487,
      "learning_rate": 0.0003,
      "loss": 4.9376,
      "step": 8925
    },
    {
      "epoch": 0.18286814244465832,
      "grad_norm": 0.38250601291656494,
      "learning_rate": 0.0003,
      "loss": 4.9658,
      "step": 8930
    },
    {
      "epoch": 0.18297053222206294,
      "grad_norm": 0.26618579030036926,
      "learning_rate": 0.0003,
      "loss": 4.9351,
      "step": 8935
    },
    {
      "epoch": 0.18307292199946756,
      "grad_norm": 0.25149083137512207,
      "learning_rate": 0.0003,
      "loss": 4.9782,
      "step": 8940
    },
    {
      "epoch": 0.18317531177687219,
      "grad_norm": 0.25862792134284973,
      "learning_rate": 0.0003,
      "loss": 4.9518,
      "step": 8945
    },
    {
      "epoch": 0.18327770155427683,
      "grad_norm": 0.23274236917495728,
      "learning_rate": 0.0003,
      "loss": 4.9542,
      "step": 8950
    },
    {
      "epoch": 0.18338009133168146,
      "grad_norm": 0.23445458710193634,
      "learning_rate": 0.0003,
      "loss": 4.9482,
      "step": 8955
    },
    {
      "epoch": 0.18348248110908608,
      "grad_norm": 0.2242894023656845,
      "learning_rate": 0.0003,
      "loss": 4.9599,
      "step": 8960
    },
    {
      "epoch": 0.1835848708864907,
      "grad_norm": 0.243010476231575,
      "learning_rate": 0.0003,
      "loss": 4.9959,
      "step": 8965
    },
    {
      "epoch": 0.18368726066389532,
      "grad_norm": 0.2701684534549713,
      "learning_rate": 0.0003,
      "loss": 4.9371,
      "step": 8970
    },
    {
      "epoch": 0.18378965044129994,
      "grad_norm": 0.2610296308994293,
      "learning_rate": 0.0003,
      "loss": 4.9253,
      "step": 8975
    },
    {
      "epoch": 0.18389204021870456,
      "grad_norm": 0.24804575741291046,
      "learning_rate": 0.0003,
      "loss": 4.8889,
      "step": 8980
    },
    {
      "epoch": 0.18399442999610918,
      "grad_norm": 0.23091092705726624,
      "learning_rate": 0.0003,
      "loss": 4.9377,
      "step": 8985
    },
    {
      "epoch": 0.1840968197735138,
      "grad_norm": 0.2444520741701126,
      "learning_rate": 0.0003,
      "loss": 4.9067,
      "step": 8990
    },
    {
      "epoch": 0.18419920955091842,
      "grad_norm": 0.2413340061903,
      "learning_rate": 0.0003,
      "loss": 4.9861,
      "step": 8995
    },
    {
      "epoch": 0.18430159932832307,
      "grad_norm": 0.23155519366264343,
      "learning_rate": 0.0003,
      "loss": 4.9702,
      "step": 9000
    },
    {
      "epoch": 0.18430159932832307,
      "eval_loss": 4.967161178588867,
      "eval_runtime": 17.2382,
      "eval_samples_per_second": 33.762,
      "eval_steps_per_second": 4.235,
      "step": 9000
    },
    {
      "epoch": 0.1844039891057277,
      "grad_norm": 0.26229947805404663,
      "learning_rate": 0.0003,
      "loss": 4.9275,
      "step": 9005
    },
    {
      "epoch": 0.18450637888313232,
      "grad_norm": 0.2596692144870758,
      "learning_rate": 0.0003,
      "loss": 4.9474,
      "step": 9010
    },
    {
      "epoch": 0.18460876866053694,
      "grad_norm": 0.24281714856624603,
      "learning_rate": 0.0003,
      "loss": 4.9183,
      "step": 9015
    },
    {
      "epoch": 0.18471115843794156,
      "grad_norm": 0.26011863350868225,
      "learning_rate": 0.0003,
      "loss": 4.9363,
      "step": 9020
    },
    {
      "epoch": 0.18481354821534618,
      "grad_norm": 0.24929960072040558,
      "learning_rate": 0.0003,
      "loss": 4.9551,
      "step": 9025
    },
    {
      "epoch": 0.1849159379927508,
      "grad_norm": 0.25172069668769836,
      "learning_rate": 0.0003,
      "loss": 4.9257,
      "step": 9030
    },
    {
      "epoch": 0.18501832777015542,
      "grad_norm": 0.2448473423719406,
      "learning_rate": 0.0003,
      "loss": 4.9615,
      "step": 9035
    },
    {
      "epoch": 0.18512071754756004,
      "grad_norm": 0.24837471544742584,
      "learning_rate": 0.0003,
      "loss": 4.9719,
      "step": 9040
    },
    {
      "epoch": 0.18522310732496466,
      "grad_norm": 0.23794622719287872,
      "learning_rate": 0.0003,
      "loss": 4.947,
      "step": 9045
    },
    {
      "epoch": 0.1853254971023693,
      "grad_norm": 0.33640262484550476,
      "learning_rate": 0.0003,
      "loss": 4.9217,
      "step": 9050
    },
    {
      "epoch": 0.18542788687977393,
      "grad_norm": 0.2828204929828644,
      "learning_rate": 0.0003,
      "loss": 4.9556,
      "step": 9055
    },
    {
      "epoch": 0.18553027665717856,
      "grad_norm": 0.2490418553352356,
      "learning_rate": 0.0003,
      "loss": 4.9401,
      "step": 9060
    },
    {
      "epoch": 0.18563266643458318,
      "grad_norm": 0.24733878672122955,
      "learning_rate": 0.0003,
      "loss": 4.9404,
      "step": 9065
    },
    {
      "epoch": 0.1857350562119878,
      "grad_norm": 0.2425927221775055,
      "learning_rate": 0.0003,
      "loss": 4.9148,
      "step": 9070
    },
    {
      "epoch": 0.18583744598939242,
      "grad_norm": 0.2517198920249939,
      "learning_rate": 0.0003,
      "loss": 4.9049,
      "step": 9075
    },
    {
      "epoch": 0.18593983576679704,
      "grad_norm": 0.2480369359254837,
      "learning_rate": 0.0003,
      "loss": 4.9222,
      "step": 9080
    },
    {
      "epoch": 0.18604222554420166,
      "grad_norm": 0.2834649980068207,
      "learning_rate": 0.0003,
      "loss": 4.9167,
      "step": 9085
    },
    {
      "epoch": 0.18614461532160628,
      "grad_norm": 0.21334508061408997,
      "learning_rate": 0.0003,
      "loss": 4.9144,
      "step": 9090
    },
    {
      "epoch": 0.1862470050990109,
      "grad_norm": 0.24133138358592987,
      "learning_rate": 0.0003,
      "loss": 4.9551,
      "step": 9095
    },
    {
      "epoch": 0.18634939487641553,
      "grad_norm": 0.2861619293689728,
      "learning_rate": 0.0003,
      "loss": 4.9004,
      "step": 9100
    },
    {
      "epoch": 0.18645178465382017,
      "grad_norm": 0.25369203090667725,
      "learning_rate": 0.0003,
      "loss": 4.9408,
      "step": 9105
    },
    {
      "epoch": 0.1865541744312248,
      "grad_norm": 0.24333781003952026,
      "learning_rate": 0.0003,
      "loss": 4.915,
      "step": 9110
    },
    {
      "epoch": 0.18665656420862942,
      "grad_norm": 0.23844999074935913,
      "learning_rate": 0.0003,
      "loss": 4.9096,
      "step": 9115
    },
    {
      "epoch": 0.18675895398603404,
      "grad_norm": 0.2520522177219391,
      "learning_rate": 0.0003,
      "loss": 4.9019,
      "step": 9120
    },
    {
      "epoch": 0.18686134376343866,
      "grad_norm": 0.25806185603141785,
      "learning_rate": 0.0003,
      "loss": 4.9489,
      "step": 9125
    },
    {
      "epoch": 0.18696373354084328,
      "grad_norm": 0.2639296352863312,
      "learning_rate": 0.0003,
      "loss": 4.9078,
      "step": 9130
    },
    {
      "epoch": 0.1870661233182479,
      "grad_norm": 0.25021064281463623,
      "learning_rate": 0.0003,
      "loss": 4.947,
      "step": 9135
    },
    {
      "epoch": 0.18716851309565252,
      "grad_norm": 0.25453871488571167,
      "learning_rate": 0.0003,
      "loss": 4.8956,
      "step": 9140
    },
    {
      "epoch": 0.18727090287305714,
      "grad_norm": 0.2519582211971283,
      "learning_rate": 0.0003,
      "loss": 4.9217,
      "step": 9145
    },
    {
      "epoch": 0.18737329265046176,
      "grad_norm": 0.2347276508808136,
      "learning_rate": 0.0003,
      "loss": 4.9446,
      "step": 9150
    },
    {
      "epoch": 0.1874756824278664,
      "grad_norm": 0.22869284451007843,
      "learning_rate": 0.0003,
      "loss": 4.9314,
      "step": 9155
    },
    {
      "epoch": 0.18757807220527103,
      "grad_norm": 0.23401141166687012,
      "learning_rate": 0.0003,
      "loss": 4.9272,
      "step": 9160
    },
    {
      "epoch": 0.18768046198267566,
      "grad_norm": 0.26685070991516113,
      "learning_rate": 0.0003,
      "loss": 4.9796,
      "step": 9165
    },
    {
      "epoch": 0.18778285176008028,
      "grad_norm": 0.2689502239227295,
      "learning_rate": 0.0003,
      "loss": 4.9418,
      "step": 9170
    },
    {
      "epoch": 0.1878852415374849,
      "grad_norm": 0.40141159296035767,
      "learning_rate": 0.0003,
      "loss": 4.9608,
      "step": 9175
    },
    {
      "epoch": 0.18798763131488952,
      "grad_norm": 0.24301853775978088,
      "learning_rate": 0.0003,
      "loss": 4.9604,
      "step": 9180
    },
    {
      "epoch": 0.18809002109229414,
      "grad_norm": 0.24619057774543762,
      "learning_rate": 0.0003,
      "loss": 4.9413,
      "step": 9185
    },
    {
      "epoch": 0.18819241086969876,
      "grad_norm": 0.23454906046390533,
      "learning_rate": 0.0003,
      "loss": 4.9371,
      "step": 9190
    },
    {
      "epoch": 0.18829480064710338,
      "grad_norm": 0.26107001304626465,
      "learning_rate": 0.0003,
      "loss": 4.9023,
      "step": 9195
    },
    {
      "epoch": 0.188397190424508,
      "grad_norm": 0.2492724508047104,
      "learning_rate": 0.0003,
      "loss": 4.9876,
      "step": 9200
    },
    {
      "epoch": 0.18849958020191265,
      "grad_norm": 0.27773427963256836,
      "learning_rate": 0.0003,
      "loss": 4.9478,
      "step": 9205
    },
    {
      "epoch": 0.18860196997931727,
      "grad_norm": 0.2248711884021759,
      "learning_rate": 0.0003,
      "loss": 4.9573,
      "step": 9210
    },
    {
      "epoch": 0.1887043597567219,
      "grad_norm": 0.2660972476005554,
      "learning_rate": 0.0003,
      "loss": 4.9328,
      "step": 9215
    },
    {
      "epoch": 0.18880674953412652,
      "grad_norm": 0.23380662500858307,
      "learning_rate": 0.0003,
      "loss": 4.9396,
      "step": 9220
    },
    {
      "epoch": 0.18890913931153114,
      "grad_norm": 0.242654487490654,
      "learning_rate": 0.0003,
      "loss": 4.896,
      "step": 9225
    },
    {
      "epoch": 0.18901152908893576,
      "grad_norm": 0.23970633745193481,
      "learning_rate": 0.0003,
      "loss": 4.9227,
      "step": 9230
    },
    {
      "epoch": 0.18911391886634038,
      "grad_norm": 0.30415284633636475,
      "learning_rate": 0.0003,
      "loss": 4.9366,
      "step": 9235
    },
    {
      "epoch": 0.189216308643745,
      "grad_norm": 0.2458249032497406,
      "learning_rate": 0.0003,
      "loss": 4.9759,
      "step": 9240
    },
    {
      "epoch": 0.18931869842114962,
      "grad_norm": 0.2422608584165573,
      "learning_rate": 0.0003,
      "loss": 4.9593,
      "step": 9245
    },
    {
      "epoch": 0.18942108819855424,
      "grad_norm": 0.25077205896377563,
      "learning_rate": 0.0003,
      "loss": 4.9483,
      "step": 9250
    },
    {
      "epoch": 0.1895234779759589,
      "grad_norm": 0.2309390902519226,
      "learning_rate": 0.0003,
      "loss": 4.9209,
      "step": 9255
    },
    {
      "epoch": 0.18962586775336351,
      "grad_norm": 0.2438856065273285,
      "learning_rate": 0.0003,
      "loss": 4.9279,
      "step": 9260
    },
    {
      "epoch": 0.18972825753076814,
      "grad_norm": 0.2440236508846283,
      "learning_rate": 0.0003,
      "loss": 4.9087,
      "step": 9265
    },
    {
      "epoch": 0.18983064730817276,
      "grad_norm": 0.2421654611825943,
      "learning_rate": 0.0003,
      "loss": 4.9221,
      "step": 9270
    },
    {
      "epoch": 0.18993303708557738,
      "grad_norm": 0.251799076795578,
      "learning_rate": 0.0003,
      "loss": 4.9246,
      "step": 9275
    },
    {
      "epoch": 0.190035426862982,
      "grad_norm": 0.24180428683757782,
      "learning_rate": 0.0003,
      "loss": 4.9738,
      "step": 9280
    },
    {
      "epoch": 0.19013781664038662,
      "grad_norm": 0.26821961998939514,
      "learning_rate": 0.0003,
      "loss": 4.925,
      "step": 9285
    },
    {
      "epoch": 0.19024020641779124,
      "grad_norm": 0.26987001299858093,
      "learning_rate": 0.0003,
      "loss": 4.9381,
      "step": 9290
    },
    {
      "epoch": 0.19034259619519586,
      "grad_norm": 0.23224502801895142,
      "learning_rate": 0.0003,
      "loss": 4.9938,
      "step": 9295
    },
    {
      "epoch": 0.19044498597260048,
      "grad_norm": 0.2341042160987854,
      "learning_rate": 0.0003,
      "loss": 4.9247,
      "step": 9300
    },
    {
      "epoch": 0.19054737575000513,
      "grad_norm": 0.2788233757019043,
      "learning_rate": 0.0003,
      "loss": 4.9378,
      "step": 9305
    },
    {
      "epoch": 0.19064976552740975,
      "grad_norm": 0.2468472719192505,
      "learning_rate": 0.0003,
      "loss": 4.9314,
      "step": 9310
    },
    {
      "epoch": 0.19075215530481437,
      "grad_norm": 0.2537922263145447,
      "learning_rate": 0.0003,
      "loss": 4.9645,
      "step": 9315
    },
    {
      "epoch": 0.190854545082219,
      "grad_norm": 0.22504645586013794,
      "learning_rate": 0.0003,
      "loss": 4.9711,
      "step": 9320
    },
    {
      "epoch": 0.19095693485962362,
      "grad_norm": 0.43650156259536743,
      "learning_rate": 0.0003,
      "loss": 4.8517,
      "step": 9325
    },
    {
      "epoch": 0.19105932463702824,
      "grad_norm": 0.22850650548934937,
      "learning_rate": 0.0003,
      "loss": 4.918,
      "step": 9330
    },
    {
      "epoch": 0.19116171441443286,
      "grad_norm": 0.24966390430927277,
      "learning_rate": 0.0003,
      "loss": 4.8949,
      "step": 9335
    },
    {
      "epoch": 0.19126410419183748,
      "grad_norm": 0.2230202704668045,
      "learning_rate": 0.0003,
      "loss": 4.9572,
      "step": 9340
    },
    {
      "epoch": 0.1913664939692421,
      "grad_norm": 0.24259352684020996,
      "learning_rate": 0.0003,
      "loss": 5.009,
      "step": 9345
    },
    {
      "epoch": 0.19146888374664672,
      "grad_norm": 0.23545870184898376,
      "learning_rate": 0.0003,
      "loss": 5.0105,
      "step": 9350
    },
    {
      "epoch": 0.19157127352405137,
      "grad_norm": 0.2519451379776001,
      "learning_rate": 0.0003,
      "loss": 4.9496,
      "step": 9355
    },
    {
      "epoch": 0.191673663301456,
      "grad_norm": 0.23786629736423492,
      "learning_rate": 0.0003,
      "loss": 4.9777,
      "step": 9360
    },
    {
      "epoch": 0.19177605307886061,
      "grad_norm": 0.24780771136283875,
      "learning_rate": 0.0003,
      "loss": 4.983,
      "step": 9365
    },
    {
      "epoch": 0.19187844285626524,
      "grad_norm": 0.23074345290660858,
      "learning_rate": 0.0003,
      "loss": 4.9321,
      "step": 9370
    },
    {
      "epoch": 0.19198083263366986,
      "grad_norm": 0.2469666451215744,
      "learning_rate": 0.0003,
      "loss": 4.8674,
      "step": 9375
    },
    {
      "epoch": 0.19208322241107448,
      "grad_norm": 0.2651034891605377,
      "learning_rate": 0.0003,
      "loss": 4.9177,
      "step": 9380
    },
    {
      "epoch": 0.1921856121884791,
      "grad_norm": 0.218081533908844,
      "learning_rate": 0.0003,
      "loss": 4.925,
      "step": 9385
    },
    {
      "epoch": 0.19228800196588372,
      "grad_norm": 0.22809651494026184,
      "learning_rate": 0.0003,
      "loss": 4.9752,
      "step": 9390
    },
    {
      "epoch": 0.19239039174328834,
      "grad_norm": 0.25222814083099365,
      "learning_rate": 0.0003,
      "loss": 4.9522,
      "step": 9395
    },
    {
      "epoch": 0.19249278152069296,
      "grad_norm": 0.23496654629707336,
      "learning_rate": 0.0003,
      "loss": 4.887,
      "step": 9400
    },
    {
      "epoch": 0.1925951712980976,
      "grad_norm": 0.2361747771501541,
      "learning_rate": 0.0003,
      "loss": 4.9137,
      "step": 9405
    },
    {
      "epoch": 0.19269756107550223,
      "grad_norm": 0.2526848316192627,
      "learning_rate": 0.0003,
      "loss": 4.9855,
      "step": 9410
    },
    {
      "epoch": 0.19279995085290685,
      "grad_norm": 0.250507116317749,
      "learning_rate": 0.0003,
      "loss": 4.9262,
      "step": 9415
    },
    {
      "epoch": 0.19290234063031148,
      "grad_norm": 0.2319098562002182,
      "learning_rate": 0.0003,
      "loss": 4.8931,
      "step": 9420
    },
    {
      "epoch": 0.1930047304077161,
      "grad_norm": 0.23578836023807526,
      "learning_rate": 0.0003,
      "loss": 4.924,
      "step": 9425
    },
    {
      "epoch": 0.19310712018512072,
      "grad_norm": 0.22996222972869873,
      "learning_rate": 0.0003,
      "loss": 4.9866,
      "step": 9430
    },
    {
      "epoch": 0.19320950996252534,
      "grad_norm": 0.2504240870475769,
      "learning_rate": 0.0003,
      "loss": 4.9552,
      "step": 9435
    },
    {
      "epoch": 0.19331189973992996,
      "grad_norm": 0.2702684998512268,
      "learning_rate": 0.0003,
      "loss": 4.9467,
      "step": 9440
    },
    {
      "epoch": 0.19341428951733458,
      "grad_norm": 0.2502329349517822,
      "learning_rate": 0.0003,
      "loss": 4.9376,
      "step": 9445
    },
    {
      "epoch": 0.1935166792947392,
      "grad_norm": 0.24836665391921997,
      "learning_rate": 0.0003,
      "loss": 4.913,
      "step": 9450
    },
    {
      "epoch": 0.19361906907214382,
      "grad_norm": 0.23882754147052765,
      "learning_rate": 0.0003,
      "loss": 4.8785,
      "step": 9455
    },
    {
      "epoch": 0.19372145884954847,
      "grad_norm": 0.2295556366443634,
      "learning_rate": 0.0003,
      "loss": 4.9463,
      "step": 9460
    },
    {
      "epoch": 0.1938238486269531,
      "grad_norm": 0.22904041409492493,
      "learning_rate": 0.0003,
      "loss": 4.9021,
      "step": 9465
    },
    {
      "epoch": 0.19392623840435771,
      "grad_norm": 0.2299875169992447,
      "learning_rate": 0.0003,
      "loss": 4.9742,
      "step": 9470
    },
    {
      "epoch": 0.19402862818176234,
      "grad_norm": 0.23397061228752136,
      "learning_rate": 0.0003,
      "loss": 4.8921,
      "step": 9475
    },
    {
      "epoch": 0.19413101795916696,
      "grad_norm": 0.25410881638526917,
      "learning_rate": 0.0003,
      "loss": 4.9202,
      "step": 9480
    },
    {
      "epoch": 0.19423340773657158,
      "grad_norm": 0.23728393018245697,
      "learning_rate": 0.0003,
      "loss": 4.947,
      "step": 9485
    },
    {
      "epoch": 0.1943357975139762,
      "grad_norm": 0.2503977417945862,
      "learning_rate": 0.0003,
      "loss": 4.9673,
      "step": 9490
    },
    {
      "epoch": 0.19443818729138082,
      "grad_norm": 0.2601408362388611,
      "learning_rate": 0.0003,
      "loss": 4.9154,
      "step": 9495
    },
    {
      "epoch": 0.19454057706878544,
      "grad_norm": 0.2707257568836212,
      "learning_rate": 0.0003,
      "loss": 4.8975,
      "step": 9500
    },
    {
      "epoch": 0.19464296684619006,
      "grad_norm": 0.25026175379753113,
      "learning_rate": 0.0003,
      "loss": 4.9179,
      "step": 9505
    },
    {
      "epoch": 0.1947453566235947,
      "grad_norm": 0.2416429966688156,
      "learning_rate": 0.0003,
      "loss": 4.9306,
      "step": 9510
    },
    {
      "epoch": 0.19484774640099933,
      "grad_norm": 0.3844289183616638,
      "learning_rate": 0.0003,
      "loss": 4.925,
      "step": 9515
    },
    {
      "epoch": 0.19495013617840395,
      "grad_norm": 0.24641218781471252,
      "learning_rate": 0.0003,
      "loss": 4.9458,
      "step": 9520
    },
    {
      "epoch": 0.19505252595580858,
      "grad_norm": 0.24381622672080994,
      "learning_rate": 0.0003,
      "loss": 4.9495,
      "step": 9525
    },
    {
      "epoch": 0.1951549157332132,
      "grad_norm": 0.2696736454963684,
      "learning_rate": 0.0003,
      "loss": 4.9343,
      "step": 9530
    },
    {
      "epoch": 0.19525730551061782,
      "grad_norm": 0.24814894795417786,
      "learning_rate": 0.0003,
      "loss": 4.9781,
      "step": 9535
    },
    {
      "epoch": 0.19535969528802244,
      "grad_norm": 0.24902163445949554,
      "learning_rate": 0.0003,
      "loss": 4.922,
      "step": 9540
    },
    {
      "epoch": 0.19546208506542706,
      "grad_norm": 0.23567412793636322,
      "learning_rate": 0.0003,
      "loss": 4.9178,
      "step": 9545
    },
    {
      "epoch": 0.19556447484283168,
      "grad_norm": 0.23954512178897858,
      "learning_rate": 0.0003,
      "loss": 4.9485,
      "step": 9550
    },
    {
      "epoch": 0.1956668646202363,
      "grad_norm": 0.26197537779808044,
      "learning_rate": 0.0003,
      "loss": 4.9247,
      "step": 9555
    },
    {
      "epoch": 0.19576925439764095,
      "grad_norm": 0.23075419664382935,
      "learning_rate": 0.0003,
      "loss": 4.946,
      "step": 9560
    },
    {
      "epoch": 0.19587164417504557,
      "grad_norm": 0.29694515466690063,
      "learning_rate": 0.0003,
      "loss": 4.9019,
      "step": 9565
    },
    {
      "epoch": 0.1959740339524502,
      "grad_norm": 0.24621470272541046,
      "learning_rate": 0.0003,
      "loss": 4.8725,
      "step": 9570
    },
    {
      "epoch": 0.19607642372985482,
      "grad_norm": 0.2673153877258301,
      "learning_rate": 0.0003,
      "loss": 4.9209,
      "step": 9575
    },
    {
      "epoch": 0.19617881350725944,
      "grad_norm": 0.23953528702259064,
      "learning_rate": 0.0003,
      "loss": 4.953,
      "step": 9580
    },
    {
      "epoch": 0.19628120328466406,
      "grad_norm": 0.23474064469337463,
      "learning_rate": 0.0003,
      "loss": 4.9477,
      "step": 9585
    },
    {
      "epoch": 0.19638359306206868,
      "grad_norm": 0.24345049262046814,
      "learning_rate": 0.0003,
      "loss": 4.9062,
      "step": 9590
    },
    {
      "epoch": 0.1964859828394733,
      "grad_norm": 0.2565402686595917,
      "learning_rate": 0.0003,
      "loss": 4.9048,
      "step": 9595
    },
    {
      "epoch": 0.19658837261687792,
      "grad_norm": 0.2397555708885193,
      "learning_rate": 0.0003,
      "loss": 4.9502,
      "step": 9600
    },
    {
      "epoch": 0.19669076239428254,
      "grad_norm": 0.26241710782051086,
      "learning_rate": 0.0003,
      "loss": 4.9442,
      "step": 9605
    },
    {
      "epoch": 0.1967931521716872,
      "grad_norm": 0.32985183596611023,
      "learning_rate": 0.0003,
      "loss": 4.8082,
      "step": 9610
    },
    {
      "epoch": 0.1968955419490918,
      "grad_norm": 0.2346508502960205,
      "learning_rate": 0.0003,
      "loss": 4.934,
      "step": 9615
    },
    {
      "epoch": 0.19699793172649643,
      "grad_norm": 0.2326173037290573,
      "learning_rate": 0.0003,
      "loss": 4.97,
      "step": 9620
    },
    {
      "epoch": 0.19710032150390105,
      "grad_norm": 0.22407038509845734,
      "learning_rate": 0.0003,
      "loss": 4.9392,
      "step": 9625
    },
    {
      "epoch": 0.19720271128130568,
      "grad_norm": 0.23821721971035004,
      "learning_rate": 0.0003,
      "loss": 4.8939,
      "step": 9630
    },
    {
      "epoch": 0.1973051010587103,
      "grad_norm": 0.3187132775783539,
      "learning_rate": 0.0003,
      "loss": 4.9532,
      "step": 9635
    },
    {
      "epoch": 0.19740749083611492,
      "grad_norm": 0.2720498740673065,
      "learning_rate": 0.0003,
      "loss": 4.9322,
      "step": 9640
    },
    {
      "epoch": 0.19750988061351954,
      "grad_norm": 0.2887953221797943,
      "learning_rate": 0.0003,
      "loss": 4.9195,
      "step": 9645
    },
    {
      "epoch": 0.19761227039092416,
      "grad_norm": 0.2539193034172058,
      "learning_rate": 0.0003,
      "loss": 4.9229,
      "step": 9650
    },
    {
      "epoch": 0.19771466016832878,
      "grad_norm": 0.2742467224597931,
      "learning_rate": 0.0003,
      "loss": 4.919,
      "step": 9655
    },
    {
      "epoch": 0.19781704994573343,
      "grad_norm": 0.23102837800979614,
      "learning_rate": 0.0003,
      "loss": 4.924,
      "step": 9660
    },
    {
      "epoch": 0.19791943972313805,
      "grad_norm": 0.2715983986854553,
      "learning_rate": 0.0003,
      "loss": 4.9037,
      "step": 9665
    },
    {
      "epoch": 0.19802182950054267,
      "grad_norm": 0.24073027074337006,
      "learning_rate": 0.0003,
      "loss": 4.9378,
      "step": 9670
    },
    {
      "epoch": 0.1981242192779473,
      "grad_norm": 0.2399149239063263,
      "learning_rate": 0.0003,
      "loss": 4.9259,
      "step": 9675
    },
    {
      "epoch": 0.19822660905535192,
      "grad_norm": 0.2453446388244629,
      "learning_rate": 0.0003,
      "loss": 4.9253,
      "step": 9680
    },
    {
      "epoch": 0.19832899883275654,
      "grad_norm": 0.22920511662960052,
      "learning_rate": 0.0003,
      "loss": 4.931,
      "step": 9685
    },
    {
      "epoch": 0.19843138861016116,
      "grad_norm": 0.24713505804538727,
      "learning_rate": 0.0003,
      "loss": 4.8891,
      "step": 9690
    },
    {
      "epoch": 0.19853377838756578,
      "grad_norm": 0.22313618659973145,
      "learning_rate": 0.0003,
      "loss": 4.9379,
      "step": 9695
    },
    {
      "epoch": 0.1986361681649704,
      "grad_norm": 0.21806932985782623,
      "learning_rate": 0.0003,
      "loss": 4.8942,
      "step": 9700
    },
    {
      "epoch": 0.19873855794237502,
      "grad_norm": 0.22739502787590027,
      "learning_rate": 0.0003,
      "loss": 4.9851,
      "step": 9705
    },
    {
      "epoch": 0.19884094771977967,
      "grad_norm": 0.22084246575832367,
      "learning_rate": 0.0003,
      "loss": 4.9006,
      "step": 9710
    },
    {
      "epoch": 0.1989433374971843,
      "grad_norm": 0.2365892231464386,
      "learning_rate": 0.0003,
      "loss": 4.887,
      "step": 9715
    },
    {
      "epoch": 0.1990457272745889,
      "grad_norm": 0.23465533554553986,
      "learning_rate": 0.0003,
      "loss": 4.9341,
      "step": 9720
    },
    {
      "epoch": 0.19914811705199353,
      "grad_norm": 0.22784896194934845,
      "learning_rate": 0.0003,
      "loss": 4.9387,
      "step": 9725
    },
    {
      "epoch": 0.19925050682939816,
      "grad_norm": 0.2590842545032501,
      "learning_rate": 0.0003,
      "loss": 4.8865,
      "step": 9730
    },
    {
      "epoch": 0.19935289660680278,
      "grad_norm": 0.25291505455970764,
      "learning_rate": 0.0003,
      "loss": 4.9147,
      "step": 9735
    },
    {
      "epoch": 0.1994552863842074,
      "grad_norm": 0.2361178994178772,
      "learning_rate": 0.0003,
      "loss": 4.8865,
      "step": 9740
    },
    {
      "epoch": 0.19955767616161202,
      "grad_norm": 0.25283992290496826,
      "learning_rate": 0.0003,
      "loss": 4.8911,
      "step": 9745
    },
    {
      "epoch": 0.19966006593901664,
      "grad_norm": 0.2415904700756073,
      "learning_rate": 0.0003,
      "loss": 4.8535,
      "step": 9750
    },
    {
      "epoch": 0.19976245571642126,
      "grad_norm": 0.249093160033226,
      "learning_rate": 0.0003,
      "loss": 4.8886,
      "step": 9755
    },
    {
      "epoch": 0.1998648454938259,
      "grad_norm": 0.23306076228618622,
      "learning_rate": 0.0003,
      "loss": 4.9647,
      "step": 9760
    },
    {
      "epoch": 0.19996723527123053,
      "grad_norm": 0.2572837471961975,
      "learning_rate": 0.0003,
      "loss": 4.8966,
      "step": 9765
    },
    {
      "epoch": 0.20006962504863515,
      "grad_norm": 0.24322238564491272,
      "learning_rate": 0.0003,
      "loss": 4.9127,
      "step": 9770
    },
    {
      "epoch": 0.20017201482603977,
      "grad_norm": 0.35075312852859497,
      "learning_rate": 0.0003,
      "loss": 4.9215,
      "step": 9775
    },
    {
      "epoch": 0.2002744046034444,
      "grad_norm": 0.2796444892883301,
      "learning_rate": 0.0003,
      "loss": 4.9081,
      "step": 9780
    },
    {
      "epoch": 0.20037679438084902,
      "grad_norm": 0.2839694619178772,
      "learning_rate": 0.0003,
      "loss": 4.9978,
      "step": 9785
    },
    {
      "epoch": 0.20047918415825364,
      "grad_norm": 0.24952860176563263,
      "learning_rate": 0.0003,
      "loss": 4.9422,
      "step": 9790
    },
    {
      "epoch": 0.20058157393565826,
      "grad_norm": 0.2544446289539337,
      "learning_rate": 0.0003,
      "loss": 4.9141,
      "step": 9795
    },
    {
      "epoch": 0.20068396371306288,
      "grad_norm": 0.2726847231388092,
      "learning_rate": 0.0003,
      "loss": 4.9476,
      "step": 9800
    },
    {
      "epoch": 0.2007863534904675,
      "grad_norm": 0.27150022983551025,
      "learning_rate": 0.0003,
      "loss": 4.9017,
      "step": 9805
    },
    {
      "epoch": 0.20088874326787212,
      "grad_norm": 0.253474622964859,
      "learning_rate": 0.0003,
      "loss": 4.9364,
      "step": 9810
    },
    {
      "epoch": 0.20099113304527677,
      "grad_norm": 0.3557785749435425,
      "learning_rate": 0.0003,
      "loss": 4.8741,
      "step": 9815
    },
    {
      "epoch": 0.2010935228226814,
      "grad_norm": 0.2604922652244568,
      "learning_rate": 0.0003,
      "loss": 4.9599,
      "step": 9820
    },
    {
      "epoch": 0.201195912600086,
      "grad_norm": 0.32582342624664307,
      "learning_rate": 0.0003,
      "loss": 4.9457,
      "step": 9825
    },
    {
      "epoch": 0.20129830237749063,
      "grad_norm": 0.2412978559732437,
      "learning_rate": 0.0003,
      "loss": 4.9239,
      "step": 9830
    },
    {
      "epoch": 0.20140069215489526,
      "grad_norm": 0.24473164975643158,
      "learning_rate": 0.0003,
      "loss": 4.9257,
      "step": 9835
    },
    {
      "epoch": 0.20150308193229988,
      "grad_norm": 0.2287527620792389,
      "learning_rate": 0.0003,
      "loss": 4.8948,
      "step": 9840
    },
    {
      "epoch": 0.2016054717097045,
      "grad_norm": 0.22850751876831055,
      "learning_rate": 0.0003,
      "loss": 4.9297,
      "step": 9845
    },
    {
      "epoch": 0.20170786148710912,
      "grad_norm": 0.23829907178878784,
      "learning_rate": 0.0003,
      "loss": 4.9754,
      "step": 9850
    },
    {
      "epoch": 0.20181025126451374,
      "grad_norm": 0.23911869525909424,
      "learning_rate": 0.0003,
      "loss": 4.9296,
      "step": 9855
    },
    {
      "epoch": 0.20191264104191836,
      "grad_norm": 0.24259842932224274,
      "learning_rate": 0.0003,
      "loss": 4.9028,
      "step": 9860
    },
    {
      "epoch": 0.202015030819323,
      "grad_norm": 0.26722252368927,
      "learning_rate": 0.0003,
      "loss": 4.8914,
      "step": 9865
    },
    {
      "epoch": 0.20211742059672763,
      "grad_norm": 0.25296705961227417,
      "learning_rate": 0.0003,
      "loss": 4.9552,
      "step": 9870
    },
    {
      "epoch": 0.20221981037413225,
      "grad_norm": 0.2500793933868408,
      "learning_rate": 0.0003,
      "loss": 4.956,
      "step": 9875
    },
    {
      "epoch": 0.20232220015153687,
      "grad_norm": 0.2558973729610443,
      "learning_rate": 0.0003,
      "loss": 4.9346,
      "step": 9880
    },
    {
      "epoch": 0.2024245899289415,
      "grad_norm": 0.25476959347724915,
      "learning_rate": 0.0003,
      "loss": 4.9359,
      "step": 9885
    },
    {
      "epoch": 0.20252697970634612,
      "grad_norm": 0.23200826346874237,
      "learning_rate": 0.0003,
      "loss": 4.925,
      "step": 9890
    },
    {
      "epoch": 0.20262936948375074,
      "grad_norm": 0.9324480891227722,
      "learning_rate": 0.0003,
      "loss": 4.9134,
      "step": 9895
    },
    {
      "epoch": 0.20273175926115536,
      "grad_norm": 0.28359031677246094,
      "learning_rate": 0.0003,
      "loss": 4.9316,
      "step": 9900
    },
    {
      "epoch": 0.20283414903855998,
      "grad_norm": 0.2840028405189514,
      "learning_rate": 0.0003,
      "loss": 4.9371,
      "step": 9905
    },
    {
      "epoch": 0.2029365388159646,
      "grad_norm": 0.2340623140335083,
      "learning_rate": 0.0003,
      "loss": 4.9496,
      "step": 9910
    },
    {
      "epoch": 0.20303892859336925,
      "grad_norm": 0.2449018359184265,
      "learning_rate": 0.0003,
      "loss": 4.9336,
      "step": 9915
    },
    {
      "epoch": 0.20314131837077387,
      "grad_norm": 0.23518037796020508,
      "learning_rate": 0.0003,
      "loss": 4.9661,
      "step": 9920
    },
    {
      "epoch": 0.2032437081481785,
      "grad_norm": 0.2316855490207672,
      "learning_rate": 0.0003,
      "loss": 4.9192,
      "step": 9925
    },
    {
      "epoch": 0.2033460979255831,
      "grad_norm": 0.23387083411216736,
      "learning_rate": 0.0003,
      "loss": 4.919,
      "step": 9930
    },
    {
      "epoch": 0.20344848770298773,
      "grad_norm": 0.22854074835777283,
      "learning_rate": 0.0003,
      "loss": 4.9261,
      "step": 9935
    },
    {
      "epoch": 0.20355087748039236,
      "grad_norm": 0.24308951199054718,
      "learning_rate": 0.0003,
      "loss": 4.8899,
      "step": 9940
    },
    {
      "epoch": 0.20365326725779698,
      "grad_norm": 0.2530437111854553,
      "learning_rate": 0.0003,
      "loss": 4.9501,
      "step": 9945
    },
    {
      "epoch": 0.2037556570352016,
      "grad_norm": 0.23754967749118805,
      "learning_rate": 0.0003,
      "loss": 4.9646,
      "step": 9950
    },
    {
      "epoch": 0.20385804681260622,
      "grad_norm": 0.28361964225769043,
      "learning_rate": 0.0003,
      "loss": 4.8772,
      "step": 9955
    },
    {
      "epoch": 0.20396043659001084,
      "grad_norm": 0.23497337102890015,
      "learning_rate": 0.0003,
      "loss": 4.9215,
      "step": 9960
    },
    {
      "epoch": 0.2040628263674155,
      "grad_norm": 0.24981237947940826,
      "learning_rate": 0.0003,
      "loss": 4.8853,
      "step": 9965
    },
    {
      "epoch": 0.2041652161448201,
      "grad_norm": 0.25962939858436584,
      "learning_rate": 0.0003,
      "loss": 4.929,
      "step": 9970
    },
    {
      "epoch": 0.20426760592222473,
      "grad_norm": 0.24603718519210815,
      "learning_rate": 0.0003,
      "loss": 4.8615,
      "step": 9975
    },
    {
      "epoch": 0.20436999569962935,
      "grad_norm": 0.23433342576026917,
      "learning_rate": 0.0003,
      "loss": 4.9285,
      "step": 9980
    },
    {
      "epoch": 0.20447238547703397,
      "grad_norm": 0.251792848110199,
      "learning_rate": 0.0003,
      "loss": 4.9216,
      "step": 9985
    },
    {
      "epoch": 0.2045747752544386,
      "grad_norm": 0.2492300122976303,
      "learning_rate": 0.0003,
      "loss": 4.9555,
      "step": 9990
    },
    {
      "epoch": 0.20467716503184322,
      "grad_norm": 0.3164156675338745,
      "learning_rate": 0.0003,
      "loss": 4.879,
      "step": 9995
    },
    {
      "epoch": 0.20477955480924784,
      "grad_norm": 0.289057195186615,
      "learning_rate": 0.0003,
      "loss": 4.9641,
      "step": 10000
    },
    {
      "epoch": 0.20477955480924784,
      "eval_loss": 4.946022033691406,
      "eval_runtime": 17.2507,
      "eval_samples_per_second": 33.738,
      "eval_steps_per_second": 4.232,
      "step": 10000
    },
    {
      "epoch": 0.20488194458665246,
      "grad_norm": 0.3100838363170624,
      "learning_rate": 0.0003,
      "loss": 4.9897,
      "step": 10005
    },
    {
      "epoch": 0.20498433436405708,
      "grad_norm": 0.2570136487483978,
      "learning_rate": 0.0003,
      "loss": 4.9455,
      "step": 10010
    },
    {
      "epoch": 0.20508672414146173,
      "grad_norm": 0.23518770933151245,
      "learning_rate": 0.0003,
      "loss": 4.9661,
      "step": 10015
    },
    {
      "epoch": 0.20518911391886635,
      "grad_norm": 0.23959320783615112,
      "learning_rate": 0.0003,
      "loss": 4.9419,
      "step": 10020
    },
    {
      "epoch": 0.20529150369627097,
      "grad_norm": 0.32576480507850647,
      "learning_rate": 0.0003,
      "loss": 4.9426,
      "step": 10025
    },
    {
      "epoch": 0.2053938934736756,
      "grad_norm": 0.2734343409538269,
      "learning_rate": 0.0003,
      "loss": 4.9268,
      "step": 10030
    },
    {
      "epoch": 0.20549628325108021,
      "grad_norm": 0.32146018743515015,
      "learning_rate": 0.0003,
      "loss": 4.9162,
      "step": 10035
    },
    {
      "epoch": 0.20559867302848484,
      "grad_norm": 0.24999310076236725,
      "learning_rate": 0.0003,
      "loss": 4.9229,
      "step": 10040
    },
    {
      "epoch": 0.20570106280588946,
      "grad_norm": 0.2612939774990082,
      "learning_rate": 0.0003,
      "loss": 4.9008,
      "step": 10045
    },
    {
      "epoch": 0.20580345258329408,
      "grad_norm": 0.23676054179668427,
      "learning_rate": 0.0003,
      "loss": 4.9459,
      "step": 10050
    },
    {
      "epoch": 0.2059058423606987,
      "grad_norm": 0.23431552946567535,
      "learning_rate": 0.0003,
      "loss": 4.9652,
      "step": 10055
    },
    {
      "epoch": 0.20600823213810332,
      "grad_norm": 0.2386656105518341,
      "learning_rate": 0.0003,
      "loss": 4.9272,
      "step": 10060
    },
    {
      "epoch": 0.20611062191550797,
      "grad_norm": 0.2440197616815567,
      "learning_rate": 0.0003,
      "loss": 4.923,
      "step": 10065
    },
    {
      "epoch": 0.2062130116929126,
      "grad_norm": 0.24586227536201477,
      "learning_rate": 0.0003,
      "loss": 4.9104,
      "step": 10070
    },
    {
      "epoch": 0.2063154014703172,
      "grad_norm": 0.2599840462207794,
      "learning_rate": 0.0003,
      "loss": 4.9388,
      "step": 10075
    },
    {
      "epoch": 0.20641779124772183,
      "grad_norm": 0.2490772008895874,
      "learning_rate": 0.0003,
      "loss": 4.9334,
      "step": 10080
    },
    {
      "epoch": 0.20652018102512645,
      "grad_norm": 0.22175948321819305,
      "learning_rate": 0.0003,
      "loss": 4.9286,
      "step": 10085
    },
    {
      "epoch": 0.20662257080253107,
      "grad_norm": 0.23675376176834106,
      "learning_rate": 0.0003,
      "loss": 4.9357,
      "step": 10090
    },
    {
      "epoch": 0.2067249605799357,
      "grad_norm": 0.2420710176229477,
      "learning_rate": 0.0003,
      "loss": 4.9201,
      "step": 10095
    },
    {
      "epoch": 0.20682735035734032,
      "grad_norm": 0.2561867833137512,
      "learning_rate": 0.0003,
      "loss": 4.9254,
      "step": 10100
    },
    {
      "epoch": 0.20692974013474494,
      "grad_norm": 0.22612594068050385,
      "learning_rate": 0.0003,
      "loss": 4.9592,
      "step": 10105
    },
    {
      "epoch": 0.20703212991214956,
      "grad_norm": 0.23760299384593964,
      "learning_rate": 0.0003,
      "loss": 4.8737,
      "step": 10110
    },
    {
      "epoch": 0.2071345196895542,
      "grad_norm": 0.26986318826675415,
      "learning_rate": 0.0003,
      "loss": 4.9264,
      "step": 10115
    },
    {
      "epoch": 0.20723690946695883,
      "grad_norm": 0.24083851277828217,
      "learning_rate": 0.0003,
      "loss": 4.8706,
      "step": 10120
    },
    {
      "epoch": 0.20733929924436345,
      "grad_norm": 0.2366999238729477,
      "learning_rate": 0.0003,
      "loss": 4.8748,
      "step": 10125
    },
    {
      "epoch": 0.20744168902176807,
      "grad_norm": 0.2339482456445694,
      "learning_rate": 0.0003,
      "loss": 4.9069,
      "step": 10130
    },
    {
      "epoch": 0.2075440787991727,
      "grad_norm": 0.2478560507297516,
      "learning_rate": 0.0003,
      "loss": 4.8885,
      "step": 10135
    },
    {
      "epoch": 0.20764646857657731,
      "grad_norm": 0.23827484250068665,
      "learning_rate": 0.0003,
      "loss": 4.9083,
      "step": 10140
    },
    {
      "epoch": 0.20774885835398194,
      "grad_norm": 0.24935999512672424,
      "learning_rate": 0.0003,
      "loss": 4.9239,
      "step": 10145
    },
    {
      "epoch": 0.20785124813138656,
      "grad_norm": 0.2415740042924881,
      "learning_rate": 0.0003,
      "loss": 4.9198,
      "step": 10150
    },
    {
      "epoch": 0.20795363790879118,
      "grad_norm": 0.2714764475822449,
      "learning_rate": 0.0003,
      "loss": 4.953,
      "step": 10155
    },
    {
      "epoch": 0.2080560276861958,
      "grad_norm": 0.24922576546669006,
      "learning_rate": 0.0003,
      "loss": 4.954,
      "step": 10160
    },
    {
      "epoch": 0.20815841746360042,
      "grad_norm": 0.2823648452758789,
      "learning_rate": 0.0003,
      "loss": 4.8743,
      "step": 10165
    },
    {
      "epoch": 0.20826080724100507,
      "grad_norm": 0.2266201674938202,
      "learning_rate": 0.0003,
      "loss": 4.9497,
      "step": 10170
    },
    {
      "epoch": 0.2083631970184097,
      "grad_norm": 0.24112917482852936,
      "learning_rate": 0.0003,
      "loss": 4.8825,
      "step": 10175
    },
    {
      "epoch": 0.2084655867958143,
      "grad_norm": 0.23484036326408386,
      "learning_rate": 0.0003,
      "loss": 4.9199,
      "step": 10180
    },
    {
      "epoch": 0.20856797657321893,
      "grad_norm": 0.24271684885025024,
      "learning_rate": 0.0003,
      "loss": 4.961,
      "step": 10185
    },
    {
      "epoch": 0.20867036635062355,
      "grad_norm": 0.24168355762958527,
      "learning_rate": 0.0003,
      "loss": 4.9199,
      "step": 10190
    },
    {
      "epoch": 0.20877275612802818,
      "grad_norm": 0.24393868446350098,
      "learning_rate": 0.0003,
      "loss": 4.9405,
      "step": 10195
    },
    {
      "epoch": 0.2088751459054328,
      "grad_norm": 0.23558944463729858,
      "learning_rate": 0.0003,
      "loss": 4.9036,
      "step": 10200
    },
    {
      "epoch": 0.20897753568283742,
      "grad_norm": 0.24454809725284576,
      "learning_rate": 0.0003,
      "loss": 4.899,
      "step": 10205
    },
    {
      "epoch": 0.20907992546024204,
      "grad_norm": 0.27660131454467773,
      "learning_rate": 0.0003,
      "loss": 4.9094,
      "step": 10210
    },
    {
      "epoch": 0.20918231523764666,
      "grad_norm": 0.2513921856880188,
      "learning_rate": 0.0003,
      "loss": 4.8493,
      "step": 10215
    },
    {
      "epoch": 0.2092847050150513,
      "grad_norm": 0.23608632385730743,
      "learning_rate": 0.0003,
      "loss": 4.9348,
      "step": 10220
    },
    {
      "epoch": 0.20938709479245593,
      "grad_norm": 0.26476597785949707,
      "learning_rate": 0.0003,
      "loss": 4.8984,
      "step": 10225
    },
    {
      "epoch": 0.20948948456986055,
      "grad_norm": 0.24490049481391907,
      "learning_rate": 0.0003,
      "loss": 4.8976,
      "step": 10230
    },
    {
      "epoch": 0.20959187434726517,
      "grad_norm": 0.23808947205543518,
      "learning_rate": 0.0003,
      "loss": 4.8992,
      "step": 10235
    },
    {
      "epoch": 0.2096942641246698,
      "grad_norm": 0.24560321867465973,
      "learning_rate": 0.0003,
      "loss": 4.8878,
      "step": 10240
    },
    {
      "epoch": 0.20979665390207441,
      "grad_norm": 0.23638848960399628,
      "learning_rate": 0.0003,
      "loss": 4.9148,
      "step": 10245
    },
    {
      "epoch": 0.20989904367947904,
      "grad_norm": 0.24860762059688568,
      "learning_rate": 0.0003,
      "loss": 4.9318,
      "step": 10250
    },
    {
      "epoch": 0.21000143345688366,
      "grad_norm": 0.2630869448184967,
      "learning_rate": 0.0003,
      "loss": 4.957,
      "step": 10255
    },
    {
      "epoch": 0.21010382323428828,
      "grad_norm": 0.348331481218338,
      "learning_rate": 0.0003,
      "loss": 4.8954,
      "step": 10260
    },
    {
      "epoch": 0.2102062130116929,
      "grad_norm": 0.23188897967338562,
      "learning_rate": 0.0003,
      "loss": 4.9356,
      "step": 10265
    },
    {
      "epoch": 0.21030860278909755,
      "grad_norm": 0.23428316414356232,
      "learning_rate": 0.0003,
      "loss": 4.9342,
      "step": 10270
    },
    {
      "epoch": 0.21041099256650217,
      "grad_norm": 0.25637906789779663,
      "learning_rate": 0.0003,
      "loss": 4.9023,
      "step": 10275
    },
    {
      "epoch": 0.2105133823439068,
      "grad_norm": 0.2521513104438782,
      "learning_rate": 0.0003,
      "loss": 4.8935,
      "step": 10280
    },
    {
      "epoch": 0.2106157721213114,
      "grad_norm": 0.23467016220092773,
      "learning_rate": 0.0003,
      "loss": 4.932,
      "step": 10285
    },
    {
      "epoch": 0.21071816189871603,
      "grad_norm": 0.23410767316818237,
      "learning_rate": 0.0003,
      "loss": 4.9049,
      "step": 10290
    },
    {
      "epoch": 0.21082055167612065,
      "grad_norm": 0.22825035452842712,
      "learning_rate": 0.0003,
      "loss": 4.9483,
      "step": 10295
    },
    {
      "epoch": 0.21092294145352528,
      "grad_norm": 0.2342665195465088,
      "learning_rate": 0.0003,
      "loss": 4.8045,
      "step": 10300
    },
    {
      "epoch": 0.2110253312309299,
      "grad_norm": 0.24313782155513763,
      "learning_rate": 0.0003,
      "loss": 4.9105,
      "step": 10305
    },
    {
      "epoch": 0.21112772100833452,
      "grad_norm": 0.24948935210704803,
      "learning_rate": 0.0003,
      "loss": 4.9,
      "step": 10310
    },
    {
      "epoch": 0.21123011078573914,
      "grad_norm": 0.24122211337089539,
      "learning_rate": 0.0003,
      "loss": 4.9418,
      "step": 10315
    },
    {
      "epoch": 0.2113325005631438,
      "grad_norm": 0.23953549563884735,
      "learning_rate": 0.0003,
      "loss": 4.923,
      "step": 10320
    },
    {
      "epoch": 0.2114348903405484,
      "grad_norm": 0.36018356680870056,
      "learning_rate": 0.0003,
      "loss": 4.9288,
      "step": 10325
    },
    {
      "epoch": 0.21153728011795303,
      "grad_norm": 0.23670226335525513,
      "learning_rate": 0.0003,
      "loss": 4.9307,
      "step": 10330
    },
    {
      "epoch": 0.21163966989535765,
      "grad_norm": 0.23643650114536285,
      "learning_rate": 0.0003,
      "loss": 4.9081,
      "step": 10335
    },
    {
      "epoch": 0.21174205967276227,
      "grad_norm": 0.24532055854797363,
      "learning_rate": 0.0003,
      "loss": 4.8971,
      "step": 10340
    },
    {
      "epoch": 0.2118444494501669,
      "grad_norm": 0.2591317594051361,
      "learning_rate": 0.0003,
      "loss": 4.9349,
      "step": 10345
    },
    {
      "epoch": 0.21194683922757152,
      "grad_norm": 0.22916889190673828,
      "learning_rate": 0.0003,
      "loss": 4.9148,
      "step": 10350
    },
    {
      "epoch": 0.21204922900497614,
      "grad_norm": 0.22888129949569702,
      "learning_rate": 0.0003,
      "loss": 4.9326,
      "step": 10355
    },
    {
      "epoch": 0.21215161878238076,
      "grad_norm": 0.22662124037742615,
      "learning_rate": 0.0003,
      "loss": 4.9134,
      "step": 10360
    },
    {
      "epoch": 0.21225400855978538,
      "grad_norm": 0.24721461534500122,
      "learning_rate": 0.0003,
      "loss": 4.8911,
      "step": 10365
    },
    {
      "epoch": 0.21235639833719003,
      "grad_norm": 0.23616601526737213,
      "learning_rate": 0.0003,
      "loss": 4.9176,
      "step": 10370
    },
    {
      "epoch": 0.21245878811459465,
      "grad_norm": 0.23695924878120422,
      "learning_rate": 0.0003,
      "loss": 4.95,
      "step": 10375
    },
    {
      "epoch": 0.21256117789199927,
      "grad_norm": 0.24746832251548767,
      "learning_rate": 0.0003,
      "loss": 4.9246,
      "step": 10380
    },
    {
      "epoch": 0.2126635676694039,
      "grad_norm": 0.24671892821788788,
      "learning_rate": 0.0003,
      "loss": 4.8928,
      "step": 10385
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 0.2777179777622223,
      "learning_rate": 0.0003,
      "loss": 4.8935,
      "step": 10390
    },
    {
      "epoch": 0.21286834722421313,
      "grad_norm": 0.23937489092350006,
      "learning_rate": 0.0003,
      "loss": 4.8945,
      "step": 10395
    },
    {
      "epoch": 0.21297073700161775,
      "grad_norm": 0.2379358410835266,
      "learning_rate": 0.0003,
      "loss": 4.8815,
      "step": 10400
    },
    {
      "epoch": 0.21307312677902238,
      "grad_norm": 0.2482132464647293,
      "learning_rate": 0.0003,
      "loss": 4.8642,
      "step": 10405
    },
    {
      "epoch": 0.213175516556427,
      "grad_norm": 0.25206390023231506,
      "learning_rate": 0.0003,
      "loss": 4.9147,
      "step": 10410
    },
    {
      "epoch": 0.21327790633383162,
      "grad_norm": 0.23628085851669312,
      "learning_rate": 0.0003,
      "loss": 4.8867,
      "step": 10415
    },
    {
      "epoch": 0.21338029611123627,
      "grad_norm": 0.22866074740886688,
      "learning_rate": 0.0003,
      "loss": 4.8927,
      "step": 10420
    },
    {
      "epoch": 0.2134826858886409,
      "grad_norm": 0.24655304849147797,
      "learning_rate": 0.0003,
      "loss": 4.8836,
      "step": 10425
    },
    {
      "epoch": 0.2135850756660455,
      "grad_norm": 0.36821508407592773,
      "learning_rate": 0.0003,
      "loss": 4.9246,
      "step": 10430
    },
    {
      "epoch": 0.21368746544345013,
      "grad_norm": 0.24489273130893707,
      "learning_rate": 0.0003,
      "loss": 4.8621,
      "step": 10435
    },
    {
      "epoch": 0.21378985522085475,
      "grad_norm": 0.22568266093730927,
      "learning_rate": 0.0003,
      "loss": 4.883,
      "step": 10440
    },
    {
      "epoch": 0.21389224499825937,
      "grad_norm": 0.24610939621925354,
      "learning_rate": 0.0003,
      "loss": 4.9333,
      "step": 10445
    },
    {
      "epoch": 0.213994634775664,
      "grad_norm": 0.2335331290960312,
      "learning_rate": 0.0003,
      "loss": 4.8785,
      "step": 10450
    },
    {
      "epoch": 0.21409702455306862,
      "grad_norm": 0.2339749038219452,
      "learning_rate": 0.0003,
      "loss": 4.9098,
      "step": 10455
    },
    {
      "epoch": 0.21419941433047324,
      "grad_norm": 0.2380892038345337,
      "learning_rate": 0.0003,
      "loss": 4.8889,
      "step": 10460
    },
    {
      "epoch": 0.21430180410787786,
      "grad_norm": 0.24296675622463226,
      "learning_rate": 0.0003,
      "loss": 4.9245,
      "step": 10465
    },
    {
      "epoch": 0.2144041938852825,
      "grad_norm": 0.24508559703826904,
      "learning_rate": 0.0003,
      "loss": 4.8672,
      "step": 10470
    },
    {
      "epoch": 0.21450658366268713,
      "grad_norm": 0.23185114562511444,
      "learning_rate": 0.0003,
      "loss": 4.896,
      "step": 10475
    },
    {
      "epoch": 0.21460897344009175,
      "grad_norm": 0.2594311237335205,
      "learning_rate": 0.0003,
      "loss": 4.8997,
      "step": 10480
    },
    {
      "epoch": 0.21471136321749637,
      "grad_norm": 0.24049535393714905,
      "learning_rate": 0.0003,
      "loss": 4.9123,
      "step": 10485
    },
    {
      "epoch": 0.214813752994901,
      "grad_norm": 0.24497857689857483,
      "learning_rate": 0.0003,
      "loss": 4.9061,
      "step": 10490
    },
    {
      "epoch": 0.2149161427723056,
      "grad_norm": 0.2623315155506134,
      "learning_rate": 0.0003,
      "loss": 4.8854,
      "step": 10495
    },
    {
      "epoch": 0.21501853254971023,
      "grad_norm": 0.2425769716501236,
      "learning_rate": 0.0003,
      "loss": 4.8957,
      "step": 10500
    },
    {
      "epoch": 0.21512092232711486,
      "grad_norm": 0.27547577023506165,
      "learning_rate": 0.0003,
      "loss": 4.9533,
      "step": 10505
    },
    {
      "epoch": 0.21522331210451948,
      "grad_norm": 0.24182581901550293,
      "learning_rate": 0.0003,
      "loss": 4.877,
      "step": 10510
    },
    {
      "epoch": 0.2153257018819241,
      "grad_norm": 0.2818247079849243,
      "learning_rate": 0.0003,
      "loss": 4.9121,
      "step": 10515
    },
    {
      "epoch": 0.21542809165932875,
      "grad_norm": 0.2383284568786621,
      "learning_rate": 0.0003,
      "loss": 4.9165,
      "step": 10520
    },
    {
      "epoch": 0.21553048143673337,
      "grad_norm": 0.2394833266735077,
      "learning_rate": 0.0003,
      "loss": 4.9228,
      "step": 10525
    },
    {
      "epoch": 0.215632871214138,
      "grad_norm": 0.2611193358898163,
      "learning_rate": 0.0003,
      "loss": 4.9472,
      "step": 10530
    },
    {
      "epoch": 0.2157352609915426,
      "grad_norm": 0.2499607503414154,
      "learning_rate": 0.0003,
      "loss": 4.9171,
      "step": 10535
    },
    {
      "epoch": 0.21583765076894723,
      "grad_norm": 0.2569971978664398,
      "learning_rate": 0.0003,
      "loss": 4.9122,
      "step": 10540
    },
    {
      "epoch": 0.21594004054635185,
      "grad_norm": 0.25128480792045593,
      "learning_rate": 0.0003,
      "loss": 4.8578,
      "step": 10545
    },
    {
      "epoch": 0.21604243032375647,
      "grad_norm": 0.2561035752296448,
      "learning_rate": 0.0003,
      "loss": 4.9266,
      "step": 10550
    },
    {
      "epoch": 0.2161448201011611,
      "grad_norm": 0.2700226604938507,
      "learning_rate": 0.0003,
      "loss": 4.8567,
      "step": 10555
    },
    {
      "epoch": 0.21624720987856572,
      "grad_norm": 0.2432553470134735,
      "learning_rate": 0.0003,
      "loss": 4.9168,
      "step": 10560
    },
    {
      "epoch": 0.21634959965597034,
      "grad_norm": 0.2681170403957367,
      "learning_rate": 0.0003,
      "loss": 4.9353,
      "step": 10565
    },
    {
      "epoch": 0.21645198943337496,
      "grad_norm": 0.2788884937763214,
      "learning_rate": 0.0003,
      "loss": 4.888,
      "step": 10570
    },
    {
      "epoch": 0.2165543792107796,
      "grad_norm": 0.24926117062568665,
      "learning_rate": 0.0003,
      "loss": 4.9003,
      "step": 10575
    },
    {
      "epoch": 0.21665676898818423,
      "grad_norm": 0.24506379663944244,
      "learning_rate": 0.0003,
      "loss": 4.8823,
      "step": 10580
    },
    {
      "epoch": 0.21675915876558885,
      "grad_norm": 0.22609646618366241,
      "learning_rate": 0.0003,
      "loss": 4.942,
      "step": 10585
    },
    {
      "epoch": 0.21686154854299347,
      "grad_norm": 0.23347771167755127,
      "learning_rate": 0.0003,
      "loss": 4.913,
      "step": 10590
    },
    {
      "epoch": 0.2169639383203981,
      "grad_norm": 0.2599465548992157,
      "learning_rate": 0.0003,
      "loss": 4.8689,
      "step": 10595
    },
    {
      "epoch": 0.2170663280978027,
      "grad_norm": 0.24910423159599304,
      "learning_rate": 0.0003,
      "loss": 4.9238,
      "step": 10600
    },
    {
      "epoch": 0.21716871787520733,
      "grad_norm": 0.2776698172092438,
      "learning_rate": 0.0003,
      "loss": 4.9034,
      "step": 10605
    },
    {
      "epoch": 0.21727110765261196,
      "grad_norm": 0.27069202065467834,
      "learning_rate": 0.0003,
      "loss": 4.9207,
      "step": 10610
    },
    {
      "epoch": 0.21737349743001658,
      "grad_norm": 0.23688560724258423,
      "learning_rate": 0.0003,
      "loss": 4.9181,
      "step": 10615
    },
    {
      "epoch": 0.2174758872074212,
      "grad_norm": 0.24356743693351746,
      "learning_rate": 0.0003,
      "loss": 4.8475,
      "step": 10620
    },
    {
      "epoch": 0.21757827698482585,
      "grad_norm": 0.27072104811668396,
      "learning_rate": 0.0003,
      "loss": 4.918,
      "step": 10625
    },
    {
      "epoch": 0.21768066676223047,
      "grad_norm": 0.23528029024600983,
      "learning_rate": 0.0003,
      "loss": 4.9577,
      "step": 10630
    },
    {
      "epoch": 0.2177830565396351,
      "grad_norm": 0.2385890930891037,
      "learning_rate": 0.0003,
      "loss": 4.8956,
      "step": 10635
    },
    {
      "epoch": 0.2178854463170397,
      "grad_norm": 0.23565299808979034,
      "learning_rate": 0.0003,
      "loss": 4.9131,
      "step": 10640
    },
    {
      "epoch": 0.21798783609444433,
      "grad_norm": 0.2600311040878296,
      "learning_rate": 0.0003,
      "loss": 4.9202,
      "step": 10645
    },
    {
      "epoch": 0.21809022587184895,
      "grad_norm": 0.25743547081947327,
      "learning_rate": 0.0003,
      "loss": 4.8687,
      "step": 10650
    },
    {
      "epoch": 0.21819261564925357,
      "grad_norm": 0.2406614124774933,
      "learning_rate": 0.0003,
      "loss": 4.8673,
      "step": 10655
    },
    {
      "epoch": 0.2182950054266582,
      "grad_norm": 0.23556888103485107,
      "learning_rate": 0.0003,
      "loss": 4.8713,
      "step": 10660
    },
    {
      "epoch": 0.21839739520406282,
      "grad_norm": 0.2352827936410904,
      "learning_rate": 0.0003,
      "loss": 4.8852,
      "step": 10665
    },
    {
      "epoch": 0.21849978498146744,
      "grad_norm": 0.23802189528942108,
      "learning_rate": 0.0003,
      "loss": 4.8555,
      "step": 10670
    },
    {
      "epoch": 0.2186021747588721,
      "grad_norm": 0.24062120914459229,
      "learning_rate": 0.0003,
      "loss": 4.8945,
      "step": 10675
    },
    {
      "epoch": 0.2187045645362767,
      "grad_norm": 0.2625252902507782,
      "learning_rate": 0.0003,
      "loss": 4.869,
      "step": 10680
    },
    {
      "epoch": 0.21880695431368133,
      "grad_norm": 0.2866668701171875,
      "learning_rate": 0.0003,
      "loss": 4.9274,
      "step": 10685
    },
    {
      "epoch": 0.21890934409108595,
      "grad_norm": 0.31666335463523865,
      "learning_rate": 0.0003,
      "loss": 4.8808,
      "step": 10690
    },
    {
      "epoch": 0.21901173386849057,
      "grad_norm": 0.2611194849014282,
      "learning_rate": 0.0003,
      "loss": 4.892,
      "step": 10695
    },
    {
      "epoch": 0.2191141236458952,
      "grad_norm": 0.3305256962776184,
      "learning_rate": 0.0003,
      "loss": 4.8437,
      "step": 10700
    },
    {
      "epoch": 0.2192165134232998,
      "grad_norm": 0.24971061944961548,
      "learning_rate": 0.0003,
      "loss": 4.8621,
      "step": 10705
    },
    {
      "epoch": 0.21931890320070443,
      "grad_norm": 0.2453843057155609,
      "learning_rate": 0.0003,
      "loss": 4.9466,
      "step": 10710
    },
    {
      "epoch": 0.21942129297810906,
      "grad_norm": 0.2334168702363968,
      "learning_rate": 0.0003,
      "loss": 4.9143,
      "step": 10715
    },
    {
      "epoch": 0.21952368275551368,
      "grad_norm": 0.24947424232959747,
      "learning_rate": 0.0003,
      "loss": 4.8124,
      "step": 10720
    },
    {
      "epoch": 0.21962607253291833,
      "grad_norm": 0.2462509423494339,
      "learning_rate": 0.0003,
      "loss": 4.8933,
      "step": 10725
    },
    {
      "epoch": 0.21972846231032295,
      "grad_norm": 0.2200932651758194,
      "learning_rate": 0.0003,
      "loss": 4.9114,
      "step": 10730
    },
    {
      "epoch": 0.21983085208772757,
      "grad_norm": 0.22099857032299042,
      "learning_rate": 0.0003,
      "loss": 4.9182,
      "step": 10735
    },
    {
      "epoch": 0.2199332418651322,
      "grad_norm": 0.25395599007606506,
      "learning_rate": 0.0003,
      "loss": 4.9027,
      "step": 10740
    },
    {
      "epoch": 0.2200356316425368,
      "grad_norm": 0.24311977624893188,
      "learning_rate": 0.0003,
      "loss": 4.8598,
      "step": 10745
    },
    {
      "epoch": 0.22013802141994143,
      "grad_norm": 0.24510981142520905,
      "learning_rate": 0.0003,
      "loss": 4.8418,
      "step": 10750
    },
    {
      "epoch": 0.22024041119734605,
      "grad_norm": 0.25221526622772217,
      "learning_rate": 0.0003,
      "loss": 4.9188,
      "step": 10755
    },
    {
      "epoch": 0.22034280097475067,
      "grad_norm": 0.22577425837516785,
      "learning_rate": 0.0003,
      "loss": 4.8804,
      "step": 10760
    },
    {
      "epoch": 0.2204451907521553,
      "grad_norm": 0.23756377398967743,
      "learning_rate": 0.0003,
      "loss": 4.8441,
      "step": 10765
    },
    {
      "epoch": 0.22054758052955992,
      "grad_norm": 0.2356066256761551,
      "learning_rate": 0.00029999999540266433,
      "loss": 4.9067,
      "step": 10770
    },
    {
      "epoch": 0.22064997030696457,
      "grad_norm": 0.23691512644290924,
      "learning_rate": 0.00029999996730783644,
      "loss": 4.9293,
      "step": 10775
    },
    {
      "epoch": 0.2207523600843692,
      "grad_norm": 0.22614943981170654,
      "learning_rate": 0.00029999991367226075,
      "loss": 4.9012,
      "step": 10780
    },
    {
      "epoch": 0.2208547498617738,
      "grad_norm": 0.23372440040111542,
      "learning_rate": 0.0002999998344959465,
      "loss": 4.9256,
      "step": 10785
    },
    {
      "epoch": 0.22095713963917843,
      "grad_norm": 0.24283495545387268,
      "learning_rate": 0.000299999729778907,
      "loss": 4.8643,
      "step": 10790
    },
    {
      "epoch": 0.22105952941658305,
      "grad_norm": 0.27330297231674194,
      "learning_rate": 0.00029999959952116027,
      "loss": 4.897,
      "step": 10795
    },
    {
      "epoch": 0.22116191919398767,
      "grad_norm": 0.25471654534339905,
      "learning_rate": 0.00029999944372272833,
      "loss": 4.8822,
      "step": 10800
    },
    {
      "epoch": 0.2212643089713923,
      "grad_norm": 0.2778959274291992,
      "learning_rate": 0.00029999926238363783,
      "loss": 4.8658,
      "step": 10805
    },
    {
      "epoch": 0.22136669874879691,
      "grad_norm": 0.2582586705684662,
      "learning_rate": 0.00029999905550391956,
      "loss": 4.9524,
      "step": 10810
    },
    {
      "epoch": 0.22146908852620154,
      "grad_norm": 0.24232889711856842,
      "learning_rate": 0.0002999988230836088,
      "loss": 4.8638,
      "step": 10815
    },
    {
      "epoch": 0.22157147830360616,
      "grad_norm": 0.25435176491737366,
      "learning_rate": 0.0002999985651227451,
      "loss": 4.8973,
      "step": 10820
    },
    {
      "epoch": 0.2216738680810108,
      "grad_norm": 0.23583756387233734,
      "learning_rate": 0.0002999982816213724,
      "loss": 4.8829,
      "step": 10825
    },
    {
      "epoch": 0.22177625785841543,
      "grad_norm": 0.25342777371406555,
      "learning_rate": 0.0002999979725795389,
      "loss": 4.8806,
      "step": 10830
    },
    {
      "epoch": 0.22187864763582005,
      "grad_norm": 0.22450606524944305,
      "learning_rate": 0.00029999763799729733,
      "loss": 4.9066,
      "step": 10835
    },
    {
      "epoch": 0.22198103741322467,
      "grad_norm": 0.23883618414402008,
      "learning_rate": 0.0002999972778747046,
      "loss": 4.9482,
      "step": 10840
    },
    {
      "epoch": 0.2220834271906293,
      "grad_norm": 0.2697833776473999,
      "learning_rate": 0.0002999968922118221,
      "loss": 4.9225,
      "step": 10845
    },
    {
      "epoch": 0.2221858169680339,
      "grad_norm": 0.23498277366161346,
      "learning_rate": 0.00029999648100871535,
      "loss": 4.8926,
      "step": 10850
    },
    {
      "epoch": 0.22228820674543853,
      "grad_norm": 0.2613791525363922,
      "learning_rate": 0.0002999960442654545,
      "loss": 4.9152,
      "step": 10855
    },
    {
      "epoch": 0.22239059652284315,
      "grad_norm": 0.264046311378479,
      "learning_rate": 0.0002999955819821138,
      "loss": 4.9276,
      "step": 10860
    },
    {
      "epoch": 0.22249298630024777,
      "grad_norm": 0.2305372655391693,
      "learning_rate": 0.00029999509415877215,
      "loss": 4.8974,
      "step": 10865
    },
    {
      "epoch": 0.2225953760776524,
      "grad_norm": 0.22589483857154846,
      "learning_rate": 0.0002999945807955124,
      "loss": 4.8986,
      "step": 10870
    },
    {
      "epoch": 0.22269776585505704,
      "grad_norm": 0.24377058446407318,
      "learning_rate": 0.0002999940418924221,
      "loss": 4.8925,
      "step": 10875
    },
    {
      "epoch": 0.22280015563246167,
      "grad_norm": 0.23469287157058716,
      "learning_rate": 0.0002999934774495929,
      "loss": 4.8825,
      "step": 10880
    },
    {
      "epoch": 0.2229025454098663,
      "grad_norm": 0.2578478455543518,
      "learning_rate": 0.00029999288746712107,
      "loss": 4.8929,
      "step": 10885
    },
    {
      "epoch": 0.2230049351872709,
      "grad_norm": 0.24005863070487976,
      "learning_rate": 0.0002999922719451069,
      "loss": 4.8768,
      "step": 10890
    },
    {
      "epoch": 0.22310732496467553,
      "grad_norm": 0.3334100544452667,
      "learning_rate": 0.0002999916308836553,
      "loss": 4.89,
      "step": 10895
    },
    {
      "epoch": 0.22320971474208015,
      "grad_norm": 0.2511036992073059,
      "learning_rate": 0.00029999096428287537,
      "loss": 4.8613,
      "step": 10900
    },
    {
      "epoch": 0.22331210451948477,
      "grad_norm": 0.2527853548526764,
      "learning_rate": 0.0002999902721428807,
      "loss": 4.8834,
      "step": 10905
    },
    {
      "epoch": 0.2234144942968894,
      "grad_norm": 0.23672787845134735,
      "learning_rate": 0.000299989554463789,
      "loss": 4.9026,
      "step": 10910
    },
    {
      "epoch": 0.22351688407429401,
      "grad_norm": 0.26204660534858704,
      "learning_rate": 0.0002999888112457226,
      "loss": 4.9342,
      "step": 10915
    },
    {
      "epoch": 0.22361927385169864,
      "grad_norm": 0.24639709293842316,
      "learning_rate": 0.000299988042488808,
      "loss": 4.8823,
      "step": 10920
    },
    {
      "epoch": 0.22372166362910326,
      "grad_norm": 0.2559228539466858,
      "learning_rate": 0.000299987248193176,
      "loss": 4.9291,
      "step": 10925
    },
    {
      "epoch": 0.2238240534065079,
      "grad_norm": 0.2215857356786728,
      "learning_rate": 0.00029998642835896204,
      "loss": 4.8747,
      "step": 10930
    },
    {
      "epoch": 0.22392644318391253,
      "grad_norm": 0.23071123659610748,
      "learning_rate": 0.00029998558298630556,
      "loss": 4.8983,
      "step": 10935
    },
    {
      "epoch": 0.22402883296131715,
      "grad_norm": 0.24900750815868378,
      "learning_rate": 0.0002999847120753506,
      "loss": 4.9054,
      "step": 10940
    },
    {
      "epoch": 0.22413122273872177,
      "grad_norm": 0.2291107475757599,
      "learning_rate": 0.0002999838156262454,
      "loss": 4.8966,
      "step": 10945
    },
    {
      "epoch": 0.2242336125161264,
      "grad_norm": 0.24814331531524658,
      "learning_rate": 0.0002999828936391426,
      "loss": 4.855,
      "step": 10950
    },
    {
      "epoch": 0.224336002293531,
      "grad_norm": 0.24090422689914703,
      "learning_rate": 0.0002999819461141992,
      "loss": 4.8942,
      "step": 10955
    },
    {
      "epoch": 0.22443839207093563,
      "grad_norm": 0.22519832849502563,
      "learning_rate": 0.0002999809730515765,
      "loss": 4.9153,
      "step": 10960
    },
    {
      "epoch": 0.22454078184834025,
      "grad_norm": 0.29087841510772705,
      "learning_rate": 0.0002999799744514403,
      "loss": 4.9382,
      "step": 10965
    },
    {
      "epoch": 0.22464317162574488,
      "grad_norm": 0.2530803084373474,
      "learning_rate": 0.0002999789503139605,
      "loss": 4.8792,
      "step": 10970
    },
    {
      "epoch": 0.2247455614031495,
      "grad_norm": 0.26962804794311523,
      "learning_rate": 0.0002999779006393115,
      "loss": 4.8897,
      "step": 10975
    },
    {
      "epoch": 0.22484795118055415,
      "grad_norm": 0.2534395754337311,
      "learning_rate": 0.0002999768254276721,
      "loss": 4.9142,
      "step": 10980
    },
    {
      "epoch": 0.22495034095795877,
      "grad_norm": 0.27273815870285034,
      "learning_rate": 0.0002999757246792254,
      "loss": 4.9112,
      "step": 10985
    },
    {
      "epoch": 0.2250527307353634,
      "grad_norm": 0.2543729543685913,
      "learning_rate": 0.0002999745983941587,
      "loss": 4.8987,
      "step": 10990
    },
    {
      "epoch": 0.225155120512768,
      "grad_norm": 0.22888705134391785,
      "learning_rate": 0.00029997344657266387,
      "loss": 4.8616,
      "step": 10995
    },
    {
      "epoch": 0.22525751029017263,
      "grad_norm": 0.24257194995880127,
      "learning_rate": 0.00029997226921493693,
      "loss": 4.868,
      "step": 11000
    },
    {
      "epoch": 0.22525751029017263,
      "eval_loss": 4.919167518615723,
      "eval_runtime": 17.2454,
      "eval_samples_per_second": 33.748,
      "eval_steps_per_second": 4.233,
      "step": 11000
    },
    {
      "epoch": 0.22535990006757725,
      "grad_norm": 0.24592594802379608,
      "learning_rate": 0.0002999710663211785,
      "loss": 4.9788,
      "step": 11005
    },
    {
      "epoch": 0.22546228984498187,
      "grad_norm": 0.2423733025789261,
      "learning_rate": 0.0002999698378915933,
      "loss": 4.8767,
      "step": 11010
    },
    {
      "epoch": 0.2255646796223865,
      "grad_norm": 0.30392077565193176,
      "learning_rate": 0.00029996858392639054,
      "loss": 4.937,
      "step": 11015
    },
    {
      "epoch": 0.22566706939979111,
      "grad_norm": 0.2611886262893677,
      "learning_rate": 0.00029996730442578373,
      "loss": 4.8765,
      "step": 11020
    },
    {
      "epoch": 0.22576945917719574,
      "grad_norm": 0.24054093658924103,
      "learning_rate": 0.0002999659993899907,
      "loss": 4.931,
      "step": 11025
    },
    {
      "epoch": 0.22587184895460038,
      "grad_norm": 0.2398342490196228,
      "learning_rate": 0.00029996466881923367,
      "loss": 4.8824,
      "step": 11030
    },
    {
      "epoch": 0.225974238732005,
      "grad_norm": 0.25371286273002625,
      "learning_rate": 0.0002999633127137392,
      "loss": 4.9421,
      "step": 11035
    },
    {
      "epoch": 0.22607662850940963,
      "grad_norm": 0.26272109150886536,
      "learning_rate": 0.00029996193107373825,
      "loss": 4.8708,
      "step": 11040
    },
    {
      "epoch": 0.22617901828681425,
      "grad_norm": 0.24237163364887238,
      "learning_rate": 0.000299960523899466,
      "loss": 4.8722,
      "step": 11045
    },
    {
      "epoch": 0.22628140806421887,
      "grad_norm": 0.2590733766555786,
      "learning_rate": 0.0002999590911911621,
      "loss": 4.9091,
      "step": 11050
    },
    {
      "epoch": 0.2263837978416235,
      "grad_norm": 0.23286749422550201,
      "learning_rate": 0.00029995763294907044,
      "loss": 4.9017,
      "step": 11055
    },
    {
      "epoch": 0.2264861876190281,
      "grad_norm": 0.2581934630870819,
      "learning_rate": 0.0002999561491734394,
      "loss": 4.8817,
      "step": 11060
    },
    {
      "epoch": 0.22658857739643273,
      "grad_norm": 0.23137080669403076,
      "learning_rate": 0.00029995463986452156,
      "loss": 4.8804,
      "step": 11065
    },
    {
      "epoch": 0.22669096717383735,
      "grad_norm": 0.23512017726898193,
      "learning_rate": 0.0002999531050225739,
      "loss": 4.8719,
      "step": 11070
    },
    {
      "epoch": 0.22679335695124198,
      "grad_norm": 0.22481665015220642,
      "learning_rate": 0.00029995154464785786,
      "loss": 4.8657,
      "step": 11075
    },
    {
      "epoch": 0.22689574672864662,
      "grad_norm": 0.2558952569961548,
      "learning_rate": 0.00029994995874063905,
      "loss": 4.9027,
      "step": 11080
    },
    {
      "epoch": 0.22699813650605125,
      "grad_norm": 0.25101906061172485,
      "learning_rate": 0.0002999483473011875,
      "loss": 4.8701,
      "step": 11085
    },
    {
      "epoch": 0.22710052628345587,
      "grad_norm": 0.23361775279045105,
      "learning_rate": 0.00029994671032977763,
      "loss": 4.9379,
      "step": 11090
    },
    {
      "epoch": 0.2272029160608605,
      "grad_norm": 0.24297501146793365,
      "learning_rate": 0.0002999450478266881,
      "loss": 4.8984,
      "step": 11095
    },
    {
      "epoch": 0.2273053058382651,
      "grad_norm": 0.23249764740467072,
      "learning_rate": 0.0002999433597922021,
      "loss": 4.8988,
      "step": 11100
    },
    {
      "epoch": 0.22740769561566973,
      "grad_norm": 0.24362662434577942,
      "learning_rate": 0.00029994164622660697,
      "loss": 4.865,
      "step": 11105
    },
    {
      "epoch": 0.22751008539307435,
      "grad_norm": 0.2715204060077667,
      "learning_rate": 0.00029993990713019446,
      "loss": 4.8854,
      "step": 11110
    },
    {
      "epoch": 0.22761247517047897,
      "grad_norm": 0.2356962114572525,
      "learning_rate": 0.00029993814250326085,
      "loss": 4.8502,
      "step": 11115
    },
    {
      "epoch": 0.2277148649478836,
      "grad_norm": 0.2375289797782898,
      "learning_rate": 0.0002999363523461064,
      "loss": 4.9038,
      "step": 11120
    },
    {
      "epoch": 0.22781725472528822,
      "grad_norm": 0.26735198497772217,
      "learning_rate": 0.000299934536659036,
      "loss": 4.8722,
      "step": 11125
    },
    {
      "epoch": 0.22791964450269286,
      "grad_norm": 0.2644442617893219,
      "learning_rate": 0.00029993269544235885,
      "loss": 4.8949,
      "step": 11130
    },
    {
      "epoch": 0.22802203428009749,
      "grad_norm": 0.23039349913597107,
      "learning_rate": 0.00029993082869638843,
      "loss": 4.9066,
      "step": 11135
    },
    {
      "epoch": 0.2281244240575021,
      "grad_norm": 0.25083819031715393,
      "learning_rate": 0.0002999289364214426,
      "loss": 4.9457,
      "step": 11140
    },
    {
      "epoch": 0.22822681383490673,
      "grad_norm": 0.22685234248638153,
      "learning_rate": 0.0002999270186178435,
      "loss": 4.8967,
      "step": 11145
    },
    {
      "epoch": 0.22832920361231135,
      "grad_norm": 0.2520279884338379,
      "learning_rate": 0.00029992507528591784,
      "loss": 4.8651,
      "step": 11150
    },
    {
      "epoch": 0.22843159338971597,
      "grad_norm": 0.25449731945991516,
      "learning_rate": 0.00029992310642599633,
      "loss": 4.8679,
      "step": 11155
    },
    {
      "epoch": 0.2285339831671206,
      "grad_norm": 0.23374278843402863,
      "learning_rate": 0.0002999211120384143,
      "loss": 4.8691,
      "step": 11160
    },
    {
      "epoch": 0.2286363729445252,
      "grad_norm": 0.24310655891895294,
      "learning_rate": 0.00029991909212351135,
      "loss": 4.8958,
      "step": 11165
    },
    {
      "epoch": 0.22873876272192983,
      "grad_norm": 0.22431394457817078,
      "learning_rate": 0.00029991704668163135,
      "loss": 4.8878,
      "step": 11170
    },
    {
      "epoch": 0.22884115249933445,
      "grad_norm": 0.22711428999900818,
      "learning_rate": 0.00029991497571312263,
      "loss": 4.8992,
      "step": 11175
    },
    {
      "epoch": 0.2289435422767391,
      "grad_norm": 0.23749734461307526,
      "learning_rate": 0.0002999128792183378,
      "loss": 4.8963,
      "step": 11180
    },
    {
      "epoch": 0.22904593205414372,
      "grad_norm": 0.24261240661144257,
      "learning_rate": 0.0002999107571976338,
      "loss": 4.9011,
      "step": 11185
    },
    {
      "epoch": 0.22914832183154835,
      "grad_norm": 0.23800702393054962,
      "learning_rate": 0.00029990860965137207,
      "loss": 4.8943,
      "step": 11190
    },
    {
      "epoch": 0.22925071160895297,
      "grad_norm": 0.24408547580242157,
      "learning_rate": 0.0002999064365799182,
      "loss": 4.8213,
      "step": 11195
    },
    {
      "epoch": 0.2293531013863576,
      "grad_norm": 0.22900575399398804,
      "learning_rate": 0.00029990423798364216,
      "loss": 4.844,
      "step": 11200
    },
    {
      "epoch": 0.2294554911637622,
      "grad_norm": 0.2334994524717331,
      "learning_rate": 0.00029990201386291836,
      "loss": 4.8588,
      "step": 11205
    },
    {
      "epoch": 0.22955788094116683,
      "grad_norm": 0.2324739694595337,
      "learning_rate": 0.0002998997642181255,
      "loss": 4.8618,
      "step": 11210
    },
    {
      "epoch": 0.22966027071857145,
      "grad_norm": 0.24837611615657806,
      "learning_rate": 0.0002998974890496466,
      "loss": 4.8935,
      "step": 11215
    },
    {
      "epoch": 0.22976266049597607,
      "grad_norm": 0.24437487125396729,
      "learning_rate": 0.0002998951883578691,
      "loss": 4.8669,
      "step": 11220
    },
    {
      "epoch": 0.2298650502733807,
      "grad_norm": 0.3906613290309906,
      "learning_rate": 0.00029989286214318476,
      "loss": 4.8749,
      "step": 11225
    },
    {
      "epoch": 0.22996744005078534,
      "grad_norm": 0.24109190702438354,
      "learning_rate": 0.0002998905104059896,
      "loss": 4.9013,
      "step": 11230
    },
    {
      "epoch": 0.23006982982818996,
      "grad_norm": 0.24165146052837372,
      "learning_rate": 0.0002998881331466841,
      "loss": 4.8549,
      "step": 11235
    },
    {
      "epoch": 0.23017221960559459,
      "grad_norm": 0.24196457862854004,
      "learning_rate": 0.000299885730365673,
      "loss": 4.8945,
      "step": 11240
    },
    {
      "epoch": 0.2302746093829992,
      "grad_norm": 0.2327190488576889,
      "learning_rate": 0.00029988330206336545,
      "loss": 4.9292,
      "step": 11245
    },
    {
      "epoch": 0.23037699916040383,
      "grad_norm": 0.2443750500679016,
      "learning_rate": 0.00029988084824017497,
      "loss": 4.8996,
      "step": 11250
    },
    {
      "epoch": 0.23047938893780845,
      "grad_norm": 0.2533819377422333,
      "learning_rate": 0.0002998783688965193,
      "loss": 4.9055,
      "step": 11255
    },
    {
      "epoch": 0.23058177871521307,
      "grad_norm": 0.6916893720626831,
      "learning_rate": 0.0002998758640328206,
      "loss": 4.78,
      "step": 11260
    },
    {
      "epoch": 0.2306841684926177,
      "grad_norm": 0.2645145654678345,
      "learning_rate": 0.0002998733336495054,
      "loss": 4.8977,
      "step": 11265
    },
    {
      "epoch": 0.2307865582700223,
      "grad_norm": 0.2584141790866852,
      "learning_rate": 0.00029987077774700464,
      "loss": 4.8775,
      "step": 11270
    },
    {
      "epoch": 0.23088894804742693,
      "grad_norm": 0.22412125766277313,
      "learning_rate": 0.0002998681963257534,
      "loss": 4.8988,
      "step": 11275
    },
    {
      "epoch": 0.23099133782483156,
      "grad_norm": 0.317967027425766,
      "learning_rate": 0.0002998655893861913,
      "loss": 4.8529,
      "step": 11280
    },
    {
      "epoch": 0.2310937276022362,
      "grad_norm": 0.2685169577598572,
      "learning_rate": 0.0002998629569287621,
      "loss": 4.8384,
      "step": 11285
    },
    {
      "epoch": 0.23119611737964083,
      "grad_norm": 0.24802027642726898,
      "learning_rate": 0.0002998602989539142,
      "loss": 4.8708,
      "step": 11290
    },
    {
      "epoch": 0.23129850715704545,
      "grad_norm": 0.2655007243156433,
      "learning_rate": 0.00029985761546210007,
      "loss": 4.9067,
      "step": 11295
    },
    {
      "epoch": 0.23140089693445007,
      "grad_norm": 0.24444621801376343,
      "learning_rate": 0.0002998549064537767,
      "loss": 4.907,
      "step": 11300
    },
    {
      "epoch": 0.2315032867118547,
      "grad_norm": 0.22952929139137268,
      "learning_rate": 0.00029985217192940526,
      "loss": 4.9045,
      "step": 11305
    },
    {
      "epoch": 0.2316056764892593,
      "grad_norm": 0.2440677136182785,
      "learning_rate": 0.00029984941188945146,
      "loss": 4.9086,
      "step": 11310
    },
    {
      "epoch": 0.23170806626666393,
      "grad_norm": 0.22809574007987976,
      "learning_rate": 0.0002998466263343853,
      "loss": 4.8561,
      "step": 11315
    },
    {
      "epoch": 0.23181045604406855,
      "grad_norm": 0.24641495943069458,
      "learning_rate": 0.0002998438152646809,
      "loss": 4.899,
      "step": 11320
    },
    {
      "epoch": 0.23191284582147317,
      "grad_norm": 0.24032865464687347,
      "learning_rate": 0.000299840978680817,
      "loss": 4.8743,
      "step": 11325
    },
    {
      "epoch": 0.2320152355988778,
      "grad_norm": 0.24213042855262756,
      "learning_rate": 0.0002998381165832766,
      "loss": 4.8848,
      "step": 11330
    },
    {
      "epoch": 0.23211762537628244,
      "grad_norm": 0.24833174049854279,
      "learning_rate": 0.00029983522897254706,
      "loss": 4.8933,
      "step": 11335
    },
    {
      "epoch": 0.23222001515368706,
      "grad_norm": 0.26272082328796387,
      "learning_rate": 0.00029983231584912006,
      "loss": 4.8843,
      "step": 11340
    },
    {
      "epoch": 0.23232240493109169,
      "grad_norm": 0.2515399754047394,
      "learning_rate": 0.00029982937721349155,
      "loss": 4.918,
      "step": 11345
    },
    {
      "epoch": 0.2324247947084963,
      "grad_norm": 0.25837281346321106,
      "learning_rate": 0.00029982641306616195,
      "loss": 4.876,
      "step": 11350
    },
    {
      "epoch": 0.23252718448590093,
      "grad_norm": 0.23617346584796906,
      "learning_rate": 0.00029982342340763594,
      "loss": 4.9494,
      "step": 11355
    },
    {
      "epoch": 0.23262957426330555,
      "grad_norm": 0.2497721016407013,
      "learning_rate": 0.0002998204082384226,
      "loss": 4.8599,
      "step": 11360
    },
    {
      "epoch": 0.23273196404071017,
      "grad_norm": 0.22829598188400269,
      "learning_rate": 0.00029981736755903537,
      "loss": 4.8906,
      "step": 11365
    },
    {
      "epoch": 0.2328343538181148,
      "grad_norm": 0.24126386642456055,
      "learning_rate": 0.0002998143013699919,
      "loss": 4.9093,
      "step": 11370
    },
    {
      "epoch": 0.2329367435955194,
      "grad_norm": 0.2352132499217987,
      "learning_rate": 0.0002998112096718143,
      "loss": 4.8942,
      "step": 11375
    },
    {
      "epoch": 0.23303913337292403,
      "grad_norm": 0.2613617479801178,
      "learning_rate": 0.00029980809246502906,
      "loss": 4.9019,
      "step": 11380
    },
    {
      "epoch": 0.23314152315032868,
      "grad_norm": 0.2374405860900879,
      "learning_rate": 0.0002998049497501669,
      "loss": 4.8281,
      "step": 11385
    },
    {
      "epoch": 0.2332439129277333,
      "grad_norm": 0.2515256702899933,
      "learning_rate": 0.0002998017815277629,
      "loss": 4.8755,
      "step": 11390
    },
    {
      "epoch": 0.23334630270513793,
      "grad_norm": 0.25903964042663574,
      "learning_rate": 0.00029979858779835664,
      "loss": 4.8912,
      "step": 11395
    },
    {
      "epoch": 0.23344869248254255,
      "grad_norm": 0.251465767621994,
      "learning_rate": 0.0002997953685624918,
      "loss": 4.8704,
      "step": 11400
    },
    {
      "epoch": 0.23355108225994717,
      "grad_norm": 0.2379695475101471,
      "learning_rate": 0.00029979212382071654,
      "loss": 4.9282,
      "step": 11405
    },
    {
      "epoch": 0.2336534720373518,
      "grad_norm": 0.23356886208057404,
      "learning_rate": 0.0002997888535735834,
      "loss": 4.8989,
      "step": 11410
    },
    {
      "epoch": 0.2337558618147564,
      "grad_norm": 0.2495730221271515,
      "learning_rate": 0.00029978555782164923,
      "loss": 4.8938,
      "step": 11415
    },
    {
      "epoch": 0.23385825159216103,
      "grad_norm": 0.24322271347045898,
      "learning_rate": 0.00029978223656547506,
      "loss": 4.924,
      "step": 11420
    },
    {
      "epoch": 0.23396064136956565,
      "grad_norm": 0.31163012981414795,
      "learning_rate": 0.0002997788898056266,
      "loss": 4.8874,
      "step": 11425
    },
    {
      "epoch": 0.23406303114697027,
      "grad_norm": 0.25091469287872314,
      "learning_rate": 0.0002997755175426736,
      "loss": 4.9591,
      "step": 11430
    },
    {
      "epoch": 0.23416542092437492,
      "grad_norm": 0.2644343674182892,
      "learning_rate": 0.00029977211977719024,
      "loss": 4.8423,
      "step": 11435
    },
    {
      "epoch": 0.23426781070177954,
      "grad_norm": 0.24409982562065125,
      "learning_rate": 0.0002997686965097551,
      "loss": 4.9034,
      "step": 11440
    },
    {
      "epoch": 0.23437020047918417,
      "grad_norm": 0.2304883897304535,
      "learning_rate": 0.00029976524774095114,
      "loss": 4.9265,
      "step": 11445
    },
    {
      "epoch": 0.2344725902565888,
      "grad_norm": 0.24100245535373688,
      "learning_rate": 0.0002997617734713654,
      "loss": 4.9006,
      "step": 11450
    },
    {
      "epoch": 0.2345749800339934,
      "grad_norm": 0.24879340827465057,
      "learning_rate": 0.00029975827370158967,
      "loss": 4.9382,
      "step": 11455
    },
    {
      "epoch": 0.23467736981139803,
      "grad_norm": 0.2458147257566452,
      "learning_rate": 0.0002997547484322197,
      "loss": 4.9253,
      "step": 11460
    },
    {
      "epoch": 0.23477975958880265,
      "grad_norm": 0.2216917872428894,
      "learning_rate": 0.0002997511976638558,
      "loss": 4.8462,
      "step": 11465
    },
    {
      "epoch": 0.23488214936620727,
      "grad_norm": 0.22838068008422852,
      "learning_rate": 0.0002997476213971025,
      "loss": 4.8981,
      "step": 11470
    },
    {
      "epoch": 0.2349845391436119,
      "grad_norm": 0.2378031313419342,
      "learning_rate": 0.00029974401963256885,
      "loss": 4.9067,
      "step": 11475
    },
    {
      "epoch": 0.2350869289210165,
      "grad_norm": 0.2917433977127075,
      "learning_rate": 0.0002997403923708681,
      "loss": 4.9242,
      "step": 11480
    },
    {
      "epoch": 0.23518931869842116,
      "grad_norm": 0.24154146015644073,
      "learning_rate": 0.0002997367396126179,
      "loss": 4.8598,
      "step": 11485
    },
    {
      "epoch": 0.23529170847582578,
      "grad_norm": 0.24472050368785858,
      "learning_rate": 0.0002997330613584401,
      "loss": 4.8583,
      "step": 11490
    },
    {
      "epoch": 0.2353940982532304,
      "grad_norm": 0.25144362449645996,
      "learning_rate": 0.00029972935760896104,
      "loss": 4.8015,
      "step": 11495
    },
    {
      "epoch": 0.23549648803063503,
      "grad_norm": 0.24405333399772644,
      "learning_rate": 0.0002997256283648114,
      "loss": 4.9228,
      "step": 11500
    },
    {
      "epoch": 0.23559887780803965,
      "grad_norm": 0.25055980682373047,
      "learning_rate": 0.00029972187362662615,
      "loss": 4.9692,
      "step": 11505
    },
    {
      "epoch": 0.23570126758544427,
      "grad_norm": 0.28883928060531616,
      "learning_rate": 0.0002997180933950446,
      "loss": 4.8954,
      "step": 11510
    },
    {
      "epoch": 0.2358036573628489,
      "grad_norm": 0.23557475209236145,
      "learning_rate": 0.00029971428767071046,
      "loss": 4.8881,
      "step": 11515
    },
    {
      "epoch": 0.2359060471402535,
      "grad_norm": 0.2456962913274765,
      "learning_rate": 0.0002997104564542717,
      "loss": 4.9171,
      "step": 11520
    },
    {
      "epoch": 0.23600843691765813,
      "grad_norm": 0.2360301911830902,
      "learning_rate": 0.00029970659974638066,
      "loss": 4.9174,
      "step": 11525
    },
    {
      "epoch": 0.23611082669506275,
      "grad_norm": 0.2407263070344925,
      "learning_rate": 0.00029970271754769406,
      "loss": 4.8916,
      "step": 11530
    },
    {
      "epoch": 0.2362132164724674,
      "grad_norm": 0.2519512176513672,
      "learning_rate": 0.00029969880985887293,
      "loss": 4.884,
      "step": 11535
    },
    {
      "epoch": 0.23631560624987202,
      "grad_norm": 0.22996272146701813,
      "learning_rate": 0.00029969487668058266,
      "loss": 4.8716,
      "step": 11540
    },
    {
      "epoch": 0.23641799602727664,
      "grad_norm": 0.2510731816291809,
      "learning_rate": 0.00029969091801349283,
      "loss": 4.9083,
      "step": 11545
    },
    {
      "epoch": 0.23652038580468127,
      "grad_norm": 0.24260838329792023,
      "learning_rate": 0.0002996869338582776,
      "loss": 4.8984,
      "step": 11550
    },
    {
      "epoch": 0.2366227755820859,
      "grad_norm": 0.255476176738739,
      "learning_rate": 0.0002996829242156154,
      "loss": 4.8608,
      "step": 11555
    },
    {
      "epoch": 0.2367251653594905,
      "grad_norm": 0.2573225796222687,
      "learning_rate": 0.0002996788890861889,
      "loss": 4.8959,
      "step": 11560
    },
    {
      "epoch": 0.23682755513689513,
      "grad_norm": 0.25678911805152893,
      "learning_rate": 0.0002996748284706851,
      "loss": 4.8688,
      "step": 11565
    },
    {
      "epoch": 0.23692994491429975,
      "grad_norm": 0.24863708019256592,
      "learning_rate": 0.0002996707423697955,
      "loss": 4.915,
      "step": 11570
    },
    {
      "epoch": 0.23703233469170437,
      "grad_norm": 0.27415382862091064,
      "learning_rate": 0.0002996666307842158,
      "loss": 4.861,
      "step": 11575
    },
    {
      "epoch": 0.237134724469109,
      "grad_norm": 0.2612413167953491,
      "learning_rate": 0.0002996624937146461,
      "loss": 4.9256,
      "step": 11580
    },
    {
      "epoch": 0.23723711424651364,
      "grad_norm": 0.234205961227417,
      "learning_rate": 0.00029965833116179087,
      "loss": 4.8574,
      "step": 11585
    },
    {
      "epoch": 0.23733950402391826,
      "grad_norm": 0.25088369846343994,
      "learning_rate": 0.00029965414312635886,
      "loss": 4.8845,
      "step": 11590
    },
    {
      "epoch": 0.23744189380132288,
      "grad_norm": 0.25705236196517944,
      "learning_rate": 0.00029964992960906315,
      "loss": 4.8452,
      "step": 11595
    },
    {
      "epoch": 0.2375442835787275,
      "grad_norm": 0.31214144825935364,
      "learning_rate": 0.0002996456906106211,
      "loss": 4.8657,
      "step": 11600
    },
    {
      "epoch": 0.23764667335613213,
      "grad_norm": 0.2581920027732849,
      "learning_rate": 0.0002996414261317547,
      "loss": 4.9491,
      "step": 11605
    },
    {
      "epoch": 0.23774906313353675,
      "grad_norm": 0.2636168897151947,
      "learning_rate": 0.00029963713617318986,
      "loss": 4.8617,
      "step": 11610
    },
    {
      "epoch": 0.23785145291094137,
      "grad_norm": 0.2412351369857788,
      "learning_rate": 0.00029963282073565715,
      "loss": 4.8962,
      "step": 11615
    },
    {
      "epoch": 0.237953842688346,
      "grad_norm": 0.22004848718643188,
      "learning_rate": 0.0002996284798198913,
      "loss": 4.8952,
      "step": 11620
    },
    {
      "epoch": 0.2380562324657506,
      "grad_norm": 0.24782150983810425,
      "learning_rate": 0.00029962411342663155,
      "loss": 4.8736,
      "step": 11625
    },
    {
      "epoch": 0.23815862224315523,
      "grad_norm": 0.23767095804214478,
      "learning_rate": 0.0002996197215566213,
      "loss": 4.9123,
      "step": 11630
    },
    {
      "epoch": 0.23826101202055985,
      "grad_norm": 0.23496316373348236,
      "learning_rate": 0.00029961530421060835,
      "loss": 4.8902,
      "step": 11635
    },
    {
      "epoch": 0.2383634017979645,
      "grad_norm": 0.2419152557849884,
      "learning_rate": 0.00029961086138934485,
      "loss": 4.9154,
      "step": 11640
    },
    {
      "epoch": 0.23846579157536912,
      "grad_norm": 0.2437465935945511,
      "learning_rate": 0.00029960639309358736,
      "loss": 4.8857,
      "step": 11645
    },
    {
      "epoch": 0.23856818135277374,
      "grad_norm": 0.2309345155954361,
      "learning_rate": 0.0002996018993240966,
      "loss": 4.8832,
      "step": 11650
    },
    {
      "epoch": 0.23867057113017837,
      "grad_norm": 0.2655540406703949,
      "learning_rate": 0.00029959738008163785,
      "loss": 4.881,
      "step": 11655
    },
    {
      "epoch": 0.238772960907583,
      "grad_norm": 0.24720311164855957,
      "learning_rate": 0.0002995928353669805,
      "loss": 4.9194,
      "step": 11660
    },
    {
      "epoch": 0.2388753506849876,
      "grad_norm": 0.2300921380519867,
      "learning_rate": 0.00029958826518089844,
      "loss": 4.8955,
      "step": 11665
    },
    {
      "epoch": 0.23897774046239223,
      "grad_norm": 0.25959837436676025,
      "learning_rate": 0.0002995836695241698,
      "loss": 4.8663,
      "step": 11670
    },
    {
      "epoch": 0.23908013023979685,
      "grad_norm": 0.3259102702140808,
      "learning_rate": 0.00029957904839757714,
      "loss": 4.8782,
      "step": 11675
    },
    {
      "epoch": 0.23918252001720147,
      "grad_norm": 0.3216576874256134,
      "learning_rate": 0.0002995744018019073,
      "loss": 4.8729,
      "step": 11680
    },
    {
      "epoch": 0.2392849097946061,
      "grad_norm": 0.2604944109916687,
      "learning_rate": 0.0002995697297379515,
      "loss": 4.9039,
      "step": 11685
    },
    {
      "epoch": 0.23938729957201074,
      "grad_norm": 0.2581782639026642,
      "learning_rate": 0.0002995650322065051,
      "loss": 4.8791,
      "step": 11690
    },
    {
      "epoch": 0.23948968934941536,
      "grad_norm": 0.24211914837360382,
      "learning_rate": 0.00029956030920836815,
      "loss": 4.8697,
      "step": 11695
    },
    {
      "epoch": 0.23959207912681998,
      "grad_norm": 0.24143648147583008,
      "learning_rate": 0.00029955556074434474,
      "loss": 4.8289,
      "step": 11700
    },
    {
      "epoch": 0.2396944689042246,
      "grad_norm": 0.2300914078950882,
      "learning_rate": 0.00029955078681524343,
      "loss": 4.908,
      "step": 11705
    },
    {
      "epoch": 0.23979685868162923,
      "grad_norm": 0.2471146285533905,
      "learning_rate": 0.00029954598742187703,
      "loss": 4.9411,
      "step": 11710
    },
    {
      "epoch": 0.23989924845903385,
      "grad_norm": 0.25336185097694397,
      "learning_rate": 0.0002995411625650628,
      "loss": 4.9059,
      "step": 11715
    },
    {
      "epoch": 0.24000163823643847,
      "grad_norm": 0.261109858751297,
      "learning_rate": 0.00029953631224562236,
      "loss": 4.8506,
      "step": 11720
    },
    {
      "epoch": 0.2401040280138431,
      "grad_norm": 0.24551954865455627,
      "learning_rate": 0.0002995314364643814,
      "loss": 4.9237,
      "step": 11725
    },
    {
      "epoch": 0.2402064177912477,
      "grad_norm": 0.30039453506469727,
      "learning_rate": 0.00029952653522217023,
      "loss": 4.9264,
      "step": 11730
    },
    {
      "epoch": 0.24030880756865233,
      "grad_norm": 0.24575817584991455,
      "learning_rate": 0.00029952160851982336,
      "loss": 4.8702,
      "step": 11735
    },
    {
      "epoch": 0.24041119734605698,
      "grad_norm": 0.24071602523326874,
      "learning_rate": 0.0002995166563581797,
      "loss": 4.9219,
      "step": 11740
    },
    {
      "epoch": 0.2405135871234616,
      "grad_norm": 0.2818033993244171,
      "learning_rate": 0.0002995116787380825,
      "loss": 4.8766,
      "step": 11745
    },
    {
      "epoch": 0.24061597690086622,
      "grad_norm": 0.2729192078113556,
      "learning_rate": 0.00029950667566037916,
      "loss": 4.8786,
      "step": 11750
    },
    {
      "epoch": 0.24071836667827085,
      "grad_norm": 0.24694868922233582,
      "learning_rate": 0.0002995016471259217,
      "loss": 4.9232,
      "step": 11755
    },
    {
      "epoch": 0.24082075645567547,
      "grad_norm": 0.3318852186203003,
      "learning_rate": 0.0002994965931355663,
      "loss": 4.8491,
      "step": 11760
    },
    {
      "epoch": 0.2409231462330801,
      "grad_norm": 0.24177590012550354,
      "learning_rate": 0.0002994915136901735,
      "loss": 4.899,
      "step": 11765
    },
    {
      "epoch": 0.2410255360104847,
      "grad_norm": 0.24748991429805756,
      "learning_rate": 0.00029948640879060815,
      "loss": 4.8468,
      "step": 11770
    },
    {
      "epoch": 0.24112792578788933,
      "grad_norm": 0.24905535578727722,
      "learning_rate": 0.0002994812784377396,
      "loss": 4.8605,
      "step": 11775
    },
    {
      "epoch": 0.24123031556529395,
      "grad_norm": 0.23898392915725708,
      "learning_rate": 0.0002994761226324412,
      "loss": 4.8881,
      "step": 11780
    },
    {
      "epoch": 0.24133270534269857,
      "grad_norm": 0.2341669201850891,
      "learning_rate": 0.00029947094137559106,
      "loss": 4.8815,
      "step": 11785
    },
    {
      "epoch": 0.24143509512010322,
      "grad_norm": 0.2497628629207611,
      "learning_rate": 0.00029946573466807123,
      "loss": 4.8997,
      "step": 11790
    },
    {
      "epoch": 0.24153748489750784,
      "grad_norm": 0.252549409866333,
      "learning_rate": 0.0002994605025107684,
      "loss": 4.9143,
      "step": 11795
    },
    {
      "epoch": 0.24163987467491246,
      "grad_norm": 0.2625248432159424,
      "learning_rate": 0.0002994552449045733,
      "loss": 4.9114,
      "step": 11800
    },
    {
      "epoch": 0.24174226445231708,
      "grad_norm": 0.2671903371810913,
      "learning_rate": 0.0002994499618503813,
      "loss": 4.8908,
      "step": 11805
    },
    {
      "epoch": 0.2418446542297217,
      "grad_norm": 0.2578773498535156,
      "learning_rate": 0.00029944465334909184,
      "loss": 4.9297,
      "step": 11810
    },
    {
      "epoch": 0.24194704400712633,
      "grad_norm": 0.2517646253108978,
      "learning_rate": 0.0002994393194016089,
      "loss": 4.9183,
      "step": 11815
    },
    {
      "epoch": 0.24204943378453095,
      "grad_norm": 0.23537741601467133,
      "learning_rate": 0.00029943396000884064,
      "loss": 4.91,
      "step": 11820
    },
    {
      "epoch": 0.24215182356193557,
      "grad_norm": 0.2289832979440689,
      "learning_rate": 0.0002994285751716997,
      "loss": 4.8988,
      "step": 11825
    },
    {
      "epoch": 0.2422542133393402,
      "grad_norm": 0.244520902633667,
      "learning_rate": 0.0002994231648911028,
      "loss": 4.8668,
      "step": 11830
    },
    {
      "epoch": 0.2423566031167448,
      "grad_norm": 0.2290887087583542,
      "learning_rate": 0.0002994177291679713,
      "loss": 4.8668,
      "step": 11835
    },
    {
      "epoch": 0.24245899289414946,
      "grad_norm": 0.23854729533195496,
      "learning_rate": 0.0002994122680032307,
      "loss": 4.8915,
      "step": 11840
    },
    {
      "epoch": 0.24256138267155408,
      "grad_norm": 0.2502431273460388,
      "learning_rate": 0.0002994067813978109,
      "loss": 4.8691,
      "step": 11845
    },
    {
      "epoch": 0.2426637724489587,
      "grad_norm": 0.22970812022686005,
      "learning_rate": 0.0002994012693526461,
      "loss": 4.8593,
      "step": 11850
    },
    {
      "epoch": 0.24276616222636332,
      "grad_norm": 0.23039467632770538,
      "learning_rate": 0.00029939573186867486,
      "loss": 4.8874,
      "step": 11855
    },
    {
      "epoch": 0.24286855200376795,
      "grad_norm": 0.2997787296772003,
      "learning_rate": 0.00029939016894684004,
      "loss": 4.8941,
      "step": 11860
    },
    {
      "epoch": 0.24297094178117257,
      "grad_norm": 0.2484385073184967,
      "learning_rate": 0.0002993845805880888,
      "loss": 4.8978,
      "step": 11865
    },
    {
      "epoch": 0.2430733315585772,
      "grad_norm": 0.2619800865650177,
      "learning_rate": 0.0002993789667933728,
      "loss": 4.9135,
      "step": 11870
    },
    {
      "epoch": 0.2431757213359818,
      "grad_norm": 0.24602265655994415,
      "learning_rate": 0.0002993733275636478,
      "loss": 4.8797,
      "step": 11875
    },
    {
      "epoch": 0.24327811111338643,
      "grad_norm": 0.2266729325056076,
      "learning_rate": 0.0002993676628998741,
      "loss": 4.8751,
      "step": 11880
    },
    {
      "epoch": 0.24338050089079105,
      "grad_norm": 0.2554974853992462,
      "learning_rate": 0.0002993619728030161,
      "loss": 4.8515,
      "step": 11885
    },
    {
      "epoch": 0.2434828906681957,
      "grad_norm": 0.24222640693187714,
      "learning_rate": 0.00029935625727404275,
      "loss": 4.9201,
      "step": 11890
    },
    {
      "epoch": 0.24358528044560032,
      "grad_norm": 0.2396879941225052,
      "learning_rate": 0.00029935051631392725,
      "loss": 4.8785,
      "step": 11895
    },
    {
      "epoch": 0.24368767022300494,
      "grad_norm": 0.26316046714782715,
      "learning_rate": 0.0002993447499236471,
      "loss": 4.7969,
      "step": 11900
    },
    {
      "epoch": 0.24379006000040956,
      "grad_norm": 0.2509313225746155,
      "learning_rate": 0.00029933895810418416,
      "loss": 4.8781,
      "step": 11905
    },
    {
      "epoch": 0.24389244977781419,
      "grad_norm": 0.24141843616962433,
      "learning_rate": 0.0002993331408565246,
      "loss": 4.8658,
      "step": 11910
    },
    {
      "epoch": 0.2439948395552188,
      "grad_norm": 0.2386726289987564,
      "learning_rate": 0.00029932729818165895,
      "loss": 4.8963,
      "step": 11915
    },
    {
      "epoch": 0.24409722933262343,
      "grad_norm": 0.22077973186969757,
      "learning_rate": 0.00029932143008058206,
      "loss": 4.933,
      "step": 11920
    },
    {
      "epoch": 0.24419961911002805,
      "grad_norm": 0.22215411067008972,
      "learning_rate": 0.00029931553655429305,
      "loss": 4.8624,
      "step": 11925
    },
    {
      "epoch": 0.24430200888743267,
      "grad_norm": 0.23376335203647614,
      "learning_rate": 0.00029930961760379546,
      "loss": 4.9226,
      "step": 11930
    },
    {
      "epoch": 0.2444043986648373,
      "grad_norm": 0.2362065464258194,
      "learning_rate": 0.00029930367323009716,
      "loss": 4.8627,
      "step": 11935
    },
    {
      "epoch": 0.24450678844224194,
      "grad_norm": 0.24486564099788666,
      "learning_rate": 0.0002992977034342102,
      "loss": 4.899,
      "step": 11940
    },
    {
      "epoch": 0.24460917821964656,
      "grad_norm": 0.24312803149223328,
      "learning_rate": 0.0002992917082171512,
      "loss": 4.9063,
      "step": 11945
    },
    {
      "epoch": 0.24471156799705118,
      "grad_norm": 0.22825714945793152,
      "learning_rate": 0.0002992856875799408,
      "loss": 4.8959,
      "step": 11950
    },
    {
      "epoch": 0.2448139577744558,
      "grad_norm": 0.25117015838623047,
      "learning_rate": 0.00029927964152360437,
      "loss": 4.9115,
      "step": 11955
    },
    {
      "epoch": 0.24491634755186042,
      "grad_norm": 0.23526737093925476,
      "learning_rate": 0.0002992735700491712,
      "loss": 4.8645,
      "step": 11960
    },
    {
      "epoch": 0.24501873732926505,
      "grad_norm": 0.23061364889144897,
      "learning_rate": 0.00029926747315767513,
      "loss": 4.8657,
      "step": 11965
    },
    {
      "epoch": 0.24512112710666967,
      "grad_norm": 0.22766825556755066,
      "learning_rate": 0.00029926135085015437,
      "loss": 4.8939,
      "step": 11970
    },
    {
      "epoch": 0.2452235168840743,
      "grad_norm": 0.24208544194698334,
      "learning_rate": 0.00029925520312765126,
      "loss": 4.8886,
      "step": 11975
    },
    {
      "epoch": 0.2453259066614789,
      "grad_norm": 0.26806017756462097,
      "learning_rate": 0.0002992490299912127,
      "loss": 4.8273,
      "step": 11980
    },
    {
      "epoch": 0.24542829643888353,
      "grad_norm": 0.24627339839935303,
      "learning_rate": 0.0002992428314418897,
      "loss": 4.9321,
      "step": 11985
    },
    {
      "epoch": 0.24553068621628815,
      "grad_norm": 0.24083460867404938,
      "learning_rate": 0.00029923660748073773,
      "loss": 4.8596,
      "step": 11990
    },
    {
      "epoch": 0.2456330759936928,
      "grad_norm": 0.235739067196846,
      "learning_rate": 0.0002992303581088166,
      "loss": 4.9229,
      "step": 11995
    },
    {
      "epoch": 0.24573546577109742,
      "grad_norm": 0.27138984203338623,
      "learning_rate": 0.00029922408332719037,
      "loss": 4.8903,
      "step": 12000
    },
    {
      "epoch": 0.24573546577109742,
      "eval_loss": 4.899479389190674,
      "eval_runtime": 17.2369,
      "eval_samples_per_second": 33.765,
      "eval_steps_per_second": 4.235,
      "step": 12000
    },
    {
      "epoch": 0.24583785554850204,
      "grad_norm": 0.2613738179206848,
      "learning_rate": 0.00029921778313692744,
      "loss": 4.8717,
      "step": 12005
    },
    {
      "epoch": 0.24594024532590666,
      "grad_norm": 0.281594842672348,
      "learning_rate": 0.0002992114575391005,
      "loss": 4.9472,
      "step": 12010
    },
    {
      "epoch": 0.24604263510331129,
      "grad_norm": 0.23299552500247955,
      "learning_rate": 0.0002992051065347868,
      "loss": 4.8476,
      "step": 12015
    },
    {
      "epoch": 0.2461450248807159,
      "grad_norm": 0.2404499500989914,
      "learning_rate": 0.00029919873012506753,
      "loss": 4.8642,
      "step": 12020
    },
    {
      "epoch": 0.24624741465812053,
      "grad_norm": 0.24060982465744019,
      "learning_rate": 0.00029919232831102853,
      "loss": 4.8646,
      "step": 12025
    },
    {
      "epoch": 0.24634980443552515,
      "grad_norm": 0.24198798835277557,
      "learning_rate": 0.0002991859010937598,
      "loss": 4.888,
      "step": 12030
    },
    {
      "epoch": 0.24645219421292977,
      "grad_norm": 0.24580661952495575,
      "learning_rate": 0.0002991794484743558,
      "loss": 4.9298,
      "step": 12035
    },
    {
      "epoch": 0.2465545839903344,
      "grad_norm": 0.26965638995170593,
      "learning_rate": 0.0002991729704539151,
      "loss": 4.8499,
      "step": 12040
    },
    {
      "epoch": 0.24665697376773904,
      "grad_norm": 0.2403213232755661,
      "learning_rate": 0.00029916646703354076,
      "loss": 4.8562,
      "step": 12045
    },
    {
      "epoch": 0.24675936354514366,
      "grad_norm": 0.2616066038608551,
      "learning_rate": 0.0002991599382143402,
      "loss": 4.9207,
      "step": 12050
    },
    {
      "epoch": 0.24686175332254828,
      "grad_norm": 0.2665911018848419,
      "learning_rate": 0.0002991533839974251,
      "loss": 4.8644,
      "step": 12055
    },
    {
      "epoch": 0.2469641430999529,
      "grad_norm": 0.2643393576145172,
      "learning_rate": 0.00029914680438391134,
      "loss": 4.8855,
      "step": 12060
    },
    {
      "epoch": 0.24706653287735753,
      "grad_norm": 0.26985055208206177,
      "learning_rate": 0.0002991401993749193,
      "loss": 4.8476,
      "step": 12065
    },
    {
      "epoch": 0.24716892265476215,
      "grad_norm": 0.24749569594860077,
      "learning_rate": 0.00029913356897157367,
      "loss": 4.8275,
      "step": 12070
    },
    {
      "epoch": 0.24727131243216677,
      "grad_norm": 0.24287645518779755,
      "learning_rate": 0.0002991269131750033,
      "loss": 4.8895,
      "step": 12075
    },
    {
      "epoch": 0.2473737022095714,
      "grad_norm": 0.2698119878768921,
      "learning_rate": 0.0002991202319863417,
      "loss": 4.8322,
      "step": 12080
    },
    {
      "epoch": 0.247476091986976,
      "grad_norm": 0.2555227279663086,
      "learning_rate": 0.00029911352540672626,
      "loss": 4.8419,
      "step": 12085
    },
    {
      "epoch": 0.24757848176438063,
      "grad_norm": 0.2494737207889557,
      "learning_rate": 0.00029910679343729904,
      "loss": 4.8357,
      "step": 12090
    },
    {
      "epoch": 0.24768087154178528,
      "grad_norm": 0.2341720312833786,
      "learning_rate": 0.0002991000360792063,
      "loss": 4.8621,
      "step": 12095
    },
    {
      "epoch": 0.2477832613191899,
      "grad_norm": 0.25251176953315735,
      "learning_rate": 0.00029909325333359857,
      "loss": 4.8641,
      "step": 12100
    },
    {
      "epoch": 0.24788565109659452,
      "grad_norm": 0.23488686978816986,
      "learning_rate": 0.0002990864452016308,
      "loss": 4.8845,
      "step": 12105
    },
    {
      "epoch": 0.24798804087399914,
      "grad_norm": 0.2652224004268646,
      "learning_rate": 0.0002990796116844622,
      "loss": 4.89,
      "step": 12110
    },
    {
      "epoch": 0.24809043065140376,
      "grad_norm": 0.23602823913097382,
      "learning_rate": 0.0002990727527832564,
      "loss": 4.8609,
      "step": 12115
    },
    {
      "epoch": 0.2481928204288084,
      "grad_norm": 0.2633868455886841,
      "learning_rate": 0.0002990658684991812,
      "loss": 4.9009,
      "step": 12120
    },
    {
      "epoch": 0.248295210206213,
      "grad_norm": 0.2638199031352997,
      "learning_rate": 0.0002990589588334088,
      "loss": 4.9046,
      "step": 12125
    },
    {
      "epoch": 0.24839759998361763,
      "grad_norm": 0.30601924657821655,
      "learning_rate": 0.00029905202378711573,
      "loss": 4.9206,
      "step": 12130
    },
    {
      "epoch": 0.24849998976102225,
      "grad_norm": 0.2439941018819809,
      "learning_rate": 0.00029904506336148284,
      "loss": 4.8901,
      "step": 12135
    },
    {
      "epoch": 0.24860237953842687,
      "grad_norm": 0.27119070291519165,
      "learning_rate": 0.00029903807755769535,
      "loss": 4.9295,
      "step": 12140
    },
    {
      "epoch": 0.24870476931583152,
      "grad_norm": 0.2307243049144745,
      "learning_rate": 0.00029903106637694263,
      "loss": 4.8929,
      "step": 12145
    },
    {
      "epoch": 0.24880715909323614,
      "grad_norm": 0.24744997918605804,
      "learning_rate": 0.00029902402982041854,
      "loss": 4.8785,
      "step": 12150
    },
    {
      "epoch": 0.24890954887064076,
      "grad_norm": 0.2281043976545334,
      "learning_rate": 0.0002990169678893212,
      "loss": 4.8468,
      "step": 12155
    },
    {
      "epoch": 0.24901193864804538,
      "grad_norm": 0.21781419217586517,
      "learning_rate": 0.0002990098805848531,
      "loss": 4.9168,
      "step": 12160
    },
    {
      "epoch": 0.24911432842545,
      "grad_norm": 0.23341822624206543,
      "learning_rate": 0.000299002767908221,
      "loss": 4.8919,
      "step": 12165
    },
    {
      "epoch": 0.24921671820285463,
      "grad_norm": 0.2326321005821228,
      "learning_rate": 0.0002989956298606359,
      "loss": 4.9025,
      "step": 12170
    },
    {
      "epoch": 0.24931910798025925,
      "grad_norm": 0.2548215985298157,
      "learning_rate": 0.0002989884664433133,
      "loss": 4.874,
      "step": 12175
    },
    {
      "epoch": 0.24942149775766387,
      "grad_norm": 0.27959537506103516,
      "learning_rate": 0.0002989812776574729,
      "loss": 4.779,
      "step": 12180
    },
    {
      "epoch": 0.2495238875350685,
      "grad_norm": 0.2326943725347519,
      "learning_rate": 0.0002989740635043387,
      "loss": 4.8494,
      "step": 12185
    },
    {
      "epoch": 0.2496262773124731,
      "grad_norm": 0.25300073623657227,
      "learning_rate": 0.00029896682398513914,
      "loss": 4.8717,
      "step": 12190
    },
    {
      "epoch": 0.24972866708987776,
      "grad_norm": 0.24167612195014954,
      "learning_rate": 0.0002989595591011069,
      "loss": 4.8367,
      "step": 12195
    },
    {
      "epoch": 0.24983105686728238,
      "grad_norm": 0.25171294808387756,
      "learning_rate": 0.00029895226885347893,
      "loss": 4.8535,
      "step": 12200
    },
    {
      "epoch": 0.249933446644687,
      "grad_norm": 0.44656553864479065,
      "learning_rate": 0.00029894495324349655,
      "loss": 4.8807,
      "step": 12205
    },
    {
      "epoch": 0.2500358364220916,
      "grad_norm": 0.24718737602233887,
      "learning_rate": 0.0002989376122724055,
      "loss": 4.8775,
      "step": 12210
    },
    {
      "epoch": 0.2501382261994962,
      "grad_norm": 0.25999680161476135,
      "learning_rate": 0.0002989302459414556,
      "loss": 4.8474,
      "step": 12215
    },
    {
      "epoch": 0.2502406159769009,
      "grad_norm": 0.22106848657131195,
      "learning_rate": 0.0002989228542519013,
      "loss": 4.9025,
      "step": 12220
    },
    {
      "epoch": 0.2503430057543055,
      "grad_norm": 0.24296283721923828,
      "learning_rate": 0.00029891543720500096,
      "loss": 4.9432,
      "step": 12225
    },
    {
      "epoch": 0.25044539553171014,
      "grad_norm": 0.22640258073806763,
      "learning_rate": 0.00029890799480201773,
      "loss": 4.8549,
      "step": 12230
    },
    {
      "epoch": 0.25054778530911476,
      "grad_norm": 0.23242172598838806,
      "learning_rate": 0.0002989005270442187,
      "loss": 4.8673,
      "step": 12235
    },
    {
      "epoch": 0.2506501750865194,
      "grad_norm": 0.23328714072704315,
      "learning_rate": 0.00029889303393287544,
      "loss": 4.8547,
      "step": 12240
    },
    {
      "epoch": 0.250752564863924,
      "grad_norm": 0.30010512471199036,
      "learning_rate": 0.0002988855154692638,
      "loss": 4.8451,
      "step": 12245
    },
    {
      "epoch": 0.2508549546413286,
      "grad_norm": 0.24082158505916595,
      "learning_rate": 0.0002988779716546641,
      "loss": 4.8461,
      "step": 12250
    },
    {
      "epoch": 0.25095734441873324,
      "grad_norm": 0.2572222948074341,
      "learning_rate": 0.00029887040249036066,
      "loss": 4.8194,
      "step": 12255
    },
    {
      "epoch": 0.25105973419613786,
      "grad_norm": 0.23628303408622742,
      "learning_rate": 0.00029886280797764236,
      "loss": 4.9751,
      "step": 12260
    },
    {
      "epoch": 0.2511621239735425,
      "grad_norm": 0.2489302158355713,
      "learning_rate": 0.0002988551881178023,
      "loss": 4.8359,
      "step": 12265
    },
    {
      "epoch": 0.2512645137509471,
      "grad_norm": 0.23080943524837494,
      "learning_rate": 0.0002988475429121381,
      "loss": 4.8837,
      "step": 12270
    },
    {
      "epoch": 0.2513669035283517,
      "grad_norm": 0.23010952770709991,
      "learning_rate": 0.00029883987236195125,
      "loss": 4.8853,
      "step": 12275
    },
    {
      "epoch": 0.25146929330575635,
      "grad_norm": 0.239060178399086,
      "learning_rate": 0.000298832176468548,
      "loss": 4.8867,
      "step": 12280
    },
    {
      "epoch": 0.25157168308316097,
      "grad_norm": 0.2550260126590729,
      "learning_rate": 0.0002988244552332387,
      "loss": 4.8565,
      "step": 12285
    },
    {
      "epoch": 0.2516740728605656,
      "grad_norm": 0.24920465052127838,
      "learning_rate": 0.0002988167086573381,
      "loss": 4.8484,
      "step": 12290
    },
    {
      "epoch": 0.2517764626379702,
      "grad_norm": 0.24076999723911285,
      "learning_rate": 0.00029880893674216514,
      "loss": 4.9091,
      "step": 12295
    },
    {
      "epoch": 0.25187885241537483,
      "grad_norm": 0.24830858409404755,
      "learning_rate": 0.0002988011394890432,
      "loss": 4.8945,
      "step": 12300
    },
    {
      "epoch": 0.25198124219277945,
      "grad_norm": 0.2224910408258438,
      "learning_rate": 0.0002987933168993,
      "loss": 4.851,
      "step": 12305
    },
    {
      "epoch": 0.2520836319701841,
      "grad_norm": 0.24562443792819977,
      "learning_rate": 0.0002987854689742674,
      "loss": 4.8528,
      "step": 12310
    },
    {
      "epoch": 0.2521860217475887,
      "grad_norm": 0.23242530226707458,
      "learning_rate": 0.00029877759571528164,
      "loss": 4.8868,
      "step": 12315
    },
    {
      "epoch": 0.2522884115249934,
      "grad_norm": 0.23901917040348053,
      "learning_rate": 0.0002987696971236835,
      "loss": 4.8363,
      "step": 12320
    },
    {
      "epoch": 0.252390801302398,
      "grad_norm": 0.2646287977695465,
      "learning_rate": 0.0002987617732008177,
      "loss": 4.8992,
      "step": 12325
    },
    {
      "epoch": 0.2524931910798026,
      "grad_norm": 0.24285374581813812,
      "learning_rate": 0.00029875382394803357,
      "loss": 4.8793,
      "step": 12330
    },
    {
      "epoch": 0.25259558085720724,
      "grad_norm": 0.2439645379781723,
      "learning_rate": 0.0002987458493666846,
      "loss": 4.8692,
      "step": 12335
    },
    {
      "epoch": 0.25269797063461186,
      "grad_norm": 0.2646964192390442,
      "learning_rate": 0.00029873784945812866,
      "loss": 4.8714,
      "step": 12340
    },
    {
      "epoch": 0.2528003604120165,
      "grad_norm": 0.25290536880493164,
      "learning_rate": 0.00029872982422372786,
      "loss": 4.8899,
      "step": 12345
    },
    {
      "epoch": 0.2529027501894211,
      "grad_norm": 0.2795182466506958,
      "learning_rate": 0.00029872177366484877,
      "loss": 4.828,
      "step": 12350
    },
    {
      "epoch": 0.2530051399668257,
      "grad_norm": 0.2529841959476471,
      "learning_rate": 0.000298713697782862,
      "loss": 4.8898,
      "step": 12355
    },
    {
      "epoch": 0.25310752974423034,
      "grad_norm": 0.24668854475021362,
      "learning_rate": 0.0002987055965791428,
      "loss": 4.865,
      "step": 12360
    },
    {
      "epoch": 0.25320991952163496,
      "grad_norm": 0.23832497000694275,
      "learning_rate": 0.0002986974700550706,
      "loss": 4.8724,
      "step": 12365
    },
    {
      "epoch": 0.2533123092990396,
      "grad_norm": 0.24035145342350006,
      "learning_rate": 0.000298689318212029,
      "loss": 4.8931,
      "step": 12370
    },
    {
      "epoch": 0.2534146990764442,
      "grad_norm": 0.2509881258010864,
      "learning_rate": 0.00029868114105140604,
      "loss": 4.9036,
      "step": 12375
    },
    {
      "epoch": 0.2535170888538488,
      "grad_norm": 0.4640673100948334,
      "learning_rate": 0.0002986729385745941,
      "loss": 4.8843,
      "step": 12380
    },
    {
      "epoch": 0.25361947863125345,
      "grad_norm": 0.24764907360076904,
      "learning_rate": 0.00029866471078298984,
      "loss": 4.8745,
      "step": 12385
    },
    {
      "epoch": 0.25372186840865807,
      "grad_norm": 0.23006363213062286,
      "learning_rate": 0.00029865645767799414,
      "loss": 4.8311,
      "step": 12390
    },
    {
      "epoch": 0.2538242581860627,
      "grad_norm": 0.23863214254379272,
      "learning_rate": 0.0002986481792610124,
      "loss": 4.8998,
      "step": 12395
    },
    {
      "epoch": 0.2539266479634673,
      "grad_norm": 0.2536652088165283,
      "learning_rate": 0.00029863987553345407,
      "loss": 4.8854,
      "step": 12400
    },
    {
      "epoch": 0.25402903774087193,
      "grad_norm": 0.25096675753593445,
      "learning_rate": 0.0002986315464967331,
      "loss": 4.8637,
      "step": 12405
    },
    {
      "epoch": 0.25413142751827655,
      "grad_norm": 0.25348666310310364,
      "learning_rate": 0.00029862319215226776,
      "loss": 4.8907,
      "step": 12410
    },
    {
      "epoch": 0.2542338172956812,
      "grad_norm": 0.24533510208129883,
      "learning_rate": 0.00029861481250148043,
      "loss": 4.8715,
      "step": 12415
    },
    {
      "epoch": 0.2543362070730858,
      "grad_norm": 0.23900151252746582,
      "learning_rate": 0.000298606407545798,
      "loss": 4.8733,
      "step": 12420
    },
    {
      "epoch": 0.2544385968504905,
      "grad_norm": 0.501625657081604,
      "learning_rate": 0.00029859797728665156,
      "loss": 4.8067,
      "step": 12425
    },
    {
      "epoch": 0.2545409866278951,
      "grad_norm": 0.24328076839447021,
      "learning_rate": 0.00029858952172547655,
      "loss": 4.8499,
      "step": 12430
    },
    {
      "epoch": 0.2546433764052997,
      "grad_norm": 0.2383161336183548,
      "learning_rate": 0.00029858104086371277,
      "loss": 4.8578,
      "step": 12435
    },
    {
      "epoch": 0.25474576618270434,
      "grad_norm": 0.22936564683914185,
      "learning_rate": 0.00029857253470280416,
      "loss": 4.8867,
      "step": 12440
    },
    {
      "epoch": 0.25484815596010896,
      "grad_norm": 0.2559540271759033,
      "learning_rate": 0.00029856400324419926,
      "loss": 4.8367,
      "step": 12445
    },
    {
      "epoch": 0.2549505457375136,
      "grad_norm": 0.25993111729621887,
      "learning_rate": 0.00029855544648935053,
      "loss": 4.8831,
      "step": 12450
    },
    {
      "epoch": 0.2550529355149182,
      "grad_norm": 0.22727420926094055,
      "learning_rate": 0.000298546864439715,
      "loss": 4.8524,
      "step": 12455
    },
    {
      "epoch": 0.2551553252923228,
      "grad_norm": 0.4473954141139984,
      "learning_rate": 0.0002985382570967541,
      "loss": 4.8277,
      "step": 12460
    },
    {
      "epoch": 0.25525771506972744,
      "grad_norm": 0.2686767876148224,
      "learning_rate": 0.00029852962446193324,
      "loss": 4.9113,
      "step": 12465
    },
    {
      "epoch": 0.25536010484713206,
      "grad_norm": 0.24050889909267426,
      "learning_rate": 0.0002985209665367224,
      "loss": 4.8886,
      "step": 12470
    },
    {
      "epoch": 0.2554624946245367,
      "grad_norm": 0.22935850918293,
      "learning_rate": 0.00029851228332259576,
      "loss": 4.8715,
      "step": 12475
    },
    {
      "epoch": 0.2555648844019413,
      "grad_norm": 0.25805386900901794,
      "learning_rate": 0.00029850357482103174,
      "loss": 4.8766,
      "step": 12480
    },
    {
      "epoch": 0.2556672741793459,
      "grad_norm": 0.24458548426628113,
      "learning_rate": 0.00029849484103351335,
      "loss": 4.8769,
      "step": 12485
    },
    {
      "epoch": 0.25576966395675055,
      "grad_norm": 0.2588340640068054,
      "learning_rate": 0.0002984860819615275,
      "loss": 4.9156,
      "step": 12490
    },
    {
      "epoch": 0.25587205373415517,
      "grad_norm": 0.23135457932949066,
      "learning_rate": 0.0002984772976065657,
      "loss": 4.875,
      "step": 12495
    },
    {
      "epoch": 0.2559744435115598,
      "grad_norm": 0.24211369454860687,
      "learning_rate": 0.0002984684879701237,
      "loss": 4.9047,
      "step": 12500
    },
    {
      "epoch": 0.2560768332889644,
      "grad_norm": 0.22892974317073822,
      "learning_rate": 0.0002984596530537015,
      "loss": 4.8486,
      "step": 12505
    },
    {
      "epoch": 0.25617922306636903,
      "grad_norm": 0.28492414951324463,
      "learning_rate": 0.0002984507928588035,
      "loss": 4.811,
      "step": 12510
    },
    {
      "epoch": 0.25628161284377365,
      "grad_norm": 0.2366408109664917,
      "learning_rate": 0.0002984419073869382,
      "loss": 4.8877,
      "step": 12515
    },
    {
      "epoch": 0.2563840026211783,
      "grad_norm": 0.3380194306373596,
      "learning_rate": 0.00029843299663961866,
      "loss": 4.8395,
      "step": 12520
    },
    {
      "epoch": 0.25648639239858295,
      "grad_norm": 0.25046271085739136,
      "learning_rate": 0.0002984240606183621,
      "loss": 4.8949,
      "step": 12525
    },
    {
      "epoch": 0.2565887821759876,
      "grad_norm": 0.23867449164390564,
      "learning_rate": 0.00029841509932469014,
      "loss": 4.8595,
      "step": 12530
    },
    {
      "epoch": 0.2566911719533922,
      "grad_norm": 0.2649063766002655,
      "learning_rate": 0.00029840611276012847,
      "loss": 4.8243,
      "step": 12535
    },
    {
      "epoch": 0.2567935617307968,
      "grad_norm": 0.2422143667936325,
      "learning_rate": 0.00029839710092620733,
      "loss": 4.8398,
      "step": 12540
    },
    {
      "epoch": 0.25689595150820144,
      "grad_norm": 0.23876677453517914,
      "learning_rate": 0.00029838806382446125,
      "loss": 4.8657,
      "step": 12545
    },
    {
      "epoch": 0.25699834128560606,
      "grad_norm": 0.23439767956733704,
      "learning_rate": 0.00029837900145642893,
      "loss": 4.8701,
      "step": 12550
    },
    {
      "epoch": 0.2571007310630107,
      "grad_norm": 0.22320416569709778,
      "learning_rate": 0.0002983699138236534,
      "loss": 4.8458,
      "step": 12555
    },
    {
      "epoch": 0.2572031208404153,
      "grad_norm": 0.2350194752216339,
      "learning_rate": 0.0002983608009276821,
      "loss": 4.8874,
      "step": 12560
    },
    {
      "epoch": 0.2573055106178199,
      "grad_norm": 0.2400473803281784,
      "learning_rate": 0.00029835166277006667,
      "loss": 4.8909,
      "step": 12565
    },
    {
      "epoch": 0.25740790039522454,
      "grad_norm": 0.3497767150402069,
      "learning_rate": 0.000298342499352363,
      "loss": 4.8771,
      "step": 12570
    },
    {
      "epoch": 0.25751029017262916,
      "grad_norm": 0.25395315885543823,
      "learning_rate": 0.0002983333106761315,
      "loss": 4.8964,
      "step": 12575
    },
    {
      "epoch": 0.2576126799500338,
      "grad_norm": 0.2592972218990326,
      "learning_rate": 0.00029832409674293667,
      "loss": 4.8609,
      "step": 12580
    },
    {
      "epoch": 0.2577150697274384,
      "grad_norm": 0.2572644352912903,
      "learning_rate": 0.0002983148575543474,
      "loss": 4.9023,
      "step": 12585
    },
    {
      "epoch": 0.257817459504843,
      "grad_norm": 0.23417991399765015,
      "learning_rate": 0.0002983055931119368,
      "loss": 4.8572,
      "step": 12590
    },
    {
      "epoch": 0.25791984928224765,
      "grad_norm": 0.22180429100990295,
      "learning_rate": 0.00029829630341728243,
      "loss": 4.8973,
      "step": 12595
    },
    {
      "epoch": 0.25802223905965227,
      "grad_norm": 0.219478040933609,
      "learning_rate": 0.000298286988471966,
      "loss": 4.8552,
      "step": 12600
    },
    {
      "epoch": 0.2581246288370569,
      "grad_norm": 0.34473809599876404,
      "learning_rate": 0.0002982776482775736,
      "loss": 4.8242,
      "step": 12605
    },
    {
      "epoch": 0.2582270186144615,
      "grad_norm": 0.24101996421813965,
      "learning_rate": 0.0002982682828356956,
      "loss": 4.8951,
      "step": 12610
    },
    {
      "epoch": 0.25832940839186613,
      "grad_norm": 0.2439136803150177,
      "learning_rate": 0.00029825889214792674,
      "loss": 4.8801,
      "step": 12615
    },
    {
      "epoch": 0.25843179816927075,
      "grad_norm": 0.2350088506937027,
      "learning_rate": 0.00029824947621586584,
      "loss": 4.8369,
      "step": 12620
    },
    {
      "epoch": 0.25853418794667543,
      "grad_norm": 0.24532537162303925,
      "learning_rate": 0.0002982400350411163,
      "loss": 4.8539,
      "step": 12625
    },
    {
      "epoch": 0.25863657772408005,
      "grad_norm": 0.2509115934371948,
      "learning_rate": 0.0002982305686252856,
      "loss": 4.8934,
      "step": 12630
    },
    {
      "epoch": 0.2587389675014847,
      "grad_norm": 0.2993391454219818,
      "learning_rate": 0.0002982210769699857,
      "loss": 4.8307,
      "step": 12635
    },
    {
      "epoch": 0.2588413572788893,
      "grad_norm": 0.24147208034992218,
      "learning_rate": 0.0002982115600768326,
      "loss": 4.9087,
      "step": 12640
    },
    {
      "epoch": 0.2589437470562939,
      "grad_norm": 0.2247917801141739,
      "learning_rate": 0.00029820201794744693,
      "loss": 4.8464,
      "step": 12645
    },
    {
      "epoch": 0.25904613683369854,
      "grad_norm": 0.24929684400558472,
      "learning_rate": 0.0002981924505834534,
      "loss": 4.9148,
      "step": 12650
    },
    {
      "epoch": 0.25914852661110316,
      "grad_norm": 0.2546851336956024,
      "learning_rate": 0.00029818285798648093,
      "loss": 4.8648,
      "step": 12655
    },
    {
      "epoch": 0.2592509163885078,
      "grad_norm": 0.2752739489078522,
      "learning_rate": 0.00029817324015816304,
      "loss": 4.8639,
      "step": 12660
    },
    {
      "epoch": 0.2593533061659124,
      "grad_norm": 0.2405664026737213,
      "learning_rate": 0.0002981635971001373,
      "loss": 4.8622,
      "step": 12665
    },
    {
      "epoch": 0.259455695943317,
      "grad_norm": 0.2586182951927185,
      "learning_rate": 0.0002981539288140456,
      "loss": 4.8757,
      "step": 12670
    },
    {
      "epoch": 0.25955808572072164,
      "grad_norm": 0.27839696407318115,
      "learning_rate": 0.00029814423530153425,
      "loss": 4.8243,
      "step": 12675
    },
    {
      "epoch": 0.25966047549812626,
      "grad_norm": 0.23663580417633057,
      "learning_rate": 0.0002981345165642538,
      "loss": 4.8768,
      "step": 12680
    },
    {
      "epoch": 0.2597628652755309,
      "grad_norm": 0.2404603362083435,
      "learning_rate": 0.00029812477260385896,
      "loss": 4.8903,
      "step": 12685
    },
    {
      "epoch": 0.2598652550529355,
      "grad_norm": 0.2810831367969513,
      "learning_rate": 0.000298115003422009,
      "loss": 4.8141,
      "step": 12690
    },
    {
      "epoch": 0.2599676448303401,
      "grad_norm": 0.2602687180042267,
      "learning_rate": 0.0002981052090203672,
      "loss": 4.8472,
      "step": 12695
    },
    {
      "epoch": 0.26007003460774475,
      "grad_norm": 0.23098993301391602,
      "learning_rate": 0.0002980953894006014,
      "loss": 4.8435,
      "step": 12700
    },
    {
      "epoch": 0.26017242438514937,
      "grad_norm": 0.24064970016479492,
      "learning_rate": 0.00029808554456438346,
      "loss": 4.8473,
      "step": 12705
    },
    {
      "epoch": 0.260274814162554,
      "grad_norm": 0.44505807757377625,
      "learning_rate": 0.00029807567451338977,
      "loss": 4.8631,
      "step": 12710
    },
    {
      "epoch": 0.2603772039399586,
      "grad_norm": 0.24504883587360382,
      "learning_rate": 0.0002980657792493009,
      "loss": 4.8666,
      "step": 12715
    },
    {
      "epoch": 0.26047959371736323,
      "grad_norm": 0.23710520565509796,
      "learning_rate": 0.0002980558587738018,
      "loss": 4.9392,
      "step": 12720
    },
    {
      "epoch": 0.26058198349476785,
      "grad_norm": 0.2329605668783188,
      "learning_rate": 0.00029804591308858145,
      "loss": 4.8668,
      "step": 12725
    },
    {
      "epoch": 0.26068437327217253,
      "grad_norm": 0.2457711100578308,
      "learning_rate": 0.0002980359421953335,
      "loss": 4.8803,
      "step": 12730
    },
    {
      "epoch": 0.26078676304957715,
      "grad_norm": 0.2757936120033264,
      "learning_rate": 0.0002980259460957557,
      "loss": 4.8527,
      "step": 12735
    },
    {
      "epoch": 0.2608891528269818,
      "grad_norm": 0.245131254196167,
      "learning_rate": 0.00029801592479155003,
      "loss": 4.8267,
      "step": 12740
    },
    {
      "epoch": 0.2609915426043864,
      "grad_norm": 0.236470028758049,
      "learning_rate": 0.00029800587828442284,
      "loss": 4.8362,
      "step": 12745
    },
    {
      "epoch": 0.261093932381791,
      "grad_norm": 0.22589100897312164,
      "learning_rate": 0.0002979958065760848,
      "loss": 4.7946,
      "step": 12750
    },
    {
      "epoch": 0.26119632215919564,
      "grad_norm": 0.2344709187746048,
      "learning_rate": 0.0002979857096682509,
      "loss": 4.8593,
      "step": 12755
    },
    {
      "epoch": 0.26129871193660026,
      "grad_norm": 0.24362529814243317,
      "learning_rate": 0.0002979755875626402,
      "loss": 4.8996,
      "step": 12760
    },
    {
      "epoch": 0.2614011017140049,
      "grad_norm": 0.2457917481660843,
      "learning_rate": 0.0002979654402609763,
      "loss": 4.8396,
      "step": 12765
    },
    {
      "epoch": 0.2615034914914095,
      "grad_norm": 0.25447165966033936,
      "learning_rate": 0.000297955267764987,
      "loss": 4.8407,
      "step": 12770
    },
    {
      "epoch": 0.2616058812688141,
      "grad_norm": 0.271729975938797,
      "learning_rate": 0.00029794507007640446,
      "loss": 4.8571,
      "step": 12775
    },
    {
      "epoch": 0.26170827104621874,
      "grad_norm": 0.23931005597114563,
      "learning_rate": 0.0002979348471969649,
      "loss": 4.8729,
      "step": 12780
    },
    {
      "epoch": 0.26181066082362336,
      "grad_norm": 0.241059809923172,
      "learning_rate": 0.000297924599128409,
      "loss": 4.8818,
      "step": 12785
    },
    {
      "epoch": 0.261913050601028,
      "grad_norm": 0.2745431661605835,
      "learning_rate": 0.0002979143258724819,
      "loss": 4.8635,
      "step": 12790
    },
    {
      "epoch": 0.2620154403784326,
      "grad_norm": 0.2509162425994873,
      "learning_rate": 0.0002979040274309327,
      "loss": 4.7888,
      "step": 12795
    },
    {
      "epoch": 0.26211783015583723,
      "grad_norm": 0.23292668163776398,
      "learning_rate": 0.000297893703805515,
      "loss": 4.8462,
      "step": 12800
    },
    {
      "epoch": 0.26222021993324185,
      "grad_norm": 0.24121513962745667,
      "learning_rate": 0.00029788335499798645,
      "loss": 4.8779,
      "step": 12805
    },
    {
      "epoch": 0.26232260971064647,
      "grad_norm": 0.25357139110565186,
      "learning_rate": 0.0002978729810101094,
      "loss": 4.9119,
      "step": 12810
    },
    {
      "epoch": 0.2624249994880511,
      "grad_norm": 0.2415647953748703,
      "learning_rate": 0.00029786258184365006,
      "loss": 4.8324,
      "step": 12815
    },
    {
      "epoch": 0.2625273892654557,
      "grad_norm": 0.2385522425174713,
      "learning_rate": 0.00029785215750037926,
      "loss": 4.8612,
      "step": 12820
    },
    {
      "epoch": 0.26262977904286033,
      "grad_norm": 0.24159187078475952,
      "learning_rate": 0.0002978417079820718,
      "loss": 4.7787,
      "step": 12825
    },
    {
      "epoch": 0.262732168820265,
      "grad_norm": 0.25553110241889954,
      "learning_rate": 0.0002978312332905071,
      "loss": 4.8671,
      "step": 12830
    },
    {
      "epoch": 0.26283455859766963,
      "grad_norm": 0.23924241960048676,
      "learning_rate": 0.0002978207334274686,
      "loss": 4.9007,
      "step": 12835
    },
    {
      "epoch": 0.26293694837507425,
      "grad_norm": 0.24613811075687408,
      "learning_rate": 0.0002978102083947442,
      "loss": 4.8378,
      "step": 12840
    },
    {
      "epoch": 0.2630393381524789,
      "grad_norm": 0.23306281864643097,
      "learning_rate": 0.000297799658194126,
      "loss": 4.8707,
      "step": 12845
    },
    {
      "epoch": 0.2631417279298835,
      "grad_norm": 0.23660875856876373,
      "learning_rate": 0.00029778908282741037,
      "loss": 4.893,
      "step": 12850
    },
    {
      "epoch": 0.2632441177072881,
      "grad_norm": 0.2504987418651581,
      "learning_rate": 0.00029777848229639803,
      "loss": 4.8757,
      "step": 12855
    },
    {
      "epoch": 0.26334650748469274,
      "grad_norm": 0.26087918877601624,
      "learning_rate": 0.00029776785660289387,
      "loss": 4.8782,
      "step": 12860
    },
    {
      "epoch": 0.26344889726209736,
      "grad_norm": 0.2325216382741928,
      "learning_rate": 0.0002977572057487073,
      "loss": 4.8199,
      "step": 12865
    },
    {
      "epoch": 0.263551287039502,
      "grad_norm": 0.2563367187976837,
      "learning_rate": 0.00029774652973565166,
      "loss": 4.8593,
      "step": 12870
    },
    {
      "epoch": 0.2636536768169066,
      "grad_norm": 0.23316645622253418,
      "learning_rate": 0.00029773582856554493,
      "loss": 4.8754,
      "step": 12875
    },
    {
      "epoch": 0.2637560665943112,
      "grad_norm": 0.24804489314556122,
      "learning_rate": 0.0002977251022402092,
      "loss": 4.8731,
      "step": 12880
    },
    {
      "epoch": 0.26385845637171584,
      "grad_norm": 0.23226231336593628,
      "learning_rate": 0.00029771435076147073,
      "loss": 4.9199,
      "step": 12885
    },
    {
      "epoch": 0.26396084614912046,
      "grad_norm": 0.22917960584163666,
      "learning_rate": 0.00029770357413116033,
      "loss": 4.8899,
      "step": 12890
    },
    {
      "epoch": 0.2640632359265251,
      "grad_norm": 0.24159789085388184,
      "learning_rate": 0.0002976927723511129,
      "loss": 4.8238,
      "step": 12895
    },
    {
      "epoch": 0.2641656257039297,
      "grad_norm": 0.23341481387615204,
      "learning_rate": 0.0002976819454231677,
      "loss": 4.8309,
      "step": 12900
    },
    {
      "epoch": 0.26426801548133433,
      "grad_norm": 0.23765237629413605,
      "learning_rate": 0.00029767109334916824,
      "loss": 4.8656,
      "step": 12905
    },
    {
      "epoch": 0.26437040525873895,
      "grad_norm": 0.2955288589000702,
      "learning_rate": 0.0002976602161309623,
      "loss": 4.8886,
      "step": 12910
    },
    {
      "epoch": 0.26447279503614357,
      "grad_norm": 0.2644578218460083,
      "learning_rate": 0.000297649313770402,
      "loss": 4.8946,
      "step": 12915
    },
    {
      "epoch": 0.2645751848135482,
      "grad_norm": 0.441896915435791,
      "learning_rate": 0.0002976383862693437,
      "loss": 4.8514,
      "step": 12920
    },
    {
      "epoch": 0.2646775745909528,
      "grad_norm": 0.27758267521858215,
      "learning_rate": 0.000297627433629648,
      "loss": 4.849,
      "step": 12925
    },
    {
      "epoch": 0.2647799643683575,
      "grad_norm": 0.24811238050460815,
      "learning_rate": 0.0002976164558531798,
      "loss": 4.8608,
      "step": 12930
    },
    {
      "epoch": 0.2648823541457621,
      "grad_norm": 0.2523718476295471,
      "learning_rate": 0.00029760545294180846,
      "loss": 4.8731,
      "step": 12935
    },
    {
      "epoch": 0.26498474392316673,
      "grad_norm": 0.23273690044879913,
      "learning_rate": 0.0002975944248974073,
      "loss": 4.8509,
      "step": 12940
    },
    {
      "epoch": 0.26508713370057135,
      "grad_norm": 0.2602105438709259,
      "learning_rate": 0.0002975833717218541,
      "loss": 4.905,
      "step": 12945
    },
    {
      "epoch": 0.265189523477976,
      "grad_norm": 0.23075635731220245,
      "learning_rate": 0.00029757229341703103,
      "loss": 4.8708,
      "step": 12950
    },
    {
      "epoch": 0.2652919132553806,
      "grad_norm": 0.24693787097930908,
      "learning_rate": 0.00029756118998482427,
      "loss": 4.8541,
      "step": 12955
    },
    {
      "epoch": 0.2653943030327852,
      "grad_norm": 0.25319182872772217,
      "learning_rate": 0.0002975500614271245,
      "loss": 4.8526,
      "step": 12960
    },
    {
      "epoch": 0.26549669281018984,
      "grad_norm": 0.23757408559322357,
      "learning_rate": 0.0002975389077458266,
      "loss": 4.8878,
      "step": 12965
    },
    {
      "epoch": 0.26559908258759446,
      "grad_norm": 0.2358894944190979,
      "learning_rate": 0.0002975277289428297,
      "loss": 4.8995,
      "step": 12970
    },
    {
      "epoch": 0.2657014723649991,
      "grad_norm": 0.23779599368572235,
      "learning_rate": 0.0002975165250200372,
      "loss": 4.8562,
      "step": 12975
    },
    {
      "epoch": 0.2658038621424037,
      "grad_norm": 0.23578375577926636,
      "learning_rate": 0.0002975052959793569,
      "loss": 4.8733,
      "step": 12980
    },
    {
      "epoch": 0.2659062519198083,
      "grad_norm": 0.22678780555725098,
      "learning_rate": 0.0002974940418227007,
      "loss": 4.8401,
      "step": 12985
    },
    {
      "epoch": 0.26600864169721294,
      "grad_norm": 0.2594931423664093,
      "learning_rate": 0.0002974827625519849,
      "loss": 4.8202,
      "step": 12990
    },
    {
      "epoch": 0.26611103147461757,
      "grad_norm": 0.3577810823917389,
      "learning_rate": 0.0002974714581691301,
      "loss": 4.8104,
      "step": 12995
    },
    {
      "epoch": 0.2662134212520222,
      "grad_norm": 0.2552907168865204,
      "learning_rate": 0.00029746012867606095,
      "loss": 4.8498,
      "step": 13000
    },
    {
      "epoch": 0.2662134212520222,
      "eval_loss": 4.883366584777832,
      "eval_runtime": 17.076,
      "eval_samples_per_second": 34.083,
      "eval_steps_per_second": 4.275,
      "step": 13000
    },
    {
      "epoch": 0.2663158110294268,
      "grad_norm": 0.30066636204719543,
      "learning_rate": 0.0002974487740747067,
      "loss": 4.8731,
      "step": 13005
    },
    {
      "epoch": 0.26641820080683143,
      "grad_norm": 0.2680542469024658,
      "learning_rate": 0.0002974373943670007,
      "loss": 4.8089,
      "step": 13010
    },
    {
      "epoch": 0.26652059058423605,
      "grad_norm": 0.256352037191391,
      "learning_rate": 0.0002974259895548805,
      "loss": 4.8462,
      "step": 13015
    },
    {
      "epoch": 0.26662298036164067,
      "grad_norm": 0.25665661692619324,
      "learning_rate": 0.00029741455964028807,
      "loss": 4.8946,
      "step": 13020
    },
    {
      "epoch": 0.2667253701390453,
      "grad_norm": 0.23544175922870636,
      "learning_rate": 0.00029740310462516965,
      "loss": 4.9251,
      "step": 13025
    },
    {
      "epoch": 0.26682775991644997,
      "grad_norm": 0.2365446239709854,
      "learning_rate": 0.00029739162451147567,
      "loss": 4.8693,
      "step": 13030
    },
    {
      "epoch": 0.2669301496938546,
      "grad_norm": 0.27612054347991943,
      "learning_rate": 0.00029738011930116083,
      "loss": 4.8541,
      "step": 13035
    },
    {
      "epoch": 0.2670325394712592,
      "grad_norm": 0.2788892090320587,
      "learning_rate": 0.00029736858899618413,
      "loss": 4.888,
      "step": 13040
    },
    {
      "epoch": 0.26713492924866383,
      "grad_norm": 0.23531077802181244,
      "learning_rate": 0.00029735703359850895,
      "loss": 4.8474,
      "step": 13045
    },
    {
      "epoch": 0.26723731902606845,
      "grad_norm": 0.22987061738967896,
      "learning_rate": 0.0002973454531101028,
      "loss": 4.846,
      "step": 13050
    },
    {
      "epoch": 0.2673397088034731,
      "grad_norm": 0.24541215598583221,
      "learning_rate": 0.00029733384753293743,
      "loss": 4.8158,
      "step": 13055
    },
    {
      "epoch": 0.2674420985808777,
      "grad_norm": 0.23491857945919037,
      "learning_rate": 0.0002973222168689891,
      "loss": 4.8532,
      "step": 13060
    },
    {
      "epoch": 0.2675444883582823,
      "grad_norm": 0.2576743960380554,
      "learning_rate": 0.00029731056112023807,
      "loss": 4.8559,
      "step": 13065
    },
    {
      "epoch": 0.26764687813568694,
      "grad_norm": 0.23715195059776306,
      "learning_rate": 0.000297298880288669,
      "loss": 4.8553,
      "step": 13070
    },
    {
      "epoch": 0.26774926791309156,
      "grad_norm": 0.24965982139110565,
      "learning_rate": 0.00029728717437627083,
      "loss": 4.8552,
      "step": 13075
    },
    {
      "epoch": 0.2678516576904962,
      "grad_norm": 0.2617979645729065,
      "learning_rate": 0.00029727544338503664,
      "loss": 4.8221,
      "step": 13080
    },
    {
      "epoch": 0.2679540474679008,
      "grad_norm": 0.28162577748298645,
      "learning_rate": 0.00029726368731696406,
      "loss": 4.9248,
      "step": 13085
    },
    {
      "epoch": 0.2680564372453054,
      "grad_norm": 0.3132152259349823,
      "learning_rate": 0.00029725190617405474,
      "loss": 4.8982,
      "step": 13090
    },
    {
      "epoch": 0.26815882702271004,
      "grad_norm": 0.3432159423828125,
      "learning_rate": 0.0002972400999583146,
      "loss": 4.8901,
      "step": 13095
    },
    {
      "epoch": 0.26826121680011467,
      "grad_norm": 0.26496621966362,
      "learning_rate": 0.00029722826867175403,
      "loss": 4.828,
      "step": 13100
    },
    {
      "epoch": 0.2683636065775193,
      "grad_norm": 0.26306840777397156,
      "learning_rate": 0.00029721641231638746,
      "loss": 4.8419,
      "step": 13105
    },
    {
      "epoch": 0.2684659963549239,
      "grad_norm": 0.24219603836536407,
      "learning_rate": 0.00029720453089423376,
      "loss": 4.8359,
      "step": 13110
    },
    {
      "epoch": 0.26856838613232853,
      "grad_norm": 0.26120564341545105,
      "learning_rate": 0.000297192624407316,
      "loss": 4.8603,
      "step": 13115
    },
    {
      "epoch": 0.26867077590973315,
      "grad_norm": 0.2502914369106293,
      "learning_rate": 0.00029718069285766146,
      "loss": 4.8762,
      "step": 13120
    },
    {
      "epoch": 0.26877316568713777,
      "grad_norm": 0.522968053817749,
      "learning_rate": 0.0002971687362473018,
      "loss": 4.8406,
      "step": 13125
    },
    {
      "epoch": 0.2688755554645424,
      "grad_norm": 0.23913554847240448,
      "learning_rate": 0.00029715675457827283,
      "loss": 4.8191,
      "step": 13130
    },
    {
      "epoch": 0.26897794524194707,
      "grad_norm": 0.2351391315460205,
      "learning_rate": 0.00029714474785261474,
      "loss": 4.8869,
      "step": 13135
    },
    {
      "epoch": 0.2690803350193517,
      "grad_norm": 0.26714980602264404,
      "learning_rate": 0.00029713271607237194,
      "loss": 4.8483,
      "step": 13140
    },
    {
      "epoch": 0.2691827247967563,
      "grad_norm": 0.23896341025829315,
      "learning_rate": 0.0002971206592395931,
      "loss": 4.8974,
      "step": 13145
    },
    {
      "epoch": 0.26928511457416093,
      "grad_norm": 0.28090232610702515,
      "learning_rate": 0.00029710857735633107,
      "loss": 4.8072,
      "step": 13150
    },
    {
      "epoch": 0.26938750435156555,
      "grad_norm": 0.252164363861084,
      "learning_rate": 0.0002970964704246432,
      "loss": 4.8944,
      "step": 13155
    },
    {
      "epoch": 0.2694898941289702,
      "grad_norm": 0.2368669956922531,
      "learning_rate": 0.0002970843384465908,
      "loss": 4.8508,
      "step": 13160
    },
    {
      "epoch": 0.2695922839063748,
      "grad_norm": 0.24469216167926788,
      "learning_rate": 0.0002970721814242398,
      "loss": 4.8878,
      "step": 13165
    },
    {
      "epoch": 0.2696946736837794,
      "grad_norm": 0.22728322446346283,
      "learning_rate": 0.00029705999935966004,
      "loss": 4.8618,
      "step": 13170
    },
    {
      "epoch": 0.26979706346118404,
      "grad_norm": 0.2591262757778168,
      "learning_rate": 0.0002970477922549258,
      "loss": 4.8821,
      "step": 13175
    },
    {
      "epoch": 0.26989945323858866,
      "grad_norm": 0.24526582658290863,
      "learning_rate": 0.0002970355601121156,
      "loss": 4.8319,
      "step": 13180
    },
    {
      "epoch": 0.2700018430159933,
      "grad_norm": 0.2433377355337143,
      "learning_rate": 0.00029702330293331227,
      "loss": 4.8802,
      "step": 13185
    },
    {
      "epoch": 0.2701042327933979,
      "grad_norm": 0.25172290205955505,
      "learning_rate": 0.00029701102072060293,
      "loss": 4.845,
      "step": 13190
    },
    {
      "epoch": 0.2702066225708025,
      "grad_norm": 0.26094940304756165,
      "learning_rate": 0.0002969987134760787,
      "loss": 4.8879,
      "step": 13195
    },
    {
      "epoch": 0.27030901234820714,
      "grad_norm": 0.2812451124191284,
      "learning_rate": 0.0002969863812018353,
      "loss": 4.8625,
      "step": 13200
    },
    {
      "epoch": 0.27041140212561177,
      "grad_norm": 0.2290884107351303,
      "learning_rate": 0.0002969740238999726,
      "loss": 4.8672,
      "step": 13205
    },
    {
      "epoch": 0.2705137919030164,
      "grad_norm": 0.3153294622898102,
      "learning_rate": 0.0002969616415725945,
      "loss": 4.8075,
      "step": 13210
    },
    {
      "epoch": 0.270616181680421,
      "grad_norm": 0.2929977476596832,
      "learning_rate": 0.0002969492342218096,
      "loss": 4.8449,
      "step": 13215
    },
    {
      "epoch": 0.27071857145782563,
      "grad_norm": 0.23602008819580078,
      "learning_rate": 0.00029693680184973036,
      "loss": 4.9074,
      "step": 13220
    },
    {
      "epoch": 0.27082096123523025,
      "grad_norm": 0.25510868430137634,
      "learning_rate": 0.00029692434445847374,
      "loss": 4.8755,
      "step": 13225
    },
    {
      "epoch": 0.27092335101263487,
      "grad_norm": 0.24809250235557556,
      "learning_rate": 0.0002969118620501608,
      "loss": 4.8583,
      "step": 13230
    },
    {
      "epoch": 0.27102574079003955,
      "grad_norm": 1.9793117046356201,
      "learning_rate": 0.000296899354626917,
      "loss": 4.7931,
      "step": 13235
    },
    {
      "epoch": 0.27112813056744417,
      "grad_norm": 0.2557344138622284,
      "learning_rate": 0.0002968868221908721,
      "loss": 4.8424,
      "step": 13240
    },
    {
      "epoch": 0.2712305203448488,
      "grad_norm": 0.2346993088722229,
      "learning_rate": 0.0002968742647441598,
      "loss": 4.8449,
      "step": 13245
    },
    {
      "epoch": 0.2713329101222534,
      "grad_norm": 0.24267706274986267,
      "learning_rate": 0.00029686168228891844,
      "loss": 4.847,
      "step": 13250
    },
    {
      "epoch": 0.27143529989965803,
      "grad_norm": 0.2332083135843277,
      "learning_rate": 0.0002968490748272903,
      "loss": 4.9079,
      "step": 13255
    },
    {
      "epoch": 0.27153768967706265,
      "grad_norm": 0.2350085824728012,
      "learning_rate": 0.0002968364423614223,
      "loss": 4.8773,
      "step": 13260
    },
    {
      "epoch": 0.2716400794544673,
      "grad_norm": 0.22749336063861847,
      "learning_rate": 0.00029682378489346526,
      "loss": 4.8448,
      "step": 13265
    },
    {
      "epoch": 0.2717424692318719,
      "grad_norm": 0.25906381011009216,
      "learning_rate": 0.00029681110242557433,
      "loss": 4.822,
      "step": 13270
    },
    {
      "epoch": 0.2718448590092765,
      "grad_norm": 0.24409353733062744,
      "learning_rate": 0.00029679839495990915,
      "loss": 4.848,
      "step": 13275
    },
    {
      "epoch": 0.27194724878668114,
      "grad_norm": 0.24792829155921936,
      "learning_rate": 0.00029678566249863325,
      "loss": 4.8405,
      "step": 13280
    },
    {
      "epoch": 0.27204963856408576,
      "grad_norm": 0.2603752017021179,
      "learning_rate": 0.0002967729050439147,
      "loss": 4.8137,
      "step": 13285
    },
    {
      "epoch": 0.2721520283414904,
      "grad_norm": 0.2420881986618042,
      "learning_rate": 0.00029676012259792576,
      "loss": 4.8403,
      "step": 13290
    },
    {
      "epoch": 0.272254418118895,
      "grad_norm": 0.2657824456691742,
      "learning_rate": 0.00029674731516284285,
      "loss": 4.8592,
      "step": 13295
    },
    {
      "epoch": 0.2723568078962996,
      "grad_norm": 0.24368324875831604,
      "learning_rate": 0.0002967344827408468,
      "loss": 4.8381,
      "step": 13300
    },
    {
      "epoch": 0.27245919767370425,
      "grad_norm": 0.2560962438583374,
      "learning_rate": 0.0002967216253341225,
      "loss": 4.8776,
      "step": 13305
    },
    {
      "epoch": 0.27256158745110887,
      "grad_norm": 0.27904170751571655,
      "learning_rate": 0.0002967087429448593,
      "loss": 4.8394,
      "step": 13310
    },
    {
      "epoch": 0.2726639772285135,
      "grad_norm": 0.28567177057266235,
      "learning_rate": 0.0002966958355752506,
      "loss": 4.8572,
      "step": 13315
    },
    {
      "epoch": 0.2727663670059181,
      "grad_norm": 0.2559954524040222,
      "learning_rate": 0.00029668290322749436,
      "loss": 4.8638,
      "step": 13320
    },
    {
      "epoch": 0.27286875678332273,
      "grad_norm": 0.2415838986635208,
      "learning_rate": 0.00029666994590379236,
      "loss": 4.8609,
      "step": 13325
    },
    {
      "epoch": 0.27297114656072735,
      "grad_norm": 0.2440149486064911,
      "learning_rate": 0.000296656963606351,
      "loss": 4.7883,
      "step": 13330
    },
    {
      "epoch": 0.27307353633813203,
      "grad_norm": 0.23534487187862396,
      "learning_rate": 0.00029664395633738073,
      "loss": 4.8313,
      "step": 13335
    },
    {
      "epoch": 0.27317592611553665,
      "grad_norm": 0.2263362854719162,
      "learning_rate": 0.00029663092409909633,
      "loss": 4.8468,
      "step": 13340
    },
    {
      "epoch": 0.27327831589294127,
      "grad_norm": 0.28955426812171936,
      "learning_rate": 0.00029661786689371685,
      "loss": 4.7981,
      "step": 13345
    },
    {
      "epoch": 0.2733807056703459,
      "grad_norm": 0.2760336101055145,
      "learning_rate": 0.00029660478472346563,
      "loss": 4.8296,
      "step": 13350
    },
    {
      "epoch": 0.2734830954477505,
      "grad_norm": 0.26077723503112793,
      "learning_rate": 0.00029659167759057004,
      "loss": 4.8306,
      "step": 13355
    },
    {
      "epoch": 0.27358548522515513,
      "grad_norm": 0.24385151267051697,
      "learning_rate": 0.000296578545497262,
      "loss": 4.846,
      "step": 13360
    },
    {
      "epoch": 0.27368787500255976,
      "grad_norm": 0.2572381794452667,
      "learning_rate": 0.0002965653884457774,
      "loss": 4.8641,
      "step": 13365
    },
    {
      "epoch": 0.2737902647799644,
      "grad_norm": 0.2537418007850647,
      "learning_rate": 0.0002965522064383566,
      "loss": 4.8958,
      "step": 13370
    },
    {
      "epoch": 0.273892654557369,
      "grad_norm": 0.2534486651420593,
      "learning_rate": 0.00029653899947724413,
      "loss": 4.8643,
      "step": 13375
    },
    {
      "epoch": 0.2739950443347736,
      "grad_norm": 0.45812422037124634,
      "learning_rate": 0.0002965257675646887,
      "loss": 4.8877,
      "step": 13380
    },
    {
      "epoch": 0.27409743411217824,
      "grad_norm": 0.2593996226787567,
      "learning_rate": 0.00029651251070294334,
      "loss": 4.8917,
      "step": 13385
    },
    {
      "epoch": 0.27419982388958286,
      "grad_norm": 0.261886328458786,
      "learning_rate": 0.00029649922889426534,
      "loss": 4.8464,
      "step": 13390
    },
    {
      "epoch": 0.2743022136669875,
      "grad_norm": 0.32401710748672485,
      "learning_rate": 0.0002964859221409163,
      "loss": 4.8962,
      "step": 13395
    },
    {
      "epoch": 0.2744046034443921,
      "grad_norm": 0.276602178812027,
      "learning_rate": 0.00029647259044516183,
      "loss": 4.8667,
      "step": 13400
    },
    {
      "epoch": 0.2745069932217967,
      "grad_norm": 0.24835945665836334,
      "learning_rate": 0.000296459233809272,
      "loss": 4.8001,
      "step": 13405
    },
    {
      "epoch": 0.27460938299920135,
      "grad_norm": 0.24881890416145325,
      "learning_rate": 0.00029644585223552114,
      "loss": 4.8348,
      "step": 13410
    },
    {
      "epoch": 0.27471177277660597,
      "grad_norm": 0.2437090426683426,
      "learning_rate": 0.0002964324457261877,
      "loss": 4.7979,
      "step": 13415
    },
    {
      "epoch": 0.2748141625540106,
      "grad_norm": 0.23669248819351196,
      "learning_rate": 0.0002964190142835543,
      "loss": 4.8229,
      "step": 13420
    },
    {
      "epoch": 0.2749165523314152,
      "grad_norm": 0.26821815967559814,
      "learning_rate": 0.0002964055579099081,
      "loss": 4.8096,
      "step": 13425
    },
    {
      "epoch": 0.27501894210881983,
      "grad_norm": 0.2394394725561142,
      "learning_rate": 0.0002963920766075404,
      "loss": 4.8859,
      "step": 13430
    },
    {
      "epoch": 0.27512133188622445,
      "grad_norm": 0.25903841853141785,
      "learning_rate": 0.0002963785703787465,
      "loss": 4.8699,
      "step": 13435
    },
    {
      "epoch": 0.27522372166362913,
      "grad_norm": 0.2664482295513153,
      "learning_rate": 0.0002963650392258262,
      "loss": 4.8241,
      "step": 13440
    },
    {
      "epoch": 0.27532611144103375,
      "grad_norm": 0.31281593441963196,
      "learning_rate": 0.00029635148315108347,
      "loss": 4.852,
      "step": 13445
    },
    {
      "epoch": 0.27542850121843837,
      "grad_norm": 0.24482332170009613,
      "learning_rate": 0.00029633790215682654,
      "loss": 4.8072,
      "step": 13450
    },
    {
      "epoch": 0.275530890995843,
      "grad_norm": 0.2669447064399719,
      "learning_rate": 0.0002963242962453679,
      "loss": 4.8428,
      "step": 13455
    },
    {
      "epoch": 0.2756332807732476,
      "grad_norm": 0.22922296822071075,
      "learning_rate": 0.0002963106654190242,
      "loss": 4.8868,
      "step": 13460
    },
    {
      "epoch": 0.27573567055065223,
      "grad_norm": 0.2323535829782486,
      "learning_rate": 0.0002962970096801164,
      "loss": 4.8734,
      "step": 13465
    },
    {
      "epoch": 0.27583806032805686,
      "grad_norm": 0.22814953327178955,
      "learning_rate": 0.0002962833290309697,
      "loss": 4.8149,
      "step": 13470
    },
    {
      "epoch": 0.2759404501054615,
      "grad_norm": 0.26555827260017395,
      "learning_rate": 0.00029626962347391344,
      "loss": 4.8785,
      "step": 13475
    },
    {
      "epoch": 0.2760428398828661,
      "grad_norm": 0.24419274926185608,
      "learning_rate": 0.0002962558930112814,
      "loss": 4.8504,
      "step": 13480
    },
    {
      "epoch": 0.2761452296602707,
      "grad_norm": 0.3241846263408661,
      "learning_rate": 0.00029624213764541145,
      "loss": 4.9261,
      "step": 13485
    },
    {
      "epoch": 0.27624761943767534,
      "grad_norm": 0.3510860204696655,
      "learning_rate": 0.00029622835737864577,
      "loss": 4.7729,
      "step": 13490
    },
    {
      "epoch": 0.27635000921507996,
      "grad_norm": 0.23222967982292175,
      "learning_rate": 0.00029621455221333067,
      "loss": 4.8048,
      "step": 13495
    },
    {
      "epoch": 0.2764523989924846,
      "grad_norm": 0.23177790641784668,
      "learning_rate": 0.00029620072215181685,
      "loss": 4.8359,
      "step": 13500
    },
    {
      "epoch": 0.2765547887698892,
      "grad_norm": 0.25533327460289,
      "learning_rate": 0.00029618686719645916,
      "loss": 4.8698,
      "step": 13505
    },
    {
      "epoch": 0.2766571785472938,
      "grad_norm": 0.23750196397304535,
      "learning_rate": 0.00029617298734961665,
      "loss": 4.8783,
      "step": 13510
    },
    {
      "epoch": 0.27675956832469845,
      "grad_norm": 0.23804759979248047,
      "learning_rate": 0.0002961590826136528,
      "loss": 4.8569,
      "step": 13515
    },
    {
      "epoch": 0.27686195810210307,
      "grad_norm": 0.2607920169830322,
      "learning_rate": 0.00029614515299093507,
      "loss": 4.9458,
      "step": 13520
    },
    {
      "epoch": 0.2769643478795077,
      "grad_norm": 0.2700970470905304,
      "learning_rate": 0.0002961311984838353,
      "loss": 4.8696,
      "step": 13525
    },
    {
      "epoch": 0.2770667376569123,
      "grad_norm": 0.23598410189151764,
      "learning_rate": 0.0002961172190947296,
      "loss": 4.8557,
      "step": 13530
    },
    {
      "epoch": 0.27716912743431693,
      "grad_norm": 0.2305101603269577,
      "learning_rate": 0.0002961032148259982,
      "loss": 4.8504,
      "step": 13535
    },
    {
      "epoch": 0.2772715172117216,
      "grad_norm": 0.24045445024967194,
      "learning_rate": 0.0002960891856800257,
      "loss": 4.8541,
      "step": 13540
    },
    {
      "epoch": 0.27737390698912623,
      "grad_norm": 0.24124599993228912,
      "learning_rate": 0.00029607513165920077,
      "loss": 4.8321,
      "step": 13545
    },
    {
      "epoch": 0.27747629676653085,
      "grad_norm": 0.25752729177474976,
      "learning_rate": 0.0002960610527659165,
      "loss": 4.7887,
      "step": 13550
    },
    {
      "epoch": 0.27757868654393547,
      "grad_norm": 0.24083749949932098,
      "learning_rate": 0.0002960469490025701,
      "loss": 4.7622,
      "step": 13555
    },
    {
      "epoch": 0.2776810763213401,
      "grad_norm": 0.25222840905189514,
      "learning_rate": 0.00029603282037156305,
      "loss": 4.8629,
      "step": 13560
    },
    {
      "epoch": 0.2777834660987447,
      "grad_norm": 0.23185598850250244,
      "learning_rate": 0.00029601866687530107,
      "loss": 4.8439,
      "step": 13565
    },
    {
      "epoch": 0.27788585587614933,
      "grad_norm": 0.2794724404811859,
      "learning_rate": 0.000296004488516194,
      "loss": 4.8413,
      "step": 13570
    },
    {
      "epoch": 0.27798824565355396,
      "grad_norm": 0.25876951217651367,
      "learning_rate": 0.00029599028529665614,
      "loss": 4.8725,
      "step": 13575
    },
    {
      "epoch": 0.2780906354309586,
      "grad_norm": 0.25357869267463684,
      "learning_rate": 0.0002959760572191058,
      "loss": 4.8585,
      "step": 13580
    },
    {
      "epoch": 0.2781930252083632,
      "grad_norm": 0.2559994161128998,
      "learning_rate": 0.00029596180428596565,
      "loss": 4.8744,
      "step": 13585
    },
    {
      "epoch": 0.2782954149857678,
      "grad_norm": 0.24752305448055267,
      "learning_rate": 0.0002959475264996626,
      "loss": 4.8185,
      "step": 13590
    },
    {
      "epoch": 0.27839780476317244,
      "grad_norm": 0.2554483711719513,
      "learning_rate": 0.0002959332238626278,
      "loss": 4.8734,
      "step": 13595
    },
    {
      "epoch": 0.27850019454057706,
      "grad_norm": 0.23777826130390167,
      "learning_rate": 0.0002959188963772964,
      "loss": 4.8903,
      "step": 13600
    },
    {
      "epoch": 0.2786025843179817,
      "grad_norm": 0.23869380354881287,
      "learning_rate": 0.0002959045440461081,
      "loss": 4.795,
      "step": 13605
    },
    {
      "epoch": 0.2787049740953863,
      "grad_norm": 0.23788928985595703,
      "learning_rate": 0.0002958901668715067,
      "loss": 4.8753,
      "step": 13610
    },
    {
      "epoch": 0.2788073638727909,
      "grad_norm": 0.25508829951286316,
      "learning_rate": 0.0002958757648559402,
      "loss": 4.8589,
      "step": 13615
    },
    {
      "epoch": 0.27890975365019555,
      "grad_norm": 0.2538926303386688,
      "learning_rate": 0.0002958613380018609,
      "loss": 4.9016,
      "step": 13620
    },
    {
      "epoch": 0.27901214342760017,
      "grad_norm": 0.23998358845710754,
      "learning_rate": 0.00029584688631172513,
      "loss": 4.87,
      "step": 13625
    },
    {
      "epoch": 0.2791145332050048,
      "grad_norm": 0.2575380504131317,
      "learning_rate": 0.00029583240978799374,
      "loss": 4.7988,
      "step": 13630
    },
    {
      "epoch": 0.2792169229824094,
      "grad_norm": 0.24486133456230164,
      "learning_rate": 0.00029581790843313165,
      "loss": 4.9109,
      "step": 13635
    },
    {
      "epoch": 0.2793193127598141,
      "grad_norm": 0.2378109097480774,
      "learning_rate": 0.0002958033822496081,
      "loss": 4.8434,
      "step": 13640
    },
    {
      "epoch": 0.2794217025372187,
      "grad_norm": 0.24347257614135742,
      "learning_rate": 0.00029578883123989626,
      "loss": 4.7965,
      "step": 13645
    },
    {
      "epoch": 0.27952409231462333,
      "grad_norm": 0.2288026362657547,
      "learning_rate": 0.000295774255406474,
      "loss": 4.8634,
      "step": 13650
    },
    {
      "epoch": 0.27962648209202795,
      "grad_norm": 0.22999413311481476,
      "learning_rate": 0.00029575965475182306,
      "loss": 4.8546,
      "step": 13655
    },
    {
      "epoch": 0.27972887186943257,
      "grad_norm": 0.28599053621292114,
      "learning_rate": 0.00029574502927842954,
      "loss": 4.8263,
      "step": 13660
    },
    {
      "epoch": 0.2798312616468372,
      "grad_norm": 0.23187308013439178,
      "learning_rate": 0.0002957303789887838,
      "loss": 4.8318,
      "step": 13665
    },
    {
      "epoch": 0.2799336514242418,
      "grad_norm": 0.25390827655792236,
      "learning_rate": 0.0002957157038853802,
      "loss": 4.8723,
      "step": 13670
    },
    {
      "epoch": 0.28003604120164644,
      "grad_norm": 0.25755733251571655,
      "learning_rate": 0.0002957010039707177,
      "loss": 4.8037,
      "step": 13675
    },
    {
      "epoch": 0.28013843097905106,
      "grad_norm": 0.26187583804130554,
      "learning_rate": 0.0002956862792472991,
      "loss": 4.8296,
      "step": 13680
    },
    {
      "epoch": 0.2802408207564557,
      "grad_norm": 0.2727029025554657,
      "learning_rate": 0.0002956715297176317,
      "loss": 4.8459,
      "step": 13685
    },
    {
      "epoch": 0.2803432105338603,
      "grad_norm": 0.23636601865291595,
      "learning_rate": 0.00029565675538422695,
      "loss": 4.8011,
      "step": 13690
    },
    {
      "epoch": 0.2804456003112649,
      "grad_norm": 0.23167629539966583,
      "learning_rate": 0.0002956419562496004,
      "loss": 4.834,
      "step": 13695
    },
    {
      "epoch": 0.28054799008866954,
      "grad_norm": 0.2523535490036011,
      "learning_rate": 0.0002956271323162721,
      "loss": 4.8502,
      "step": 13700
    },
    {
      "epoch": 0.28065037986607416,
      "grad_norm": 0.23229660093784332,
      "learning_rate": 0.00029561228358676594,
      "loss": 4.8333,
      "step": 13705
    },
    {
      "epoch": 0.2807527696434788,
      "grad_norm": 0.24005243182182312,
      "learning_rate": 0.0002955974100636104,
      "loss": 4.8577,
      "step": 13710
    },
    {
      "epoch": 0.2808551594208834,
      "grad_norm": 0.2420342117547989,
      "learning_rate": 0.0002955825117493379,
      "loss": 4.8757,
      "step": 13715
    },
    {
      "epoch": 0.280957549198288,
      "grad_norm": 0.2549287676811218,
      "learning_rate": 0.0002955675886464853,
      "loss": 4.8719,
      "step": 13720
    },
    {
      "epoch": 0.28105993897569265,
      "grad_norm": 0.26122424006462097,
      "learning_rate": 0.00029555264075759345,
      "loss": 4.8487,
      "step": 13725
    },
    {
      "epoch": 0.28116232875309727,
      "grad_norm": 0.24591730535030365,
      "learning_rate": 0.0002955376680852077,
      "loss": 4.8564,
      "step": 13730
    },
    {
      "epoch": 0.2812647185305019,
      "grad_norm": 0.24601192772388458,
      "learning_rate": 0.00029552267063187744,
      "loss": 4.8877,
      "step": 13735
    },
    {
      "epoch": 0.28136710830790657,
      "grad_norm": 0.31291663646698,
      "learning_rate": 0.00029550764840015624,
      "loss": 4.7991,
      "step": 13740
    },
    {
      "epoch": 0.2814694980853112,
      "grad_norm": 0.28130653500556946,
      "learning_rate": 0.000295492601392602,
      "loss": 4.8278,
      "step": 13745
    },
    {
      "epoch": 0.2815718878627158,
      "grad_norm": 0.23157049715518951,
      "learning_rate": 0.00029547752961177684,
      "loss": 4.9106,
      "step": 13750
    },
    {
      "epoch": 0.28167427764012043,
      "grad_norm": 0.25931575894355774,
      "learning_rate": 0.000295462433060247,
      "loss": 4.8456,
      "step": 13755
    },
    {
      "epoch": 0.28177666741752505,
      "grad_norm": 0.2281424105167389,
      "learning_rate": 0.00029544731174058295,
      "loss": 4.8622,
      "step": 13760
    },
    {
      "epoch": 0.28187905719492967,
      "grad_norm": 0.2262158840894699,
      "learning_rate": 0.0002954321656553596,
      "loss": 4.8308,
      "step": 13765
    },
    {
      "epoch": 0.2819814469723343,
      "grad_norm": 0.23673996329307556,
      "learning_rate": 0.0002954169948071557,
      "loss": 4.828,
      "step": 13770
    },
    {
      "epoch": 0.2820838367497389,
      "grad_norm": 0.2535232901573181,
      "learning_rate": 0.00029540179919855454,
      "loss": 4.8334,
      "step": 13775
    },
    {
      "epoch": 0.28218622652714354,
      "grad_norm": 0.25045135617256165,
      "learning_rate": 0.0002953865788321434,
      "loss": 4.8997,
      "step": 13780
    },
    {
      "epoch": 0.28228861630454816,
      "grad_norm": 0.24860410392284393,
      "learning_rate": 0.00029537133371051403,
      "loss": 4.8624,
      "step": 13785
    },
    {
      "epoch": 0.2823910060819528,
      "grad_norm": 0.240471750497818,
      "learning_rate": 0.00029535606383626216,
      "loss": 4.8817,
      "step": 13790
    },
    {
      "epoch": 0.2824933958593574,
      "grad_norm": 0.2637968957424164,
      "learning_rate": 0.0002953407692119878,
      "loss": 4.8445,
      "step": 13795
    },
    {
      "epoch": 0.282595785636762,
      "grad_norm": 0.24042394757270813,
      "learning_rate": 0.00029532544984029515,
      "loss": 4.8302,
      "step": 13800
    },
    {
      "epoch": 0.28269817541416664,
      "grad_norm": 0.2555772364139557,
      "learning_rate": 0.00029531010572379275,
      "loss": 4.8607,
      "step": 13805
    },
    {
      "epoch": 0.28280056519157126,
      "grad_norm": 0.24765047430992126,
      "learning_rate": 0.00029529473686509323,
      "loss": 4.8576,
      "step": 13810
    },
    {
      "epoch": 0.2829029549689759,
      "grad_norm": 0.2348913550376892,
      "learning_rate": 0.0002952793432668135,
      "loss": 4.8319,
      "step": 13815
    },
    {
      "epoch": 0.2830053447463805,
      "grad_norm": 0.2364525943994522,
      "learning_rate": 0.00029526392493157456,
      "loss": 4.8501,
      "step": 13820
    },
    {
      "epoch": 0.2831077345237851,
      "grad_norm": 0.2292618453502655,
      "learning_rate": 0.00029524848186200187,
      "loss": 4.8406,
      "step": 13825
    },
    {
      "epoch": 0.28321012430118975,
      "grad_norm": 0.2529616057872772,
      "learning_rate": 0.00029523301406072486,
      "loss": 4.9217,
      "step": 13830
    },
    {
      "epoch": 0.28331251407859437,
      "grad_norm": 0.2316058874130249,
      "learning_rate": 0.0002952175215303773,
      "loss": 4.8789,
      "step": 13835
    },
    {
      "epoch": 0.283414903855999,
      "grad_norm": 0.28541165590286255,
      "learning_rate": 0.000295202004273597,
      "loss": 4.786,
      "step": 13840
    },
    {
      "epoch": 0.28351729363340367,
      "grad_norm": 0.22854579985141754,
      "learning_rate": 0.00029518646229302625,
      "loss": 4.8737,
      "step": 13845
    },
    {
      "epoch": 0.2836196834108083,
      "grad_norm": 0.2418985515832901,
      "learning_rate": 0.0002951708955913113,
      "loss": 4.8302,
      "step": 13850
    },
    {
      "epoch": 0.2837220731882129,
      "grad_norm": 0.2630443871021271,
      "learning_rate": 0.0002951553041711029,
      "loss": 4.8347,
      "step": 13855
    },
    {
      "epoch": 0.28382446296561753,
      "grad_norm": 0.2776937782764435,
      "learning_rate": 0.00029513968803505567,
      "loss": 4.8634,
      "step": 13860
    },
    {
      "epoch": 0.28392685274302215,
      "grad_norm": 0.26442214846611023,
      "learning_rate": 0.0002951240471858286,
      "loss": 4.8737,
      "step": 13865
    },
    {
      "epoch": 0.2840292425204268,
      "grad_norm": 0.35495322942733765,
      "learning_rate": 0.000295108381626085,
      "loss": 4.7939,
      "step": 13870
    },
    {
      "epoch": 0.2841316322978314,
      "grad_norm": 0.22926858067512512,
      "learning_rate": 0.0002950926913584921,
      "loss": 4.8723,
      "step": 13875
    },
    {
      "epoch": 0.284234022075236,
      "grad_norm": 0.22714608907699585,
      "learning_rate": 0.00029507697638572164,
      "loss": 4.8262,
      "step": 13880
    },
    {
      "epoch": 0.28433641185264064,
      "grad_norm": 0.2394273579120636,
      "learning_rate": 0.00029506123671044936,
      "loss": 4.8742,
      "step": 13885
    },
    {
      "epoch": 0.28443880163004526,
      "grad_norm": 0.2603166103363037,
      "learning_rate": 0.0002950454723353553,
      "loss": 4.8836,
      "step": 13890
    },
    {
      "epoch": 0.2845411914074499,
      "grad_norm": 0.2738681435585022,
      "learning_rate": 0.0002950296832631238,
      "loss": 4.8528,
      "step": 13895
    },
    {
      "epoch": 0.2846435811848545,
      "grad_norm": 0.25415581464767456,
      "learning_rate": 0.0002950138694964431,
      "loss": 4.7952,
      "step": 13900
    },
    {
      "epoch": 0.2847459709622591,
      "grad_norm": 0.23789221048355103,
      "learning_rate": 0.00029499803103800597,
      "loss": 4.8537,
      "step": 13905
    },
    {
      "epoch": 0.28484836073966374,
      "grad_norm": 0.28230541944503784,
      "learning_rate": 0.0002949821678905092,
      "loss": 4.8322,
      "step": 13910
    },
    {
      "epoch": 0.28495075051706836,
      "grad_norm": 0.28278619050979614,
      "learning_rate": 0.00029496628005665383,
      "loss": 4.8186,
      "step": 13915
    },
    {
      "epoch": 0.285053140294473,
      "grad_norm": 0.2463081181049347,
      "learning_rate": 0.0002949503675391452,
      "loss": 4.8754,
      "step": 13920
    },
    {
      "epoch": 0.2851555300718776,
      "grad_norm": 0.2316562682390213,
      "learning_rate": 0.0002949344303406926,
      "loss": 4.8454,
      "step": 13925
    },
    {
      "epoch": 0.2852579198492822,
      "grad_norm": 0.2407759726047516,
      "learning_rate": 0.0002949184684640098,
      "loss": 4.8185,
      "step": 13930
    },
    {
      "epoch": 0.28536030962668685,
      "grad_norm": 0.34712404012680054,
      "learning_rate": 0.00029490248191181464,
      "loss": 4.863,
      "step": 13935
    },
    {
      "epoch": 0.28546269940409147,
      "grad_norm": 0.25207993388175964,
      "learning_rate": 0.00029488647068682915,
      "loss": 4.8482,
      "step": 13940
    },
    {
      "epoch": 0.28556508918149615,
      "grad_norm": 0.23798006772994995,
      "learning_rate": 0.0002948704347917796,
      "loss": 4.8646,
      "step": 13945
    },
    {
      "epoch": 0.28566747895890077,
      "grad_norm": 0.24008354544639587,
      "learning_rate": 0.00029485437422939643,
      "loss": 4.9122,
      "step": 13950
    },
    {
      "epoch": 0.2857698687363054,
      "grad_norm": 0.25269708037376404,
      "learning_rate": 0.00029483828900241435,
      "loss": 4.8838,
      "step": 13955
    },
    {
      "epoch": 0.28587225851371,
      "grad_norm": 0.26204586029052734,
      "learning_rate": 0.0002948221791135722,
      "loss": 4.8624,
      "step": 13960
    },
    {
      "epoch": 0.28597464829111463,
      "grad_norm": 0.2489648014307022,
      "learning_rate": 0.00029480604456561297,
      "loss": 4.8598,
      "step": 13965
    },
    {
      "epoch": 0.28607703806851925,
      "grad_norm": 0.256124883890152,
      "learning_rate": 0.000294789885361284,
      "loss": 4.8417,
      "step": 13970
    },
    {
      "epoch": 0.2861794278459239,
      "grad_norm": 0.25204986333847046,
      "learning_rate": 0.0002947737015033367,
      "loss": 4.8558,
      "step": 13975
    },
    {
      "epoch": 0.2862818176233285,
      "grad_norm": 0.23605899512767792,
      "learning_rate": 0.0002947574929945268,
      "loss": 4.8755,
      "step": 13980
    },
    {
      "epoch": 0.2863842074007331,
      "grad_norm": 0.2371349036693573,
      "learning_rate": 0.00029474125983761405,
      "loss": 4.8112,
      "step": 13985
    },
    {
      "epoch": 0.28648659717813774,
      "grad_norm": 0.2582697868347168,
      "learning_rate": 0.0002947250020353625,
      "loss": 4.7941,
      "step": 13990
    },
    {
      "epoch": 0.28658898695554236,
      "grad_norm": 0.25083622336387634,
      "learning_rate": 0.0002947087195905405,
      "loss": 4.8397,
      "step": 13995
    },
    {
      "epoch": 0.286691376732947,
      "grad_norm": 0.2635503113269806,
      "learning_rate": 0.0002946924125059204,
      "loss": 4.8535,
      "step": 14000
    },
    {
      "epoch": 0.286691376732947,
      "eval_loss": 4.864429950714111,
      "eval_runtime": 17.2602,
      "eval_samples_per_second": 33.719,
      "eval_steps_per_second": 4.229,
      "step": 14000
    },
    {
      "epoch": 0.2867937665103516,
      "grad_norm": 0.26862239837646484,
      "learning_rate": 0.00029467608078427885,
      "loss": 4.8708,
      "step": 14005
    },
    {
      "epoch": 0.2868961562877562,
      "grad_norm": 0.25329339504241943,
      "learning_rate": 0.0002946597244283967,
      "loss": 4.8642,
      "step": 14010
    },
    {
      "epoch": 0.28699854606516084,
      "grad_norm": 0.24351774156093597,
      "learning_rate": 0.00029464334344105894,
      "loss": 4.8595,
      "step": 14015
    },
    {
      "epoch": 0.28710093584256546,
      "grad_norm": 0.23963317275047302,
      "learning_rate": 0.0002946269378250548,
      "loss": 4.8637,
      "step": 14020
    },
    {
      "epoch": 0.2872033256199701,
      "grad_norm": 0.29429295659065247,
      "learning_rate": 0.0002946105075831777,
      "loss": 4.8212,
      "step": 14025
    },
    {
      "epoch": 0.2873057153973747,
      "grad_norm": 0.28569620847702026,
      "learning_rate": 0.00029459405271822526,
      "loss": 4.8509,
      "step": 14030
    },
    {
      "epoch": 0.2874081051747793,
      "grad_norm": 0.23280423879623413,
      "learning_rate": 0.00029457757323299923,
      "loss": 4.866,
      "step": 14035
    },
    {
      "epoch": 0.28751049495218395,
      "grad_norm": 0.2550869584083557,
      "learning_rate": 0.00029456106913030563,
      "loss": 4.8956,
      "step": 14040
    },
    {
      "epoch": 0.2876128847295886,
      "grad_norm": 0.3817024230957031,
      "learning_rate": 0.0002945445404129547,
      "loss": 4.8865,
      "step": 14045
    },
    {
      "epoch": 0.28771527450699325,
      "grad_norm": 0.298347145318985,
      "learning_rate": 0.00029452798708376075,
      "loss": 4.8385,
      "step": 14050
    },
    {
      "epoch": 0.28781766428439787,
      "grad_norm": 0.24775293469429016,
      "learning_rate": 0.0002945114091455423,
      "loss": 4.808,
      "step": 14055
    },
    {
      "epoch": 0.2879200540618025,
      "grad_norm": 0.24405379593372345,
      "learning_rate": 0.0002944948066011222,
      "loss": 4.82,
      "step": 14060
    },
    {
      "epoch": 0.2880224438392071,
      "grad_norm": 0.24029859900474548,
      "learning_rate": 0.0002944781794533273,
      "loss": 4.8412,
      "step": 14065
    },
    {
      "epoch": 0.28812483361661173,
      "grad_norm": 0.23408369719982147,
      "learning_rate": 0.00029446152770498885,
      "loss": 4.8452,
      "step": 14070
    },
    {
      "epoch": 0.28822722339401635,
      "grad_norm": 0.22656705975532532,
      "learning_rate": 0.00029444485135894205,
      "loss": 4.8105,
      "step": 14075
    },
    {
      "epoch": 0.288329613171421,
      "grad_norm": 0.2891269028186798,
      "learning_rate": 0.0002944281504180264,
      "loss": 4.8055,
      "step": 14080
    },
    {
      "epoch": 0.2884320029488256,
      "grad_norm": 0.23477844893932343,
      "learning_rate": 0.00029441142488508573,
      "loss": 4.8317,
      "step": 14085
    },
    {
      "epoch": 0.2885343927262302,
      "grad_norm": 0.2410440742969513,
      "learning_rate": 0.00029439467476296785,
      "loss": 4.8635,
      "step": 14090
    },
    {
      "epoch": 0.28863678250363484,
      "grad_norm": 0.2463831752538681,
      "learning_rate": 0.0002943779000545248,
      "loss": 4.8682,
      "step": 14095
    },
    {
      "epoch": 0.28873917228103946,
      "grad_norm": 0.2598053216934204,
      "learning_rate": 0.0002943611007626129,
      "loss": 4.8892,
      "step": 14100
    },
    {
      "epoch": 0.2888415620584441,
      "grad_norm": 0.260195255279541,
      "learning_rate": 0.0002943442768900926,
      "loss": 4.8778,
      "step": 14105
    },
    {
      "epoch": 0.2889439518358487,
      "grad_norm": 0.2583038806915283,
      "learning_rate": 0.00029432742843982847,
      "loss": 4.7925,
      "step": 14110
    },
    {
      "epoch": 0.2890463416132533,
      "grad_norm": 0.23612385988235474,
      "learning_rate": 0.00029431055541468933,
      "loss": 4.8462,
      "step": 14115
    },
    {
      "epoch": 0.28914873139065794,
      "grad_norm": 0.25369763374328613,
      "learning_rate": 0.0002942936578175482,
      "loss": 4.8755,
      "step": 14120
    },
    {
      "epoch": 0.28925112116806256,
      "grad_norm": 0.23460794985294342,
      "learning_rate": 0.0002942767356512823,
      "loss": 4.7736,
      "step": 14125
    },
    {
      "epoch": 0.2893535109454672,
      "grad_norm": 0.2332044243812561,
      "learning_rate": 0.00029425978891877287,
      "loss": 4.8565,
      "step": 14130
    },
    {
      "epoch": 0.2894559007228718,
      "grad_norm": 0.23149293661117554,
      "learning_rate": 0.0002942428176229056,
      "loss": 4.8087,
      "step": 14135
    },
    {
      "epoch": 0.2895582905002764,
      "grad_norm": 0.27413442730903625,
      "learning_rate": 0.0002942258217665701,
      "loss": 4.8628,
      "step": 14140
    },
    {
      "epoch": 0.2896606802776811,
      "grad_norm": 0.2448301911354065,
      "learning_rate": 0.0002942088013526604,
      "loss": 4.8488,
      "step": 14145
    },
    {
      "epoch": 0.2897630700550857,
      "grad_norm": 0.24141919612884521,
      "learning_rate": 0.0002941917563840746,
      "loss": 4.8163,
      "step": 14150
    },
    {
      "epoch": 0.28986545983249035,
      "grad_norm": 0.23401378095149994,
      "learning_rate": 0.00029417468686371485,
      "loss": 4.872,
      "step": 14155
    },
    {
      "epoch": 0.28996784960989497,
      "grad_norm": 0.23848387598991394,
      "learning_rate": 0.0002941575927944877,
      "loss": 4.8367,
      "step": 14160
    },
    {
      "epoch": 0.2900702393872996,
      "grad_norm": 0.2547352910041809,
      "learning_rate": 0.00029414047417930376,
      "loss": 4.8486,
      "step": 14165
    },
    {
      "epoch": 0.2901726291647042,
      "grad_norm": 0.2544548213481903,
      "learning_rate": 0.0002941233310210778,
      "loss": 4.8464,
      "step": 14170
    },
    {
      "epoch": 0.29027501894210883,
      "grad_norm": 0.24377994239330292,
      "learning_rate": 0.00029410616332272896,
      "loss": 4.8373,
      "step": 14175
    },
    {
      "epoch": 0.29037740871951345,
      "grad_norm": 0.2727868854999542,
      "learning_rate": 0.0002940889710871802,
      "loss": 4.836,
      "step": 14180
    },
    {
      "epoch": 0.2904797984969181,
      "grad_norm": 0.24912340939044952,
      "learning_rate": 0.00029407175431735903,
      "loss": 4.8477,
      "step": 14185
    },
    {
      "epoch": 0.2905821882743227,
      "grad_norm": 0.24339748919010162,
      "learning_rate": 0.00029405451301619693,
      "loss": 4.7966,
      "step": 14190
    },
    {
      "epoch": 0.2906845780517273,
      "grad_norm": 0.26348021626472473,
      "learning_rate": 0.00029403724718662965,
      "loss": 4.8565,
      "step": 14195
    },
    {
      "epoch": 0.29078696782913194,
      "grad_norm": 0.24618175625801086,
      "learning_rate": 0.000294019956831597,
      "loss": 4.8437,
      "step": 14200
    },
    {
      "epoch": 0.29088935760653656,
      "grad_norm": 0.24233630299568176,
      "learning_rate": 0.00029400264195404306,
      "loss": 4.8455,
      "step": 14205
    },
    {
      "epoch": 0.2909917473839412,
      "grad_norm": 0.2411973625421524,
      "learning_rate": 0.00029398530255691607,
      "loss": 4.8245,
      "step": 14210
    },
    {
      "epoch": 0.2910941371613458,
      "grad_norm": 0.27775901556015015,
      "learning_rate": 0.00029396793864316847,
      "loss": 4.8144,
      "step": 14215
    },
    {
      "epoch": 0.2911965269387504,
      "grad_norm": 0.2820267081260681,
      "learning_rate": 0.00029395055021575676,
      "loss": 4.8259,
      "step": 14220
    },
    {
      "epoch": 0.29129891671615504,
      "grad_norm": 0.26808738708496094,
      "learning_rate": 0.0002939331372776418,
      "loss": 4.8684,
      "step": 14225
    },
    {
      "epoch": 0.29140130649355966,
      "grad_norm": 0.26385441422462463,
      "learning_rate": 0.00029391569983178845,
      "loss": 4.8392,
      "step": 14230
    },
    {
      "epoch": 0.2915036962709643,
      "grad_norm": 0.2610601782798767,
      "learning_rate": 0.0002938982378811658,
      "loss": 4.8455,
      "step": 14235
    },
    {
      "epoch": 0.2916060860483689,
      "grad_norm": 0.2848125100135803,
      "learning_rate": 0.0002938807514287472,
      "loss": 4.8184,
      "step": 14240
    },
    {
      "epoch": 0.2917084758257735,
      "grad_norm": 0.2858230471611023,
      "learning_rate": 0.0002938632404775101,
      "loss": 4.7833,
      "step": 14245
    },
    {
      "epoch": 0.2918108656031782,
      "grad_norm": 0.2503651976585388,
      "learning_rate": 0.000293845705030436,
      "loss": 4.8225,
      "step": 14250
    },
    {
      "epoch": 0.2919132553805828,
      "grad_norm": 0.22477854788303375,
      "learning_rate": 0.0002938281450905108,
      "loss": 4.8776,
      "step": 14255
    },
    {
      "epoch": 0.29201564515798745,
      "grad_norm": 0.2450319230556488,
      "learning_rate": 0.0002938105606607244,
      "loss": 4.8536,
      "step": 14260
    },
    {
      "epoch": 0.29211803493539207,
      "grad_norm": 0.2509637773036957,
      "learning_rate": 0.000293792951744071,
      "loss": 4.8472,
      "step": 14265
    },
    {
      "epoch": 0.2922204247127967,
      "grad_norm": 0.24204738438129425,
      "learning_rate": 0.00029377531834354887,
      "loss": 4.8526,
      "step": 14270
    },
    {
      "epoch": 0.2923228144902013,
      "grad_norm": 0.24912716448307037,
      "learning_rate": 0.0002937576604621604,
      "loss": 4.8502,
      "step": 14275
    },
    {
      "epoch": 0.29242520426760593,
      "grad_norm": 0.25580674409866333,
      "learning_rate": 0.0002937399781029123,
      "loss": 4.861,
      "step": 14280
    },
    {
      "epoch": 0.29252759404501055,
      "grad_norm": 0.2348536103963852,
      "learning_rate": 0.0002937222712688154,
      "loss": 4.8442,
      "step": 14285
    },
    {
      "epoch": 0.2926299838224152,
      "grad_norm": 0.2764236330986023,
      "learning_rate": 0.0002937045399628847,
      "loss": 4.8065,
      "step": 14290
    },
    {
      "epoch": 0.2927323735998198,
      "grad_norm": 0.3763650357723236,
      "learning_rate": 0.0002936867841881392,
      "loss": 4.822,
      "step": 14295
    },
    {
      "epoch": 0.2928347633772244,
      "grad_norm": 0.22743472456932068,
      "learning_rate": 0.0002936690039476023,
      "loss": 4.8379,
      "step": 14300
    },
    {
      "epoch": 0.29293715315462904,
      "grad_norm": 0.23576591908931732,
      "learning_rate": 0.0002936511992443015,
      "loss": 4.8546,
      "step": 14305
    },
    {
      "epoch": 0.29303954293203366,
      "grad_norm": 0.2846321761608124,
      "learning_rate": 0.0002936333700812684,
      "loss": 4.8515,
      "step": 14310
    },
    {
      "epoch": 0.2931419327094383,
      "grad_norm": 0.2503238022327423,
      "learning_rate": 0.00029361551646153874,
      "loss": 4.8576,
      "step": 14315
    },
    {
      "epoch": 0.2932443224868429,
      "grad_norm": 0.2515864372253418,
      "learning_rate": 0.00029359763838815255,
      "loss": 4.8107,
      "step": 14320
    },
    {
      "epoch": 0.2933467122642475,
      "grad_norm": 0.25199994444847107,
      "learning_rate": 0.00029357973586415393,
      "loss": 4.8071,
      "step": 14325
    },
    {
      "epoch": 0.29344910204165214,
      "grad_norm": 0.24264228343963623,
      "learning_rate": 0.00029356180889259127,
      "loss": 4.7938,
      "step": 14330
    },
    {
      "epoch": 0.29355149181905676,
      "grad_norm": 0.2395656406879425,
      "learning_rate": 0.0002935438574765169,
      "loss": 4.8557,
      "step": 14335
    },
    {
      "epoch": 0.2936538815964614,
      "grad_norm": 0.2653576731681824,
      "learning_rate": 0.00029352588161898755,
      "loss": 4.8134,
      "step": 14340
    },
    {
      "epoch": 0.293756271373866,
      "grad_norm": 0.2938157021999359,
      "learning_rate": 0.0002935078813230639,
      "loss": 4.8175,
      "step": 14345
    },
    {
      "epoch": 0.2938586611512707,
      "grad_norm": 0.2888815999031067,
      "learning_rate": 0.0002934898565918109,
      "loss": 4.7882,
      "step": 14350
    },
    {
      "epoch": 0.2939610509286753,
      "grad_norm": 0.26561999320983887,
      "learning_rate": 0.00029347180742829773,
      "loss": 4.8123,
      "step": 14355
    },
    {
      "epoch": 0.2940634407060799,
      "grad_norm": 0.23468156158924103,
      "learning_rate": 0.00029345373383559754,
      "loss": 4.8979,
      "step": 14360
    },
    {
      "epoch": 0.29416583048348455,
      "grad_norm": 0.23520389199256897,
      "learning_rate": 0.0002934356358167879,
      "loss": 4.8703,
      "step": 14365
    },
    {
      "epoch": 0.29426822026088917,
      "grad_norm": 0.26250308752059937,
      "learning_rate": 0.0002934175133749502,
      "loss": 4.8335,
      "step": 14370
    },
    {
      "epoch": 0.2943706100382938,
      "grad_norm": 0.24580250680446625,
      "learning_rate": 0.00029339936651317036,
      "loss": 4.8437,
      "step": 14375
    },
    {
      "epoch": 0.2944729998156984,
      "grad_norm": 0.2702936828136444,
      "learning_rate": 0.0002933811952345382,
      "loss": 4.8213,
      "step": 14380
    },
    {
      "epoch": 0.29457538959310303,
      "grad_norm": 0.2317942976951599,
      "learning_rate": 0.00029336299954214775,
      "loss": 4.7984,
      "step": 14385
    },
    {
      "epoch": 0.29467777937050765,
      "grad_norm": 0.24207837879657745,
      "learning_rate": 0.00029334477943909727,
      "loss": 4.8218,
      "step": 14390
    },
    {
      "epoch": 0.2947801691479123,
      "grad_norm": 0.2444547861814499,
      "learning_rate": 0.00029332653492848904,
      "loss": 4.7892,
      "step": 14395
    },
    {
      "epoch": 0.2948825589253169,
      "grad_norm": 0.24976152181625366,
      "learning_rate": 0.0002933082660134297,
      "loss": 4.8267,
      "step": 14400
    },
    {
      "epoch": 0.2949849487027215,
      "grad_norm": 0.25552061200141907,
      "learning_rate": 0.00029328997269702987,
      "loss": 4.8644,
      "step": 14405
    },
    {
      "epoch": 0.29508733848012614,
      "grad_norm": 0.23503315448760986,
      "learning_rate": 0.0002932716549824044,
      "loss": 4.8656,
      "step": 14410
    },
    {
      "epoch": 0.29518972825753076,
      "grad_norm": 0.23296082019805908,
      "learning_rate": 0.00029325331287267226,
      "loss": 4.8821,
      "step": 14415
    },
    {
      "epoch": 0.2952921180349354,
      "grad_norm": 0.23490382730960846,
      "learning_rate": 0.0002932349463709566,
      "loss": 4.8248,
      "step": 14420
    },
    {
      "epoch": 0.29539450781234,
      "grad_norm": 0.240232452750206,
      "learning_rate": 0.00029321655548038473,
      "loss": 4.8924,
      "step": 14425
    },
    {
      "epoch": 0.2954968975897446,
      "grad_norm": 0.2854427397251129,
      "learning_rate": 0.000293198140204088,
      "loss": 4.8129,
      "step": 14430
    },
    {
      "epoch": 0.29559928736714924,
      "grad_norm": 0.24002832174301147,
      "learning_rate": 0.0002931797005452022,
      "loss": 4.8113,
      "step": 14435
    },
    {
      "epoch": 0.29570167714455386,
      "grad_norm": 0.33993467688560486,
      "learning_rate": 0.00029316123650686694,
      "loss": 4.8248,
      "step": 14440
    },
    {
      "epoch": 0.2958040669219585,
      "grad_norm": 0.23282067477703094,
      "learning_rate": 0.00029314274809222617,
      "loss": 4.8296,
      "step": 14445
    },
    {
      "epoch": 0.29590645669936316,
      "grad_norm": 0.359872967004776,
      "learning_rate": 0.00029312423530442793,
      "loss": 4.8899,
      "step": 14450
    },
    {
      "epoch": 0.2960088464767678,
      "grad_norm": 0.23167221248149872,
      "learning_rate": 0.00029310569814662443,
      "loss": 4.8045,
      "step": 14455
    },
    {
      "epoch": 0.2961112362541724,
      "grad_norm": 0.25053495168685913,
      "learning_rate": 0.00029308713662197196,
      "loss": 4.7976,
      "step": 14460
    },
    {
      "epoch": 0.296213626031577,
      "grad_norm": 0.22132182121276855,
      "learning_rate": 0.0002930685507336311,
      "loss": 4.8098,
      "step": 14465
    },
    {
      "epoch": 0.29631601580898165,
      "grad_norm": 0.2549269497394562,
      "learning_rate": 0.00029304994048476654,
      "loss": 4.8825,
      "step": 14470
    },
    {
      "epoch": 0.29641840558638627,
      "grad_norm": 0.21993915736675262,
      "learning_rate": 0.00029303130587854696,
      "loss": 4.8111,
      "step": 14475
    },
    {
      "epoch": 0.2965207953637909,
      "grad_norm": 0.24117755889892578,
      "learning_rate": 0.0002930126469181454,
      "loss": 4.8284,
      "step": 14480
    },
    {
      "epoch": 0.2966231851411955,
      "grad_norm": 0.2512553632259369,
      "learning_rate": 0.0002929939636067389,
      "loss": 4.7943,
      "step": 14485
    },
    {
      "epoch": 0.29672557491860013,
      "grad_norm": 0.4628860652446747,
      "learning_rate": 0.00029297525594750874,
      "loss": 4.8659,
      "step": 14490
    },
    {
      "epoch": 0.29682796469600475,
      "grad_norm": 0.27421388030052185,
      "learning_rate": 0.00029295652394364024,
      "loss": 4.7795,
      "step": 14495
    },
    {
      "epoch": 0.2969303544734094,
      "grad_norm": 0.23088233172893524,
      "learning_rate": 0.000292937767598323,
      "loss": 4.8242,
      "step": 14500
    },
    {
      "epoch": 0.297032744250814,
      "grad_norm": 0.2422669231891632,
      "learning_rate": 0.0002929189869147507,
      "loss": 4.8384,
      "step": 14505
    },
    {
      "epoch": 0.2971351340282186,
      "grad_norm": 0.24192313849925995,
      "learning_rate": 0.00029290018189612107,
      "loss": 4.8465,
      "step": 14510
    },
    {
      "epoch": 0.29723752380562324,
      "grad_norm": 0.2521132230758667,
      "learning_rate": 0.0002928813525456362,
      "loss": 4.8136,
      "step": 14515
    },
    {
      "epoch": 0.29733991358302786,
      "grad_norm": 0.2397879958152771,
      "learning_rate": 0.000292862498866502,
      "loss": 4.8538,
      "step": 14520
    },
    {
      "epoch": 0.2974423033604325,
      "grad_norm": 0.2553879916667938,
      "learning_rate": 0.00029284362086192897,
      "loss": 4.8258,
      "step": 14525
    },
    {
      "epoch": 0.2975446931378371,
      "grad_norm": 0.2838282287120819,
      "learning_rate": 0.0002928247185351313,
      "loss": 4.881,
      "step": 14530
    },
    {
      "epoch": 0.2976470829152417,
      "grad_norm": 0.24460811913013458,
      "learning_rate": 0.0002928057918893276,
      "loss": 4.8377,
      "step": 14535
    },
    {
      "epoch": 0.29774947269264634,
      "grad_norm": 0.2518077492713928,
      "learning_rate": 0.00029278684092774054,
      "loss": 4.8534,
      "step": 14540
    },
    {
      "epoch": 0.29785186247005097,
      "grad_norm": 0.32397088408470154,
      "learning_rate": 0.0002927678656535969,
      "loss": 4.8601,
      "step": 14545
    },
    {
      "epoch": 0.2979542522474556,
      "grad_norm": 0.25878608226776123,
      "learning_rate": 0.00029274886607012773,
      "loss": 4.8296,
      "step": 14550
    },
    {
      "epoch": 0.29805664202486026,
      "grad_norm": 0.2423200160264969,
      "learning_rate": 0.000292729842180568,
      "loss": 4.8986,
      "step": 14555
    },
    {
      "epoch": 0.2981590318022649,
      "grad_norm": 0.23571667075157166,
      "learning_rate": 0.00029271079398815697,
      "loss": 4.8347,
      "step": 14560
    },
    {
      "epoch": 0.2982614215796695,
      "grad_norm": 0.22638769447803497,
      "learning_rate": 0.00029269172149613804,
      "loss": 4.8315,
      "step": 14565
    },
    {
      "epoch": 0.2983638113570741,
      "grad_norm": 0.29264798760414124,
      "learning_rate": 0.00029267262470775873,
      "loss": 4.7274,
      "step": 14570
    },
    {
      "epoch": 0.29846620113447875,
      "grad_norm": 0.25082406401634216,
      "learning_rate": 0.0002926535036262706,
      "loss": 4.8894,
      "step": 14575
    },
    {
      "epoch": 0.29856859091188337,
      "grad_norm": 0.2945980131626129,
      "learning_rate": 0.0002926343582549295,
      "loss": 4.7709,
      "step": 14580
    },
    {
      "epoch": 0.298670980689288,
      "grad_norm": 0.253180593252182,
      "learning_rate": 0.00029261518859699536,
      "loss": 4.8286,
      "step": 14585
    },
    {
      "epoch": 0.2987733704666926,
      "grad_norm": 0.250714510679245,
      "learning_rate": 0.0002925959946557322,
      "loss": 4.8418,
      "step": 14590
    },
    {
      "epoch": 0.29887576024409723,
      "grad_norm": 0.2238171398639679,
      "learning_rate": 0.0002925767764344082,
      "loss": 4.8599,
      "step": 14595
    },
    {
      "epoch": 0.29897815002150185,
      "grad_norm": 0.2589882016181946,
      "learning_rate": 0.00029255753393629566,
      "loss": 4.8653,
      "step": 14600
    },
    {
      "epoch": 0.2990805397989065,
      "grad_norm": 0.23701760172843933,
      "learning_rate": 0.0002925382671646711,
      "loss": 4.828,
      "step": 14605
    },
    {
      "epoch": 0.2991829295763111,
      "grad_norm": 0.2523413896560669,
      "learning_rate": 0.00029251897612281503,
      "loss": 4.7975,
      "step": 14610
    },
    {
      "epoch": 0.2992853193537157,
      "grad_norm": 0.2446378916501999,
      "learning_rate": 0.00029249966081401217,
      "loss": 4.8171,
      "step": 14615
    },
    {
      "epoch": 0.29938770913112034,
      "grad_norm": 0.24434706568717957,
      "learning_rate": 0.00029248032124155144,
      "loss": 4.8778,
      "step": 14620
    },
    {
      "epoch": 0.29949009890852496,
      "grad_norm": 0.2354562133550644,
      "learning_rate": 0.0002924609574087258,
      "loss": 4.8285,
      "step": 14625
    },
    {
      "epoch": 0.2995924886859296,
      "grad_norm": 0.26459982991218567,
      "learning_rate": 0.00029244156931883226,
      "loss": 4.8441,
      "step": 14630
    },
    {
      "epoch": 0.2996948784633342,
      "grad_norm": 0.23767104744911194,
      "learning_rate": 0.0002924221569751722,
      "loss": 4.8784,
      "step": 14635
    },
    {
      "epoch": 0.2997972682407388,
      "grad_norm": 0.22710983455181122,
      "learning_rate": 0.0002924027203810509,
      "loss": 4.8236,
      "step": 14640
    },
    {
      "epoch": 0.29989965801814344,
      "grad_norm": 0.27179643511772156,
      "learning_rate": 0.00029238325953977797,
      "loss": 4.8766,
      "step": 14645
    },
    {
      "epoch": 0.30000204779554807,
      "grad_norm": 0.2793235778808594,
      "learning_rate": 0.0002923637744546669,
      "loss": 4.831,
      "step": 14650
    },
    {
      "epoch": 0.30010443757295274,
      "grad_norm": 0.23421184718608856,
      "learning_rate": 0.0002923442651290356,
      "loss": 4.8611,
      "step": 14655
    },
    {
      "epoch": 0.30020682735035736,
      "grad_norm": 0.25046485662460327,
      "learning_rate": 0.0002923247315662058,
      "loss": 4.8713,
      "step": 14660
    },
    {
      "epoch": 0.300309217127762,
      "grad_norm": 0.24280314147472382,
      "learning_rate": 0.0002923051737695036,
      "loss": 4.7816,
      "step": 14665
    },
    {
      "epoch": 0.3004116069051666,
      "grad_norm": 0.24065694212913513,
      "learning_rate": 0.0002922855917422591,
      "loss": 4.8155,
      "step": 14670
    },
    {
      "epoch": 0.3005139966825712,
      "grad_norm": 0.23348140716552734,
      "learning_rate": 0.0002922659854878067,
      "loss": 4.8206,
      "step": 14675
    },
    {
      "epoch": 0.30061638645997585,
      "grad_norm": 0.23345831036567688,
      "learning_rate": 0.0002922463550094846,
      "loss": 4.7384,
      "step": 14680
    },
    {
      "epoch": 0.30071877623738047,
      "grad_norm": 0.23435938358306885,
      "learning_rate": 0.0002922267003106354,
      "loss": 4.8781,
      "step": 14685
    },
    {
      "epoch": 0.3008211660147851,
      "grad_norm": 0.2535153925418854,
      "learning_rate": 0.00029220702139460576,
      "loss": 4.8395,
      "step": 14690
    },
    {
      "epoch": 0.3009235557921897,
      "grad_norm": 0.23041364550590515,
      "learning_rate": 0.0002921873182647464,
      "loss": 4.8476,
      "step": 14695
    },
    {
      "epoch": 0.30102594556959433,
      "grad_norm": 0.2340639978647232,
      "learning_rate": 0.00029216759092441224,
      "loss": 4.8033,
      "step": 14700
    },
    {
      "epoch": 0.30112833534699895,
      "grad_norm": 0.23467111587524414,
      "learning_rate": 0.00029214783937696223,
      "loss": 4.8225,
      "step": 14705
    },
    {
      "epoch": 0.3012307251244036,
      "grad_norm": 0.2542557120323181,
      "learning_rate": 0.00029212806362575956,
      "loss": 4.8653,
      "step": 14710
    },
    {
      "epoch": 0.3013331149018082,
      "grad_norm": 0.25412076711654663,
      "learning_rate": 0.00029210826367417143,
      "loss": 4.8282,
      "step": 14715
    },
    {
      "epoch": 0.3014355046792128,
      "grad_norm": 0.23797860741615295,
      "learning_rate": 0.0002920884395255693,
      "loss": 4.8049,
      "step": 14720
    },
    {
      "epoch": 0.30153789445661744,
      "grad_norm": 0.2716714143753052,
      "learning_rate": 0.0002920685911833286,
      "loss": 4.8646,
      "step": 14725
    },
    {
      "epoch": 0.30164028423402206,
      "grad_norm": 0.23290424048900604,
      "learning_rate": 0.0002920487186508289,
      "loss": 4.8195,
      "step": 14730
    },
    {
      "epoch": 0.3017426740114267,
      "grad_norm": 0.23280218243598938,
      "learning_rate": 0.000292028821931454,
      "loss": 4.833,
      "step": 14735
    },
    {
      "epoch": 0.3018450637888313,
      "grad_norm": 0.2538447380065918,
      "learning_rate": 0.0002920089010285917,
      "loss": 4.8369,
      "step": 14740
    },
    {
      "epoch": 0.3019474535662359,
      "grad_norm": 0.24996918439865112,
      "learning_rate": 0.00029198895594563404,
      "loss": 4.8894,
      "step": 14745
    },
    {
      "epoch": 0.30204984334364054,
      "grad_norm": 0.2496694177389145,
      "learning_rate": 0.000291968986685977,
      "loss": 4.8217,
      "step": 14750
    },
    {
      "epoch": 0.3021522331210452,
      "grad_norm": 0.24708005785942078,
      "learning_rate": 0.00029194899325302084,
      "loss": 4.8339,
      "step": 14755
    },
    {
      "epoch": 0.30225462289844984,
      "grad_norm": 0.2296389639377594,
      "learning_rate": 0.0002919289756501698,
      "loss": 4.799,
      "step": 14760
    },
    {
      "epoch": 0.30235701267585446,
      "grad_norm": 0.24229353666305542,
      "learning_rate": 0.00029190893388083243,
      "loss": 4.7975,
      "step": 14765
    },
    {
      "epoch": 0.3024594024532591,
      "grad_norm": 0.2631322145462036,
      "learning_rate": 0.00029188886794842127,
      "loss": 4.7853,
      "step": 14770
    },
    {
      "epoch": 0.3025617922306637,
      "grad_norm": 0.2699955105781555,
      "learning_rate": 0.0002918687778563529,
      "loss": 4.8729,
      "step": 14775
    },
    {
      "epoch": 0.3026641820080683,
      "grad_norm": 0.2579222321510315,
      "learning_rate": 0.0002918486636080481,
      "loss": 4.8245,
      "step": 14780
    },
    {
      "epoch": 0.30276657178547295,
      "grad_norm": 0.23751991987228394,
      "learning_rate": 0.0002918285252069318,
      "loss": 4.8287,
      "step": 14785
    },
    {
      "epoch": 0.30286896156287757,
      "grad_norm": 0.24596819281578064,
      "learning_rate": 0.000291808362656433,
      "loss": 4.85,
      "step": 14790
    },
    {
      "epoch": 0.3029713513402822,
      "grad_norm": 0.24527274072170258,
      "learning_rate": 0.0002917881759599848,
      "loss": 4.8079,
      "step": 14795
    },
    {
      "epoch": 0.3030737411176868,
      "grad_norm": 0.2361818104982376,
      "learning_rate": 0.00029176796512102435,
      "loss": 4.8187,
      "step": 14800
    },
    {
      "epoch": 0.30317613089509143,
      "grad_norm": 0.2400951236486435,
      "learning_rate": 0.0002917477301429931,
      "loss": 4.8427,
      "step": 14805
    },
    {
      "epoch": 0.30327852067249605,
      "grad_norm": 0.23457743227481842,
      "learning_rate": 0.0002917274710293364,
      "loss": 4.8784,
      "step": 14810
    },
    {
      "epoch": 0.3033809104499007,
      "grad_norm": 0.23099301755428314,
      "learning_rate": 0.0002917071877835039,
      "loss": 4.8478,
      "step": 14815
    },
    {
      "epoch": 0.3034833002273053,
      "grad_norm": 0.29923340678215027,
      "learning_rate": 0.00029168688040894923,
      "loss": 4.8072,
      "step": 14820
    },
    {
      "epoch": 0.3035856900047099,
      "grad_norm": 0.23164720833301544,
      "learning_rate": 0.00029166654890913006,
      "loss": 4.82,
      "step": 14825
    },
    {
      "epoch": 0.30368807978211454,
      "grad_norm": 0.2423633486032486,
      "learning_rate": 0.0002916461932875084,
      "loss": 4.7873,
      "step": 14830
    },
    {
      "epoch": 0.30379046955951916,
      "grad_norm": 0.4875575304031372,
      "learning_rate": 0.0002916258135475502,
      "loss": 4.8053,
      "step": 14835
    },
    {
      "epoch": 0.3038928593369238,
      "grad_norm": 0.2630945146083832,
      "learning_rate": 0.0002916054096927255,
      "loss": 4.8467,
      "step": 14840
    },
    {
      "epoch": 0.3039952491143284,
      "grad_norm": 0.24198538064956665,
      "learning_rate": 0.0002915849817265085,
      "loss": 4.81,
      "step": 14845
    },
    {
      "epoch": 0.304097638891733,
      "grad_norm": 0.2396472841501236,
      "learning_rate": 0.0002915645296523777,
      "loss": 4.8306,
      "step": 14850
    },
    {
      "epoch": 0.3042000286691377,
      "grad_norm": 0.267425000667572,
      "learning_rate": 0.0002915440534738152,
      "loss": 4.8202,
      "step": 14855
    },
    {
      "epoch": 0.3043024184465423,
      "grad_norm": 0.25262871384620667,
      "learning_rate": 0.0002915235531943077,
      "loss": 4.8405,
      "step": 14860
    },
    {
      "epoch": 0.30440480822394694,
      "grad_norm": 0.25218236446380615,
      "learning_rate": 0.00029150302881734584,
      "loss": 4.8296,
      "step": 14865
    },
    {
      "epoch": 0.30450719800135156,
      "grad_norm": 0.2474459409713745,
      "learning_rate": 0.00029148248034642423,
      "loss": 4.8403,
      "step": 14870
    },
    {
      "epoch": 0.3046095877787562,
      "grad_norm": 0.27986475825309753,
      "learning_rate": 0.00029146190778504176,
      "loss": 4.7722,
      "step": 14875
    },
    {
      "epoch": 0.3047119775561608,
      "grad_norm": 0.24763359129428864,
      "learning_rate": 0.00029144131113670134,
      "loss": 4.8281,
      "step": 14880
    },
    {
      "epoch": 0.30481436733356543,
      "grad_norm": 0.23273034393787384,
      "learning_rate": 0.00029142069040490997,
      "loss": 4.7693,
      "step": 14885
    },
    {
      "epoch": 0.30491675711097005,
      "grad_norm": 0.26339665055274963,
      "learning_rate": 0.00029140004559317884,
      "loss": 4.8096,
      "step": 14890
    },
    {
      "epoch": 0.30501914688837467,
      "grad_norm": 0.2356628030538559,
      "learning_rate": 0.0002913793767050231,
      "loss": 4.8152,
      "step": 14895
    },
    {
      "epoch": 0.3051215366657793,
      "grad_norm": 0.23748421669006348,
      "learning_rate": 0.00029135868374396214,
      "loss": 4.8254,
      "step": 14900
    },
    {
      "epoch": 0.3052239264431839,
      "grad_norm": 0.23817649483680725,
      "learning_rate": 0.00029133796671351937,
      "loss": 4.851,
      "step": 14905
    },
    {
      "epoch": 0.30532631622058853,
      "grad_norm": 0.23400908708572388,
      "learning_rate": 0.0002913172256172223,
      "loss": 4.8095,
      "step": 14910
    },
    {
      "epoch": 0.30542870599799316,
      "grad_norm": 0.2905957102775574,
      "learning_rate": 0.00029129646045860253,
      "loss": 4.8246,
      "step": 14915
    },
    {
      "epoch": 0.3055310957753978,
      "grad_norm": 0.2719612419605255,
      "learning_rate": 0.00029127567124119584,
      "loss": 4.7834,
      "step": 14920
    },
    {
      "epoch": 0.3056334855528024,
      "grad_norm": 0.2600967586040497,
      "learning_rate": 0.00029125485796854204,
      "loss": 4.8093,
      "step": 14925
    },
    {
      "epoch": 0.305735875330207,
      "grad_norm": 0.2781115472316742,
      "learning_rate": 0.00029123402064418495,
      "loss": 4.8964,
      "step": 14930
    },
    {
      "epoch": 0.30583826510761164,
      "grad_norm": 0.23672987520694733,
      "learning_rate": 0.00029121315927167264,
      "loss": 4.7962,
      "step": 14935
    },
    {
      "epoch": 0.30594065488501626,
      "grad_norm": 0.2530001401901245,
      "learning_rate": 0.00029119227385455726,
      "loss": 4.8588,
      "step": 14940
    },
    {
      "epoch": 0.3060430446624209,
      "grad_norm": 0.2906632423400879,
      "learning_rate": 0.0002911713643963949,
      "loss": 4.8293,
      "step": 14945
    },
    {
      "epoch": 0.3061454344398255,
      "grad_norm": 0.30496054887771606,
      "learning_rate": 0.0002911504309007459,
      "loss": 4.8093,
      "step": 14950
    },
    {
      "epoch": 0.3062478242172301,
      "grad_norm": 0.24770379066467285,
      "learning_rate": 0.00029112947337117467,
      "loss": 4.8286,
      "step": 14955
    },
    {
      "epoch": 0.3063502139946348,
      "grad_norm": 0.23954397439956665,
      "learning_rate": 0.00029110849181124965,
      "loss": 4.8892,
      "step": 14960
    },
    {
      "epoch": 0.3064526037720394,
      "grad_norm": 0.23769916594028473,
      "learning_rate": 0.00029108748622454347,
      "loss": 4.8276,
      "step": 14965
    },
    {
      "epoch": 0.30655499354944404,
      "grad_norm": 0.24345830082893372,
      "learning_rate": 0.0002910664566146327,
      "loss": 4.7806,
      "step": 14970
    },
    {
      "epoch": 0.30665738332684866,
      "grad_norm": 0.25595176219940186,
      "learning_rate": 0.00029104540298509807,
      "loss": 4.8298,
      "step": 14975
    },
    {
      "epoch": 0.3067597731042533,
      "grad_norm": 0.28201404213905334,
      "learning_rate": 0.0002910243253395245,
      "loss": 4.8136,
      "step": 14980
    },
    {
      "epoch": 0.3068621628816579,
      "grad_norm": 0.24373804032802582,
      "learning_rate": 0.0002910032236815009,
      "loss": 4.833,
      "step": 14985
    },
    {
      "epoch": 0.30696455265906253,
      "grad_norm": 0.2403092235326767,
      "learning_rate": 0.00029098209801462025,
      "loss": 4.8195,
      "step": 14990
    },
    {
      "epoch": 0.30706694243646715,
      "grad_norm": 0.2325127273797989,
      "learning_rate": 0.0002909609483424796,
      "loss": 4.8381,
      "step": 14995
    },
    {
      "epoch": 0.30716933221387177,
      "grad_norm": 0.24421493709087372,
      "learning_rate": 0.0002909397746686803,
      "loss": 4.7955,
      "step": 15000
    },
    {
      "epoch": 0.30716933221387177,
      "eval_loss": 4.847011089324951,
      "eval_runtime": 17.2506,
      "eval_samples_per_second": 33.738,
      "eval_steps_per_second": 4.232,
      "step": 15000
    },
    {
      "epoch": 0.3072717219912764,
      "grad_norm": 0.2475053369998932,
      "learning_rate": 0.0002909185769968275,
      "loss": 4.8524,
      "step": 15005
    },
    {
      "epoch": 0.307374111768681,
      "grad_norm": 0.2444298416376114,
      "learning_rate": 0.0002908973553305307,
      "loss": 4.8274,
      "step": 15010
    },
    {
      "epoch": 0.30747650154608563,
      "grad_norm": 0.25403663516044617,
      "learning_rate": 0.0002908761096734031,
      "loss": 4.848,
      "step": 15015
    },
    {
      "epoch": 0.30757889132349026,
      "grad_norm": 0.2355048805475235,
      "learning_rate": 0.0002908548400290625,
      "loss": 4.8068,
      "step": 15020
    },
    {
      "epoch": 0.3076812811008949,
      "grad_norm": 0.2571440041065216,
      "learning_rate": 0.00029083354640113034,
      "loss": 4.8013,
      "step": 15025
    },
    {
      "epoch": 0.3077836708782995,
      "grad_norm": 0.2513095736503601,
      "learning_rate": 0.0002908122287932324,
      "loss": 4.808,
      "step": 15030
    },
    {
      "epoch": 0.3078860606557041,
      "grad_norm": 0.2425457090139389,
      "learning_rate": 0.00029079088720899846,
      "loss": 4.8485,
      "step": 15035
    },
    {
      "epoch": 0.30798845043310874,
      "grad_norm": 0.2379244714975357,
      "learning_rate": 0.00029076952165206233,
      "loss": 4.8135,
      "step": 15040
    },
    {
      "epoch": 0.30809084021051336,
      "grad_norm": 0.2496187686920166,
      "learning_rate": 0.00029074813212606206,
      "loss": 4.7596,
      "step": 15045
    },
    {
      "epoch": 0.308193229987918,
      "grad_norm": 0.293245792388916,
      "learning_rate": 0.0002907267186346396,
      "loss": 4.8237,
      "step": 15050
    },
    {
      "epoch": 0.3082956197653226,
      "grad_norm": 0.26889434456825256,
      "learning_rate": 0.0002907052811814411,
      "loss": 4.7961,
      "step": 15055
    },
    {
      "epoch": 0.3083980095427273,
      "grad_norm": 0.24317938089370728,
      "learning_rate": 0.0002906838197701167,
      "loss": 4.7785,
      "step": 15060
    },
    {
      "epoch": 0.3085003993201319,
      "grad_norm": 0.2693104147911072,
      "learning_rate": 0.00029066233440432077,
      "loss": 4.7457,
      "step": 15065
    },
    {
      "epoch": 0.3086027890975365,
      "grad_norm": 0.24795982241630554,
      "learning_rate": 0.0002906408250877116,
      "loss": 4.8225,
      "step": 15070
    },
    {
      "epoch": 0.30870517887494114,
      "grad_norm": 0.2525266408920288,
      "learning_rate": 0.00029061929182395155,
      "loss": 4.84,
      "step": 15075
    },
    {
      "epoch": 0.30880756865234577,
      "grad_norm": 0.2872849702835083,
      "learning_rate": 0.0002905977346167072,
      "loss": 4.8114,
      "step": 15080
    },
    {
      "epoch": 0.3089099584297504,
      "grad_norm": 0.256111204624176,
      "learning_rate": 0.00029057615346964917,
      "loss": 4.7953,
      "step": 15085
    },
    {
      "epoch": 0.309012348207155,
      "grad_norm": 0.25090688467025757,
      "learning_rate": 0.0002905545483864521,
      "loss": 4.8147,
      "step": 15090
    },
    {
      "epoch": 0.30911473798455963,
      "grad_norm": 0.24256761372089386,
      "learning_rate": 0.00029053291937079464,
      "loss": 4.8341,
      "step": 15095
    },
    {
      "epoch": 0.30921712776196425,
      "grad_norm": 0.29538941383361816,
      "learning_rate": 0.0002905112664263597,
      "loss": 4.8359,
      "step": 15100
    },
    {
      "epoch": 0.30931951753936887,
      "grad_norm": 0.3012793958187103,
      "learning_rate": 0.000290489589556834,
      "loss": 4.7685,
      "step": 15105
    },
    {
      "epoch": 0.3094219073167735,
      "grad_norm": 0.24757792055606842,
      "learning_rate": 0.0002904678887659088,
      "loss": 4.8886,
      "step": 15110
    },
    {
      "epoch": 0.3095242970941781,
      "grad_norm": 0.2345486283302307,
      "learning_rate": 0.0002904461640572788,
      "loss": 4.8281,
      "step": 15115
    },
    {
      "epoch": 0.30962668687158273,
      "grad_norm": 0.252042293548584,
      "learning_rate": 0.00029042441543464335,
      "loss": 4.8532,
      "step": 15120
    },
    {
      "epoch": 0.30972907664898736,
      "grad_norm": 0.23863114416599274,
      "learning_rate": 0.00029040264290170544,
      "loss": 4.8753,
      "step": 15125
    },
    {
      "epoch": 0.309831466426392,
      "grad_norm": 0.25628069043159485,
      "learning_rate": 0.00029038084646217246,
      "loss": 4.8412,
      "step": 15130
    },
    {
      "epoch": 0.3099338562037966,
      "grad_norm": 0.24536699056625366,
      "learning_rate": 0.0002903590261197557,
      "loss": 4.8406,
      "step": 15135
    },
    {
      "epoch": 0.3100362459812012,
      "grad_norm": 0.23125344514846802,
      "learning_rate": 0.0002903371818781704,
      "loss": 4.7983,
      "step": 15140
    },
    {
      "epoch": 0.31013863575860584,
      "grad_norm": 0.3031942546367645,
      "learning_rate": 0.00029031531374113624,
      "loss": 4.7796,
      "step": 15145
    },
    {
      "epoch": 0.31024102553601046,
      "grad_norm": 0.23981033265590668,
      "learning_rate": 0.00029029342171237656,
      "loss": 4.8364,
      "step": 15150
    },
    {
      "epoch": 0.3103434153134151,
      "grad_norm": 0.24899478256702423,
      "learning_rate": 0.000290271505795619,
      "loss": 4.8795,
      "step": 15155
    },
    {
      "epoch": 0.31044580509081976,
      "grad_norm": 0.23420363664627075,
      "learning_rate": 0.00029024956599459534,
      "loss": 4.8527,
      "step": 15160
    },
    {
      "epoch": 0.3105481948682244,
      "grad_norm": 0.22677026689052582,
      "learning_rate": 0.00029022760231304113,
      "loss": 4.8321,
      "step": 15165
    },
    {
      "epoch": 0.310650584645629,
      "grad_norm": 0.25254639983177185,
      "learning_rate": 0.0002902056147546963,
      "loss": 4.7307,
      "step": 15170
    },
    {
      "epoch": 0.3107529744230336,
      "grad_norm": 0.25167813897132874,
      "learning_rate": 0.0002901836033233047,
      "loss": 4.8039,
      "step": 15175
    },
    {
      "epoch": 0.31085536420043824,
      "grad_norm": 0.2575048506259918,
      "learning_rate": 0.0002901615680226141,
      "loss": 4.7933,
      "step": 15180
    },
    {
      "epoch": 0.31095775397784287,
      "grad_norm": 0.2427930384874344,
      "learning_rate": 0.0002901395088563767,
      "loss": 4.826,
      "step": 15185
    },
    {
      "epoch": 0.3110601437552475,
      "grad_norm": 0.2652527391910553,
      "learning_rate": 0.0002901174258283484,
      "loss": 4.78,
      "step": 15190
    },
    {
      "epoch": 0.3111625335326521,
      "grad_norm": 0.23744867742061615,
      "learning_rate": 0.00029009531894228933,
      "loss": 4.7958,
      "step": 15195
    },
    {
      "epoch": 0.31126492331005673,
      "grad_norm": 0.23774176836013794,
      "learning_rate": 0.0002900731882019638,
      "loss": 4.8793,
      "step": 15200
    },
    {
      "epoch": 0.31136731308746135,
      "grad_norm": 0.23590290546417236,
      "learning_rate": 0.0002900510336111398,
      "loss": 4.7966,
      "step": 15205
    },
    {
      "epoch": 0.31146970286486597,
      "grad_norm": 0.24009494483470917,
      "learning_rate": 0.0002900288551735899,
      "loss": 4.8661,
      "step": 15210
    },
    {
      "epoch": 0.3115720926422706,
      "grad_norm": 0.23347415030002594,
      "learning_rate": 0.00029000665289309033,
      "loss": 4.8165,
      "step": 15215
    },
    {
      "epoch": 0.3116744824196752,
      "grad_norm": 0.2547255754470825,
      "learning_rate": 0.0002899844267734214,
      "loss": 4.7903,
      "step": 15220
    },
    {
      "epoch": 0.31177687219707984,
      "grad_norm": 0.2570987343788147,
      "learning_rate": 0.00028996217681836784,
      "loss": 4.8647,
      "step": 15225
    },
    {
      "epoch": 0.31187926197448446,
      "grad_norm": 0.2709431052207947,
      "learning_rate": 0.000289939903031718,
      "loss": 4.844,
      "step": 15230
    },
    {
      "epoch": 0.3119816517518891,
      "grad_norm": 0.2512519955635071,
      "learning_rate": 0.00028991760541726455,
      "loss": 4.8062,
      "step": 15235
    },
    {
      "epoch": 0.3120840415292937,
      "grad_norm": 0.26664140820503235,
      "learning_rate": 0.00028989528397880415,
      "loss": 4.8416,
      "step": 15240
    },
    {
      "epoch": 0.3121864313066983,
      "grad_norm": 0.25389641523361206,
      "learning_rate": 0.00028987293872013737,
      "loss": 4.83,
      "step": 15245
    },
    {
      "epoch": 0.31228882108410294,
      "grad_norm": 0.2916410565376282,
      "learning_rate": 0.0002898505696450692,
      "loss": 4.8773,
      "step": 15250
    },
    {
      "epoch": 0.31239121086150756,
      "grad_norm": 0.24153843522071838,
      "learning_rate": 0.0002898281767574083,
      "loss": 4.8245,
      "step": 15255
    },
    {
      "epoch": 0.3124936006389122,
      "grad_norm": 0.23637597262859344,
      "learning_rate": 0.0002898057600609676,
      "loss": 4.828,
      "step": 15260
    },
    {
      "epoch": 0.31259599041631686,
      "grad_norm": 0.2532103359699249,
      "learning_rate": 0.000289783319559564,
      "loss": 4.8363,
      "step": 15265
    },
    {
      "epoch": 0.3126983801937215,
      "grad_norm": 0.3597238063812256,
      "learning_rate": 0.0002897608552570185,
      "loss": 4.8332,
      "step": 15270
    },
    {
      "epoch": 0.3128007699711261,
      "grad_norm": 0.2457640916109085,
      "learning_rate": 0.0002897383671571562,
      "loss": 4.8603,
      "step": 15275
    },
    {
      "epoch": 0.3129031597485307,
      "grad_norm": 0.23612958192825317,
      "learning_rate": 0.000289715855263806,
      "loss": 4.8515,
      "step": 15280
    },
    {
      "epoch": 0.31300554952593534,
      "grad_norm": 0.23397192358970642,
      "learning_rate": 0.00028969331958080127,
      "loss": 4.787,
      "step": 15285
    },
    {
      "epoch": 0.31310793930333997,
      "grad_norm": 0.29118800163269043,
      "learning_rate": 0.00028967076011197907,
      "loss": 4.8675,
      "step": 15290
    },
    {
      "epoch": 0.3132103290807446,
      "grad_norm": 0.23778823018074036,
      "learning_rate": 0.0002896481768611807,
      "loss": 4.8126,
      "step": 15295
    },
    {
      "epoch": 0.3133127188581492,
      "grad_norm": 0.2586337625980377,
      "learning_rate": 0.0002896255698322513,
      "loss": 4.8179,
      "step": 15300
    },
    {
      "epoch": 0.31341510863555383,
      "grad_norm": 0.3234255909919739,
      "learning_rate": 0.00028960293902904037,
      "loss": 4.8404,
      "step": 15305
    },
    {
      "epoch": 0.31351749841295845,
      "grad_norm": 0.24055859446525574,
      "learning_rate": 0.00028958028445540124,
      "loss": 4.7472,
      "step": 15310
    },
    {
      "epoch": 0.31361988819036307,
      "grad_norm": 0.23130378127098083,
      "learning_rate": 0.00028955760611519137,
      "loss": 4.84,
      "step": 15315
    },
    {
      "epoch": 0.3137222779677677,
      "grad_norm": 0.22796563804149628,
      "learning_rate": 0.0002895349040122722,
      "loss": 4.7846,
      "step": 15320
    },
    {
      "epoch": 0.3138246677451723,
      "grad_norm": 0.24777504801750183,
      "learning_rate": 0.0002895121781505092,
      "loss": 4.8461,
      "step": 15325
    },
    {
      "epoch": 0.31392705752257694,
      "grad_norm": 0.26132479310035706,
      "learning_rate": 0.0002894894285337721,
      "loss": 4.7783,
      "step": 15330
    },
    {
      "epoch": 0.31402944729998156,
      "grad_norm": 0.268828809261322,
      "learning_rate": 0.00028946665516593436,
      "loss": 4.822,
      "step": 15335
    },
    {
      "epoch": 0.3141318370773862,
      "grad_norm": 0.2682000398635864,
      "learning_rate": 0.0002894438580508737,
      "loss": 4.7456,
      "step": 15340
    },
    {
      "epoch": 0.3142342268547908,
      "grad_norm": 0.24134917557239532,
      "learning_rate": 0.00028942103719247186,
      "loss": 4.8673,
      "step": 15345
    },
    {
      "epoch": 0.3143366166321954,
      "grad_norm": 0.2589450776576996,
      "learning_rate": 0.00028939819259461453,
      "loss": 4.8995,
      "step": 15350
    },
    {
      "epoch": 0.31443900640960004,
      "grad_norm": 0.22643651068210602,
      "learning_rate": 0.00028937532426119155,
      "loss": 4.8269,
      "step": 15355
    },
    {
      "epoch": 0.31454139618700466,
      "grad_norm": 0.24379104375839233,
      "learning_rate": 0.0002893524321960967,
      "loss": 4.8361,
      "step": 15360
    },
    {
      "epoch": 0.31464378596440934,
      "grad_norm": 0.25083568692207336,
      "learning_rate": 0.0002893295164032279,
      "loss": 4.7882,
      "step": 15365
    },
    {
      "epoch": 0.31474617574181396,
      "grad_norm": 0.2672452926635742,
      "learning_rate": 0.00028930657688648696,
      "loss": 4.791,
      "step": 15370
    },
    {
      "epoch": 0.3148485655192186,
      "grad_norm": 1.3716267347335815,
      "learning_rate": 0.00028928361364977996,
      "loss": 4.8938,
      "step": 15375
    },
    {
      "epoch": 0.3149509552966232,
      "grad_norm": 0.23629119992256165,
      "learning_rate": 0.0002892606266970168,
      "loss": 4.8081,
      "step": 15380
    },
    {
      "epoch": 0.3150533450740278,
      "grad_norm": 0.2300252914428711,
      "learning_rate": 0.0002892376160321116,
      "loss": 4.8424,
      "step": 15385
    },
    {
      "epoch": 0.31515573485143245,
      "grad_norm": 0.2472316175699234,
      "learning_rate": 0.0002892145816589823,
      "loss": 4.8456,
      "step": 15390
    },
    {
      "epoch": 0.31525812462883707,
      "grad_norm": 0.2334398329257965,
      "learning_rate": 0.0002891915235815511,
      "loss": 4.8625,
      "step": 15395
    },
    {
      "epoch": 0.3153605144062417,
      "grad_norm": 0.27647051215171814,
      "learning_rate": 0.00028916844180374405,
      "loss": 4.8315,
      "step": 15400
    },
    {
      "epoch": 0.3154629041836463,
      "grad_norm": 0.24235866963863373,
      "learning_rate": 0.0002891453363294915,
      "loss": 4.8063,
      "step": 15405
    },
    {
      "epoch": 0.31556529396105093,
      "grad_norm": 0.2546367049217224,
      "learning_rate": 0.0002891222071627274,
      "loss": 4.7442,
      "step": 15410
    },
    {
      "epoch": 0.31566768373845555,
      "grad_norm": 0.25487855076789856,
      "learning_rate": 0.00028909905430739015,
      "loss": 4.7989,
      "step": 15415
    },
    {
      "epoch": 0.3157700735158602,
      "grad_norm": 0.26692599058151245,
      "learning_rate": 0.0002890758777674221,
      "loss": 4.8926,
      "step": 15420
    },
    {
      "epoch": 0.3158724632932648,
      "grad_norm": 0.2569117844104767,
      "learning_rate": 0.0002890526775467694,
      "loss": 4.8236,
      "step": 15425
    },
    {
      "epoch": 0.3159748530706694,
      "grad_norm": 0.24728327989578247,
      "learning_rate": 0.0002890294536493824,
      "loss": 4.7589,
      "step": 15430
    },
    {
      "epoch": 0.31607724284807404,
      "grad_norm": 0.27122560143470764,
      "learning_rate": 0.0002890062060792155,
      "loss": 4.8202,
      "step": 15435
    },
    {
      "epoch": 0.31617963262547866,
      "grad_norm": 0.26875802874565125,
      "learning_rate": 0.0002889829348402272,
      "loss": 4.7956,
      "step": 15440
    },
    {
      "epoch": 0.3162820224028833,
      "grad_norm": 0.26880842447280884,
      "learning_rate": 0.00028895963993637986,
      "loss": 4.8323,
      "step": 15445
    },
    {
      "epoch": 0.3163844121802879,
      "grad_norm": 0.2605154514312744,
      "learning_rate": 0.0002889363213716399,
      "loss": 4.8423,
      "step": 15450
    },
    {
      "epoch": 0.3164868019576925,
      "grad_norm": 0.25857120752334595,
      "learning_rate": 0.00028891297914997795,
      "loss": 4.7802,
      "step": 15455
    },
    {
      "epoch": 0.31658919173509714,
      "grad_norm": 0.26708948612213135,
      "learning_rate": 0.00028888961327536833,
      "loss": 4.7449,
      "step": 15460
    },
    {
      "epoch": 0.3166915815125018,
      "grad_norm": 0.23551607131958008,
      "learning_rate": 0.00028886622375178976,
      "loss": 4.8423,
      "step": 15465
    },
    {
      "epoch": 0.31679397128990644,
      "grad_norm": 0.24214282631874084,
      "learning_rate": 0.0002888428105832247,
      "loss": 4.7936,
      "step": 15470
    },
    {
      "epoch": 0.31689636106731106,
      "grad_norm": 0.23676030337810516,
      "learning_rate": 0.0002888193737736598,
      "loss": 4.8165,
      "step": 15475
    },
    {
      "epoch": 0.3169987508447157,
      "grad_norm": 0.6792005896568298,
      "learning_rate": 0.0002887959133270857,
      "loss": 4.8453,
      "step": 15480
    },
    {
      "epoch": 0.3171011406221203,
      "grad_norm": 0.2418385148048401,
      "learning_rate": 0.000288772429247497,
      "loss": 4.8382,
      "step": 15485
    },
    {
      "epoch": 0.3172035303995249,
      "grad_norm": 0.2369857281446457,
      "learning_rate": 0.0002887489215388924,
      "loss": 4.7731,
      "step": 15490
    },
    {
      "epoch": 0.31730592017692955,
      "grad_norm": 0.23815925419330597,
      "learning_rate": 0.00028872539020527465,
      "loss": 4.8536,
      "step": 15495
    },
    {
      "epoch": 0.31740830995433417,
      "grad_norm": 0.24218904972076416,
      "learning_rate": 0.00028870183525065037,
      "loss": 4.8051,
      "step": 15500
    },
    {
      "epoch": 0.3175106997317388,
      "grad_norm": 0.22303232550621033,
      "learning_rate": 0.00028867825667903036,
      "loss": 4.8295,
      "step": 15505
    },
    {
      "epoch": 0.3176130895091434,
      "grad_norm": 0.2368650883436203,
      "learning_rate": 0.00028865465449442936,
      "loss": 4.8234,
      "step": 15510
    },
    {
      "epoch": 0.31771547928654803,
      "grad_norm": 0.23938125371932983,
      "learning_rate": 0.0002886310287008662,
      "loss": 4.8194,
      "step": 15515
    },
    {
      "epoch": 0.31781786906395265,
      "grad_norm": 0.33564168214797974,
      "learning_rate": 0.0002886073793023637,
      "loss": 4.8487,
      "step": 15520
    },
    {
      "epoch": 0.3179202588413573,
      "grad_norm": 0.2706722319126129,
      "learning_rate": 0.00028858370630294855,
      "loss": 4.8275,
      "step": 15525
    },
    {
      "epoch": 0.3180226486187619,
      "grad_norm": 0.24289891123771667,
      "learning_rate": 0.0002885600097066517,
      "loss": 4.8184,
      "step": 15530
    },
    {
      "epoch": 0.3181250383961665,
      "grad_norm": 0.2418476790189743,
      "learning_rate": 0.000288536289517508,
      "loss": 4.7873,
      "step": 15535
    },
    {
      "epoch": 0.31822742817357114,
      "grad_norm": 0.2375362068414688,
      "learning_rate": 0.0002885125457395563,
      "loss": 4.8121,
      "step": 15540
    },
    {
      "epoch": 0.31832981795097576,
      "grad_norm": 0.24532565474510193,
      "learning_rate": 0.00028848877837683954,
      "loss": 4.8509,
      "step": 15545
    },
    {
      "epoch": 0.3184322077283804,
      "grad_norm": 0.2715478241443634,
      "learning_rate": 0.0002884649874334045,
      "loss": 4.847,
      "step": 15550
    },
    {
      "epoch": 0.318534597505785,
      "grad_norm": 0.2489650547504425,
      "learning_rate": 0.00028844117291330234,
      "loss": 4.7647,
      "step": 15555
    },
    {
      "epoch": 0.3186369872831896,
      "grad_norm": 0.24650003015995026,
      "learning_rate": 0.00028841733482058777,
      "loss": 4.8137,
      "step": 15560
    },
    {
      "epoch": 0.3187393770605943,
      "grad_norm": 0.2365977168083191,
      "learning_rate": 0.00028839347315931985,
      "loss": 4.8598,
      "step": 15565
    },
    {
      "epoch": 0.3188417668379989,
      "grad_norm": 0.25365331768989563,
      "learning_rate": 0.0002883695879335615,
      "loss": 4.8118,
      "step": 15570
    },
    {
      "epoch": 0.31894415661540354,
      "grad_norm": 0.24908719956874847,
      "learning_rate": 0.00028834567914737974,
      "loss": 4.8367,
      "step": 15575
    },
    {
      "epoch": 0.31904654639280816,
      "grad_norm": 0.24689409136772156,
      "learning_rate": 0.0002883217468048456,
      "loss": 4.8306,
      "step": 15580
    },
    {
      "epoch": 0.3191489361702128,
      "grad_norm": 0.23742550611495972,
      "learning_rate": 0.0002882977909100339,
      "loss": 4.81,
      "step": 15585
    },
    {
      "epoch": 0.3192513259476174,
      "grad_norm": 0.23478001356124878,
      "learning_rate": 0.00028827381146702386,
      "loss": 4.8192,
      "step": 15590
    },
    {
      "epoch": 0.319353715725022,
      "grad_norm": 0.2573738992214203,
      "learning_rate": 0.00028824980847989845,
      "loss": 4.839,
      "step": 15595
    },
    {
      "epoch": 0.31945610550242665,
      "grad_norm": 0.3302048444747925,
      "learning_rate": 0.0002882257819527446,
      "loss": 4.7653,
      "step": 15600
    },
    {
      "epoch": 0.31955849527983127,
      "grad_norm": 0.24334406852722168,
      "learning_rate": 0.0002882017318896534,
      "loss": 4.8084,
      "step": 15605
    },
    {
      "epoch": 0.3196608850572359,
      "grad_norm": 0.2528078258037567,
      "learning_rate": 0.0002881776582947199,
      "loss": 4.7564,
      "step": 15610
    },
    {
      "epoch": 0.3197632748346405,
      "grad_norm": 0.23048368096351624,
      "learning_rate": 0.0002881535611720432,
      "loss": 4.8018,
      "step": 15615
    },
    {
      "epoch": 0.31986566461204513,
      "grad_norm": 0.2321743220090866,
      "learning_rate": 0.0002881294405257263,
      "loss": 4.8037,
      "step": 15620
    },
    {
      "epoch": 0.31996805438944975,
      "grad_norm": 0.2332994043827057,
      "learning_rate": 0.0002881052963598763,
      "loss": 4.7536,
      "step": 15625
    },
    {
      "epoch": 0.3200704441668544,
      "grad_norm": 0.2396085262298584,
      "learning_rate": 0.0002880811286786042,
      "loss": 4.8436,
      "step": 15630
    },
    {
      "epoch": 0.320172833944259,
      "grad_norm": 0.4317147433757782,
      "learning_rate": 0.00028805693748602514,
      "loss": 4.8264,
      "step": 15635
    },
    {
      "epoch": 0.3202752237216636,
      "grad_norm": 0.23182272911071777,
      "learning_rate": 0.0002880327227862582,
      "loss": 4.8014,
      "step": 15640
    },
    {
      "epoch": 0.32037761349906824,
      "grad_norm": 0.24353723227977753,
      "learning_rate": 0.0002880084845834264,
      "loss": 4.8086,
      "step": 15645
    },
    {
      "epoch": 0.32048000327647286,
      "grad_norm": 0.2554088234901428,
      "learning_rate": 0.00028798422288165684,
      "loss": 4.845,
      "step": 15650
    },
    {
      "epoch": 0.3205823930538775,
      "grad_norm": 0.23942938446998596,
      "learning_rate": 0.0002879599376850807,
      "loss": 4.7911,
      "step": 15655
    },
    {
      "epoch": 0.3206847828312821,
      "grad_norm": 0.2640797197818756,
      "learning_rate": 0.00028793562899783284,
      "loss": 4.858,
      "step": 15660
    },
    {
      "epoch": 0.3207871726086867,
      "grad_norm": 0.2859177589416504,
      "learning_rate": 0.00028791129682405255,
      "loss": 4.8502,
      "step": 15665
    },
    {
      "epoch": 0.3208895623860914,
      "grad_norm": 0.24477677047252655,
      "learning_rate": 0.00028788694116788283,
      "loss": 4.8044,
      "step": 15670
    },
    {
      "epoch": 0.320991952163496,
      "grad_norm": 0.22414985299110413,
      "learning_rate": 0.00028786256203347076,
      "loss": 4.7959,
      "step": 15675
    },
    {
      "epoch": 0.32109434194090064,
      "grad_norm": 0.2535495460033417,
      "learning_rate": 0.00028783815942496737,
      "loss": 4.8198,
      "step": 15680
    },
    {
      "epoch": 0.32119673171830526,
      "grad_norm": 0.27068763971328735,
      "learning_rate": 0.00028781373334652786,
      "loss": 4.8548,
      "step": 15685
    },
    {
      "epoch": 0.3212991214957099,
      "grad_norm": 0.2561670243740082,
      "learning_rate": 0.0002877892838023112,
      "loss": 4.8365,
      "step": 15690
    },
    {
      "epoch": 0.3214015112731145,
      "grad_norm": 0.2500530481338501,
      "learning_rate": 0.0002877648107964805,
      "loss": 4.8265,
      "step": 15695
    },
    {
      "epoch": 0.3215039010505191,
      "grad_norm": 0.23399963974952698,
      "learning_rate": 0.00028774031433320273,
      "loss": 4.81,
      "step": 15700
    },
    {
      "epoch": 0.32160629082792375,
      "grad_norm": 0.22941014170646667,
      "learning_rate": 0.0002877157944166491,
      "loss": 4.8312,
      "step": 15705
    },
    {
      "epoch": 0.32170868060532837,
      "grad_norm": 0.2397286742925644,
      "learning_rate": 0.00028769125105099447,
      "loss": 4.7979,
      "step": 15710
    },
    {
      "epoch": 0.321811070382733,
      "grad_norm": 0.25594308972358704,
      "learning_rate": 0.000287666684240418,
      "loss": 4.8217,
      "step": 15715
    },
    {
      "epoch": 0.3219134601601376,
      "grad_norm": 0.24058277904987335,
      "learning_rate": 0.0002876420939891028,
      "loss": 4.8679,
      "step": 15720
    },
    {
      "epoch": 0.32201584993754223,
      "grad_norm": 0.24760036170482635,
      "learning_rate": 0.0002876174803012357,
      "loss": 4.7753,
      "step": 15725
    },
    {
      "epoch": 0.32211823971494685,
      "grad_norm": 0.249937504529953,
      "learning_rate": 0.00028759284318100785,
      "loss": 4.8472,
      "step": 15730
    },
    {
      "epoch": 0.3222206294923515,
      "grad_norm": 0.241938978433609,
      "learning_rate": 0.0002875681826326142,
      "loss": 4.7948,
      "step": 15735
    },
    {
      "epoch": 0.3223230192697561,
      "grad_norm": 0.2601494789123535,
      "learning_rate": 0.00028754349866025374,
      "loss": 4.826,
      "step": 15740
    },
    {
      "epoch": 0.3224254090471607,
      "grad_norm": 0.2735443115234375,
      "learning_rate": 0.0002875187912681295,
      "loss": 4.8471,
      "step": 15745
    },
    {
      "epoch": 0.32252779882456534,
      "grad_norm": 0.24875079095363617,
      "learning_rate": 0.00028749406046044845,
      "loss": 4.8263,
      "step": 15750
    },
    {
      "epoch": 0.32263018860196996,
      "grad_norm": 0.24867716431617737,
      "learning_rate": 0.00028746930624142143,
      "loss": 4.8503,
      "step": 15755
    },
    {
      "epoch": 0.3227325783793746,
      "grad_norm": 0.27599936723709106,
      "learning_rate": 0.0002874445286152635,
      "loss": 4.8091,
      "step": 15760
    },
    {
      "epoch": 0.3228349681567792,
      "grad_norm": 0.2574349641799927,
      "learning_rate": 0.00028741972758619357,
      "loss": 4.8073,
      "step": 15765
    },
    {
      "epoch": 0.3229373579341839,
      "grad_norm": 0.24333839118480682,
      "learning_rate": 0.00028739490315843455,
      "loss": 4.8281,
      "step": 15770
    },
    {
      "epoch": 0.3230397477115885,
      "grad_norm": 0.2696617543697357,
      "learning_rate": 0.0002873700553362133,
      "loss": 4.8326,
      "step": 15775
    },
    {
      "epoch": 0.3231421374889931,
      "grad_norm": 0.2507472336292267,
      "learning_rate": 0.0002873451841237607,
      "loss": 4.8961,
      "step": 15780
    },
    {
      "epoch": 0.32324452726639774,
      "grad_norm": 0.269682914018631,
      "learning_rate": 0.00028732028952531175,
      "loss": 4.8254,
      "step": 15785
    },
    {
      "epoch": 0.32334691704380236,
      "grad_norm": 0.2574978470802307,
      "learning_rate": 0.0002872953715451051,
      "loss": 4.7598,
      "step": 15790
    },
    {
      "epoch": 0.323449306821207,
      "grad_norm": 0.22101959586143494,
      "learning_rate": 0.00028727043018738366,
      "loss": 4.8443,
      "step": 15795
    },
    {
      "epoch": 0.3235516965986116,
      "grad_norm": 0.23810577392578125,
      "learning_rate": 0.0002872454654563943,
      "loss": 4.8215,
      "step": 15800
    },
    {
      "epoch": 0.3236540863760162,
      "grad_norm": 0.2515735924243927,
      "learning_rate": 0.0002872204773563877,
      "loss": 4.8268,
      "step": 15805
    },
    {
      "epoch": 0.32375647615342085,
      "grad_norm": 0.25298553705215454,
      "learning_rate": 0.00028719546589161865,
      "loss": 4.8677,
      "step": 15810
    },
    {
      "epoch": 0.32385886593082547,
      "grad_norm": 0.24797645211219788,
      "learning_rate": 0.0002871704310663459,
      "loss": 4.807,
      "step": 15815
    },
    {
      "epoch": 0.3239612557082301,
      "grad_norm": 0.2690029740333557,
      "learning_rate": 0.00028714537288483227,
      "loss": 4.7658,
      "step": 15820
    },
    {
      "epoch": 0.3240636454856347,
      "grad_norm": 0.26422470808029175,
      "learning_rate": 0.00028712029135134434,
      "loss": 4.8034,
      "step": 15825
    },
    {
      "epoch": 0.32416603526303933,
      "grad_norm": 0.27222222089767456,
      "learning_rate": 0.0002870951864701528,
      "loss": 4.8401,
      "step": 15830
    },
    {
      "epoch": 0.32426842504044395,
      "grad_norm": 0.23684225976467133,
      "learning_rate": 0.00028707005824553237,
      "loss": 4.8393,
      "step": 15835
    },
    {
      "epoch": 0.3243708148178486,
      "grad_norm": 0.2376585751771927,
      "learning_rate": 0.00028704490668176154,
      "loss": 4.8081,
      "step": 15840
    },
    {
      "epoch": 0.3244732045952532,
      "grad_norm": 0.25881075859069824,
      "learning_rate": 0.0002870197317831231,
      "loss": 4.808,
      "step": 15845
    },
    {
      "epoch": 0.3245755943726578,
      "grad_norm": 0.24747461080551147,
      "learning_rate": 0.0002869945335539035,
      "loss": 4.7748,
      "step": 15850
    },
    {
      "epoch": 0.32467798415006244,
      "grad_norm": 0.23320181667804718,
      "learning_rate": 0.00028696931199839325,
      "loss": 4.7779,
      "step": 15855
    },
    {
      "epoch": 0.32478037392746706,
      "grad_norm": 0.23846809566020966,
      "learning_rate": 0.00028694406712088694,
      "loss": 4.8357,
      "step": 15860
    },
    {
      "epoch": 0.3248827637048717,
      "grad_norm": 0.241712749004364,
      "learning_rate": 0.0002869187989256831,
      "loss": 4.8271,
      "step": 15865
    },
    {
      "epoch": 0.32498515348227636,
      "grad_norm": 0.2403462678194046,
      "learning_rate": 0.0002868935074170841,
      "loss": 4.8015,
      "step": 15870
    },
    {
      "epoch": 0.325087543259681,
      "grad_norm": 0.2553309202194214,
      "learning_rate": 0.00028686819259939636,
      "loss": 4.7891,
      "step": 15875
    },
    {
      "epoch": 0.3251899330370856,
      "grad_norm": 0.24671103060245514,
      "learning_rate": 0.00028684285447693034,
      "loss": 4.7901,
      "step": 15880
    },
    {
      "epoch": 0.3252923228144902,
      "grad_norm": 0.24250692129135132,
      "learning_rate": 0.0002868174930540004,
      "loss": 4.7807,
      "step": 15885
    },
    {
      "epoch": 0.32539471259189484,
      "grad_norm": 0.2553572654724121,
      "learning_rate": 0.00028679210833492476,
      "loss": 4.8591,
      "step": 15890
    },
    {
      "epoch": 0.32549710236929946,
      "grad_norm": 0.24860507249832153,
      "learning_rate": 0.00028676670032402586,
      "loss": 4.8435,
      "step": 15895
    },
    {
      "epoch": 0.3255994921467041,
      "grad_norm": 0.2537551820278168,
      "learning_rate": 0.0002867412690256299,
      "loss": 4.8049,
      "step": 15900
    },
    {
      "epoch": 0.3257018819241087,
      "grad_norm": 0.3558686375617981,
      "learning_rate": 0.0002867158144440672,
      "loss": 4.81,
      "step": 15905
    },
    {
      "epoch": 0.3258042717015133,
      "grad_norm": 0.24676696956157684,
      "learning_rate": 0.00028669033658367183,
      "loss": 4.8159,
      "step": 15910
    },
    {
      "epoch": 0.32590666147891795,
      "grad_norm": 0.23601582646369934,
      "learning_rate": 0.00028666483544878194,
      "loss": 4.8348,
      "step": 15915
    },
    {
      "epoch": 0.32600905125632257,
      "grad_norm": 0.22834748029708862,
      "learning_rate": 0.0002866393110437398,
      "loss": 4.7828,
      "step": 15920
    },
    {
      "epoch": 0.3261114410337272,
      "grad_norm": 0.23815374076366425,
      "learning_rate": 0.0002866137633728914,
      "loss": 4.833,
      "step": 15925
    },
    {
      "epoch": 0.3262138308111318,
      "grad_norm": 0.24401137232780457,
      "learning_rate": 0.0002865881924405867,
      "loss": 4.7999,
      "step": 15930
    },
    {
      "epoch": 0.32631622058853643,
      "grad_norm": 0.23739293217658997,
      "learning_rate": 0.00028656259825117985,
      "loss": 4.8044,
      "step": 15935
    },
    {
      "epoch": 0.32641861036594105,
      "grad_norm": 0.24494828283786774,
      "learning_rate": 0.00028653698080902877,
      "loss": 4.7714,
      "step": 15940
    },
    {
      "epoch": 0.3265210001433457,
      "grad_norm": 0.269447386264801,
      "learning_rate": 0.0002865113401184953,
      "loss": 4.8062,
      "step": 15945
    },
    {
      "epoch": 0.3266233899207503,
      "grad_norm": 0.2683109641075134,
      "learning_rate": 0.0002864856761839455,
      "loss": 4.7858,
      "step": 15950
    },
    {
      "epoch": 0.3267257796981549,
      "grad_norm": 0.24812977015972137,
      "learning_rate": 0.000286459989009749,
      "loss": 4.7581,
      "step": 15955
    },
    {
      "epoch": 0.32682816947555954,
      "grad_norm": 0.24161896109580994,
      "learning_rate": 0.0002864342786002798,
      "loss": 4.852,
      "step": 15960
    },
    {
      "epoch": 0.32693055925296416,
      "grad_norm": 0.22758758068084717,
      "learning_rate": 0.0002864085449599155,
      "loss": 4.8442,
      "step": 15965
    },
    {
      "epoch": 0.3270329490303688,
      "grad_norm": 0.2524091601371765,
      "learning_rate": 0.00028638278809303794,
      "loss": 4.8593,
      "step": 15970
    },
    {
      "epoch": 0.32713533880777346,
      "grad_norm": 0.2220301330089569,
      "learning_rate": 0.00028635700800403265,
      "loss": 4.8399,
      "step": 15975
    },
    {
      "epoch": 0.3272377285851781,
      "grad_norm": 0.27116480469703674,
      "learning_rate": 0.00028633120469728936,
      "loss": 4.8445,
      "step": 15980
    },
    {
      "epoch": 0.3273401183625827,
      "grad_norm": 0.25042563676834106,
      "learning_rate": 0.0002863053781772016,
      "loss": 4.8223,
      "step": 15985
    },
    {
      "epoch": 0.3274425081399873,
      "grad_norm": 0.22962479293346405,
      "learning_rate": 0.00028627952844816686,
      "loss": 4.8336,
      "step": 15990
    },
    {
      "epoch": 0.32754489791739194,
      "grad_norm": 0.2547265291213989,
      "learning_rate": 0.0002862536555145867,
      "loss": 4.8092,
      "step": 15995
    },
    {
      "epoch": 0.32764728769479656,
      "grad_norm": 0.24327710270881653,
      "learning_rate": 0.00028622775938086647,
      "loss": 4.8828,
      "step": 16000
    },
    {
      "epoch": 0.32764728769479656,
      "eval_loss": 4.832391738891602,
      "eval_runtime": 17.2365,
      "eval_samples_per_second": 33.765,
      "eval_steps_per_second": 4.235,
      "step": 16000
    },
    {
      "epoch": 0.3277496774722012,
      "grad_norm": 0.23021191358566284,
      "learning_rate": 0.00028620184005141556,
      "loss": 4.7917,
      "step": 16005
    },
    {
      "epoch": 0.3278520672496058,
      "grad_norm": 0.2317049205303192,
      "learning_rate": 0.00028617589753064735,
      "loss": 4.8406,
      "step": 16010
    },
    {
      "epoch": 0.3279544570270104,
      "grad_norm": 0.24882535636425018,
      "learning_rate": 0.00028614993182297905,
      "loss": 4.7988,
      "step": 16015
    },
    {
      "epoch": 0.32805684680441505,
      "grad_norm": 0.40424302220344543,
      "learning_rate": 0.00028612394293283197,
      "loss": 4.836,
      "step": 16020
    },
    {
      "epoch": 0.32815923658181967,
      "grad_norm": 0.2324492335319519,
      "learning_rate": 0.0002860979308646312,
      "loss": 4.8052,
      "step": 16025
    },
    {
      "epoch": 0.3282616263592243,
      "grad_norm": 0.2379092574119568,
      "learning_rate": 0.0002860718956228059,
      "loss": 4.7905,
      "step": 16030
    },
    {
      "epoch": 0.3283640161366289,
      "grad_norm": 0.23237118124961853,
      "learning_rate": 0.00028604583721178906,
      "loss": 4.8486,
      "step": 16035
    },
    {
      "epoch": 0.32846640591403353,
      "grad_norm": 0.24024012684822083,
      "learning_rate": 0.0002860197556360178,
      "loss": 4.7788,
      "step": 16040
    },
    {
      "epoch": 0.32856879569143815,
      "grad_norm": 0.26056140661239624,
      "learning_rate": 0.0002859936508999331,
      "loss": 4.7711,
      "step": 16045
    },
    {
      "epoch": 0.3286711854688428,
      "grad_norm": 0.2680843472480774,
      "learning_rate": 0.0002859675230079797,
      "loss": 4.8395,
      "step": 16050
    },
    {
      "epoch": 0.3287735752462474,
      "grad_norm": 0.23419305682182312,
      "learning_rate": 0.00028594137196460653,
      "loss": 4.8007,
      "step": 16055
    },
    {
      "epoch": 0.328875965023652,
      "grad_norm": 0.2893793284893036,
      "learning_rate": 0.0002859151977742664,
      "loss": 4.8185,
      "step": 16060
    },
    {
      "epoch": 0.32897835480105664,
      "grad_norm": 0.29999348521232605,
      "learning_rate": 0.00028588900044141595,
      "loss": 4.8263,
      "step": 16065
    },
    {
      "epoch": 0.32908074457846126,
      "grad_norm": 0.2500661611557007,
      "learning_rate": 0.0002858627799705159,
      "loss": 4.812,
      "step": 16070
    },
    {
      "epoch": 0.32918313435586594,
      "grad_norm": 0.2984530031681061,
      "learning_rate": 0.0002858365363660309,
      "loss": 4.7767,
      "step": 16075
    },
    {
      "epoch": 0.32928552413327056,
      "grad_norm": 0.22807948291301727,
      "learning_rate": 0.0002858102696324294,
      "loss": 4.7975,
      "step": 16080
    },
    {
      "epoch": 0.3293879139106752,
      "grad_norm": 0.2420845627784729,
      "learning_rate": 0.00028578397977418394,
      "loss": 4.8134,
      "step": 16085
    },
    {
      "epoch": 0.3294903036880798,
      "grad_norm": 0.24594122171401978,
      "learning_rate": 0.0002857576667957709,
      "loss": 4.789,
      "step": 16090
    },
    {
      "epoch": 0.3295926934654844,
      "grad_norm": 0.22552920877933502,
      "learning_rate": 0.00028573133070167075,
      "loss": 4.8137,
      "step": 16095
    },
    {
      "epoch": 0.32969508324288904,
      "grad_norm": 0.25765693187713623,
      "learning_rate": 0.0002857049714963676,
      "loss": 4.827,
      "step": 16100
    },
    {
      "epoch": 0.32979747302029366,
      "grad_norm": 0.25722846388816833,
      "learning_rate": 0.0002856785891843498,
      "loss": 4.8175,
      "step": 16105
    },
    {
      "epoch": 0.3298998627976983,
      "grad_norm": 0.2697278559207916,
      "learning_rate": 0.00028565218377010944,
      "loss": 4.8756,
      "step": 16110
    },
    {
      "epoch": 0.3300022525751029,
      "grad_norm": 0.24712322652339935,
      "learning_rate": 0.00028562575525814266,
      "loss": 4.8043,
      "step": 16115
    },
    {
      "epoch": 0.3301046423525075,
      "grad_norm": 0.23435640335083008,
      "learning_rate": 0.00028559930365294945,
      "loss": 4.8202,
      "step": 16120
    },
    {
      "epoch": 0.33020703212991215,
      "grad_norm": 0.2309722602367401,
      "learning_rate": 0.00028557282895903384,
      "loss": 4.8489,
      "step": 16125
    },
    {
      "epoch": 0.33030942190731677,
      "grad_norm": 0.24916371703147888,
      "learning_rate": 0.0002855463311809037,
      "loss": 4.7592,
      "step": 16130
    },
    {
      "epoch": 0.3304118116847214,
      "grad_norm": 0.2296198457479477,
      "learning_rate": 0.0002855198103230708,
      "loss": 4.7814,
      "step": 16135
    },
    {
      "epoch": 0.330514201462126,
      "grad_norm": 0.24471209943294525,
      "learning_rate": 0.00028549326639005086,
      "loss": 4.8081,
      "step": 16140
    },
    {
      "epoch": 0.33061659123953063,
      "grad_norm": 0.25593388080596924,
      "learning_rate": 0.0002854666993863637,
      "loss": 4.8149,
      "step": 16145
    },
    {
      "epoch": 0.33071898101693525,
      "grad_norm": 0.27216002345085144,
      "learning_rate": 0.00028544010931653285,
      "loss": 4.8326,
      "step": 16150
    },
    {
      "epoch": 0.3308213707943399,
      "grad_norm": 0.2517782151699066,
      "learning_rate": 0.00028541349618508584,
      "loss": 4.7682,
      "step": 16155
    },
    {
      "epoch": 0.3309237605717445,
      "grad_norm": 0.24866561591625214,
      "learning_rate": 0.0002853868599965541,
      "loss": 4.7985,
      "step": 16160
    },
    {
      "epoch": 0.3310261503491491,
      "grad_norm": 0.24357706308364868,
      "learning_rate": 0.0002853602007554731,
      "loss": 4.7943,
      "step": 16165
    },
    {
      "epoch": 0.33112854012655374,
      "grad_norm": 0.25437769293785095,
      "learning_rate": 0.00028533351846638205,
      "loss": 4.8135,
      "step": 16170
    },
    {
      "epoch": 0.3312309299039584,
      "grad_norm": 0.26568377017974854,
      "learning_rate": 0.0002853068131338243,
      "loss": 4.8396,
      "step": 16175
    },
    {
      "epoch": 0.33133331968136304,
      "grad_norm": 0.2585509717464447,
      "learning_rate": 0.00028528008476234694,
      "loss": 4.8335,
      "step": 16180
    },
    {
      "epoch": 0.33143570945876766,
      "grad_norm": 0.23786552250385284,
      "learning_rate": 0.0002852533333565011,
      "loss": 4.8348,
      "step": 16185
    },
    {
      "epoch": 0.3315380992361723,
      "grad_norm": 0.2424546480178833,
      "learning_rate": 0.0002852265589208417,
      "loss": 4.7527,
      "step": 16190
    },
    {
      "epoch": 0.3316404890135769,
      "grad_norm": 0.2438025027513504,
      "learning_rate": 0.0002851997614599278,
      "loss": 4.7954,
      "step": 16195
    },
    {
      "epoch": 0.3317428787909815,
      "grad_norm": 0.5066136121749878,
      "learning_rate": 0.00028517294097832216,
      "loss": 4.7434,
      "step": 16200
    },
    {
      "epoch": 0.33184526856838614,
      "grad_norm": 0.25915953516960144,
      "learning_rate": 0.0002851460974805916,
      "loss": 4.8257,
      "step": 16205
    },
    {
      "epoch": 0.33194765834579076,
      "grad_norm": 0.23952145874500275,
      "learning_rate": 0.0002851192309713067,
      "loss": 4.797,
      "step": 16210
    },
    {
      "epoch": 0.3320500481231954,
      "grad_norm": 0.28268367052078247,
      "learning_rate": 0.0002850923414550422,
      "loss": 4.7946,
      "step": 16215
    },
    {
      "epoch": 0.3321524379006,
      "grad_norm": 0.2623506188392639,
      "learning_rate": 0.0002850654289363766,
      "loss": 4.7887,
      "step": 16220
    },
    {
      "epoch": 0.3322548276780046,
      "grad_norm": 0.2400224804878235,
      "learning_rate": 0.00028503849341989223,
      "loss": 4.863,
      "step": 16225
    },
    {
      "epoch": 0.33235721745540925,
      "grad_norm": 0.23497094213962555,
      "learning_rate": 0.00028501153491017556,
      "loss": 4.8554,
      "step": 16230
    },
    {
      "epoch": 0.33245960723281387,
      "grad_norm": 0.23780083656311035,
      "learning_rate": 0.0002849845534118168,
      "loss": 4.8084,
      "step": 16235
    },
    {
      "epoch": 0.3325619970102185,
      "grad_norm": 0.26179882884025574,
      "learning_rate": 0.0002849575489294102,
      "loss": 4.8035,
      "step": 16240
    },
    {
      "epoch": 0.3326643867876231,
      "grad_norm": 0.2360720932483673,
      "learning_rate": 0.0002849305214675538,
      "loss": 4.7602,
      "step": 16245
    },
    {
      "epoch": 0.33276677656502773,
      "grad_norm": 0.2859875559806824,
      "learning_rate": 0.00028490347103084956,
      "loss": 4.8084,
      "step": 16250
    },
    {
      "epoch": 0.33286916634243235,
      "grad_norm": 0.2751981317996979,
      "learning_rate": 0.00028487639762390357,
      "loss": 4.8233,
      "step": 16255
    },
    {
      "epoch": 0.332971556119837,
      "grad_norm": 0.2609846889972687,
      "learning_rate": 0.00028484930125132553,
      "loss": 4.7336,
      "step": 16260
    },
    {
      "epoch": 0.3330739458972416,
      "grad_norm": 0.2535109221935272,
      "learning_rate": 0.0002848221819177292,
      "loss": 4.8385,
      "step": 16265
    },
    {
      "epoch": 0.3331763356746462,
      "grad_norm": 0.25714072585105896,
      "learning_rate": 0.0002847950396277323,
      "loss": 4.8218,
      "step": 16270
    },
    {
      "epoch": 0.3332787254520509,
      "grad_norm": 0.23650965094566345,
      "learning_rate": 0.00028476787438595633,
      "loss": 4.8234,
      "step": 16275
    },
    {
      "epoch": 0.3333811152294555,
      "grad_norm": 0.26363539695739746,
      "learning_rate": 0.0002847406861970268,
      "loss": 4.808,
      "step": 16280
    },
    {
      "epoch": 0.33348350500686014,
      "grad_norm": 0.2578558623790741,
      "learning_rate": 0.000284713475065573,
      "loss": 4.7717,
      "step": 16285
    },
    {
      "epoch": 0.33358589478426476,
      "grad_norm": 0.2533988654613495,
      "learning_rate": 0.0002846862409962284,
      "loss": 4.8438,
      "step": 16290
    },
    {
      "epoch": 0.3336882845616694,
      "grad_norm": 0.3612011969089508,
      "learning_rate": 0.00028465898399363,
      "loss": 4.8041,
      "step": 16295
    },
    {
      "epoch": 0.333790674339074,
      "grad_norm": 0.23150953650474548,
      "learning_rate": 0.00028463170406241897,
      "loss": 4.7842,
      "step": 16300
    },
    {
      "epoch": 0.3338930641164786,
      "grad_norm": 0.24778206646442413,
      "learning_rate": 0.0002846044012072403,
      "loss": 4.7862,
      "step": 16305
    },
    {
      "epoch": 0.33399545389388324,
      "grad_norm": 0.24051642417907715,
      "learning_rate": 0.00028457707543274295,
      "loss": 4.767,
      "step": 16310
    },
    {
      "epoch": 0.33409784367128786,
      "grad_norm": 0.23826947808265686,
      "learning_rate": 0.00028454972674357975,
      "loss": 4.8691,
      "step": 16315
    },
    {
      "epoch": 0.3342002334486925,
      "grad_norm": 0.2339300811290741,
      "learning_rate": 0.0002845223551444072,
      "loss": 4.8082,
      "step": 16320
    },
    {
      "epoch": 0.3343026232260971,
      "grad_norm": 0.2331026792526245,
      "learning_rate": 0.0002844949606398861,
      "loss": 4.8197,
      "step": 16325
    },
    {
      "epoch": 0.3344050130035017,
      "grad_norm": 0.22589321434497833,
      "learning_rate": 0.0002844675432346809,
      "loss": 4.8737,
      "step": 16330
    },
    {
      "epoch": 0.33450740278090635,
      "grad_norm": 0.24397091567516327,
      "learning_rate": 0.00028444010293346005,
      "loss": 4.8021,
      "step": 16335
    },
    {
      "epoch": 0.33460979255831097,
      "grad_norm": 0.24349862337112427,
      "learning_rate": 0.00028441263974089584,
      "loss": 4.791,
      "step": 16340
    },
    {
      "epoch": 0.3347121823357156,
      "grad_norm": 0.23792225122451782,
      "learning_rate": 0.0002843851536616644,
      "loss": 4.7971,
      "step": 16345
    },
    {
      "epoch": 0.3348145721131202,
      "grad_norm": 0.26191046833992004,
      "learning_rate": 0.0002843576447004459,
      "loss": 4.79,
      "step": 16350
    },
    {
      "epoch": 0.33491696189052483,
      "grad_norm": 0.27435147762298584,
      "learning_rate": 0.0002843301128619243,
      "loss": 4.8152,
      "step": 16355
    },
    {
      "epoch": 0.33501935166792945,
      "grad_norm": 0.2583461105823517,
      "learning_rate": 0.0002843025581507876,
      "loss": 4.7318,
      "step": 16360
    },
    {
      "epoch": 0.3351217414453341,
      "grad_norm": 0.2498168796300888,
      "learning_rate": 0.00028427498057172744,
      "loss": 4.8408,
      "step": 16365
    },
    {
      "epoch": 0.3352241312227387,
      "grad_norm": 0.2474629431962967,
      "learning_rate": 0.00028424738012943954,
      "loss": 4.7808,
      "step": 16370
    },
    {
      "epoch": 0.3353265210001433,
      "grad_norm": 0.2743283212184906,
      "learning_rate": 0.00028421975682862353,
      "loss": 4.8157,
      "step": 16375
    },
    {
      "epoch": 0.335428910777548,
      "grad_norm": 0.2380060851573944,
      "learning_rate": 0.00028419211067398284,
      "loss": 4.7808,
      "step": 16380
    },
    {
      "epoch": 0.3355313005549526,
      "grad_norm": 0.2658345699310303,
      "learning_rate": 0.0002841644416702248,
      "loss": 4.8096,
      "step": 16385
    },
    {
      "epoch": 0.33563369033235724,
      "grad_norm": 0.24597862362861633,
      "learning_rate": 0.0002841367498220607,
      "loss": 4.8153,
      "step": 16390
    },
    {
      "epoch": 0.33573608010976186,
      "grad_norm": 0.2558453381061554,
      "learning_rate": 0.0002841090351342057,
      "loss": 4.7838,
      "step": 16395
    },
    {
      "epoch": 0.3358384698871665,
      "grad_norm": 0.22768911719322205,
      "learning_rate": 0.0002840812976113788,
      "loss": 4.8122,
      "step": 16400
    },
    {
      "epoch": 0.3359408596645711,
      "grad_norm": 0.24096199870109558,
      "learning_rate": 0.00028405353725830294,
      "loss": 4.8449,
      "step": 16405
    },
    {
      "epoch": 0.3360432494419757,
      "grad_norm": 0.23811636865139008,
      "learning_rate": 0.0002840257540797048,
      "loss": 4.7902,
      "step": 16410
    },
    {
      "epoch": 0.33614563921938034,
      "grad_norm": 0.25714996457099915,
      "learning_rate": 0.00028399794808031524,
      "loss": 4.816,
      "step": 16415
    },
    {
      "epoch": 0.33624802899678496,
      "grad_norm": 0.28754061460494995,
      "learning_rate": 0.00028397011926486874,
      "loss": 4.7533,
      "step": 16420
    },
    {
      "epoch": 0.3363504187741896,
      "grad_norm": 0.27357572317123413,
      "learning_rate": 0.0002839422676381037,
      "loss": 4.7692,
      "step": 16425
    },
    {
      "epoch": 0.3364528085515942,
      "grad_norm": 0.23396272957324982,
      "learning_rate": 0.0002839143932047626,
      "loss": 4.8036,
      "step": 16430
    },
    {
      "epoch": 0.33655519832899883,
      "grad_norm": 0.24027958512306213,
      "learning_rate": 0.0002838864959695916,
      "loss": 4.7694,
      "step": 16435
    },
    {
      "epoch": 0.33665758810640345,
      "grad_norm": 0.2548987865447998,
      "learning_rate": 0.00028385857593734085,
      "loss": 4.8462,
      "step": 16440
    },
    {
      "epoch": 0.33675997788380807,
      "grad_norm": 0.24632757902145386,
      "learning_rate": 0.00028383063311276425,
      "loss": 4.7882,
      "step": 16445
    },
    {
      "epoch": 0.3368623676612127,
      "grad_norm": 0.22980652749538422,
      "learning_rate": 0.0002838026675006197,
      "loss": 4.8391,
      "step": 16450
    },
    {
      "epoch": 0.3369647574386173,
      "grad_norm": 0.2512337863445282,
      "learning_rate": 0.00028377467910566904,
      "loss": 4.7639,
      "step": 16455
    },
    {
      "epoch": 0.33706714721602193,
      "grad_norm": 0.2517258822917938,
      "learning_rate": 0.0002837466679326778,
      "loss": 4.8728,
      "step": 16460
    },
    {
      "epoch": 0.33716953699342656,
      "grad_norm": 0.2762230932712555,
      "learning_rate": 0.0002837186339864156,
      "loss": 4.8328,
      "step": 16465
    },
    {
      "epoch": 0.3372719267708312,
      "grad_norm": 0.2519664466381073,
      "learning_rate": 0.0002836905772716556,
      "loss": 4.8005,
      "step": 16470
    },
    {
      "epoch": 0.3373743165482358,
      "grad_norm": 0.28721392154693604,
      "learning_rate": 0.0002836624977931754,
      "loss": 4.7947,
      "step": 16475
    },
    {
      "epoch": 0.3374767063256405,
      "grad_norm": 0.25575771927833557,
      "learning_rate": 0.0002836343955557559,
      "loss": 4.8228,
      "step": 16480
    },
    {
      "epoch": 0.3375790961030451,
      "grad_norm": 0.2361585646867752,
      "learning_rate": 0.00028360627056418216,
      "loss": 4.8759,
      "step": 16485
    },
    {
      "epoch": 0.3376814858804497,
      "grad_norm": 0.2382163107395172,
      "learning_rate": 0.0002835781228232431,
      "loss": 4.804,
      "step": 16490
    },
    {
      "epoch": 0.33778387565785434,
      "grad_norm": 0.24462802708148956,
      "learning_rate": 0.0002835499523377314,
      "loss": 4.8405,
      "step": 16495
    },
    {
      "epoch": 0.33788626543525896,
      "grad_norm": 0.2449190318584442,
      "learning_rate": 0.0002835217591124438,
      "loss": 4.8327,
      "step": 16500
    },
    {
      "epoch": 0.3379886552126636,
      "grad_norm": 0.25404995679855347,
      "learning_rate": 0.00028349354315218084,
      "loss": 4.8044,
      "step": 16505
    },
    {
      "epoch": 0.3380910449900682,
      "grad_norm": 0.24927467107772827,
      "learning_rate": 0.00028346530446174677,
      "loss": 4.8553,
      "step": 16510
    },
    {
      "epoch": 0.3381934347674728,
      "grad_norm": 0.25905531644821167,
      "learning_rate": 0.00028343704304594993,
      "loss": 4.8353,
      "step": 16515
    },
    {
      "epoch": 0.33829582454487744,
      "grad_norm": 0.22443336248397827,
      "learning_rate": 0.0002834087589096024,
      "loss": 4.7935,
      "step": 16520
    },
    {
      "epoch": 0.33839821432228206,
      "grad_norm": 0.24087576568126678,
      "learning_rate": 0.00028338045205752023,
      "loss": 4.7795,
      "step": 16525
    },
    {
      "epoch": 0.3385006040996867,
      "grad_norm": 0.2546943128108978,
      "learning_rate": 0.00028335212249452316,
      "loss": 4.8057,
      "step": 16530
    },
    {
      "epoch": 0.3386029938770913,
      "grad_norm": 0.23839382827281952,
      "learning_rate": 0.000283323770225435,
      "loss": 4.8209,
      "step": 16535
    },
    {
      "epoch": 0.33870538365449593,
      "grad_norm": 0.29359230399131775,
      "learning_rate": 0.0002832953952550833,
      "loss": 4.8573,
      "step": 16540
    },
    {
      "epoch": 0.33880777343190055,
      "grad_norm": 0.2874332070350647,
      "learning_rate": 0.0002832669975882996,
      "loss": 4.84,
      "step": 16545
    },
    {
      "epoch": 0.33891016320930517,
      "grad_norm": 0.2460402101278305,
      "learning_rate": 0.0002832385772299191,
      "loss": 4.6877,
      "step": 16550
    },
    {
      "epoch": 0.3390125529867098,
      "grad_norm": 0.23004408180713654,
      "learning_rate": 0.000283210134184781,
      "loss": 4.793,
      "step": 16555
    },
    {
      "epoch": 0.3391149427641144,
      "grad_norm": 0.25409266352653503,
      "learning_rate": 0.00028318166845772843,
      "loss": 4.7968,
      "step": 16560
    },
    {
      "epoch": 0.33921733254151903,
      "grad_norm": 0.2459293007850647,
      "learning_rate": 0.0002831531800536082,
      "loss": 4.8258,
      "step": 16565
    },
    {
      "epoch": 0.33931972231892366,
      "grad_norm": 0.25078505277633667,
      "learning_rate": 0.0002831246689772712,
      "loss": 4.7815,
      "step": 16570
    },
    {
      "epoch": 0.3394221120963283,
      "grad_norm": 0.2511816620826721,
      "learning_rate": 0.0002830961352335719,
      "loss": 4.8209,
      "step": 16575
    },
    {
      "epoch": 0.33952450187373295,
      "grad_norm": 0.24562671780586243,
      "learning_rate": 0.00028306757882736887,
      "loss": 4.8267,
      "step": 16580
    },
    {
      "epoch": 0.3396268916511376,
      "grad_norm": 0.23564331233501434,
      "learning_rate": 0.0002830389997635245,
      "loss": 4.8405,
      "step": 16585
    },
    {
      "epoch": 0.3397292814285422,
      "grad_norm": 0.24330298602581024,
      "learning_rate": 0.00028301039804690494,
      "loss": 4.8183,
      "step": 16590
    },
    {
      "epoch": 0.3398316712059468,
      "grad_norm": 0.22338157892227173,
      "learning_rate": 0.0002829817736823802,
      "loss": 4.8447,
      "step": 16595
    },
    {
      "epoch": 0.33993406098335144,
      "grad_norm": 0.235164612531662,
      "learning_rate": 0.0002829531266748244,
      "loss": 4.7876,
      "step": 16600
    },
    {
      "epoch": 0.34003645076075606,
      "grad_norm": 0.2610068619251251,
      "learning_rate": 0.0002829244570291151,
      "loss": 4.8369,
      "step": 16605
    },
    {
      "epoch": 0.3401388405381607,
      "grad_norm": 0.23874932527542114,
      "learning_rate": 0.000282895764750134,
      "loss": 4.7797,
      "step": 16610
    },
    {
      "epoch": 0.3402412303155653,
      "grad_norm": 0.2389604151248932,
      "learning_rate": 0.00028286704984276665,
      "loss": 4.7596,
      "step": 16615
    },
    {
      "epoch": 0.3403436200929699,
      "grad_norm": 0.24898886680603027,
      "learning_rate": 0.00028283831231190224,
      "loss": 4.8472,
      "step": 16620
    },
    {
      "epoch": 0.34044600987037454,
      "grad_norm": 0.24569012224674225,
      "learning_rate": 0.0002828095521624341,
      "loss": 4.7886,
      "step": 16625
    },
    {
      "epoch": 0.34054839964777917,
      "grad_norm": 0.2282789796590805,
      "learning_rate": 0.00028278076939925926,
      "loss": 4.8108,
      "step": 16630
    },
    {
      "epoch": 0.3406507894251838,
      "grad_norm": 0.3270750343799591,
      "learning_rate": 0.0002827519640272785,
      "loss": 4.7373,
      "step": 16635
    },
    {
      "epoch": 0.3407531792025884,
      "grad_norm": 0.2560490369796753,
      "learning_rate": 0.0002827231360513966,
      "loss": 4.7895,
      "step": 16640
    },
    {
      "epoch": 0.34085556897999303,
      "grad_norm": 0.2517138123512268,
      "learning_rate": 0.0002826942854765222,
      "loss": 4.7704,
      "step": 16645
    },
    {
      "epoch": 0.34095795875739765,
      "grad_norm": 0.24209338426589966,
      "learning_rate": 0.00028266541230756773,
      "loss": 4.8018,
      "step": 16650
    },
    {
      "epoch": 0.34106034853480227,
      "grad_norm": 0.23627488315105438,
      "learning_rate": 0.00028263651654944944,
      "loss": 4.8031,
      "step": 16655
    },
    {
      "epoch": 0.3411627383122069,
      "grad_norm": 0.24600344896316528,
      "learning_rate": 0.0002826075982070875,
      "loss": 4.817,
      "step": 16660
    },
    {
      "epoch": 0.3412651280896115,
      "grad_norm": 0.23830489814281464,
      "learning_rate": 0.0002825786572854058,
      "loss": 4.8191,
      "step": 16665
    },
    {
      "epoch": 0.34136751786701613,
      "grad_norm": 0.2401708960533142,
      "learning_rate": 0.0002825496937893322,
      "loss": 4.7781,
      "step": 16670
    },
    {
      "epoch": 0.34146990764442076,
      "grad_norm": 0.23110555112361908,
      "learning_rate": 0.0002825207077237985,
      "loss": 4.7622,
      "step": 16675
    },
    {
      "epoch": 0.34157229742182543,
      "grad_norm": 0.23407591879367828,
      "learning_rate": 0.00028249169909373993,
      "loss": 4.8343,
      "step": 16680
    },
    {
      "epoch": 0.34167468719923005,
      "grad_norm": 0.258706659078598,
      "learning_rate": 0.0002824626679040961,
      "loss": 4.7401,
      "step": 16685
    },
    {
      "epoch": 0.3417770769766347,
      "grad_norm": 0.2502451539039612,
      "learning_rate": 0.00028243361415981004,
      "loss": 4.8391,
      "step": 16690
    },
    {
      "epoch": 0.3418794667540393,
      "grad_norm": 0.26287874579429626,
      "learning_rate": 0.00028240453786582885,
      "loss": 4.8496,
      "step": 16695
    },
    {
      "epoch": 0.3419818565314439,
      "grad_norm": 0.24094346165657043,
      "learning_rate": 0.0002823754390271034,
      "loss": 4.7413,
      "step": 16700
    },
    {
      "epoch": 0.34208424630884854,
      "grad_norm": 0.2397972196340561,
      "learning_rate": 0.0002823463176485883,
      "loss": 4.8279,
      "step": 16705
    },
    {
      "epoch": 0.34218663608625316,
      "grad_norm": 0.42894208431243896,
      "learning_rate": 0.0002823171737352423,
      "loss": 4.779,
      "step": 16710
    },
    {
      "epoch": 0.3422890258636578,
      "grad_norm": 0.24300961196422577,
      "learning_rate": 0.00028228800729202756,
      "loss": 4.7593,
      "step": 16715
    },
    {
      "epoch": 0.3423914156410624,
      "grad_norm": 0.2818509638309479,
      "learning_rate": 0.0002822588183239105,
      "loss": 4.774,
      "step": 16720
    },
    {
      "epoch": 0.342493805418467,
      "grad_norm": 0.2569493055343628,
      "learning_rate": 0.00028222960683586096,
      "loss": 4.8116,
      "step": 16725
    },
    {
      "epoch": 0.34259619519587164,
      "grad_norm": 0.24437934160232544,
      "learning_rate": 0.00028220037283285305,
      "loss": 4.8085,
      "step": 16730
    },
    {
      "epoch": 0.34269858497327627,
      "grad_norm": 0.31071799993515015,
      "learning_rate": 0.00028217111631986434,
      "loss": 4.8283,
      "step": 16735
    },
    {
      "epoch": 0.3428009747506809,
      "grad_norm": 0.2863980829715729,
      "learning_rate": 0.0002821418373018764,
      "loss": 4.83,
      "step": 16740
    },
    {
      "epoch": 0.3429033645280855,
      "grad_norm": 0.25225335359573364,
      "learning_rate": 0.0002821125357838747,
      "loss": 4.8388,
      "step": 16745
    },
    {
      "epoch": 0.34300575430549013,
      "grad_norm": 0.2698221206665039,
      "learning_rate": 0.0002820832117708484,
      "loss": 4.8276,
      "step": 16750
    },
    {
      "epoch": 0.34310814408289475,
      "grad_norm": 2.018913745880127,
      "learning_rate": 0.0002820538652677905,
      "loss": 4.8773,
      "step": 16755
    },
    {
      "epoch": 0.34321053386029937,
      "grad_norm": 0.28117841482162476,
      "learning_rate": 0.0002820244962796981,
      "loss": 4.7727,
      "step": 16760
    },
    {
      "epoch": 0.343312923637704,
      "grad_norm": 0.26004379987716675,
      "learning_rate": 0.0002819951048115716,
      "loss": 4.8188,
      "step": 16765
    },
    {
      "epoch": 0.3434153134151086,
      "grad_norm": 0.23128695785999298,
      "learning_rate": 0.00028196569086841566,
      "loss": 4.7956,
      "step": 16770
    },
    {
      "epoch": 0.34351770319251324,
      "grad_norm": 0.25050005316734314,
      "learning_rate": 0.00028193625445523877,
      "loss": 4.8066,
      "step": 16775
    },
    {
      "epoch": 0.34362009296991786,
      "grad_norm": 0.27198299765586853,
      "learning_rate": 0.0002819067955770529,
      "loss": 4.8297,
      "step": 16780
    },
    {
      "epoch": 0.34372248274732253,
      "grad_norm": 0.3195876181125641,
      "learning_rate": 0.00028187731423887427,
      "loss": 4.8254,
      "step": 16785
    },
    {
      "epoch": 0.34382487252472715,
      "grad_norm": 0.2708126902580261,
      "learning_rate": 0.0002818478104457226,
      "loss": 4.8057,
      "step": 16790
    },
    {
      "epoch": 0.3439272623021318,
      "grad_norm": 0.23975291848182678,
      "learning_rate": 0.00028181828420262154,
      "loss": 4.7726,
      "step": 16795
    },
    {
      "epoch": 0.3440296520795364,
      "grad_norm": 0.23758544027805328,
      "learning_rate": 0.00028178873551459866,
      "loss": 4.8273,
      "step": 16800
    },
    {
      "epoch": 0.344132041856941,
      "grad_norm": 0.24509690701961517,
      "learning_rate": 0.00028175916438668517,
      "loss": 4.8156,
      "step": 16805
    },
    {
      "epoch": 0.34423443163434564,
      "grad_norm": 0.24822518229484558,
      "learning_rate": 0.00028172957082391617,
      "loss": 4.7679,
      "step": 16810
    },
    {
      "epoch": 0.34433682141175026,
      "grad_norm": 0.24076735973358154,
      "learning_rate": 0.0002816999548313308,
      "loss": 4.8053,
      "step": 16815
    },
    {
      "epoch": 0.3444392111891549,
      "grad_norm": 0.24137085676193237,
      "learning_rate": 0.0002816703164139716,
      "loss": 4.7941,
      "step": 16820
    },
    {
      "epoch": 0.3445416009665595,
      "grad_norm": 0.26467767357826233,
      "learning_rate": 0.00028164065557688525,
      "loss": 4.8041,
      "step": 16825
    },
    {
      "epoch": 0.3446439907439641,
      "grad_norm": 0.3696797490119934,
      "learning_rate": 0.00028161097232512223,
      "loss": 4.8209,
      "step": 16830
    },
    {
      "epoch": 0.34474638052136874,
      "grad_norm": 0.24373362958431244,
      "learning_rate": 0.0002815812666637366,
      "loss": 4.7785,
      "step": 16835
    },
    {
      "epoch": 0.34484877029877337,
      "grad_norm": 0.2379792183637619,
      "learning_rate": 0.00028155153859778654,
      "loss": 4.7788,
      "step": 16840
    },
    {
      "epoch": 0.344951160076178,
      "grad_norm": 0.2604856491088867,
      "learning_rate": 0.00028152178813233377,
      "loss": 4.7654,
      "step": 16845
    },
    {
      "epoch": 0.3450535498535826,
      "grad_norm": 0.32233041524887085,
      "learning_rate": 0.000281492015272444,
      "loss": 4.7641,
      "step": 16850
    },
    {
      "epoch": 0.34515593963098723,
      "grad_norm": 0.2897481918334961,
      "learning_rate": 0.00028146222002318675,
      "loss": 4.8001,
      "step": 16855
    },
    {
      "epoch": 0.34525832940839185,
      "grad_norm": 0.2490362972021103,
      "learning_rate": 0.00028143240238963526,
      "loss": 4.8008,
      "step": 16860
    },
    {
      "epoch": 0.34536071918579647,
      "grad_norm": 0.24507276713848114,
      "learning_rate": 0.00028140256237686663,
      "loss": 4.696,
      "step": 16865
    },
    {
      "epoch": 0.3454631089632011,
      "grad_norm": 0.238366961479187,
      "learning_rate": 0.00028137269998996183,
      "loss": 4.8042,
      "step": 16870
    },
    {
      "epoch": 0.3455654987406057,
      "grad_norm": 0.25932565331459045,
      "learning_rate": 0.0002813428152340055,
      "loss": 4.7444,
      "step": 16875
    },
    {
      "epoch": 0.34566788851801034,
      "grad_norm": 0.22153566777706146,
      "learning_rate": 0.0002813129081140862,
      "loss": 4.8033,
      "step": 16880
    },
    {
      "epoch": 0.345770278295415,
      "grad_norm": 0.26651623845100403,
      "learning_rate": 0.0002812829786352963,
      "loss": 4.8489,
      "step": 16885
    },
    {
      "epoch": 0.34587266807281963,
      "grad_norm": 0.23320205509662628,
      "learning_rate": 0.0002812530268027318,
      "loss": 4.8217,
      "step": 16890
    },
    {
      "epoch": 0.34597505785022425,
      "grad_norm": 0.27574557065963745,
      "learning_rate": 0.0002812230526214928,
      "loss": 4.7301,
      "step": 16895
    },
    {
      "epoch": 0.3460774476276289,
      "grad_norm": 0.4646994173526764,
      "learning_rate": 0.0002811930560966831,
      "loss": 4.7735,
      "step": 16900
    },
    {
      "epoch": 0.3461798374050335,
      "grad_norm": 0.268422394990921,
      "learning_rate": 0.0002811630372334101,
      "loss": 4.7691,
      "step": 16905
    },
    {
      "epoch": 0.3462822271824381,
      "grad_norm": 0.2489718496799469,
      "learning_rate": 0.00028113299603678524,
      "loss": 4.7403,
      "step": 16910
    },
    {
      "epoch": 0.34638461695984274,
      "grad_norm": 0.2449781596660614,
      "learning_rate": 0.00028110293251192367,
      "loss": 4.7402,
      "step": 16915
    },
    {
      "epoch": 0.34648700673724736,
      "grad_norm": 0.2449360489845276,
      "learning_rate": 0.00028107284666394435,
      "loss": 4.8301,
      "step": 16920
    },
    {
      "epoch": 0.346589396514652,
      "grad_norm": 0.24715960025787354,
      "learning_rate": 0.00028104273849797014,
      "loss": 4.7768,
      "step": 16925
    },
    {
      "epoch": 0.3466917862920566,
      "grad_norm": 0.24429172277450562,
      "learning_rate": 0.0002810126080191275,
      "loss": 4.7959,
      "step": 16930
    },
    {
      "epoch": 0.3467941760694612,
      "grad_norm": 0.2417449951171875,
      "learning_rate": 0.00028098245523254674,
      "loss": 4.8027,
      "step": 16935
    },
    {
      "epoch": 0.34689656584686585,
      "grad_norm": 0.25111472606658936,
      "learning_rate": 0.0002809522801433622,
      "loss": 4.7777,
      "step": 16940
    },
    {
      "epoch": 0.34699895562427047,
      "grad_norm": 0.2312670648097992,
      "learning_rate": 0.0002809220827567117,
      "loss": 4.7974,
      "step": 16945
    },
    {
      "epoch": 0.3471013454016751,
      "grad_norm": 0.2415151447057724,
      "learning_rate": 0.0002808918630777371,
      "loss": 4.8084,
      "step": 16950
    },
    {
      "epoch": 0.3472037351790797,
      "grad_norm": 0.28880220651626587,
      "learning_rate": 0.0002808616211115839,
      "loss": 4.7984,
      "step": 16955
    },
    {
      "epoch": 0.34730612495648433,
      "grad_norm": 0.2454739809036255,
      "learning_rate": 0.0002808313568634015,
      "loss": 4.7436,
      "step": 16960
    },
    {
      "epoch": 0.34740851473388895,
      "grad_norm": 0.2548595368862152,
      "learning_rate": 0.00028080107033834296,
      "loss": 4.83,
      "step": 16965
    },
    {
      "epoch": 0.3475109045112936,
      "grad_norm": 0.2426546961069107,
      "learning_rate": 0.0002807707615415653,
      "loss": 4.7768,
      "step": 16970
    },
    {
      "epoch": 0.3476132942886982,
      "grad_norm": 0.27079081535339355,
      "learning_rate": 0.0002807404304782292,
      "loss": 4.7945,
      "step": 16975
    },
    {
      "epoch": 0.3477156840661028,
      "grad_norm": 0.23837393522262573,
      "learning_rate": 0.00028071007715349923,
      "loss": 4.8218,
      "step": 16980
    },
    {
      "epoch": 0.3478180738435075,
      "grad_norm": 0.25045403838157654,
      "learning_rate": 0.00028067970157254365,
      "loss": 4.7609,
      "step": 16985
    },
    {
      "epoch": 0.3479204636209121,
      "grad_norm": 0.4101354777812958,
      "learning_rate": 0.00028064930374053457,
      "loss": 4.757,
      "step": 16990
    },
    {
      "epoch": 0.34802285339831673,
      "grad_norm": 0.2520548701286316,
      "learning_rate": 0.0002806188836626479,
      "loss": 4.8007,
      "step": 16995
    },
    {
      "epoch": 0.34812524317572136,
      "grad_norm": 0.2483699470758438,
      "learning_rate": 0.00028058844134406326,
      "loss": 4.812,
      "step": 17000
    },
    {
      "epoch": 0.34812524317572136,
      "eval_loss": 4.819680690765381,
      "eval_runtime": 17.2382,
      "eval_samples_per_second": 33.762,
      "eval_steps_per_second": 4.235,
      "step": 17000
    },
    {
      "epoch": 0.348227632953126,
      "grad_norm": 0.24901323020458221,
      "learning_rate": 0.0002805579767899642,
      "loss": 4.8143,
      "step": 17005
    },
    {
      "epoch": 0.3483300227305306,
      "grad_norm": 0.22599072754383087,
      "learning_rate": 0.00028052749000553797,
      "loss": 4.7996,
      "step": 17010
    },
    {
      "epoch": 0.3484324125079352,
      "grad_norm": 0.26469743251800537,
      "learning_rate": 0.00028049698099597557,
      "loss": 4.824,
      "step": 17015
    },
    {
      "epoch": 0.34853480228533984,
      "grad_norm": 0.2392892986536026,
      "learning_rate": 0.0002804664497664718,
      "loss": 4.8004,
      "step": 17020
    },
    {
      "epoch": 0.34863719206274446,
      "grad_norm": 0.2556057870388031,
      "learning_rate": 0.00028043589632222526,
      "loss": 4.8016,
      "step": 17025
    },
    {
      "epoch": 0.3487395818401491,
      "grad_norm": 0.24753303825855255,
      "learning_rate": 0.00028040532066843837,
      "loss": 4.8308,
      "step": 17030
    },
    {
      "epoch": 0.3488419716175537,
      "grad_norm": 0.2821330726146698,
      "learning_rate": 0.0002803747228103173,
      "loss": 4.8326,
      "step": 17035
    },
    {
      "epoch": 0.3489443613949583,
      "grad_norm": 0.25612354278564453,
      "learning_rate": 0.000280344102753072,
      "loss": 4.8169,
      "step": 17040
    },
    {
      "epoch": 0.34904675117236295,
      "grad_norm": 0.3134109079837799,
      "learning_rate": 0.00028031346050191625,
      "loss": 4.7893,
      "step": 17045
    },
    {
      "epoch": 0.34914914094976757,
      "grad_norm": 0.2650546729564667,
      "learning_rate": 0.0002802827960620674,
      "loss": 4.7929,
      "step": 17050
    },
    {
      "epoch": 0.3492515307271722,
      "grad_norm": 0.3147332966327667,
      "learning_rate": 0.0002802521094387469,
      "loss": 4.7455,
      "step": 17055
    },
    {
      "epoch": 0.3493539205045768,
      "grad_norm": 0.2313137948513031,
      "learning_rate": 0.0002802214006371797,
      "loss": 4.8454,
      "step": 17060
    },
    {
      "epoch": 0.34945631028198143,
      "grad_norm": 0.26301109790802,
      "learning_rate": 0.0002801906696625947,
      "loss": 4.8467,
      "step": 17065
    },
    {
      "epoch": 0.34955870005938605,
      "grad_norm": 0.24875031411647797,
      "learning_rate": 0.0002801599165202245,
      "loss": 4.7881,
      "step": 17070
    },
    {
      "epoch": 0.3496610898367907,
      "grad_norm": 0.23368525505065918,
      "learning_rate": 0.0002801291412153055,
      "loss": 4.7991,
      "step": 17075
    },
    {
      "epoch": 0.3497634796141953,
      "grad_norm": 0.2556806206703186,
      "learning_rate": 0.0002800983437530778,
      "loss": 4.741,
      "step": 17080
    },
    {
      "epoch": 0.3498658693915999,
      "grad_norm": 0.2457631677389145,
      "learning_rate": 0.00028006752413878543,
      "loss": 4.819,
      "step": 17085
    },
    {
      "epoch": 0.3499682591690046,
      "grad_norm": 0.24281609058380127,
      "learning_rate": 0.0002800366823776761,
      "loss": 4.8416,
      "step": 17090
    },
    {
      "epoch": 0.3500706489464092,
      "grad_norm": 0.23999285697937012,
      "learning_rate": 0.00028000581847500126,
      "loss": 4.8062,
      "step": 17095
    },
    {
      "epoch": 0.35017303872381383,
      "grad_norm": 0.2510368227958679,
      "learning_rate": 0.0002799749324360161,
      "loss": 4.8053,
      "step": 17100
    },
    {
      "epoch": 0.35027542850121846,
      "grad_norm": 0.25193095207214355,
      "learning_rate": 0.00027994402426597965,
      "loss": 4.7974,
      "step": 17105
    },
    {
      "epoch": 0.3503778182786231,
      "grad_norm": 0.23930718004703522,
      "learning_rate": 0.00027991309397015475,
      "loss": 4.7517,
      "step": 17110
    },
    {
      "epoch": 0.3504802080560277,
      "grad_norm": 0.25219717621803284,
      "learning_rate": 0.00027988214155380797,
      "loss": 4.7869,
      "step": 17115
    },
    {
      "epoch": 0.3505825978334323,
      "grad_norm": 0.2704419493675232,
      "learning_rate": 0.00027985116702220955,
      "loss": 4.8298,
      "step": 17120
    },
    {
      "epoch": 0.35068498761083694,
      "grad_norm": 0.24905303120613098,
      "learning_rate": 0.0002798201703806336,
      "loss": 4.7257,
      "step": 17125
    },
    {
      "epoch": 0.35078737738824156,
      "grad_norm": 0.2514397203922272,
      "learning_rate": 0.000279789151634358,
      "loss": 4.7664,
      "step": 17130
    },
    {
      "epoch": 0.3508897671656462,
      "grad_norm": 0.26159238815307617,
      "learning_rate": 0.0002797581107886644,
      "loss": 4.7936,
      "step": 17135
    },
    {
      "epoch": 0.3509921569430508,
      "grad_norm": 0.24079452455043793,
      "learning_rate": 0.0002797270478488381,
      "loss": 4.7738,
      "step": 17140
    },
    {
      "epoch": 0.3510945467204554,
      "grad_norm": 0.2545469403266907,
      "learning_rate": 0.00027969596282016826,
      "loss": 4.8097,
      "step": 17145
    },
    {
      "epoch": 0.35119693649786005,
      "grad_norm": 0.22803018987178802,
      "learning_rate": 0.0002796648557079478,
      "loss": 4.8116,
      "step": 17150
    },
    {
      "epoch": 0.35129932627526467,
      "grad_norm": 0.25125497579574585,
      "learning_rate": 0.00027963372651747335,
      "loss": 4.7974,
      "step": 17155
    },
    {
      "epoch": 0.3514017160526693,
      "grad_norm": 0.24489642679691315,
      "learning_rate": 0.0002796025752540454,
      "loss": 4.8409,
      "step": 17160
    },
    {
      "epoch": 0.3515041058300739,
      "grad_norm": 0.2647576332092285,
      "learning_rate": 0.00027957140192296805,
      "loss": 4.7582,
      "step": 17165
    },
    {
      "epoch": 0.35160649560747853,
      "grad_norm": 0.25523173809051514,
      "learning_rate": 0.0002795402065295493,
      "loss": 4.7607,
      "step": 17170
    },
    {
      "epoch": 0.35170888538488315,
      "grad_norm": 0.29808324575424194,
      "learning_rate": 0.0002795089890791007,
      "loss": 4.7946,
      "step": 17175
    },
    {
      "epoch": 0.3518112751622878,
      "grad_norm": 0.25081580877304077,
      "learning_rate": 0.00027947774957693794,
      "loss": 4.8322,
      "step": 17180
    },
    {
      "epoch": 0.3519136649396924,
      "grad_norm": 0.23519980907440186,
      "learning_rate": 0.00027944648802838,
      "loss": 4.8076,
      "step": 17185
    },
    {
      "epoch": 0.35201605471709707,
      "grad_norm": 0.24327452480793,
      "learning_rate": 0.0002794152044387499,
      "loss": 4.761,
      "step": 17190
    },
    {
      "epoch": 0.3521184444945017,
      "grad_norm": 0.22986432909965515,
      "learning_rate": 0.0002793838988133745,
      "loss": 4.8265,
      "step": 17195
    },
    {
      "epoch": 0.3522208342719063,
      "grad_norm": 0.24117164313793182,
      "learning_rate": 0.00027935257115758403,
      "loss": 4.7674,
      "step": 17200
    },
    {
      "epoch": 0.35232322404931093,
      "grad_norm": 0.26325535774230957,
      "learning_rate": 0.00027932122147671286,
      "loss": 4.7465,
      "step": 17205
    },
    {
      "epoch": 0.35242561382671556,
      "grad_norm": 0.23438550531864166,
      "learning_rate": 0.0002792898497760989,
      "loss": 4.7598,
      "step": 17210
    },
    {
      "epoch": 0.3525280036041202,
      "grad_norm": 0.25688278675079346,
      "learning_rate": 0.0002792584560610838,
      "loss": 4.8517,
      "step": 17215
    },
    {
      "epoch": 0.3526303933815248,
      "grad_norm": 0.2520984411239624,
      "learning_rate": 0.0002792270403370132,
      "loss": 4.7871,
      "step": 17220
    },
    {
      "epoch": 0.3527327831589294,
      "grad_norm": 0.24474072456359863,
      "learning_rate": 0.0002791956026092361,
      "loss": 4.8164,
      "step": 17225
    },
    {
      "epoch": 0.35283517293633404,
      "grad_norm": 0.23748761415481567,
      "learning_rate": 0.00027916414288310556,
      "loss": 4.7758,
      "step": 17230
    },
    {
      "epoch": 0.35293756271373866,
      "grad_norm": 0.25483253598213196,
      "learning_rate": 0.00027913266116397824,
      "loss": 4.7706,
      "step": 17235
    },
    {
      "epoch": 0.3530399524911433,
      "grad_norm": 0.23142124712467194,
      "learning_rate": 0.00027910115745721466,
      "loss": 4.8061,
      "step": 17240
    },
    {
      "epoch": 0.3531423422685479,
      "grad_norm": 0.23943887650966644,
      "learning_rate": 0.00027906963176817896,
      "loss": 4.7508,
      "step": 17245
    },
    {
      "epoch": 0.3532447320459525,
      "grad_norm": 0.24452316761016846,
      "learning_rate": 0.0002790380841022391,
      "loss": 4.7999,
      "step": 17250
    },
    {
      "epoch": 0.35334712182335715,
      "grad_norm": 0.2289561629295349,
      "learning_rate": 0.00027900651446476673,
      "loss": 4.7526,
      "step": 17255
    },
    {
      "epoch": 0.35344951160076177,
      "grad_norm": 0.2562633752822876,
      "learning_rate": 0.0002789749228611372,
      "loss": 4.7962,
      "step": 17260
    },
    {
      "epoch": 0.3535519013781664,
      "grad_norm": 0.2539807856082916,
      "learning_rate": 0.0002789433092967298,
      "loss": 4.8088,
      "step": 17265
    },
    {
      "epoch": 0.353654291155571,
      "grad_norm": 0.2711130380630493,
      "learning_rate": 0.00027891167377692733,
      "loss": 4.7739,
      "step": 17270
    },
    {
      "epoch": 0.35375668093297563,
      "grad_norm": 0.23005102574825287,
      "learning_rate": 0.0002788800163071164,
      "loss": 4.7649,
      "step": 17275
    },
    {
      "epoch": 0.35385907071038025,
      "grad_norm": 0.27644792199134827,
      "learning_rate": 0.0002788483368926875,
      "loss": 4.811,
      "step": 17280
    },
    {
      "epoch": 0.3539614604877849,
      "grad_norm": 0.23939038813114166,
      "learning_rate": 0.0002788166355390347,
      "loss": 4.8298,
      "step": 17285
    },
    {
      "epoch": 0.35406385026518955,
      "grad_norm": 0.2482702136039734,
      "learning_rate": 0.00027878491225155573,
      "loss": 4.7856,
      "step": 17290
    },
    {
      "epoch": 0.35416624004259417,
      "grad_norm": 0.2748696208000183,
      "learning_rate": 0.00027875316703565226,
      "loss": 4.8458,
      "step": 17295
    },
    {
      "epoch": 0.3542686298199988,
      "grad_norm": 0.2343509942293167,
      "learning_rate": 0.00027872139989672967,
      "loss": 4.7806,
      "step": 17300
    },
    {
      "epoch": 0.3543710195974034,
      "grad_norm": 0.2615395188331604,
      "learning_rate": 0.0002786896108401968,
      "loss": 4.7473,
      "step": 17305
    },
    {
      "epoch": 0.35447340937480804,
      "grad_norm": 0.2535521388053894,
      "learning_rate": 0.0002786577998714666,
      "loss": 4.8143,
      "step": 17310
    },
    {
      "epoch": 0.35457579915221266,
      "grad_norm": 0.25139519572257996,
      "learning_rate": 0.0002786259669959556,
      "loss": 4.7792,
      "step": 17315
    },
    {
      "epoch": 0.3546781889296173,
      "grad_norm": 0.23582305014133453,
      "learning_rate": 0.0002785941122190839,
      "loss": 4.8135,
      "step": 17320
    },
    {
      "epoch": 0.3547805787070219,
      "grad_norm": 0.243644580245018,
      "learning_rate": 0.00027856223554627557,
      "loss": 4.8175,
      "step": 17325
    },
    {
      "epoch": 0.3548829684844265,
      "grad_norm": 0.226680725812912,
      "learning_rate": 0.0002785303369829582,
      "loss": 4.7672,
      "step": 17330
    },
    {
      "epoch": 0.35498535826183114,
      "grad_norm": 0.2555479109287262,
      "learning_rate": 0.00027849841653456337,
      "loss": 4.792,
      "step": 17335
    },
    {
      "epoch": 0.35508774803923576,
      "grad_norm": 0.24544550478458405,
      "learning_rate": 0.0002784664742065261,
      "loss": 4.758,
      "step": 17340
    },
    {
      "epoch": 0.3551901378166404,
      "grad_norm": 0.2856253981590271,
      "learning_rate": 0.00027843451000428526,
      "loss": 4.7925,
      "step": 17345
    },
    {
      "epoch": 0.355292527594045,
      "grad_norm": 0.33989331126213074,
      "learning_rate": 0.00027840252393328355,
      "loss": 4.7906,
      "step": 17350
    },
    {
      "epoch": 0.3553949173714496,
      "grad_norm": 0.2488049864768982,
      "learning_rate": 0.0002783705159989672,
      "loss": 4.8246,
      "step": 17355
    },
    {
      "epoch": 0.35549730714885425,
      "grad_norm": 0.2608996629714966,
      "learning_rate": 0.0002783384862067863,
      "loss": 4.8372,
      "step": 17360
    },
    {
      "epoch": 0.35559969692625887,
      "grad_norm": 0.24256552755832672,
      "learning_rate": 0.00027830643456219463,
      "loss": 4.7457,
      "step": 17365
    },
    {
      "epoch": 0.3557020867036635,
      "grad_norm": 0.23565340042114258,
      "learning_rate": 0.00027827436107064963,
      "loss": 4.8171,
      "step": 17370
    },
    {
      "epoch": 0.3558044764810681,
      "grad_norm": 0.23600706458091736,
      "learning_rate": 0.0002782422657376125,
      "loss": 4.7826,
      "step": 17375
    },
    {
      "epoch": 0.35590686625847273,
      "grad_norm": 0.25355175137519836,
      "learning_rate": 0.00027821014856854823,
      "loss": 4.8079,
      "step": 17380
    },
    {
      "epoch": 0.35600925603587735,
      "grad_norm": 0.24057556688785553,
      "learning_rate": 0.0002781780095689255,
      "loss": 4.7874,
      "step": 17385
    },
    {
      "epoch": 0.35611164581328203,
      "grad_norm": 0.2481364607810974,
      "learning_rate": 0.00027814584874421644,
      "loss": 4.7971,
      "step": 17390
    },
    {
      "epoch": 0.35621403559068665,
      "grad_norm": 0.23850895464420319,
      "learning_rate": 0.00027811366609989743,
      "loss": 4.8081,
      "step": 17395
    },
    {
      "epoch": 0.35631642536809127,
      "grad_norm": 0.24444949626922607,
      "learning_rate": 0.00027808146164144806,
      "loss": 4.8147,
      "step": 17400
    },
    {
      "epoch": 0.3564188151454959,
      "grad_norm": 0.24324198067188263,
      "learning_rate": 0.00027804923537435195,
      "loss": 4.8222,
      "step": 17405
    },
    {
      "epoch": 0.3565212049229005,
      "grad_norm": 0.33767616748809814,
      "learning_rate": 0.00027801698730409624,
      "loss": 4.7202,
      "step": 17410
    },
    {
      "epoch": 0.35662359470030514,
      "grad_norm": 0.24210509657859802,
      "learning_rate": 0.0002779847174361719,
      "loss": 4.7669,
      "step": 17415
    },
    {
      "epoch": 0.35672598447770976,
      "grad_norm": 0.2612592875957489,
      "learning_rate": 0.00027795242577607356,
      "loss": 4.7519,
      "step": 17420
    },
    {
      "epoch": 0.3568283742551144,
      "grad_norm": 0.2495550662279129,
      "learning_rate": 0.0002779201123292996,
      "loss": 4.8081,
      "step": 17425
    },
    {
      "epoch": 0.356930764032519,
      "grad_norm": 0.2539116144180298,
      "learning_rate": 0.0002778877771013521,
      "loss": 4.7959,
      "step": 17430
    },
    {
      "epoch": 0.3570331538099236,
      "grad_norm": 0.25741246342658997,
      "learning_rate": 0.0002778554200977368,
      "loss": 4.8034,
      "step": 17435
    },
    {
      "epoch": 0.35713554358732824,
      "grad_norm": 0.2593652009963989,
      "learning_rate": 0.00027782304132396314,
      "loss": 4.7875,
      "step": 17440
    },
    {
      "epoch": 0.35723793336473286,
      "grad_norm": 0.24774949252605438,
      "learning_rate": 0.0002777906407855444,
      "loss": 4.7676,
      "step": 17445
    },
    {
      "epoch": 0.3573403231421375,
      "grad_norm": 0.24466589093208313,
      "learning_rate": 0.00027775821848799746,
      "loss": 4.7929,
      "step": 17450
    },
    {
      "epoch": 0.3574427129195421,
      "grad_norm": 0.24842630326747894,
      "learning_rate": 0.0002777257744368428,
      "loss": 4.8114,
      "step": 17455
    },
    {
      "epoch": 0.3575451026969467,
      "grad_norm": 0.242864191532135,
      "learning_rate": 0.00027769330863760487,
      "loss": 4.7947,
      "step": 17460
    },
    {
      "epoch": 0.35764749247435135,
      "grad_norm": 0.24956855177879333,
      "learning_rate": 0.0002776608210958116,
      "loss": 4.7834,
      "step": 17465
    },
    {
      "epoch": 0.35774988225175597,
      "grad_norm": 0.25236186385154724,
      "learning_rate": 0.00027762831181699473,
      "loss": 4.7886,
      "step": 17470
    },
    {
      "epoch": 0.3578522720291606,
      "grad_norm": 0.2560493052005768,
      "learning_rate": 0.0002775957808066897,
      "loss": 4.7606,
      "step": 17475
    },
    {
      "epoch": 0.3579546618065652,
      "grad_norm": 0.27371951937675476,
      "learning_rate": 0.0002775632280704355,
      "loss": 4.774,
      "step": 17480
    },
    {
      "epoch": 0.35805705158396983,
      "grad_norm": 0.24178016185760498,
      "learning_rate": 0.0002775306536137751,
      "loss": 4.752,
      "step": 17485
    },
    {
      "epoch": 0.35815944136137445,
      "grad_norm": 0.24859295785427094,
      "learning_rate": 0.0002774980574422549,
      "loss": 4.8077,
      "step": 17490
    },
    {
      "epoch": 0.35826183113877913,
      "grad_norm": 0.2725426256656647,
      "learning_rate": 0.0002774654395614251,
      "loss": 4.8069,
      "step": 17495
    },
    {
      "epoch": 0.35836422091618375,
      "grad_norm": 0.25182583928108215,
      "learning_rate": 0.00027743279997683964,
      "loss": 4.7768,
      "step": 17500
    },
    {
      "epoch": 0.3584666106935884,
      "grad_norm": 0.2423197627067566,
      "learning_rate": 0.0002774001386940561,
      "loss": 4.78,
      "step": 17505
    },
    {
      "epoch": 0.358569000470993,
      "grad_norm": 0.2437697947025299,
      "learning_rate": 0.00027736745571863583,
      "loss": 4.8309,
      "step": 17510
    },
    {
      "epoch": 0.3586713902483976,
      "grad_norm": 0.25134211778640747,
      "learning_rate": 0.00027733475105614367,
      "loss": 4.7802,
      "step": 17515
    },
    {
      "epoch": 0.35877378002580224,
      "grad_norm": 0.2694650888442993,
      "learning_rate": 0.0002773020247121485,
      "loss": 4.772,
      "step": 17520
    },
    {
      "epoch": 0.35887616980320686,
      "grad_norm": 0.26452839374542236,
      "learning_rate": 0.00027726927669222254,
      "loss": 4.7724,
      "step": 17525
    },
    {
      "epoch": 0.3589785595806115,
      "grad_norm": 0.2345585823059082,
      "learning_rate": 0.0002772365070019419,
      "loss": 4.8204,
      "step": 17530
    },
    {
      "epoch": 0.3590809493580161,
      "grad_norm": 0.25094476342201233,
      "learning_rate": 0.0002772037156468863,
      "loss": 4.7761,
      "step": 17535
    },
    {
      "epoch": 0.3591833391354207,
      "grad_norm": 0.2713305652141571,
      "learning_rate": 0.00027717090263263924,
      "loss": 4.832,
      "step": 17540
    },
    {
      "epoch": 0.35928572891282534,
      "grad_norm": 0.2531552016735077,
      "learning_rate": 0.00027713806796478784,
      "loss": 4.7771,
      "step": 17545
    },
    {
      "epoch": 0.35938811869022996,
      "grad_norm": 0.2462606132030487,
      "learning_rate": 0.00027710521164892276,
      "loss": 4.7297,
      "step": 17550
    },
    {
      "epoch": 0.3594905084676346,
      "grad_norm": 0.23273545503616333,
      "learning_rate": 0.00027707233369063877,
      "loss": 4.7601,
      "step": 17555
    },
    {
      "epoch": 0.3595928982450392,
      "grad_norm": 0.2425830215215683,
      "learning_rate": 0.00027703943409553385,
      "loss": 4.8245,
      "step": 17560
    },
    {
      "epoch": 0.3596952880224438,
      "grad_norm": 0.25369793176651,
      "learning_rate": 0.00027700651286920993,
      "loss": 4.7523,
      "step": 17565
    },
    {
      "epoch": 0.35979767779984845,
      "grad_norm": 0.35871362686157227,
      "learning_rate": 0.0002769735700172726,
      "loss": 4.8501,
      "step": 17570
    },
    {
      "epoch": 0.35990006757725307,
      "grad_norm": 0.2735600173473358,
      "learning_rate": 0.00027694060554533104,
      "loss": 4.7961,
      "step": 17575
    },
    {
      "epoch": 0.3600024573546577,
      "grad_norm": 0.2527202367782593,
      "learning_rate": 0.0002769076194589982,
      "loss": 4.8039,
      "step": 17580
    },
    {
      "epoch": 0.3601048471320623,
      "grad_norm": 0.23288428783416748,
      "learning_rate": 0.0002768746117638906,
      "loss": 4.7856,
      "step": 17585
    },
    {
      "epoch": 0.36020723690946693,
      "grad_norm": 0.24862439930438995,
      "learning_rate": 0.0002768415824656287,
      "loss": 4.7309,
      "step": 17590
    },
    {
      "epoch": 0.3603096266868716,
      "grad_norm": 0.25607404112815857,
      "learning_rate": 0.00027680853156983624,
      "loss": 4.7547,
      "step": 17595
    },
    {
      "epoch": 0.36041201646427623,
      "grad_norm": 0.25481516122817993,
      "learning_rate": 0.00027677545908214095,
      "loss": 4.7573,
      "step": 17600
    },
    {
      "epoch": 0.36051440624168085,
      "grad_norm": 0.25555238127708435,
      "learning_rate": 0.00027674236500817415,
      "loss": 4.8065,
      "step": 17605
    },
    {
      "epoch": 0.3606167960190855,
      "grad_norm": 0.23385271430015564,
      "learning_rate": 0.0002767092493535708,
      "loss": 4.795,
      "step": 17610
    },
    {
      "epoch": 0.3607191857964901,
      "grad_norm": 0.24765196442604065,
      "learning_rate": 0.0002766761121239696,
      "loss": 4.8329,
      "step": 17615
    },
    {
      "epoch": 0.3608215755738947,
      "grad_norm": 0.2550165355205536,
      "learning_rate": 0.0002766429533250128,
      "loss": 4.793,
      "step": 17620
    },
    {
      "epoch": 0.36092396535129934,
      "grad_norm": 0.25023287534713745,
      "learning_rate": 0.0002766097729623465,
      "loss": 4.7971,
      "step": 17625
    },
    {
      "epoch": 0.36102635512870396,
      "grad_norm": 0.25133055448532104,
      "learning_rate": 0.0002765765710416203,
      "loss": 4.8053,
      "step": 17630
    },
    {
      "epoch": 0.3611287449061086,
      "grad_norm": 0.24358288943767548,
      "learning_rate": 0.00027654334756848763,
      "loss": 4.8307,
      "step": 17635
    },
    {
      "epoch": 0.3612311346835132,
      "grad_norm": 0.25189048051834106,
      "learning_rate": 0.0002765101025486054,
      "loss": 4.8124,
      "step": 17640
    },
    {
      "epoch": 0.3613335244609178,
      "grad_norm": 0.23731786012649536,
      "learning_rate": 0.0002764768359876343,
      "loss": 4.7909,
      "step": 17645
    },
    {
      "epoch": 0.36143591423832244,
      "grad_norm": 0.24712516367435455,
      "learning_rate": 0.00027644354789123886,
      "loss": 4.8159,
      "step": 17650
    },
    {
      "epoch": 0.36153830401572706,
      "grad_norm": 0.2414264678955078,
      "learning_rate": 0.0002764102382650869,
      "loss": 4.7784,
      "step": 17655
    },
    {
      "epoch": 0.3616406937931317,
      "grad_norm": 0.23889292776584625,
      "learning_rate": 0.0002763769071148502,
      "loss": 4.8693,
      "step": 17660
    },
    {
      "epoch": 0.3617430835705363,
      "grad_norm": 0.2331574559211731,
      "learning_rate": 0.00027634355444620403,
      "loss": 4.8343,
      "step": 17665
    },
    {
      "epoch": 0.3618454733479409,
      "grad_norm": 0.24381151795387268,
      "learning_rate": 0.0002763101802648275,
      "loss": 4.8148,
      "step": 17670
    },
    {
      "epoch": 0.36194786312534555,
      "grad_norm": 0.24279746413230896,
      "learning_rate": 0.0002762767845764033,
      "loss": 4.7826,
      "step": 17675
    },
    {
      "epoch": 0.36205025290275017,
      "grad_norm": 0.24833084642887115,
      "learning_rate": 0.0002762433673866177,
      "loss": 4.8797,
      "step": 17680
    },
    {
      "epoch": 0.3621526426801548,
      "grad_norm": 0.23561091721057892,
      "learning_rate": 0.0002762099287011607,
      "loss": 4.8394,
      "step": 17685
    },
    {
      "epoch": 0.3622550324575594,
      "grad_norm": 0.24129699170589447,
      "learning_rate": 0.00027617646852572597,
      "loss": 4.8028,
      "step": 17690
    },
    {
      "epoch": 0.3623574222349641,
      "grad_norm": 0.24755354225635529,
      "learning_rate": 0.0002761429868660109,
      "loss": 4.7667,
      "step": 17695
    },
    {
      "epoch": 0.3624598120123687,
      "grad_norm": 0.2509928047657013,
      "learning_rate": 0.00027610948372771643,
      "loss": 4.7892,
      "step": 17700
    },
    {
      "epoch": 0.36256220178977333,
      "grad_norm": 0.2602989971637726,
      "learning_rate": 0.0002760759591165471,
      "loss": 4.8036,
      "step": 17705
    },
    {
      "epoch": 0.36266459156717795,
      "grad_norm": 0.25540390610694885,
      "learning_rate": 0.0002760424130382113,
      "loss": 4.7706,
      "step": 17710
    },
    {
      "epoch": 0.3627669813445826,
      "grad_norm": 0.24948926270008087,
      "learning_rate": 0.000276008845498421,
      "loss": 4.8129,
      "step": 17715
    },
    {
      "epoch": 0.3628693711219872,
      "grad_norm": 0.2444363236427307,
      "learning_rate": 0.0002759752565028918,
      "loss": 4.8407,
      "step": 17720
    },
    {
      "epoch": 0.3629717608993918,
      "grad_norm": 0.2489061802625656,
      "learning_rate": 0.00027594164605734286,
      "loss": 4.7935,
      "step": 17725
    },
    {
      "epoch": 0.36307415067679644,
      "grad_norm": 0.25345760583877563,
      "learning_rate": 0.0002759080141674972,
      "loss": 4.773,
      "step": 17730
    },
    {
      "epoch": 0.36317654045420106,
      "grad_norm": 0.23497188091278076,
      "learning_rate": 0.0002758743608390813,
      "loss": 4.8028,
      "step": 17735
    },
    {
      "epoch": 0.3632789302316057,
      "grad_norm": 0.23632681369781494,
      "learning_rate": 0.00027584068607782537,
      "loss": 4.8242,
      "step": 17740
    },
    {
      "epoch": 0.3633813200090103,
      "grad_norm": 0.22838786244392395,
      "learning_rate": 0.0002758069898894633,
      "loss": 4.7714,
      "step": 17745
    },
    {
      "epoch": 0.3634837097864149,
      "grad_norm": 0.2365570217370987,
      "learning_rate": 0.00027577327227973255,
      "loss": 4.758,
      "step": 17750
    },
    {
      "epoch": 0.36358609956381954,
      "grad_norm": 0.22895501554012299,
      "learning_rate": 0.0002757395332543744,
      "loss": 4.7893,
      "step": 17755
    },
    {
      "epoch": 0.36368848934122416,
      "grad_norm": 0.2436460256576538,
      "learning_rate": 0.0002757057728191335,
      "loss": 4.7621,
      "step": 17760
    },
    {
      "epoch": 0.3637908791186288,
      "grad_norm": 0.2596724033355713,
      "learning_rate": 0.0002756719909797584,
      "loss": 4.7822,
      "step": 17765
    },
    {
      "epoch": 0.3638932688960334,
      "grad_norm": 0.24657489359378815,
      "learning_rate": 0.0002756381877420011,
      "loss": 4.7821,
      "step": 17770
    },
    {
      "epoch": 0.363995658673438,
      "grad_norm": 0.2288065105676651,
      "learning_rate": 0.0002756043631116175,
      "loss": 4.7956,
      "step": 17775
    },
    {
      "epoch": 0.36409804845084265,
      "grad_norm": 0.24526286125183105,
      "learning_rate": 0.0002755705170943667,
      "loss": 4.7953,
      "step": 17780
    },
    {
      "epoch": 0.36420043822824727,
      "grad_norm": 0.27630555629730225,
      "learning_rate": 0.000275536649696012,
      "loss": 4.8206,
      "step": 17785
    },
    {
      "epoch": 0.3643028280056519,
      "grad_norm": 0.24275371432304382,
      "learning_rate": 0.0002755027609223199,
      "loss": 4.7961,
      "step": 17790
    },
    {
      "epoch": 0.3644052177830565,
      "grad_norm": 0.25726425647735596,
      "learning_rate": 0.0002754688507790607,
      "loss": 4.8007,
      "step": 17795
    },
    {
      "epoch": 0.3645076075604612,
      "grad_norm": 0.26756492257118225,
      "learning_rate": 0.0002754349192720084,
      "loss": 4.846,
      "step": 17800
    },
    {
      "epoch": 0.3646099973378658,
      "grad_norm": 0.2504918873310089,
      "learning_rate": 0.0002754009664069406,
      "loss": 4.7891,
      "step": 17805
    },
    {
      "epoch": 0.36471238711527043,
      "grad_norm": 0.25351810455322266,
      "learning_rate": 0.0002753669921896384,
      "loss": 4.7709,
      "step": 17810
    },
    {
      "epoch": 0.36481477689267505,
      "grad_norm": 0.26117199659347534,
      "learning_rate": 0.0002753329966258868,
      "loss": 4.8621,
      "step": 17815
    },
    {
      "epoch": 0.3649171666700797,
      "grad_norm": 0.22292082011699677,
      "learning_rate": 0.00027529897972147413,
      "loss": 4.7652,
      "step": 17820
    },
    {
      "epoch": 0.3650195564474843,
      "grad_norm": 0.24843989312648773,
      "learning_rate": 0.0002752649414821926,
      "loss": 4.7668,
      "step": 17825
    },
    {
      "epoch": 0.3651219462248889,
      "grad_norm": 0.24700407683849335,
      "learning_rate": 0.0002752308819138379,
      "loss": 4.8542,
      "step": 17830
    },
    {
      "epoch": 0.36522433600229354,
      "grad_norm": 0.2571440041065216,
      "learning_rate": 0.00027519680102220946,
      "loss": 4.7304,
      "step": 17835
    },
    {
      "epoch": 0.36532672577969816,
      "grad_norm": 0.2338082641363144,
      "learning_rate": 0.0002751626988131102,
      "loss": 4.772,
      "step": 17840
    },
    {
      "epoch": 0.3654291155571028,
      "grad_norm": 0.24531778693199158,
      "learning_rate": 0.0002751285752923469,
      "loss": 4.7017,
      "step": 17845
    },
    {
      "epoch": 0.3655315053345074,
      "grad_norm": 0.25323912501335144,
      "learning_rate": 0.0002750944304657297,
      "loss": 4.8142,
      "step": 17850
    },
    {
      "epoch": 0.365633895111912,
      "grad_norm": 0.25650572776794434,
      "learning_rate": 0.00027506026433907264,
      "loss": 4.7979,
      "step": 17855
    },
    {
      "epoch": 0.36573628488931664,
      "grad_norm": 0.2664353549480438,
      "learning_rate": 0.0002750260769181931,
      "loss": 4.7791,
      "step": 17860
    },
    {
      "epoch": 0.36583867466672126,
      "grad_norm": 0.2455276995897293,
      "learning_rate": 0.0002749918682089123,
      "loss": 4.7854,
      "step": 17865
    },
    {
      "epoch": 0.3659410644441259,
      "grad_norm": 0.2544846832752228,
      "learning_rate": 0.000274957638217055,
      "loss": 4.7654,
      "step": 17870
    },
    {
      "epoch": 0.3660434542215305,
      "grad_norm": 0.2378363460302353,
      "learning_rate": 0.00027492338694844957,
      "loss": 4.763,
      "step": 17875
    },
    {
      "epoch": 0.3661458439989351,
      "grad_norm": 0.26293930411338806,
      "learning_rate": 0.00027488911440892814,
      "loss": 4.8123,
      "step": 17880
    },
    {
      "epoch": 0.36624823377633975,
      "grad_norm": 0.2636087238788605,
      "learning_rate": 0.00027485482060432624,
      "loss": 4.7265,
      "step": 17885
    },
    {
      "epoch": 0.36635062355374437,
      "grad_norm": 0.23257797956466675,
      "learning_rate": 0.0002748205055404832,
      "loss": 4.8015,
      "step": 17890
    },
    {
      "epoch": 0.366453013331149,
      "grad_norm": 0.25716593861579895,
      "learning_rate": 0.0002747861692232418,
      "loss": 4.793,
      "step": 17895
    },
    {
      "epoch": 0.36655540310855367,
      "grad_norm": 0.24235756695270538,
      "learning_rate": 0.0002747518116584487,
      "loss": 4.7958,
      "step": 17900
    },
    {
      "epoch": 0.3666577928859583,
      "grad_norm": 0.24651765823364258,
      "learning_rate": 0.00027471743285195383,
      "loss": 4.7845,
      "step": 17905
    },
    {
      "epoch": 0.3667601826633629,
      "grad_norm": 0.23017224669456482,
      "learning_rate": 0.00027468303280961116,
      "loss": 4.8313,
      "step": 17910
    },
    {
      "epoch": 0.36686257244076753,
      "grad_norm": 0.3161824941635132,
      "learning_rate": 0.00027464861153727785,
      "loss": 4.8361,
      "step": 17915
    },
    {
      "epoch": 0.36696496221817215,
      "grad_norm": 0.2504814565181732,
      "learning_rate": 0.00027461416904081494,
      "loss": 4.7972,
      "step": 17920
    },
    {
      "epoch": 0.3670673519955768,
      "grad_norm": 0.24127408862113953,
      "learning_rate": 0.00027457970532608703,
      "loss": 4.8106,
      "step": 17925
    },
    {
      "epoch": 0.3671697417729814,
      "grad_norm": 0.26364776492118835,
      "learning_rate": 0.00027454522039896227,
      "loss": 4.8322,
      "step": 17930
    },
    {
      "epoch": 0.367272131550386,
      "grad_norm": 0.6151398420333862,
      "learning_rate": 0.00027451071426531253,
      "loss": 4.8723,
      "step": 17935
    },
    {
      "epoch": 0.36737452132779064,
      "grad_norm": 0.2452814280986786,
      "learning_rate": 0.00027447618693101316,
      "loss": 4.7882,
      "step": 17940
    },
    {
      "epoch": 0.36747691110519526,
      "grad_norm": 0.2335476130247116,
      "learning_rate": 0.00027444163840194323,
      "loss": 4.8472,
      "step": 17945
    },
    {
      "epoch": 0.3675793008825999,
      "grad_norm": 0.2404443323612213,
      "learning_rate": 0.00027440706868398537,
      "loss": 4.6705,
      "step": 17950
    },
    {
      "epoch": 0.3676816906600045,
      "grad_norm": 0.2449806034564972,
      "learning_rate": 0.0002743724777830258,
      "loss": 4.8098,
      "step": 17955
    },
    {
      "epoch": 0.3677840804374091,
      "grad_norm": 0.2550186812877655,
      "learning_rate": 0.0002743378657049544,
      "loss": 4.7936,
      "step": 17960
    },
    {
      "epoch": 0.36788647021481374,
      "grad_norm": 0.24325023591518402,
      "learning_rate": 0.00027430323245566463,
      "loss": 4.7399,
      "step": 17965
    },
    {
      "epoch": 0.36798885999221836,
      "grad_norm": 0.25150230526924133,
      "learning_rate": 0.0002742685780410535,
      "loss": 4.8023,
      "step": 17970
    },
    {
      "epoch": 0.368091249769623,
      "grad_norm": 0.26221171021461487,
      "learning_rate": 0.0002742339024670217,
      "loss": 4.7608,
      "step": 17975
    },
    {
      "epoch": 0.3681936395470276,
      "grad_norm": 0.2458525151014328,
      "learning_rate": 0.00027419920573947356,
      "loss": 4.7636,
      "step": 17980
    },
    {
      "epoch": 0.36829602932443223,
      "grad_norm": 0.24908113479614258,
      "learning_rate": 0.00027416448786431684,
      "loss": 4.8019,
      "step": 17985
    },
    {
      "epoch": 0.36839841910183685,
      "grad_norm": 0.34281638264656067,
      "learning_rate": 0.00027412974884746317,
      "loss": 4.7678,
      "step": 17990
    },
    {
      "epoch": 0.36850080887924147,
      "grad_norm": 0.2360275685787201,
      "learning_rate": 0.00027409498869482746,
      "loss": 4.7555,
      "step": 17995
    },
    {
      "epoch": 0.36860319865664615,
      "grad_norm": 0.24784208834171295,
      "learning_rate": 0.00027406020741232843,
      "loss": 4.7721,
      "step": 18000
    },
    {
      "epoch": 0.36860319865664615,
      "eval_loss": 4.805654048919678,
      "eval_runtime": 17.2368,
      "eval_samples_per_second": 33.765,
      "eval_steps_per_second": 4.235,
      "step": 18000
    },
    {
      "epoch": 0.36870558843405077,
      "grad_norm": 0.2545544505119324,
      "learning_rate": 0.0002740254050058884,
      "loss": 4.8206,
      "step": 18005
    },
    {
      "epoch": 0.3688079782114554,
      "grad_norm": 0.2245105654001236,
      "learning_rate": 0.0002739905814814332,
      "loss": 4.8171,
      "step": 18010
    },
    {
      "epoch": 0.36891036798886,
      "grad_norm": 0.2906739413738251,
      "learning_rate": 0.0002739557368448923,
      "loss": 4.7607,
      "step": 18015
    },
    {
      "epoch": 0.36901275776626463,
      "grad_norm": 0.2782283425331116,
      "learning_rate": 0.0002739208711021986,
      "loss": 4.7538,
      "step": 18020
    },
    {
      "epoch": 0.36911514754366925,
      "grad_norm": 0.25412094593048096,
      "learning_rate": 0.000273885984259289,
      "loss": 4.7402,
      "step": 18025
    },
    {
      "epoch": 0.3692175373210739,
      "grad_norm": 0.2368987798690796,
      "learning_rate": 0.0002738510763221037,
      "loss": 4.815,
      "step": 18030
    },
    {
      "epoch": 0.3693199270984785,
      "grad_norm": 0.25744420289993286,
      "learning_rate": 0.00027381614729658635,
      "loss": 4.7611,
      "step": 18035
    },
    {
      "epoch": 0.3694223168758831,
      "grad_norm": 0.2621389925479889,
      "learning_rate": 0.0002737811971886846,
      "loss": 4.774,
      "step": 18040
    },
    {
      "epoch": 0.36952470665328774,
      "grad_norm": 0.24370889365673065,
      "learning_rate": 0.00027374622600434925,
      "loss": 4.8189,
      "step": 18045
    },
    {
      "epoch": 0.36962709643069236,
      "grad_norm": 0.24246658384799957,
      "learning_rate": 0.000273711233749535,
      "loss": 4.7838,
      "step": 18050
    },
    {
      "epoch": 0.369729486208097,
      "grad_norm": 0.28695717453956604,
      "learning_rate": 0.0002736762204302001,
      "loss": 4.8252,
      "step": 18055
    },
    {
      "epoch": 0.3698318759855016,
      "grad_norm": 0.2438812553882599,
      "learning_rate": 0.0002736411860523063,
      "loss": 4.8097,
      "step": 18060
    },
    {
      "epoch": 0.3699342657629062,
      "grad_norm": 0.2786584496498108,
      "learning_rate": 0.0002736061306218189,
      "loss": 4.7911,
      "step": 18065
    },
    {
      "epoch": 0.37003665554031084,
      "grad_norm": 0.2415219396352768,
      "learning_rate": 0.0002735710541447069,
      "loss": 4.796,
      "step": 18070
    },
    {
      "epoch": 0.37013904531771546,
      "grad_norm": 0.24593287706375122,
      "learning_rate": 0.0002735359566269428,
      "loss": 4.7672,
      "step": 18075
    },
    {
      "epoch": 0.3702414350951201,
      "grad_norm": 0.31544220447540283,
      "learning_rate": 0.0002735008380745027,
      "loss": 4.7847,
      "step": 18080
    },
    {
      "epoch": 0.3703438248725247,
      "grad_norm": 0.2856111228466034,
      "learning_rate": 0.00027346569849336636,
      "loss": 4.7575,
      "step": 18085
    },
    {
      "epoch": 0.37044621464992933,
      "grad_norm": 0.24270612001419067,
      "learning_rate": 0.000273430537889517,
      "loss": 4.8106,
      "step": 18090
    },
    {
      "epoch": 0.37054860442733395,
      "grad_norm": 0.24650530517101288,
      "learning_rate": 0.00027339535626894155,
      "loss": 4.787,
      "step": 18095
    },
    {
      "epoch": 0.3706509942047386,
      "grad_norm": 0.2618081867694855,
      "learning_rate": 0.00027336015363763034,
      "loss": 4.8112,
      "step": 18100
    },
    {
      "epoch": 0.37075338398214325,
      "grad_norm": 0.23776638507843018,
      "learning_rate": 0.0002733249300015774,
      "loss": 4.8249,
      "step": 18105
    },
    {
      "epoch": 0.37085577375954787,
      "grad_norm": 0.2623469829559326,
      "learning_rate": 0.0002732896853667804,
      "loss": 4.7815,
      "step": 18110
    },
    {
      "epoch": 0.3709581635369525,
      "grad_norm": 0.2656545341014862,
      "learning_rate": 0.00027325441973924045,
      "loss": 4.7354,
      "step": 18115
    },
    {
      "epoch": 0.3710605533143571,
      "grad_norm": 0.2753200829029083,
      "learning_rate": 0.0002732191331249622,
      "loss": 4.8207,
      "step": 18120
    },
    {
      "epoch": 0.37116294309176173,
      "grad_norm": 0.24409864842891693,
      "learning_rate": 0.00027318382552995416,
      "loss": 4.7847,
      "step": 18125
    },
    {
      "epoch": 0.37126533286916635,
      "grad_norm": 0.24308210611343384,
      "learning_rate": 0.0002731484969602281,
      "loss": 4.7126,
      "step": 18130
    },
    {
      "epoch": 0.371367722646571,
      "grad_norm": 0.2436607927083969,
      "learning_rate": 0.00027311314742179944,
      "loss": 4.7639,
      "step": 18135
    },
    {
      "epoch": 0.3714701124239756,
      "grad_norm": 0.23844410479068756,
      "learning_rate": 0.0002730777769206872,
      "loss": 4.7766,
      "step": 18140
    },
    {
      "epoch": 0.3715725022013802,
      "grad_norm": 0.23728613555431366,
      "learning_rate": 0.00027304238546291417,
      "loss": 4.7252,
      "step": 18145
    },
    {
      "epoch": 0.37167489197878484,
      "grad_norm": 0.25353166460990906,
      "learning_rate": 0.00027300697305450623,
      "loss": 4.7806,
      "step": 18150
    },
    {
      "epoch": 0.37177728175618946,
      "grad_norm": 0.23095594346523285,
      "learning_rate": 0.00027297153970149324,
      "loss": 4.8203,
      "step": 18155
    },
    {
      "epoch": 0.3718796715335941,
      "grad_norm": 0.23115617036819458,
      "learning_rate": 0.00027293608540990863,
      "loss": 4.7845,
      "step": 18160
    },
    {
      "epoch": 0.3719820613109987,
      "grad_norm": 0.2482440024614334,
      "learning_rate": 0.000272900610185789,
      "loss": 4.8398,
      "step": 18165
    },
    {
      "epoch": 0.3720844510884033,
      "grad_norm": 0.2375240921974182,
      "learning_rate": 0.000272865114035175,
      "loss": 4.7989,
      "step": 18170
    },
    {
      "epoch": 0.37218684086580794,
      "grad_norm": 0.26746800541877747,
      "learning_rate": 0.00027282959696411045,
      "loss": 4.8617,
      "step": 18175
    },
    {
      "epoch": 0.37228923064321257,
      "grad_norm": 0.2613542079925537,
      "learning_rate": 0.000272794058978643,
      "loss": 4.7514,
      "step": 18180
    },
    {
      "epoch": 0.3723916204206172,
      "grad_norm": 0.2713029086589813,
      "learning_rate": 0.00027275850008482376,
      "loss": 4.8221,
      "step": 18185
    },
    {
      "epoch": 0.3724940101980218,
      "grad_norm": 0.24614305794239044,
      "learning_rate": 0.0002727229202887074,
      "loss": 4.7811,
      "step": 18190
    },
    {
      "epoch": 0.37259639997542643,
      "grad_norm": 0.24781306087970734,
      "learning_rate": 0.0002726873195963521,
      "loss": 4.7565,
      "step": 18195
    },
    {
      "epoch": 0.37269878975283105,
      "grad_norm": 0.23562413454055786,
      "learning_rate": 0.00027265169801381976,
      "loss": 4.7957,
      "step": 18200
    },
    {
      "epoch": 0.3728011795302357,
      "grad_norm": 0.2612961530685425,
      "learning_rate": 0.00027261605554717563,
      "loss": 4.8467,
      "step": 18205
    },
    {
      "epoch": 0.37290356930764035,
      "grad_norm": 0.24376554787158966,
      "learning_rate": 0.0002725803922024886,
      "loss": 4.7415,
      "step": 18210
    },
    {
      "epoch": 0.37300595908504497,
      "grad_norm": 0.24161337316036224,
      "learning_rate": 0.0002725447079858312,
      "loss": 4.7421,
      "step": 18215
    },
    {
      "epoch": 0.3731083488624496,
      "grad_norm": 0.24609360098838806,
      "learning_rate": 0.0002725090029032794,
      "loss": 4.7671,
      "step": 18220
    },
    {
      "epoch": 0.3732107386398542,
      "grad_norm": 0.25302135944366455,
      "learning_rate": 0.00027247327696091277,
      "loss": 4.8153,
      "step": 18225
    },
    {
      "epoch": 0.37331312841725883,
      "grad_norm": 0.25249356031417847,
      "learning_rate": 0.00027243753016481444,
      "loss": 4.7782,
      "step": 18230
    },
    {
      "epoch": 0.37341551819466345,
      "grad_norm": 0.25398945808410645,
      "learning_rate": 0.00027240176252107105,
      "loss": 4.792,
      "step": 18235
    },
    {
      "epoch": 0.3735179079720681,
      "grad_norm": 0.23390665650367737,
      "learning_rate": 0.0002723659740357728,
      "loss": 4.798,
      "step": 18240
    },
    {
      "epoch": 0.3736202977494727,
      "grad_norm": 0.2441662847995758,
      "learning_rate": 0.00027233016471501354,
      "loss": 4.7666,
      "step": 18245
    },
    {
      "epoch": 0.3737226875268773,
      "grad_norm": 0.5501225590705872,
      "learning_rate": 0.00027229433456489057,
      "loss": 4.81,
      "step": 18250
    },
    {
      "epoch": 0.37382507730428194,
      "grad_norm": 0.25560975074768066,
      "learning_rate": 0.0002722584835915046,
      "loss": 4.8233,
      "step": 18255
    },
    {
      "epoch": 0.37392746708168656,
      "grad_norm": 0.29762735962867737,
      "learning_rate": 0.0002722226118009603,
      "loss": 4.8386,
      "step": 18260
    },
    {
      "epoch": 0.3740298568590912,
      "grad_norm": 0.28740373253822327,
      "learning_rate": 0.0002721867191993653,
      "loss": 4.8693,
      "step": 18265
    },
    {
      "epoch": 0.3741322466364958,
      "grad_norm": 0.31148993968963623,
      "learning_rate": 0.00027215080579283136,
      "loss": 4.7801,
      "step": 18270
    },
    {
      "epoch": 0.3742346364139004,
      "grad_norm": 0.30883365869522095,
      "learning_rate": 0.0002721148715874735,
      "loss": 4.8184,
      "step": 18275
    },
    {
      "epoch": 0.37433702619130504,
      "grad_norm": 0.2357916384935379,
      "learning_rate": 0.0002720789165894101,
      "loss": 4.7911,
      "step": 18280
    },
    {
      "epoch": 0.37443941596870967,
      "grad_norm": 0.28309592604637146,
      "learning_rate": 0.0002720429408047634,
      "loss": 4.8236,
      "step": 18285
    },
    {
      "epoch": 0.3745418057461143,
      "grad_norm": 0.23531167209148407,
      "learning_rate": 0.0002720069442396591,
      "loss": 4.7503,
      "step": 18290
    },
    {
      "epoch": 0.3746441955235189,
      "grad_norm": 0.2579338550567627,
      "learning_rate": 0.0002719709269002263,
      "loss": 4.8113,
      "step": 18295
    },
    {
      "epoch": 0.37474658530092353,
      "grad_norm": 0.23504072427749634,
      "learning_rate": 0.00027193488879259783,
      "loss": 4.7818,
      "step": 18300
    },
    {
      "epoch": 0.3748489750783282,
      "grad_norm": 0.23527486622333527,
      "learning_rate": 0.0002718988299229099,
      "loss": 4.8111,
      "step": 18305
    },
    {
      "epoch": 0.3749513648557328,
      "grad_norm": 0.2614368796348572,
      "learning_rate": 0.0002718627502973023,
      "loss": 4.7983,
      "step": 18310
    },
    {
      "epoch": 0.37505375463313745,
      "grad_norm": 0.25063908100128174,
      "learning_rate": 0.0002718266499219184,
      "loss": 4.7955,
      "step": 18315
    },
    {
      "epoch": 0.37515614441054207,
      "grad_norm": 0.24770070612430573,
      "learning_rate": 0.0002717905288029051,
      "loss": 4.7598,
      "step": 18320
    },
    {
      "epoch": 0.3752585341879467,
      "grad_norm": 0.2598324120044708,
      "learning_rate": 0.00027175438694641276,
      "loss": 4.7927,
      "step": 18325
    },
    {
      "epoch": 0.3753609239653513,
      "grad_norm": 0.23545841872692108,
      "learning_rate": 0.0002717182243585953,
      "loss": 4.759,
      "step": 18330
    },
    {
      "epoch": 0.37546331374275593,
      "grad_norm": 0.24323679506778717,
      "learning_rate": 0.00027168204104561025,
      "loss": 4.7429,
      "step": 18335
    },
    {
      "epoch": 0.37556570352016055,
      "grad_norm": 0.32741808891296387,
      "learning_rate": 0.00027164583701361856,
      "loss": 4.8504,
      "step": 18340
    },
    {
      "epoch": 0.3756680932975652,
      "grad_norm": 0.23775732517242432,
      "learning_rate": 0.00027160961226878474,
      "loss": 4.7991,
      "step": 18345
    },
    {
      "epoch": 0.3757704830749698,
      "grad_norm": 0.24511213600635529,
      "learning_rate": 0.0002715733668172769,
      "loss": 4.7407,
      "step": 18350
    },
    {
      "epoch": 0.3758728728523744,
      "grad_norm": 0.24636994302272797,
      "learning_rate": 0.00027153710066526646,
      "loss": 4.7733,
      "step": 18355
    },
    {
      "epoch": 0.37597526262977904,
      "grad_norm": 0.22909529507160187,
      "learning_rate": 0.00027150081381892874,
      "loss": 4.7889,
      "step": 18360
    },
    {
      "epoch": 0.37607765240718366,
      "grad_norm": 0.25632724165916443,
      "learning_rate": 0.0002714645062844422,
      "loss": 4.7977,
      "step": 18365
    },
    {
      "epoch": 0.3761800421845883,
      "grad_norm": 0.2533828914165497,
      "learning_rate": 0.00027142817806798906,
      "loss": 4.7285,
      "step": 18370
    },
    {
      "epoch": 0.3762824319619929,
      "grad_norm": 0.25821688771247864,
      "learning_rate": 0.0002713918291757549,
      "loss": 4.7774,
      "step": 18375
    },
    {
      "epoch": 0.3763848217393975,
      "grad_norm": 0.2438954859972,
      "learning_rate": 0.0002713554596139291,
      "loss": 4.7832,
      "step": 18380
    },
    {
      "epoch": 0.37648721151680214,
      "grad_norm": 0.2397865653038025,
      "learning_rate": 0.0002713190693887041,
      "loss": 4.8666,
      "step": 18385
    },
    {
      "epoch": 0.37658960129420677,
      "grad_norm": 0.2682671844959259,
      "learning_rate": 0.00027128265850627633,
      "loss": 4.7679,
      "step": 18390
    },
    {
      "epoch": 0.3766919910716114,
      "grad_norm": 0.2713569104671478,
      "learning_rate": 0.00027124622697284546,
      "loss": 4.7427,
      "step": 18395
    },
    {
      "epoch": 0.376794380849016,
      "grad_norm": 0.25199782848358154,
      "learning_rate": 0.00027120977479461476,
      "loss": 4.7515,
      "step": 18400
    },
    {
      "epoch": 0.3768967706264207,
      "grad_norm": 0.2549890875816345,
      "learning_rate": 0.000271173301977791,
      "loss": 4.8203,
      "step": 18405
    },
    {
      "epoch": 0.3769991604038253,
      "grad_norm": 0.25833216309547424,
      "learning_rate": 0.0002711368085285844,
      "loss": 4.7878,
      "step": 18410
    },
    {
      "epoch": 0.3771015501812299,
      "grad_norm": 0.2570663094520569,
      "learning_rate": 0.0002711002944532089,
      "loss": 4.8165,
      "step": 18415
    },
    {
      "epoch": 0.37720393995863455,
      "grad_norm": 0.24693752825260162,
      "learning_rate": 0.0002710637597578817,
      "loss": 4.794,
      "step": 18420
    },
    {
      "epoch": 0.37730632973603917,
      "grad_norm": 0.23728980123996735,
      "learning_rate": 0.0002710272044488237,
      "loss": 4.8135,
      "step": 18425
    },
    {
      "epoch": 0.3774087195134438,
      "grad_norm": 0.2658339738845825,
      "learning_rate": 0.0002709906285322592,
      "loss": 4.7731,
      "step": 18430
    },
    {
      "epoch": 0.3775111092908484,
      "grad_norm": 0.23852306604385376,
      "learning_rate": 0.00027095403201441603,
      "loss": 4.7737,
      "step": 18435
    },
    {
      "epoch": 0.37761349906825303,
      "grad_norm": 0.24067999422550201,
      "learning_rate": 0.00027091741490152554,
      "loss": 4.7919,
      "step": 18440
    },
    {
      "epoch": 0.37771588884565765,
      "grad_norm": 0.2539689540863037,
      "learning_rate": 0.0002708807771998226,
      "loss": 4.8109,
      "step": 18445
    },
    {
      "epoch": 0.3778182786230623,
      "grad_norm": 0.2400183230638504,
      "learning_rate": 0.0002708441189155456,
      "loss": 4.807,
      "step": 18450
    },
    {
      "epoch": 0.3779206684004669,
      "grad_norm": 0.26198068261146545,
      "learning_rate": 0.0002708074400549363,
      "loss": 4.7751,
      "step": 18455
    },
    {
      "epoch": 0.3780230581778715,
      "grad_norm": 0.23185764253139496,
      "learning_rate": 0.00027077074062424025,
      "loss": 4.7895,
      "step": 18460
    },
    {
      "epoch": 0.37812544795527614,
      "grad_norm": 0.23769047856330872,
      "learning_rate": 0.00027073402062970624,
      "loss": 4.7822,
      "step": 18465
    },
    {
      "epoch": 0.37822783773268076,
      "grad_norm": 0.24097198247909546,
      "learning_rate": 0.0002706972800775866,
      "loss": 4.8151,
      "step": 18470
    },
    {
      "epoch": 0.3783302275100854,
      "grad_norm": 0.25598886609077454,
      "learning_rate": 0.0002706605189741372,
      "loss": 4.7853,
      "step": 18475
    },
    {
      "epoch": 0.37843261728749,
      "grad_norm": 0.2380874752998352,
      "learning_rate": 0.00027062373732561745,
      "loss": 4.8168,
      "step": 18480
    },
    {
      "epoch": 0.3785350070648946,
      "grad_norm": 0.24340538680553436,
      "learning_rate": 0.0002705869351382903,
      "loss": 4.7703,
      "step": 18485
    },
    {
      "epoch": 0.37863739684229925,
      "grad_norm": 0.2664043605327606,
      "learning_rate": 0.00027055011241842195,
      "loss": 4.884,
      "step": 18490
    },
    {
      "epoch": 0.37873978661970387,
      "grad_norm": 0.2321472465991974,
      "learning_rate": 0.00027051326917228245,
      "loss": 4.7799,
      "step": 18495
    },
    {
      "epoch": 0.3788421763971085,
      "grad_norm": 0.2406431883573532,
      "learning_rate": 0.00027047640540614503,
      "loss": 4.7492,
      "step": 18500
    },
    {
      "epoch": 0.3789445661745131,
      "grad_norm": 0.2702486217021942,
      "learning_rate": 0.0002704395211262866,
      "loss": 4.7795,
      "step": 18505
    },
    {
      "epoch": 0.3790469559519178,
      "grad_norm": 0.2503416836261749,
      "learning_rate": 0.0002704026163389875,
      "loss": 4.7224,
      "step": 18510
    },
    {
      "epoch": 0.3791493457293224,
      "grad_norm": 0.25886625051498413,
      "learning_rate": 0.0002703656910505316,
      "loss": 4.7999,
      "step": 18515
    },
    {
      "epoch": 0.37925173550672703,
      "grad_norm": 0.22973226010799408,
      "learning_rate": 0.00027032874526720615,
      "loss": 4.7718,
      "step": 18520
    },
    {
      "epoch": 0.37935412528413165,
      "grad_norm": 0.27013394236564636,
      "learning_rate": 0.00027029177899530205,
      "loss": 4.81,
      "step": 18525
    },
    {
      "epoch": 0.37945651506153627,
      "grad_norm": 0.24470940232276917,
      "learning_rate": 0.0002702547922411136,
      "loss": 4.8061,
      "step": 18530
    },
    {
      "epoch": 0.3795589048389409,
      "grad_norm": 0.28320327401161194,
      "learning_rate": 0.0002702177850109386,
      "loss": 4.7697,
      "step": 18535
    },
    {
      "epoch": 0.3796612946163455,
      "grad_norm": 0.28084442019462585,
      "learning_rate": 0.00027018075731107823,
      "loss": 4.7264,
      "step": 18540
    },
    {
      "epoch": 0.37976368439375013,
      "grad_norm": 0.24529646337032318,
      "learning_rate": 0.0002701437091478374,
      "loss": 4.8008,
      "step": 18545
    },
    {
      "epoch": 0.37986607417115476,
      "grad_norm": 0.23677638173103333,
      "learning_rate": 0.00027010664052752424,
      "loss": 4.7524,
      "step": 18550
    },
    {
      "epoch": 0.3799684639485594,
      "grad_norm": 0.2633301615715027,
      "learning_rate": 0.00027006955145645066,
      "loss": 4.8039,
      "step": 18555
    },
    {
      "epoch": 0.380070853725964,
      "grad_norm": 0.2448771446943283,
      "learning_rate": 0.0002700324419409317,
      "loss": 4.8341,
      "step": 18560
    },
    {
      "epoch": 0.3801732435033686,
      "grad_norm": 0.30636823177337646,
      "learning_rate": 0.00026999531198728614,
      "loss": 4.703,
      "step": 18565
    },
    {
      "epoch": 0.38027563328077324,
      "grad_norm": 0.24533076584339142,
      "learning_rate": 0.0002699581616018362,
      "loss": 4.7632,
      "step": 18570
    },
    {
      "epoch": 0.38037802305817786,
      "grad_norm": 0.2559785544872284,
      "learning_rate": 0.0002699209907909074,
      "loss": 4.777,
      "step": 18575
    },
    {
      "epoch": 0.3804804128355825,
      "grad_norm": 0.2614833116531372,
      "learning_rate": 0.000269883799560829,
      "loss": 4.7689,
      "step": 18580
    },
    {
      "epoch": 0.3805828026129871,
      "grad_norm": 0.24599692225456238,
      "learning_rate": 0.0002698465879179336,
      "loss": 4.7589,
      "step": 18585
    },
    {
      "epoch": 0.3806851923903917,
      "grad_norm": 0.2603289783000946,
      "learning_rate": 0.00026980935586855726,
      "loss": 4.7759,
      "step": 18590
    },
    {
      "epoch": 0.38078758216779635,
      "grad_norm": 0.2516210079193115,
      "learning_rate": 0.0002697721034190396,
      "loss": 4.7509,
      "step": 18595
    },
    {
      "epoch": 0.38088997194520097,
      "grad_norm": 0.25227412581443787,
      "learning_rate": 0.00026973483057572353,
      "loss": 4.7953,
      "step": 18600
    },
    {
      "epoch": 0.3809923617226056,
      "grad_norm": 0.23142880201339722,
      "learning_rate": 0.0002696975373449557,
      "loss": 4.7931,
      "step": 18605
    },
    {
      "epoch": 0.38109475150001026,
      "grad_norm": 0.25117403268814087,
      "learning_rate": 0.000269660223733086,
      "loss": 4.7722,
      "step": 18610
    },
    {
      "epoch": 0.3811971412774149,
      "grad_norm": 0.26767903566360474,
      "learning_rate": 0.0002696228897464679,
      "loss": 4.7954,
      "step": 18615
    },
    {
      "epoch": 0.3812995310548195,
      "grad_norm": 0.30506137013435364,
      "learning_rate": 0.0002695855353914584,
      "loss": 4.7892,
      "step": 18620
    },
    {
      "epoch": 0.38140192083222413,
      "grad_norm": 0.2507057189941406,
      "learning_rate": 0.00026954816067441777,
      "loss": 4.8157,
      "step": 18625
    },
    {
      "epoch": 0.38150431060962875,
      "grad_norm": 0.24486896395683289,
      "learning_rate": 0.0002695107656017099,
      "loss": 4.7802,
      "step": 18630
    },
    {
      "epoch": 0.38160670038703337,
      "grad_norm": 0.23862124979496002,
      "learning_rate": 0.0002694733501797022,
      "loss": 4.7237,
      "step": 18635
    },
    {
      "epoch": 0.381709090164438,
      "grad_norm": 0.2557026743888855,
      "learning_rate": 0.00026943591441476537,
      "loss": 4.7983,
      "step": 18640
    },
    {
      "epoch": 0.3818114799418426,
      "grad_norm": 0.2563668489456177,
      "learning_rate": 0.00026939845831327363,
      "loss": 4.8076,
      "step": 18645
    },
    {
      "epoch": 0.38191386971924723,
      "grad_norm": 0.2395094484090805,
      "learning_rate": 0.0002693609818816048,
      "loss": 4.8071,
      "step": 18650
    },
    {
      "epoch": 0.38201625949665186,
      "grad_norm": 0.25724026560783386,
      "learning_rate": 0.00026932348512613997,
      "loss": 4.7311,
      "step": 18655
    },
    {
      "epoch": 0.3821186492740565,
      "grad_norm": 0.2464665174484253,
      "learning_rate": 0.0002692859680532638,
      "loss": 4.8108,
      "step": 18660
    },
    {
      "epoch": 0.3822210390514611,
      "grad_norm": 0.24866892397403717,
      "learning_rate": 0.0002692484306693644,
      "loss": 4.7831,
      "step": 18665
    },
    {
      "epoch": 0.3823234288288657,
      "grad_norm": 0.29046374559402466,
      "learning_rate": 0.0002692108729808333,
      "loss": 4.8312,
      "step": 18670
    },
    {
      "epoch": 0.38242581860627034,
      "grad_norm": 0.24792906641960144,
      "learning_rate": 0.00026917329499406557,
      "loss": 4.791,
      "step": 18675
    },
    {
      "epoch": 0.38252820838367496,
      "grad_norm": 0.2569676339626312,
      "learning_rate": 0.0002691356967154596,
      "loss": 4.7365,
      "step": 18680
    },
    {
      "epoch": 0.3826305981610796,
      "grad_norm": 0.2488475888967514,
      "learning_rate": 0.00026909807815141736,
      "loss": 4.7725,
      "step": 18685
    },
    {
      "epoch": 0.3827329879384842,
      "grad_norm": 0.24012289941310883,
      "learning_rate": 0.00026906043930834415,
      "loss": 4.7432,
      "step": 18690
    },
    {
      "epoch": 0.3828353777158888,
      "grad_norm": 0.28020796179771423,
      "learning_rate": 0.00026902278019264893,
      "loss": 4.7817,
      "step": 18695
    },
    {
      "epoch": 0.38293776749329345,
      "grad_norm": 0.24647168815135956,
      "learning_rate": 0.00026898510081074394,
      "loss": 4.7638,
      "step": 18700
    },
    {
      "epoch": 0.38304015727069807,
      "grad_norm": 0.24646998941898346,
      "learning_rate": 0.0002689474011690448,
      "loss": 4.8024,
      "step": 18705
    },
    {
      "epoch": 0.38314254704810274,
      "grad_norm": 0.2355469912290573,
      "learning_rate": 0.0002689096812739708,
      "loss": 4.7861,
      "step": 18710
    },
    {
      "epoch": 0.38324493682550737,
      "grad_norm": 0.2523438334465027,
      "learning_rate": 0.0002688719411319446,
      "loss": 4.7592,
      "step": 18715
    },
    {
      "epoch": 0.383347326602912,
      "grad_norm": 0.24443495273590088,
      "learning_rate": 0.00026883418074939215,
      "loss": 4.7463,
      "step": 18720
    },
    {
      "epoch": 0.3834497163803166,
      "grad_norm": 0.24216555058956146,
      "learning_rate": 0.00026879640013274304,
      "loss": 4.7913,
      "step": 18725
    },
    {
      "epoch": 0.38355210615772123,
      "grad_norm": 0.2408372312784195,
      "learning_rate": 0.0002687585992884303,
      "loss": 4.763,
      "step": 18730
    },
    {
      "epoch": 0.38365449593512585,
      "grad_norm": 0.2512153685092926,
      "learning_rate": 0.0002687207782228903,
      "loss": 4.7654,
      "step": 18735
    },
    {
      "epoch": 0.38375688571253047,
      "grad_norm": 0.2301095724105835,
      "learning_rate": 0.00026868293694256283,
      "loss": 4.7823,
      "step": 18740
    },
    {
      "epoch": 0.3838592754899351,
      "grad_norm": 0.2779386341571808,
      "learning_rate": 0.0002686450754538913,
      "loss": 4.7686,
      "step": 18745
    },
    {
      "epoch": 0.3839616652673397,
      "grad_norm": 0.24197597801685333,
      "learning_rate": 0.00026860719376332236,
      "loss": 4.7133,
      "step": 18750
    },
    {
      "epoch": 0.38406405504474433,
      "grad_norm": 0.40180352330207825,
      "learning_rate": 0.00026856929187730626,
      "loss": 4.7492,
      "step": 18755
    },
    {
      "epoch": 0.38416644482214896,
      "grad_norm": 0.2413758635520935,
      "learning_rate": 0.00026853136980229655,
      "loss": 4.7975,
      "step": 18760
    },
    {
      "epoch": 0.3842688345995536,
      "grad_norm": 0.24081037938594818,
      "learning_rate": 0.00026849342754475033,
      "loss": 4.7742,
      "step": 18765
    },
    {
      "epoch": 0.3843712243769582,
      "grad_norm": 0.27904465794563293,
      "learning_rate": 0.0002684554651111281,
      "loss": 4.7133,
      "step": 18770
    },
    {
      "epoch": 0.3844736141543628,
      "grad_norm": 0.23528353869915009,
      "learning_rate": 0.00026841748250789374,
      "loss": 4.7873,
      "step": 18775
    },
    {
      "epoch": 0.38457600393176744,
      "grad_norm": 0.23735223710536957,
      "learning_rate": 0.00026837947974151465,
      "loss": 4.8222,
      "step": 18780
    },
    {
      "epoch": 0.38467839370917206,
      "grad_norm": 0.2439945489168167,
      "learning_rate": 0.00026834145681846165,
      "loss": 4.7803,
      "step": 18785
    },
    {
      "epoch": 0.3847807834865767,
      "grad_norm": 0.2637277841567993,
      "learning_rate": 0.0002683034137452089,
      "loss": 4.8156,
      "step": 18790
    },
    {
      "epoch": 0.3848831732639813,
      "grad_norm": 0.2374177873134613,
      "learning_rate": 0.0002682653505282341,
      "loss": 4.7726,
      "step": 18795
    },
    {
      "epoch": 0.3849855630413859,
      "grad_norm": 0.24194015562534332,
      "learning_rate": 0.0002682272671740183,
      "loss": 4.8126,
      "step": 18800
    },
    {
      "epoch": 0.38508795281879055,
      "grad_norm": 0.2640838325023651,
      "learning_rate": 0.00026818916368904604,
      "loss": 4.8057,
      "step": 18805
    },
    {
      "epoch": 0.3851903425961952,
      "grad_norm": 0.24625428020954132,
      "learning_rate": 0.00026815104007980536,
      "loss": 4.8369,
      "step": 18810
    },
    {
      "epoch": 0.38529273237359984,
      "grad_norm": 0.24794188141822815,
      "learning_rate": 0.00026811289635278744,
      "loss": 4.816,
      "step": 18815
    },
    {
      "epoch": 0.38539512215100447,
      "grad_norm": 0.2448667287826538,
      "learning_rate": 0.00026807473251448725,
      "loss": 4.7601,
      "step": 18820
    },
    {
      "epoch": 0.3854975119284091,
      "grad_norm": 0.23488222062587738,
      "learning_rate": 0.0002680365485714029,
      "loss": 4.7819,
      "step": 18825
    },
    {
      "epoch": 0.3855999017058137,
      "grad_norm": 0.2530573010444641,
      "learning_rate": 0.00026799834453003615,
      "loss": 4.8268,
      "step": 18830
    },
    {
      "epoch": 0.38570229148321833,
      "grad_norm": 0.22342631220817566,
      "learning_rate": 0.0002679601203968919,
      "loss": 4.7262,
      "step": 18835
    },
    {
      "epoch": 0.38580468126062295,
      "grad_norm": 0.2440759539604187,
      "learning_rate": 0.00026792187617847874,
      "loss": 4.7564,
      "step": 18840
    },
    {
      "epoch": 0.38590707103802757,
      "grad_norm": 0.26362764835357666,
      "learning_rate": 0.0002678836118813086,
      "loss": 4.7602,
      "step": 18845
    },
    {
      "epoch": 0.3860094608154322,
      "grad_norm": 0.24483011662960052,
      "learning_rate": 0.0002678453275118968,
      "loss": 4.8533,
      "step": 18850
    },
    {
      "epoch": 0.3861118505928368,
      "grad_norm": 0.24865449965000153,
      "learning_rate": 0.000267807023076762,
      "loss": 4.7969,
      "step": 18855
    },
    {
      "epoch": 0.38621424037024144,
      "grad_norm": 0.23510952293872833,
      "learning_rate": 0.00026776869858242643,
      "loss": 4.7496,
      "step": 18860
    },
    {
      "epoch": 0.38631663014764606,
      "grad_norm": 0.24842463433742523,
      "learning_rate": 0.0002677303540354157,
      "loss": 4.7977,
      "step": 18865
    },
    {
      "epoch": 0.3864190199250507,
      "grad_norm": 0.2571222186088562,
      "learning_rate": 0.0002676919894422587,
      "loss": 4.7884,
      "step": 18870
    },
    {
      "epoch": 0.3865214097024553,
      "grad_norm": 0.2501884698867798,
      "learning_rate": 0.00026765360480948794,
      "loss": 4.8327,
      "step": 18875
    },
    {
      "epoch": 0.3866237994798599,
      "grad_norm": 0.462527871131897,
      "learning_rate": 0.00026761520014363915,
      "loss": 4.7341,
      "step": 18880
    },
    {
      "epoch": 0.38672618925726454,
      "grad_norm": 0.23223496973514557,
      "learning_rate": 0.00026757677545125166,
      "loss": 4.8168,
      "step": 18885
    },
    {
      "epoch": 0.38682857903466916,
      "grad_norm": 0.2507539987564087,
      "learning_rate": 0.000267538330738868,
      "loss": 4.777,
      "step": 18890
    },
    {
      "epoch": 0.3869309688120738,
      "grad_norm": 0.2501585781574249,
      "learning_rate": 0.0002674998660130342,
      "loss": 4.7874,
      "step": 18895
    },
    {
      "epoch": 0.3870333585894784,
      "grad_norm": 0.24588614702224731,
      "learning_rate": 0.00026746138128029985,
      "loss": 4.7358,
      "step": 18900
    },
    {
      "epoch": 0.387135748366883,
      "grad_norm": 0.2335381805896759,
      "learning_rate": 0.00026742287654721764,
      "loss": 4.7622,
      "step": 18905
    },
    {
      "epoch": 0.38723813814428765,
      "grad_norm": 1.543792724609375,
      "learning_rate": 0.00026738435182034403,
      "loss": 4.7863,
      "step": 18910
    },
    {
      "epoch": 0.3873405279216923,
      "grad_norm": 0.23785871267318726,
      "learning_rate": 0.0002673458071062385,
      "loss": 4.7614,
      "step": 18915
    },
    {
      "epoch": 0.38744291769909694,
      "grad_norm": 0.33225584030151367,
      "learning_rate": 0.0002673072424114643,
      "loss": 4.6968,
      "step": 18920
    },
    {
      "epoch": 0.38754530747650157,
      "grad_norm": 0.2608654201030731,
      "learning_rate": 0.00026726865774258775,
      "loss": 4.7292,
      "step": 18925
    },
    {
      "epoch": 0.3876476972539062,
      "grad_norm": 0.25150445103645325,
      "learning_rate": 0.0002672300531061788,
      "loss": 4.7513,
      "step": 18930
    },
    {
      "epoch": 0.3877500870313108,
      "grad_norm": 0.25601643323898315,
      "learning_rate": 0.00026719142850881075,
      "loss": 4.7742,
      "step": 18935
    },
    {
      "epoch": 0.38785247680871543,
      "grad_norm": 0.26587796211242676,
      "learning_rate": 0.0002671527839570603,
      "loss": 4.7692,
      "step": 18940
    },
    {
      "epoch": 0.38795486658612005,
      "grad_norm": 0.27763983607292175,
      "learning_rate": 0.0002671141194575074,
      "loss": 4.8354,
      "step": 18945
    },
    {
      "epoch": 0.38805725636352467,
      "grad_norm": 0.24708032608032227,
      "learning_rate": 0.00026707543501673555,
      "loss": 4.841,
      "step": 18950
    },
    {
      "epoch": 0.3881596461409293,
      "grad_norm": 0.25299885869026184,
      "learning_rate": 0.00026703673064133173,
      "loss": 4.7919,
      "step": 18955
    },
    {
      "epoch": 0.3882620359183339,
      "grad_norm": 0.24397580325603485,
      "learning_rate": 0.00026699800633788606,
      "loss": 4.7481,
      "step": 18960
    },
    {
      "epoch": 0.38836442569573854,
      "grad_norm": 0.23750890791416168,
      "learning_rate": 0.0002669592621129923,
      "loss": 4.7634,
      "step": 18965
    },
    {
      "epoch": 0.38846681547314316,
      "grad_norm": 0.2606174647808075,
      "learning_rate": 0.00026692049797324746,
      "loss": 4.7251,
      "step": 18970
    },
    {
      "epoch": 0.3885692052505478,
      "grad_norm": 0.26604393124580383,
      "learning_rate": 0.00026688171392525193,
      "loss": 4.7558,
      "step": 18975
    },
    {
      "epoch": 0.3886715950279524,
      "grad_norm": 0.23531261086463928,
      "learning_rate": 0.00026684290997560956,
      "loss": 4.8159,
      "step": 18980
    },
    {
      "epoch": 0.388773984805357,
      "grad_norm": 0.22895264625549316,
      "learning_rate": 0.00026680408613092763,
      "loss": 4.8155,
      "step": 18985
    },
    {
      "epoch": 0.38887637458276164,
      "grad_norm": 0.3297937512397766,
      "learning_rate": 0.00026676524239781663,
      "loss": 4.7798,
      "step": 18990
    },
    {
      "epoch": 0.38897876436016626,
      "grad_norm": 0.3952682912349701,
      "learning_rate": 0.00026672637878289064,
      "loss": 4.7737,
      "step": 18995
    },
    {
      "epoch": 0.3890811541375709,
      "grad_norm": 0.25012969970703125,
      "learning_rate": 0.000266687495292767,
      "loss": 4.7927,
      "step": 19000
    },
    {
      "epoch": 0.3890811541375709,
      "eval_loss": 4.8032073974609375,
      "eval_runtime": 22.9094,
      "eval_samples_per_second": 25.404,
      "eval_steps_per_second": 3.186,
      "step": 19000
    },
    {
      "epoch": 0.3891835439149755,
      "grad_norm": 0.2581552565097809,
      "learning_rate": 0.00026664859193406646,
      "loss": 4.7909,
      "step": 19005
    },
    {
      "epoch": 0.3892859336923801,
      "grad_norm": 0.26446202397346497,
      "learning_rate": 0.00026660966871341316,
      "loss": 4.7531,
      "step": 19010
    },
    {
      "epoch": 0.3893883234697848,
      "grad_norm": 0.24724797904491425,
      "learning_rate": 0.0002665707256374346,
      "loss": 4.799,
      "step": 19015
    },
    {
      "epoch": 0.3894907132471894,
      "grad_norm": 0.27956023812294006,
      "learning_rate": 0.00026653176271276174,
      "loss": 4.7934,
      "step": 19020
    },
    {
      "epoch": 0.38959310302459405,
      "grad_norm": 0.2637579143047333,
      "learning_rate": 0.00026649277994602883,
      "loss": 4.7787,
      "step": 19025
    },
    {
      "epoch": 0.38969549280199867,
      "grad_norm": 0.24771881103515625,
      "learning_rate": 0.00026645377734387356,
      "loss": 4.7993,
      "step": 19030
    },
    {
      "epoch": 0.3897978825794033,
      "grad_norm": 0.2541210949420929,
      "learning_rate": 0.00026641475491293693,
      "loss": 4.7925,
      "step": 19035
    },
    {
      "epoch": 0.3899002723568079,
      "grad_norm": 0.2655090093612671,
      "learning_rate": 0.00026637571265986337,
      "loss": 4.7878,
      "step": 19040
    },
    {
      "epoch": 0.39000266213421253,
      "grad_norm": 0.25652801990509033,
      "learning_rate": 0.0002663366505913007,
      "loss": 4.7968,
      "step": 19045
    },
    {
      "epoch": 0.39010505191161715,
      "grad_norm": 0.23862235248088837,
      "learning_rate": 0.00026629756871390006,
      "loss": 4.7485,
      "step": 19050
    },
    {
      "epoch": 0.3902074416890218,
      "grad_norm": 0.25127172470092773,
      "learning_rate": 0.0002662584670343159,
      "loss": 4.7509,
      "step": 19055
    },
    {
      "epoch": 0.3903098314664264,
      "grad_norm": 0.2848736047744751,
      "learning_rate": 0.0002662193455592063,
      "loss": 4.749,
      "step": 19060
    },
    {
      "epoch": 0.390412221243831,
      "grad_norm": 0.4911057651042938,
      "learning_rate": 0.00026618020429523235,
      "loss": 4.7249,
      "step": 19065
    },
    {
      "epoch": 0.39051461102123564,
      "grad_norm": 0.2578085958957672,
      "learning_rate": 0.0002661410432490589,
      "loss": 4.7815,
      "step": 19070
    },
    {
      "epoch": 0.39061700079864026,
      "grad_norm": 0.23536942899227142,
      "learning_rate": 0.00026610186242735385,
      "loss": 4.841,
      "step": 19075
    },
    {
      "epoch": 0.3907193905760449,
      "grad_norm": 0.2647155821323395,
      "learning_rate": 0.00026606266183678857,
      "loss": 4.7802,
      "step": 19080
    },
    {
      "epoch": 0.3908217803534495,
      "grad_norm": 0.2519124150276184,
      "learning_rate": 0.00026602344148403786,
      "loss": 4.7654,
      "step": 19085
    },
    {
      "epoch": 0.3909241701308541,
      "grad_norm": 0.24083812534809113,
      "learning_rate": 0.0002659842013757798,
      "loss": 4.7989,
      "step": 19090
    },
    {
      "epoch": 0.39102655990825874,
      "grad_norm": 0.26382654905319214,
      "learning_rate": 0.0002659449415186959,
      "loss": 4.78,
      "step": 19095
    },
    {
      "epoch": 0.39112894968566336,
      "grad_norm": 0.24345767498016357,
      "learning_rate": 0.000265905661919471,
      "loss": 4.8257,
      "step": 19100
    },
    {
      "epoch": 0.391231339463068,
      "grad_norm": 0.24560435116291046,
      "learning_rate": 0.0002658663625847933,
      "loss": 4.7817,
      "step": 19105
    },
    {
      "epoch": 0.3913337292404726,
      "grad_norm": 0.2307986617088318,
      "learning_rate": 0.0002658270435213544,
      "loss": 4.815,
      "step": 19110
    },
    {
      "epoch": 0.3914361190178773,
      "grad_norm": 0.24508611857891083,
      "learning_rate": 0.0002657877047358491,
      "loss": 4.7599,
      "step": 19115
    },
    {
      "epoch": 0.3915385087952819,
      "grad_norm": 0.25299644470214844,
      "learning_rate": 0.0002657483462349758,
      "loss": 4.8406,
      "step": 19120
    },
    {
      "epoch": 0.3916408985726865,
      "grad_norm": 0.2539313733577728,
      "learning_rate": 0.0002657089680254361,
      "loss": 4.7867,
      "step": 19125
    },
    {
      "epoch": 0.39174328835009115,
      "grad_norm": 0.23021337389945984,
      "learning_rate": 0.000265669570113935,
      "loss": 4.791,
      "step": 19130
    },
    {
      "epoch": 0.39184567812749577,
      "grad_norm": 0.25902292132377625,
      "learning_rate": 0.00026563015250718086,
      "loss": 4.7243,
      "step": 19135
    },
    {
      "epoch": 0.3919480679049004,
      "grad_norm": 0.25899389386177063,
      "learning_rate": 0.00026559071521188533,
      "loss": 4.8535,
      "step": 19140
    },
    {
      "epoch": 0.392050457682305,
      "grad_norm": 0.23541128635406494,
      "learning_rate": 0.00026555125823476353,
      "loss": 4.7362,
      "step": 19145
    },
    {
      "epoch": 0.39215284745970963,
      "grad_norm": 0.24286217987537384,
      "learning_rate": 0.0002655117815825338,
      "loss": 4.8065,
      "step": 19150
    },
    {
      "epoch": 0.39225523723711425,
      "grad_norm": 0.263705849647522,
      "learning_rate": 0.00026547228526191803,
      "loss": 4.749,
      "step": 19155
    },
    {
      "epoch": 0.3923576270145189,
      "grad_norm": 0.25759607553482056,
      "learning_rate": 0.00026543276927964114,
      "loss": 4.8114,
      "step": 19160
    },
    {
      "epoch": 0.3924600167919235,
      "grad_norm": 0.3094196021556854,
      "learning_rate": 0.0002653932336424317,
      "loss": 4.7475,
      "step": 19165
    },
    {
      "epoch": 0.3925624065693281,
      "grad_norm": 0.25293731689453125,
      "learning_rate": 0.00026535367835702145,
      "loss": 4.7546,
      "step": 19170
    },
    {
      "epoch": 0.39266479634673274,
      "grad_norm": 0.30960583686828613,
      "learning_rate": 0.00026531410343014563,
      "loss": 4.7823,
      "step": 19175
    },
    {
      "epoch": 0.39276718612413736,
      "grad_norm": 0.24387523531913757,
      "learning_rate": 0.0002652745088685426,
      "loss": 4.819,
      "step": 19180
    },
    {
      "epoch": 0.392869575901542,
      "grad_norm": 0.24154560267925262,
      "learning_rate": 0.0002652348946789543,
      "loss": 4.7687,
      "step": 19185
    },
    {
      "epoch": 0.3929719656789466,
      "grad_norm": 0.2489997297525406,
      "learning_rate": 0.0002651952608681259,
      "loss": 4.7429,
      "step": 19190
    },
    {
      "epoch": 0.3930743554563512,
      "grad_norm": 0.27558526396751404,
      "learning_rate": 0.00026515560744280585,
      "loss": 4.7931,
      "step": 19195
    },
    {
      "epoch": 0.39317674523375584,
      "grad_norm": 0.2533782422542572,
      "learning_rate": 0.000265115934409746,
      "loss": 4.7954,
      "step": 19200
    },
    {
      "epoch": 0.39327913501116046,
      "grad_norm": 0.24310936033725739,
      "learning_rate": 0.0002650762417757016,
      "loss": 4.762,
      "step": 19205
    },
    {
      "epoch": 0.3933815247885651,
      "grad_norm": 0.328246533870697,
      "learning_rate": 0.0002650365295474312,
      "loss": 4.7286,
      "step": 19210
    },
    {
      "epoch": 0.39348391456596976,
      "grad_norm": 0.2466014325618744,
      "learning_rate": 0.0002649967977316966,
      "loss": 4.7366,
      "step": 19215
    },
    {
      "epoch": 0.3935863043433744,
      "grad_norm": 0.2853134572505951,
      "learning_rate": 0.00026495704633526306,
      "loss": 4.7904,
      "step": 19220
    },
    {
      "epoch": 0.393688694120779,
      "grad_norm": 0.26766377687454224,
      "learning_rate": 0.00026491727536489905,
      "loss": 4.7574,
      "step": 19225
    },
    {
      "epoch": 0.3937910838981836,
      "grad_norm": 0.255753755569458,
      "learning_rate": 0.00026487748482737655,
      "loss": 4.7345,
      "step": 19230
    },
    {
      "epoch": 0.39389347367558825,
      "grad_norm": 0.24622055888175964,
      "learning_rate": 0.0002648376747294706,
      "loss": 4.8189,
      "step": 19235
    },
    {
      "epoch": 0.39399586345299287,
      "grad_norm": 0.23744042217731476,
      "learning_rate": 0.0002647978450779599,
      "loss": 4.7383,
      "step": 19240
    },
    {
      "epoch": 0.3940982532303975,
      "grad_norm": 0.23252515494823456,
      "learning_rate": 0.00026475799587962623,
      "loss": 4.7573,
      "step": 19245
    },
    {
      "epoch": 0.3942006430078021,
      "grad_norm": 0.2647829055786133,
      "learning_rate": 0.00026471812714125477,
      "loss": 4.8206,
      "step": 19250
    },
    {
      "epoch": 0.39430303278520673,
      "grad_norm": 0.2558216452598572,
      "learning_rate": 0.00026467823886963405,
      "loss": 4.7839,
      "step": 19255
    },
    {
      "epoch": 0.39440542256261135,
      "grad_norm": 0.46287602186203003,
      "learning_rate": 0.00026463833107155595,
      "loss": 4.7605,
      "step": 19260
    },
    {
      "epoch": 0.394507812340016,
      "grad_norm": 0.2570454776287079,
      "learning_rate": 0.0002645984037538156,
      "loss": 4.749,
      "step": 19265
    },
    {
      "epoch": 0.3946102021174206,
      "grad_norm": 0.2836085855960846,
      "learning_rate": 0.0002645584569232115,
      "loss": 4.8117,
      "step": 19270
    },
    {
      "epoch": 0.3947125918948252,
      "grad_norm": 0.2433154135942459,
      "learning_rate": 0.00026451849058654544,
      "loss": 4.7985,
      "step": 19275
    },
    {
      "epoch": 0.39481498167222984,
      "grad_norm": 0.2622375786304474,
      "learning_rate": 0.0002644785047506226,
      "loss": 4.8153,
      "step": 19280
    },
    {
      "epoch": 0.39491737144963446,
      "grad_norm": 0.2493102252483368,
      "learning_rate": 0.0002644384994222514,
      "loss": 4.7749,
      "step": 19285
    },
    {
      "epoch": 0.3950197612270391,
      "grad_norm": 0.24479986727237701,
      "learning_rate": 0.00026439847460824365,
      "loss": 4.7593,
      "step": 19290
    },
    {
      "epoch": 0.3951221510044437,
      "grad_norm": 0.2591876685619354,
      "learning_rate": 0.00026435843031541445,
      "loss": 4.7828,
      "step": 19295
    },
    {
      "epoch": 0.3952245407818483,
      "grad_norm": 0.2384757548570633,
      "learning_rate": 0.00026431836655058214,
      "loss": 4.7337,
      "step": 19300
    },
    {
      "epoch": 0.39532693055925294,
      "grad_norm": 0.23687398433685303,
      "learning_rate": 0.0002642782833205685,
      "loss": 4.7749,
      "step": 19305
    },
    {
      "epoch": 0.39542932033665756,
      "grad_norm": 0.24877889454364777,
      "learning_rate": 0.00026423818063219857,
      "loss": 4.7949,
      "step": 19310
    },
    {
      "epoch": 0.3955317101140622,
      "grad_norm": 0.24068373441696167,
      "learning_rate": 0.0002641980584923007,
      "loss": 4.7942,
      "step": 19315
    },
    {
      "epoch": 0.39563409989146686,
      "grad_norm": 0.232746422290802,
      "learning_rate": 0.0002641579169077065,
      "loss": 4.7772,
      "step": 19320
    },
    {
      "epoch": 0.3957364896688715,
      "grad_norm": 0.24453406035900116,
      "learning_rate": 0.000264117755885251,
      "loss": 4.7589,
      "step": 19325
    },
    {
      "epoch": 0.3958388794462761,
      "grad_norm": 0.2468212991952896,
      "learning_rate": 0.0002640775754317725,
      "loss": 4.7092,
      "step": 19330
    },
    {
      "epoch": 0.3959412692236807,
      "grad_norm": 0.27851566672325134,
      "learning_rate": 0.00026403737555411255,
      "loss": 4.7822,
      "step": 19335
    },
    {
      "epoch": 0.39604365900108535,
      "grad_norm": 0.25844505429267883,
      "learning_rate": 0.0002639971562591161,
      "loss": 4.7588,
      "step": 19340
    },
    {
      "epoch": 0.39614604877848997,
      "grad_norm": 0.23280110955238342,
      "learning_rate": 0.00026395691755363125,
      "loss": 4.7725,
      "step": 19345
    },
    {
      "epoch": 0.3962484385558946,
      "grad_norm": 0.2477782517671585,
      "learning_rate": 0.0002639166594445097,
      "loss": 4.7506,
      "step": 19350
    },
    {
      "epoch": 0.3963508283332992,
      "grad_norm": 0.23398302495479584,
      "learning_rate": 0.0002638763819386061,
      "loss": 4.7248,
      "step": 19355
    },
    {
      "epoch": 0.39645321811070383,
      "grad_norm": 0.24845682084560394,
      "learning_rate": 0.0002638360850427786,
      "loss": 4.7614,
      "step": 19360
    },
    {
      "epoch": 0.39655560788810845,
      "grad_norm": 0.2504514753818512,
      "learning_rate": 0.0002637957687638887,
      "loss": 4.744,
      "step": 19365
    },
    {
      "epoch": 0.3966579976655131,
      "grad_norm": 0.24058574438095093,
      "learning_rate": 0.00026375543310880095,
      "loss": 4.8003,
      "step": 19370
    },
    {
      "epoch": 0.3967603874429177,
      "grad_norm": 0.25265824794769287,
      "learning_rate": 0.0002637150780843836,
      "loss": 4.8176,
      "step": 19375
    },
    {
      "epoch": 0.3968627772203223,
      "grad_norm": 0.2544262707233429,
      "learning_rate": 0.0002636747036975078,
      "loss": 4.7584,
      "step": 19380
    },
    {
      "epoch": 0.39696516699772694,
      "grad_norm": 0.2531717121601105,
      "learning_rate": 0.00026363430995504823,
      "loss": 4.7528,
      "step": 19385
    },
    {
      "epoch": 0.39706755677513156,
      "grad_norm": 0.2622533440589905,
      "learning_rate": 0.0002635938968638828,
      "loss": 4.7799,
      "step": 19390
    },
    {
      "epoch": 0.3971699465525362,
      "grad_norm": 0.28076672554016113,
      "learning_rate": 0.0002635534644308927,
      "loss": 4.7745,
      "step": 19395
    },
    {
      "epoch": 0.3972723363299408,
      "grad_norm": 0.2232881784439087,
      "learning_rate": 0.00026351301266296247,
      "loss": 4.7611,
      "step": 19400
    },
    {
      "epoch": 0.3973747261073454,
      "grad_norm": 0.23840515315532684,
      "learning_rate": 0.0002634725415669798,
      "loss": 4.7596,
      "step": 19405
    },
    {
      "epoch": 0.39747711588475004,
      "grad_norm": 0.24760614335536957,
      "learning_rate": 0.0002634320511498359,
      "loss": 4.764,
      "step": 19410
    },
    {
      "epoch": 0.39757950566215466,
      "grad_norm": 0.2354411631822586,
      "learning_rate": 0.00026339154141842504,
      "loss": 4.758,
      "step": 19415
    },
    {
      "epoch": 0.39768189543955934,
      "grad_norm": 0.2832263708114624,
      "learning_rate": 0.00026335101237964494,
      "loss": 4.7504,
      "step": 19420
    },
    {
      "epoch": 0.39778428521696396,
      "grad_norm": 0.24699002504348755,
      "learning_rate": 0.0002633104640403965,
      "loss": 4.739,
      "step": 19425
    },
    {
      "epoch": 0.3978866749943686,
      "grad_norm": 0.23946571350097656,
      "learning_rate": 0.00026326989640758396,
      "loss": 4.7457,
      "step": 19430
    },
    {
      "epoch": 0.3979890647717732,
      "grad_norm": 0.2437146008014679,
      "learning_rate": 0.0002632293094881149,
      "loss": 4.7888,
      "step": 19435
    },
    {
      "epoch": 0.3980914545491778,
      "grad_norm": 0.2592165172100067,
      "learning_rate": 0.0002631887032889001,
      "loss": 4.7477,
      "step": 19440
    },
    {
      "epoch": 0.39819384432658245,
      "grad_norm": 0.24634739756584167,
      "learning_rate": 0.0002631480778168536,
      "loss": 4.8172,
      "step": 19445
    },
    {
      "epoch": 0.39829623410398707,
      "grad_norm": 0.2822176218032837,
      "learning_rate": 0.0002631074330788928,
      "loss": 4.7945,
      "step": 19450
    },
    {
      "epoch": 0.3983986238813917,
      "grad_norm": 0.26324695348739624,
      "learning_rate": 0.00026306676908193833,
      "loss": 4.8088,
      "step": 19455
    },
    {
      "epoch": 0.3985010136587963,
      "grad_norm": 0.249222069978714,
      "learning_rate": 0.00026302608583291415,
      "loss": 4.7186,
      "step": 19460
    },
    {
      "epoch": 0.39860340343620093,
      "grad_norm": 0.2556007504463196,
      "learning_rate": 0.00026298538333874744,
      "loss": 4.7699,
      "step": 19465
    },
    {
      "epoch": 0.39870579321360555,
      "grad_norm": 0.24117238819599152,
      "learning_rate": 0.00026294466160636874,
      "loss": 4.7778,
      "step": 19470
    },
    {
      "epoch": 0.3988081829910102,
      "grad_norm": 0.2642318606376648,
      "learning_rate": 0.0002629039206427117,
      "loss": 4.7332,
      "step": 19475
    },
    {
      "epoch": 0.3989105727684148,
      "grad_norm": 0.2724672555923462,
      "learning_rate": 0.0002628631604547134,
      "loss": 4.7972,
      "step": 19480
    },
    {
      "epoch": 0.3990129625458194,
      "grad_norm": 0.25742900371551514,
      "learning_rate": 0.0002628223810493142,
      "loss": 4.7281,
      "step": 19485
    },
    {
      "epoch": 0.39911535232322404,
      "grad_norm": 0.24829146265983582,
      "learning_rate": 0.0002627815824334576,
      "loss": 4.7601,
      "step": 19490
    },
    {
      "epoch": 0.39921774210062866,
      "grad_norm": 0.24778318405151367,
      "learning_rate": 0.00026274076461409054,
      "loss": 4.7738,
      "step": 19495
    },
    {
      "epoch": 0.3993201318780333,
      "grad_norm": 0.2703503966331482,
      "learning_rate": 0.000262699927598163,
      "loss": 4.7489,
      "step": 19500
    },
    {
      "epoch": 0.3994225216554379,
      "grad_norm": 0.24909624457359314,
      "learning_rate": 0.00026265907139262856,
      "loss": 4.7699,
      "step": 19505
    },
    {
      "epoch": 0.3995249114328425,
      "grad_norm": 0.27507728338241577,
      "learning_rate": 0.0002626181960044437,
      "loss": 4.8126,
      "step": 19510
    },
    {
      "epoch": 0.39962730121024714,
      "grad_norm": 0.26537808775901794,
      "learning_rate": 0.00026257730144056837,
      "loss": 4.7417,
      "step": 19515
    },
    {
      "epoch": 0.3997296909876518,
      "grad_norm": 0.27283066511154175,
      "learning_rate": 0.0002625363877079659,
      "loss": 4.7776,
      "step": 19520
    },
    {
      "epoch": 0.39983208076505644,
      "grad_norm": 0.2485787272453308,
      "learning_rate": 0.00026249545481360254,
      "loss": 4.7465,
      "step": 19525
    },
    {
      "epoch": 0.39993447054246106,
      "grad_norm": 0.24497090280056,
      "learning_rate": 0.00026245450276444815,
      "loss": 4.7873,
      "step": 19530
    },
    {
      "epoch": 0.4000368603198657,
      "grad_norm": 0.2390383630990982,
      "learning_rate": 0.00026241353156747566,
      "loss": 4.7217,
      "step": 19535
    },
    {
      "epoch": 0.4001392500972703,
      "grad_norm": 0.2366401106119156,
      "learning_rate": 0.00026237254122966127,
      "loss": 4.7863,
      "step": 19540
    },
    {
      "epoch": 0.4002416398746749,
      "grad_norm": 0.24308615922927856,
      "learning_rate": 0.00026233153175798447,
      "loss": 4.7597,
      "step": 19545
    },
    {
      "epoch": 0.40034402965207955,
      "grad_norm": 0.2379208356142044,
      "learning_rate": 0.0002622905031594281,
      "loss": 4.8152,
      "step": 19550
    },
    {
      "epoch": 0.40044641942948417,
      "grad_norm": 0.27165481448173523,
      "learning_rate": 0.0002622494554409781,
      "loss": 4.7024,
      "step": 19555
    },
    {
      "epoch": 0.4005488092068888,
      "grad_norm": 0.27947288751602173,
      "learning_rate": 0.00026220838860962375,
      "loss": 4.7907,
      "step": 19560
    },
    {
      "epoch": 0.4006511989842934,
      "grad_norm": 0.23369461297988892,
      "learning_rate": 0.00026216730267235755,
      "loss": 4.7771,
      "step": 19565
    },
    {
      "epoch": 0.40075358876169803,
      "grad_norm": 0.23187683522701263,
      "learning_rate": 0.0002621261976361753,
      "loss": 4.7554,
      "step": 19570
    },
    {
      "epoch": 0.40085597853910265,
      "grad_norm": 0.23148415982723236,
      "learning_rate": 0.000262085073508076,
      "loss": 4.769,
      "step": 19575
    },
    {
      "epoch": 0.4009583683165073,
      "grad_norm": 0.2489568144083023,
      "learning_rate": 0.0002620439302950619,
      "loss": 4.8002,
      "step": 19580
    },
    {
      "epoch": 0.4010607580939119,
      "grad_norm": 0.25447413325309753,
      "learning_rate": 0.0002620027680041386,
      "loss": 4.7785,
      "step": 19585
    },
    {
      "epoch": 0.4011631478713165,
      "grad_norm": 0.2597976326942444,
      "learning_rate": 0.0002619615866423148,
      "loss": 4.7758,
      "step": 19590
    },
    {
      "epoch": 0.40126553764872114,
      "grad_norm": 0.2571601867675781,
      "learning_rate": 0.00026192038621660266,
      "loss": 4.7568,
      "step": 19595
    },
    {
      "epoch": 0.40136792742612576,
      "grad_norm": 0.2353305220603943,
      "learning_rate": 0.00026187916673401724,
      "loss": 4.7823,
      "step": 19600
    },
    {
      "epoch": 0.4014703172035304,
      "grad_norm": 0.2579090893268585,
      "learning_rate": 0.0002618379282015772,
      "loss": 4.8025,
      "step": 19605
    },
    {
      "epoch": 0.401572706980935,
      "grad_norm": 0.25471198558807373,
      "learning_rate": 0.0002617966706263042,
      "loss": 4.7409,
      "step": 19610
    },
    {
      "epoch": 0.4016750967583396,
      "grad_norm": 0.2436644434928894,
      "learning_rate": 0.0002617553940152233,
      "loss": 4.7414,
      "step": 19615
    },
    {
      "epoch": 0.40177748653574424,
      "grad_norm": 0.2647945284843445,
      "learning_rate": 0.00026171409837536273,
      "loss": 4.7688,
      "step": 19620
    },
    {
      "epoch": 0.4018798763131489,
      "grad_norm": 0.2901569604873657,
      "learning_rate": 0.00026167278371375395,
      "loss": 4.763,
      "step": 19625
    },
    {
      "epoch": 0.40198226609055354,
      "grad_norm": 0.2863835096359253,
      "learning_rate": 0.00026163145003743165,
      "loss": 4.7224,
      "step": 19630
    },
    {
      "epoch": 0.40208465586795816,
      "grad_norm": 0.2397838532924652,
      "learning_rate": 0.0002615900973534339,
      "loss": 4.773,
      "step": 19635
    },
    {
      "epoch": 0.4021870456453628,
      "grad_norm": 0.2545151114463806,
      "learning_rate": 0.0002615487256688017,
      "loss": 4.7594,
      "step": 19640
    },
    {
      "epoch": 0.4022894354227674,
      "grad_norm": 0.2644392251968384,
      "learning_rate": 0.0002615073349905797,
      "loss": 4.8706,
      "step": 19645
    },
    {
      "epoch": 0.402391825200172,
      "grad_norm": 0.2648373246192932,
      "learning_rate": 0.0002614659253258154,
      "loss": 4.7467,
      "step": 19650
    },
    {
      "epoch": 0.40249421497757665,
      "grad_norm": 0.23473845422267914,
      "learning_rate": 0.0002614244966815597,
      "loss": 4.7412,
      "step": 19655
    },
    {
      "epoch": 0.40259660475498127,
      "grad_norm": 0.25363534688949585,
      "learning_rate": 0.0002613830490648669,
      "loss": 4.7506,
      "step": 19660
    },
    {
      "epoch": 0.4026989945323859,
      "grad_norm": 0.28193849325180054,
      "learning_rate": 0.0002613415824827941,
      "loss": 4.7699,
      "step": 19665
    },
    {
      "epoch": 0.4028013843097905,
      "grad_norm": 0.2774384617805481,
      "learning_rate": 0.00026130009694240203,
      "loss": 4.7576,
      "step": 19670
    },
    {
      "epoch": 0.40290377408719513,
      "grad_norm": 0.27691593766212463,
      "learning_rate": 0.00026125859245075447,
      "loss": 4.7844,
      "step": 19675
    },
    {
      "epoch": 0.40300616386459975,
      "grad_norm": 0.23163090646266937,
      "learning_rate": 0.00026121706901491857,
      "loss": 4.733,
      "step": 19680
    },
    {
      "epoch": 0.4031085536420044,
      "grad_norm": 0.23813559114933014,
      "learning_rate": 0.0002611755266419644,
      "loss": 4.7045,
      "step": 19685
    },
    {
      "epoch": 0.403210943419409,
      "grad_norm": 0.2549150586128235,
      "learning_rate": 0.0002611339653389656,
      "loss": 4.7576,
      "step": 19690
    },
    {
      "epoch": 0.4033133331968136,
      "grad_norm": 0.24459238350391388,
      "learning_rate": 0.0002610923851129988,
      "loss": 4.7647,
      "step": 19695
    },
    {
      "epoch": 0.40341572297421824,
      "grad_norm": 0.2550506889820099,
      "learning_rate": 0.000261050785971144,
      "loss": 4.7009,
      "step": 19700
    },
    {
      "epoch": 0.40351811275162286,
      "grad_norm": 0.23717327415943146,
      "learning_rate": 0.00026100916792048425,
      "loss": 4.7817,
      "step": 19705
    },
    {
      "epoch": 0.4036205025290275,
      "grad_norm": 0.27429184317588806,
      "learning_rate": 0.000260967530968106,
      "loss": 4.7967,
      "step": 19710
    },
    {
      "epoch": 0.4037228923064321,
      "grad_norm": 0.2548692226409912,
      "learning_rate": 0.0002609258751210989,
      "loss": 4.7328,
      "step": 19715
    },
    {
      "epoch": 0.4038252820838367,
      "grad_norm": 0.25398963689804077,
      "learning_rate": 0.00026088420038655573,
      "loss": 4.8318,
      "step": 19720
    },
    {
      "epoch": 0.4039276718612414,
      "grad_norm": 0.23279903829097748,
      "learning_rate": 0.0002608425067715724,
      "loss": 4.7604,
      "step": 19725
    },
    {
      "epoch": 0.404030061638646,
      "grad_norm": 0.2645326554775238,
      "learning_rate": 0.00026080079428324833,
      "loss": 4.7832,
      "step": 19730
    },
    {
      "epoch": 0.40413245141605064,
      "grad_norm": 0.26203349232673645,
      "learning_rate": 0.0002607590629286858,
      "loss": 4.7994,
      "step": 19735
    },
    {
      "epoch": 0.40423484119345526,
      "grad_norm": 0.256180077791214,
      "learning_rate": 0.00026071731271499067,
      "loss": 4.781,
      "step": 19740
    },
    {
      "epoch": 0.4043372309708599,
      "grad_norm": 0.257884681224823,
      "learning_rate": 0.00026067554364927163,
      "loss": 4.7952,
      "step": 19745
    },
    {
      "epoch": 0.4044396207482645,
      "grad_norm": 0.2481538951396942,
      "learning_rate": 0.00026063375573864085,
      "loss": 4.8146,
      "step": 19750
    },
    {
      "epoch": 0.4045420105256691,
      "grad_norm": 0.2525789141654968,
      "learning_rate": 0.0002605919489902137,
      "loss": 4.6853,
      "step": 19755
    },
    {
      "epoch": 0.40464440030307375,
      "grad_norm": 0.24754178524017334,
      "learning_rate": 0.0002605501234111086,
      "loss": 4.7383,
      "step": 19760
    },
    {
      "epoch": 0.40474679008047837,
      "grad_norm": 0.2472517043352127,
      "learning_rate": 0.00026050827900844727,
      "loss": 4.7838,
      "step": 19765
    },
    {
      "epoch": 0.404849179857883,
      "grad_norm": 0.24312861263751984,
      "learning_rate": 0.0002604664157893546,
      "loss": 4.8092,
      "step": 19770
    },
    {
      "epoch": 0.4049515696352876,
      "grad_norm": 0.2669880986213684,
      "learning_rate": 0.0002604245337609588,
      "loss": 4.7519,
      "step": 19775
    },
    {
      "epoch": 0.40505395941269223,
      "grad_norm": 0.27648940682411194,
      "learning_rate": 0.0002603826329303911,
      "loss": 4.7815,
      "step": 19780
    },
    {
      "epoch": 0.40515634919009685,
      "grad_norm": 0.251446008682251,
      "learning_rate": 0.00026034071330478613,
      "loss": 4.7885,
      "step": 19785
    },
    {
      "epoch": 0.4052587389675015,
      "grad_norm": 0.2564874589443207,
      "learning_rate": 0.0002602987748912816,
      "loss": 4.7329,
      "step": 19790
    },
    {
      "epoch": 0.4053611287449061,
      "grad_norm": 0.25271502137184143,
      "learning_rate": 0.0002602568176970182,
      "loss": 4.777,
      "step": 19795
    },
    {
      "epoch": 0.4054635185223107,
      "grad_norm": 0.24098694324493408,
      "learning_rate": 0.0002602148417291404,
      "loss": 4.7122,
      "step": 19800
    },
    {
      "epoch": 0.40556590829971534,
      "grad_norm": 0.2511517107486725,
      "learning_rate": 0.0002601728469947953,
      "loss": 4.7755,
      "step": 19805
    },
    {
      "epoch": 0.40566829807711996,
      "grad_norm": 0.23859263956546783,
      "learning_rate": 0.0002601308335011335,
      "loss": 4.8393,
      "step": 19810
    },
    {
      "epoch": 0.4057706878545246,
      "grad_norm": 0.27305012941360474,
      "learning_rate": 0.0002600888012553087,
      "loss": 4.7756,
      "step": 19815
    },
    {
      "epoch": 0.4058730776319292,
      "grad_norm": 0.296753466129303,
      "learning_rate": 0.0002600467502644777,
      "loss": 4.7139,
      "step": 19820
    },
    {
      "epoch": 0.4059754674093339,
      "grad_norm": 0.24156303703784943,
      "learning_rate": 0.0002600046805358007,
      "loss": 4.7338,
      "step": 19825
    },
    {
      "epoch": 0.4060778571867385,
      "grad_norm": 0.2724151611328125,
      "learning_rate": 0.000259962592076441,
      "loss": 4.7976,
      "step": 19830
    },
    {
      "epoch": 0.4061802469641431,
      "grad_norm": 0.2427988201379776,
      "learning_rate": 0.00025992048489356504,
      "loss": 4.7571,
      "step": 19835
    },
    {
      "epoch": 0.40628263674154774,
      "grad_norm": 0.24802915751934052,
      "learning_rate": 0.00025987835899434245,
      "loss": 4.7916,
      "step": 19840
    },
    {
      "epoch": 0.40638502651895236,
      "grad_norm": 0.2772034704685211,
      "learning_rate": 0.0002598362143859461,
      "loss": 4.7342,
      "step": 19845
    },
    {
      "epoch": 0.406487416296357,
      "grad_norm": 0.2626076340675354,
      "learning_rate": 0.0002597940510755521,
      "loss": 4.7238,
      "step": 19850
    },
    {
      "epoch": 0.4065898060737616,
      "grad_norm": 0.27586403489112854,
      "learning_rate": 0.0002597518690703395,
      "loss": 4.7822,
      "step": 19855
    },
    {
      "epoch": 0.4066921958511662,
      "grad_norm": 0.33457404375076294,
      "learning_rate": 0.00025970966837749083,
      "loss": 4.7788,
      "step": 19860
    },
    {
      "epoch": 0.40679458562857085,
      "grad_norm": 0.23057791590690613,
      "learning_rate": 0.0002596674490041916,
      "loss": 4.742,
      "step": 19865
    },
    {
      "epoch": 0.40689697540597547,
      "grad_norm": 0.2495986521244049,
      "learning_rate": 0.0002596252109576307,
      "loss": 4.7231,
      "step": 19870
    },
    {
      "epoch": 0.4069993651833801,
      "grad_norm": 0.23995955288410187,
      "learning_rate": 0.0002595829542449999,
      "loss": 4.7509,
      "step": 19875
    },
    {
      "epoch": 0.4071017549607847,
      "grad_norm": 0.32748815417289734,
      "learning_rate": 0.0002595406788734945,
      "loss": 4.7365,
      "step": 19880
    },
    {
      "epoch": 0.40720414473818933,
      "grad_norm": 0.24457184970378876,
      "learning_rate": 0.00025949838485031263,
      "loss": 4.8283,
      "step": 19885
    },
    {
      "epoch": 0.40730653451559395,
      "grad_norm": 0.23557250201702118,
      "learning_rate": 0.0002594560721826559,
      "loss": 4.7784,
      "step": 19890
    },
    {
      "epoch": 0.4074089242929986,
      "grad_norm": 0.2428516447544098,
      "learning_rate": 0.00025941374087772885,
      "loss": 4.7659,
      "step": 19895
    },
    {
      "epoch": 0.4075113140704032,
      "grad_norm": 0.24987810850143433,
      "learning_rate": 0.0002593713909427394,
      "loss": 4.7948,
      "step": 19900
    },
    {
      "epoch": 0.4076137038478078,
      "grad_norm": 0.2510607838630676,
      "learning_rate": 0.00025932902238489844,
      "loss": 4.7555,
      "step": 19905
    },
    {
      "epoch": 0.40771609362521244,
      "grad_norm": 0.2503218948841095,
      "learning_rate": 0.00025928663521142026,
      "loss": 4.7913,
      "step": 19910
    },
    {
      "epoch": 0.40781848340261706,
      "grad_norm": 0.2557161748409271,
      "learning_rate": 0.0002592442294295221,
      "loss": 4.7973,
      "step": 19915
    },
    {
      "epoch": 0.4079208731800217,
      "grad_norm": 0.25850632786750793,
      "learning_rate": 0.00025920180504642446,
      "loss": 4.7331,
      "step": 19920
    },
    {
      "epoch": 0.40802326295742636,
      "grad_norm": 0.25395697355270386,
      "learning_rate": 0.0002591593620693511,
      "loss": 4.7265,
      "step": 19925
    },
    {
      "epoch": 0.408125652734831,
      "grad_norm": 0.24870486557483673,
      "learning_rate": 0.0002591169005055288,
      "loss": 4.7729,
      "step": 19930
    },
    {
      "epoch": 0.4082280425122356,
      "grad_norm": 0.26790231466293335,
      "learning_rate": 0.00025907442036218753,
      "loss": 4.7463,
      "step": 19935
    },
    {
      "epoch": 0.4083304322896402,
      "grad_norm": 0.24698667228221893,
      "learning_rate": 0.0002590319216465605,
      "loss": 4.7259,
      "step": 19940
    },
    {
      "epoch": 0.40843282206704484,
      "grad_norm": 0.2523781359195709,
      "learning_rate": 0.0002589894043658841,
      "loss": 4.7681,
      "step": 19945
    },
    {
      "epoch": 0.40853521184444946,
      "grad_norm": 0.26594412326812744,
      "learning_rate": 0.00025894686852739765,
      "loss": 4.7841,
      "step": 19950
    },
    {
      "epoch": 0.4086376016218541,
      "grad_norm": 0.25075480341911316,
      "learning_rate": 0.00025890431413834393,
      "loss": 4.7966,
      "step": 19955
    },
    {
      "epoch": 0.4087399913992587,
      "grad_norm": 0.25574803352355957,
      "learning_rate": 0.0002588617412059687,
      "loss": 4.8014,
      "step": 19960
    },
    {
      "epoch": 0.4088423811766633,
      "grad_norm": 0.26465269923210144,
      "learning_rate": 0.00025881914973752097,
      "loss": 4.7336,
      "step": 19965
    },
    {
      "epoch": 0.40894477095406795,
      "grad_norm": 0.24622943997383118,
      "learning_rate": 0.0002587765397402529,
      "loss": 4.8056,
      "step": 19970
    },
    {
      "epoch": 0.40904716073147257,
      "grad_norm": 0.26525789499282837,
      "learning_rate": 0.0002587339112214196,
      "loss": 4.7851,
      "step": 19975
    },
    {
      "epoch": 0.4091495505088772,
      "grad_norm": 0.2609120309352875,
      "learning_rate": 0.00025869126418827956,
      "loss": 4.7708,
      "step": 19980
    },
    {
      "epoch": 0.4092519402862818,
      "grad_norm": 0.24709618091583252,
      "learning_rate": 0.0002586485986480945,
      "loss": 4.7359,
      "step": 19985
    },
    {
      "epoch": 0.40935433006368643,
      "grad_norm": 0.3181145191192627,
      "learning_rate": 0.00025860591460812905,
      "loss": 4.7702,
      "step": 19990
    },
    {
      "epoch": 0.40945671984109105,
      "grad_norm": 0.24727404117584229,
      "learning_rate": 0.00025856321207565105,
      "loss": 4.7353,
      "step": 19995
    },
    {
      "epoch": 0.4095591096184957,
      "grad_norm": 0.22693198919296265,
      "learning_rate": 0.0002585204910579316,
      "loss": 4.7821,
      "step": 20000
    },
    {
      "epoch": 0.4095591096184957,
      "eval_loss": 4.7799859046936035,
      "eval_runtime": 19.6546,
      "eval_samples_per_second": 29.611,
      "eval_steps_per_second": 3.714,
      "step": 20000
    },
    {
      "epoch": 0.4096614993959003,
      "grad_norm": 0.26770642399787903,
      "learning_rate": 0.00025847775156224485,
      "loss": 4.7473,
      "step": 20005
    },
    {
      "epoch": 0.4097638891733049,
      "grad_norm": 0.2308717966079712,
      "learning_rate": 0.00025843499359586817,
      "loss": 4.8047,
      "step": 20010
    },
    {
      "epoch": 0.40986627895070954,
      "grad_norm": 0.2362947165966034,
      "learning_rate": 0.00025839221716608197,
      "loss": 4.7703,
      "step": 20015
    },
    {
      "epoch": 0.40996866872811416,
      "grad_norm": 0.2500995397567749,
      "learning_rate": 0.0002583494222801699,
      "loss": 4.8034,
      "step": 20020
    },
    {
      "epoch": 0.4100710585055188,
      "grad_norm": 0.23485814034938812,
      "learning_rate": 0.0002583066089454187,
      "loss": 4.762,
      "step": 20025
    },
    {
      "epoch": 0.41017344828292346,
      "grad_norm": 0.23973077535629272,
      "learning_rate": 0.00025826377716911823,
      "loss": 4.7755,
      "step": 20030
    },
    {
      "epoch": 0.4102758380603281,
      "grad_norm": 0.2536141574382782,
      "learning_rate": 0.00025822092695856164,
      "loss": 4.7591,
      "step": 20035
    },
    {
      "epoch": 0.4103782278377327,
      "grad_norm": 0.23926681280136108,
      "learning_rate": 0.000258178058321045,
      "loss": 4.754,
      "step": 20040
    },
    {
      "epoch": 0.4104806176151373,
      "grad_norm": 0.24142976105213165,
      "learning_rate": 0.0002581351712638678,
      "loss": 4.7588,
      "step": 20045
    },
    {
      "epoch": 0.41058300739254194,
      "grad_norm": 0.24843670427799225,
      "learning_rate": 0.0002580922657943322,
      "loss": 4.7554,
      "step": 20050
    },
    {
      "epoch": 0.41068539716994656,
      "grad_norm": 0.2508491277694702,
      "learning_rate": 0.00025804934191974403,
      "loss": 4.7792,
      "step": 20055
    },
    {
      "epoch": 0.4107877869473512,
      "grad_norm": 0.24192366003990173,
      "learning_rate": 0.0002580063996474119,
      "loss": 4.7405,
      "step": 20060
    },
    {
      "epoch": 0.4108901767247558,
      "grad_norm": 0.25192683935165405,
      "learning_rate": 0.00025796343898464773,
      "loss": 4.7642,
      "step": 20065
    },
    {
      "epoch": 0.41099256650216043,
      "grad_norm": 0.24169355630874634,
      "learning_rate": 0.0002579204599387664,
      "loss": 4.7833,
      "step": 20070
    },
    {
      "epoch": 0.41109495627956505,
      "grad_norm": 0.27194544672966003,
      "learning_rate": 0.00025787746251708615,
      "loss": 4.721,
      "step": 20075
    },
    {
      "epoch": 0.41119734605696967,
      "grad_norm": 0.28333792090415955,
      "learning_rate": 0.00025783444672692815,
      "loss": 4.7635,
      "step": 20080
    },
    {
      "epoch": 0.4112997358343743,
      "grad_norm": 0.24671564996242523,
      "learning_rate": 0.00025779141257561684,
      "loss": 4.7218,
      "step": 20085
    },
    {
      "epoch": 0.4114021256117789,
      "grad_norm": 0.26813265681266785,
      "learning_rate": 0.0002577483600704796,
      "loss": 4.6989,
      "step": 20090
    },
    {
      "epoch": 0.41150451538918353,
      "grad_norm": 0.26639458537101746,
      "learning_rate": 0.00025770528921884717,
      "loss": 4.7613,
      "step": 20095
    },
    {
      "epoch": 0.41160690516658816,
      "grad_norm": 0.34087416529655457,
      "learning_rate": 0.0002576622000280532,
      "loss": 4.7471,
      "step": 20100
    },
    {
      "epoch": 0.4117092949439928,
      "grad_norm": 0.32042837142944336,
      "learning_rate": 0.0002576190925054346,
      "loss": 4.7166,
      "step": 20105
    },
    {
      "epoch": 0.4118116847213974,
      "grad_norm": 0.26964765787124634,
      "learning_rate": 0.00025757596665833136,
      "loss": 4.7626,
      "step": 20110
    },
    {
      "epoch": 0.411914074498802,
      "grad_norm": 0.24823670089244843,
      "learning_rate": 0.00025753282249408655,
      "loss": 4.7184,
      "step": 20115
    },
    {
      "epoch": 0.41201646427620664,
      "grad_norm": 0.24069556593894958,
      "learning_rate": 0.00025748966002004655,
      "loss": 4.7405,
      "step": 20120
    },
    {
      "epoch": 0.41211885405361126,
      "grad_norm": 0.24267128109931946,
      "learning_rate": 0.00025744647924356057,
      "loss": 4.7585,
      "step": 20125
    },
    {
      "epoch": 0.41222124383101594,
      "grad_norm": 0.27272501587867737,
      "learning_rate": 0.00025740328017198106,
      "loss": 4.7625,
      "step": 20130
    },
    {
      "epoch": 0.41232363360842056,
      "grad_norm": 0.2577752470970154,
      "learning_rate": 0.0002573600628126637,
      "loss": 4.7882,
      "step": 20135
    },
    {
      "epoch": 0.4124260233858252,
      "grad_norm": 0.26629868149757385,
      "learning_rate": 0.00025731682717296706,
      "loss": 4.7195,
      "step": 20140
    },
    {
      "epoch": 0.4125284131632298,
      "grad_norm": 0.23330718278884888,
      "learning_rate": 0.00025727357326025304,
      "loss": 4.7635,
      "step": 20145
    },
    {
      "epoch": 0.4126308029406344,
      "grad_norm": 0.2559824585914612,
      "learning_rate": 0.00025723030108188653,
      "loss": 4.7506,
      "step": 20150
    },
    {
      "epoch": 0.41273319271803904,
      "grad_norm": 0.2495318204164505,
      "learning_rate": 0.0002571870106452356,
      "loss": 4.7664,
      "step": 20155
    },
    {
      "epoch": 0.41283558249544366,
      "grad_norm": 0.2817722260951996,
      "learning_rate": 0.0002571437019576713,
      "loss": 4.7162,
      "step": 20160
    },
    {
      "epoch": 0.4129379722728483,
      "grad_norm": 0.26086512207984924,
      "learning_rate": 0.0002571003750265679,
      "loss": 4.7575,
      "step": 20165
    },
    {
      "epoch": 0.4130403620502529,
      "grad_norm": 0.3186280131340027,
      "learning_rate": 0.0002570570298593028,
      "loss": 4.7355,
      "step": 20170
    },
    {
      "epoch": 0.41314275182765753,
      "grad_norm": 0.33195388317108154,
      "learning_rate": 0.00025701366646325644,
      "loss": 4.7356,
      "step": 20175
    },
    {
      "epoch": 0.41324514160506215,
      "grad_norm": 0.2980879545211792,
      "learning_rate": 0.0002569702848458123,
      "loss": 4.763,
      "step": 20180
    },
    {
      "epoch": 0.41334753138246677,
      "grad_norm": 0.26233193278312683,
      "learning_rate": 0.00025692688501435714,
      "loss": 4.7139,
      "step": 20185
    },
    {
      "epoch": 0.4134499211598714,
      "grad_norm": 0.2996671795845032,
      "learning_rate": 0.0002568834669762807,
      "loss": 4.7229,
      "step": 20190
    },
    {
      "epoch": 0.413552310937276,
      "grad_norm": 0.24259331822395325,
      "learning_rate": 0.00025684003073897575,
      "loss": 4.7319,
      "step": 20195
    },
    {
      "epoch": 0.41365470071468063,
      "grad_norm": 0.2737988233566284,
      "learning_rate": 0.00025679657630983834,
      "loss": 4.6717,
      "step": 20200
    },
    {
      "epoch": 0.41375709049208526,
      "grad_norm": 0.23778383433818817,
      "learning_rate": 0.0002567531036962676,
      "loss": 4.7576,
      "step": 20205
    },
    {
      "epoch": 0.4138594802694899,
      "grad_norm": 0.24559806287288666,
      "learning_rate": 0.00025670961290566555,
      "loss": 4.781,
      "step": 20210
    },
    {
      "epoch": 0.4139618700468945,
      "grad_norm": 0.23543305695056915,
      "learning_rate": 0.00025666610394543745,
      "loss": 4.7329,
      "step": 20215
    },
    {
      "epoch": 0.4140642598242991,
      "grad_norm": 0.23338939249515533,
      "learning_rate": 0.00025662257682299174,
      "loss": 4.7182,
      "step": 20220
    },
    {
      "epoch": 0.41416664960170374,
      "grad_norm": 0.23803578317165375,
      "learning_rate": 0.00025657903154573976,
      "loss": 4.7271,
      "step": 20225
    },
    {
      "epoch": 0.4142690393791084,
      "grad_norm": 0.24348607659339905,
      "learning_rate": 0.0002565354681210961,
      "loss": 4.7413,
      "step": 20230
    },
    {
      "epoch": 0.41437142915651304,
      "grad_norm": 0.24406488239765167,
      "learning_rate": 0.00025649188655647837,
      "loss": 4.7605,
      "step": 20235
    },
    {
      "epoch": 0.41447381893391766,
      "grad_norm": 0.23760120570659637,
      "learning_rate": 0.0002564482868593073,
      "loss": 4.7714,
      "step": 20240
    },
    {
      "epoch": 0.4145762087113223,
      "grad_norm": 0.25055694580078125,
      "learning_rate": 0.00025640466903700663,
      "loss": 4.7181,
      "step": 20245
    },
    {
      "epoch": 0.4146785984887269,
      "grad_norm": 0.2715224325656891,
      "learning_rate": 0.0002563610330970032,
      "loss": 4.7606,
      "step": 20250
    },
    {
      "epoch": 0.4147809882661315,
      "grad_norm": 0.246859610080719,
      "learning_rate": 0.00025631737904672705,
      "loss": 4.7482,
      "step": 20255
    },
    {
      "epoch": 0.41488337804353614,
      "grad_norm": 0.25667813420295715,
      "learning_rate": 0.00025627370689361125,
      "loss": 4.7504,
      "step": 20260
    },
    {
      "epoch": 0.41498576782094077,
      "grad_norm": 0.24133725464344025,
      "learning_rate": 0.00025623001664509186,
      "loss": 4.7251,
      "step": 20265
    },
    {
      "epoch": 0.4150881575983454,
      "grad_norm": 0.25921839475631714,
      "learning_rate": 0.0002561863083086082,
      "loss": 4.7509,
      "step": 20270
    },
    {
      "epoch": 0.41519054737575,
      "grad_norm": 0.23406857252120972,
      "learning_rate": 0.0002561425818916024,
      "loss": 4.7663,
      "step": 20275
    },
    {
      "epoch": 0.41529293715315463,
      "grad_norm": 0.2401791661977768,
      "learning_rate": 0.00025609883740151994,
      "loss": 4.7768,
      "step": 20280
    },
    {
      "epoch": 0.41539532693055925,
      "grad_norm": 0.25224336981773376,
      "learning_rate": 0.0002560550748458093,
      "loss": 4.7404,
      "step": 20285
    },
    {
      "epoch": 0.41549771670796387,
      "grad_norm": 0.2500647008419037,
      "learning_rate": 0.00025601129423192187,
      "loss": 4.7744,
      "step": 20290
    },
    {
      "epoch": 0.4156001064853685,
      "grad_norm": 0.24695204198360443,
      "learning_rate": 0.00025596749556731235,
      "loss": 4.7639,
      "step": 20295
    },
    {
      "epoch": 0.4157024962627731,
      "grad_norm": 0.2416968047618866,
      "learning_rate": 0.00025592367885943847,
      "loss": 4.7445,
      "step": 20300
    },
    {
      "epoch": 0.41580488604017773,
      "grad_norm": 0.2485630363225937,
      "learning_rate": 0.0002558798441157608,
      "loss": 4.7599,
      "step": 20305
    },
    {
      "epoch": 0.41590727581758236,
      "grad_norm": 0.2630630135536194,
      "learning_rate": 0.0002558359913437433,
      "loss": 4.7542,
      "step": 20310
    },
    {
      "epoch": 0.416009665594987,
      "grad_norm": 0.2438594102859497,
      "learning_rate": 0.00025579212055085285,
      "loss": 4.7407,
      "step": 20315
    },
    {
      "epoch": 0.4161120553723916,
      "grad_norm": 0.2848489284515381,
      "learning_rate": 0.00025574823174455923,
      "loss": 4.766,
      "step": 20320
    },
    {
      "epoch": 0.4162144451497962,
      "grad_norm": 0.24623343348503113,
      "learning_rate": 0.00025570432493233575,
      "loss": 4.7318,
      "step": 20325
    },
    {
      "epoch": 0.41631683492720084,
      "grad_norm": 0.24903014302253723,
      "learning_rate": 0.00025566040012165825,
      "loss": 4.7153,
      "step": 20330
    },
    {
      "epoch": 0.4164192247046055,
      "grad_norm": 0.24170295894145966,
      "learning_rate": 0.00025561645732000604,
      "loss": 4.7581,
      "step": 20335
    },
    {
      "epoch": 0.41652161448201014,
      "grad_norm": 0.24682767689228058,
      "learning_rate": 0.00025557249653486127,
      "loss": 4.6759,
      "step": 20340
    },
    {
      "epoch": 0.41662400425941476,
      "grad_norm": 0.2513619363307953,
      "learning_rate": 0.0002555285177737091,
      "loss": 4.7425,
      "step": 20345
    },
    {
      "epoch": 0.4167263940368194,
      "grad_norm": 0.257161021232605,
      "learning_rate": 0.0002554845210440381,
      "loss": 4.7456,
      "step": 20350
    },
    {
      "epoch": 0.416828783814224,
      "grad_norm": 0.25595664978027344,
      "learning_rate": 0.00025544050635333953,
      "loss": 4.763,
      "step": 20355
    },
    {
      "epoch": 0.4169311735916286,
      "grad_norm": 0.26715409755706787,
      "learning_rate": 0.0002553964737091078,
      "loss": 4.7702,
      "step": 20360
    },
    {
      "epoch": 0.41703356336903324,
      "grad_norm": 0.25047749280929565,
      "learning_rate": 0.00025535242311884056,
      "loss": 4.7518,
      "step": 20365
    },
    {
      "epoch": 0.41713595314643787,
      "grad_norm": 0.25475624203681946,
      "learning_rate": 0.0002553083545900383,
      "loss": 4.7446,
      "step": 20370
    },
    {
      "epoch": 0.4172383429238425,
      "grad_norm": 0.25762513279914856,
      "learning_rate": 0.0002552642681302046,
      "loss": 4.7581,
      "step": 20375
    },
    {
      "epoch": 0.4173407327012471,
      "grad_norm": 0.24678002297878265,
      "learning_rate": 0.0002552201637468463,
      "loss": 4.7205,
      "step": 20380
    },
    {
      "epoch": 0.41744312247865173,
      "grad_norm": 0.2610575556755066,
      "learning_rate": 0.0002551760414474729,
      "loss": 4.793,
      "step": 20385
    },
    {
      "epoch": 0.41754551225605635,
      "grad_norm": 0.2367083877325058,
      "learning_rate": 0.00025513190123959735,
      "loss": 4.7223,
      "step": 20390
    },
    {
      "epoch": 0.41764790203346097,
      "grad_norm": 0.2624305486679077,
      "learning_rate": 0.0002550877431307354,
      "loss": 4.8046,
      "step": 20395
    },
    {
      "epoch": 0.4177502918108656,
      "grad_norm": 0.25630515813827515,
      "learning_rate": 0.00025504356712840597,
      "loss": 4.7957,
      "step": 20400
    },
    {
      "epoch": 0.4178526815882702,
      "grad_norm": 0.2714109420776367,
      "learning_rate": 0.0002549993732401309,
      "loss": 4.7571,
      "step": 20405
    },
    {
      "epoch": 0.41795507136567484,
      "grad_norm": 0.2652490437030792,
      "learning_rate": 0.00025495516147343533,
      "loss": 4.7841,
      "step": 20410
    },
    {
      "epoch": 0.41805746114307946,
      "grad_norm": 0.27956244349479675,
      "learning_rate": 0.00025491093183584705,
      "loss": 4.7662,
      "step": 20415
    },
    {
      "epoch": 0.4181598509204841,
      "grad_norm": 0.25715363025665283,
      "learning_rate": 0.00025486668433489726,
      "loss": 4.7849,
      "step": 20420
    },
    {
      "epoch": 0.4182622406978887,
      "grad_norm": 0.24916450679302216,
      "learning_rate": 0.0002548224189781201,
      "loss": 4.7661,
      "step": 20425
    },
    {
      "epoch": 0.4183646304752933,
      "grad_norm": 0.24065448343753815,
      "learning_rate": 0.0002547781357730525,
      "loss": 4.7972,
      "step": 20430
    },
    {
      "epoch": 0.418467020252698,
      "grad_norm": 0.3225402235984802,
      "learning_rate": 0.00025473383472723486,
      "loss": 4.7533,
      "step": 20435
    },
    {
      "epoch": 0.4185694100301026,
      "grad_norm": 0.25148963928222656,
      "learning_rate": 0.00025468951584821023,
      "loss": 4.7826,
      "step": 20440
    },
    {
      "epoch": 0.41867179980750724,
      "grad_norm": 0.25360456109046936,
      "learning_rate": 0.00025464517914352495,
      "loss": 4.7068,
      "step": 20445
    },
    {
      "epoch": 0.41877418958491186,
      "grad_norm": 0.2516668438911438,
      "learning_rate": 0.0002546008246207283,
      "loss": 4.7593,
      "step": 20450
    },
    {
      "epoch": 0.4188765793623165,
      "grad_norm": 0.2328442484140396,
      "learning_rate": 0.0002545564522873726,
      "loss": 4.7209,
      "step": 20455
    },
    {
      "epoch": 0.4189789691397211,
      "grad_norm": 0.25439298152923584,
      "learning_rate": 0.00025451206215101316,
      "loss": 4.7671,
      "step": 20460
    },
    {
      "epoch": 0.4190813589171257,
      "grad_norm": 0.30829206109046936,
      "learning_rate": 0.0002544676542192084,
      "loss": 4.6941,
      "step": 20465
    },
    {
      "epoch": 0.41918374869453034,
      "grad_norm": 0.2314857691526413,
      "learning_rate": 0.00025442322849951975,
      "loss": 4.7617,
      "step": 20470
    },
    {
      "epoch": 0.41928613847193497,
      "grad_norm": 0.27061936259269714,
      "learning_rate": 0.00025437878499951156,
      "loss": 4.7803,
      "step": 20475
    },
    {
      "epoch": 0.4193885282493396,
      "grad_norm": 0.24506454169750214,
      "learning_rate": 0.0002543343237267514,
      "loss": 4.7624,
      "step": 20480
    },
    {
      "epoch": 0.4194909180267442,
      "grad_norm": 0.26191064715385437,
      "learning_rate": 0.0002542898446888097,
      "loss": 4.769,
      "step": 20485
    },
    {
      "epoch": 0.41959330780414883,
      "grad_norm": 0.2626371681690216,
      "learning_rate": 0.00025424534789326005,
      "loss": 4.7364,
      "step": 20490
    },
    {
      "epoch": 0.41969569758155345,
      "grad_norm": 0.2845771610736847,
      "learning_rate": 0.00025420083334767895,
      "loss": 4.7913,
      "step": 20495
    },
    {
      "epoch": 0.41979808735895807,
      "grad_norm": 0.369780957698822,
      "learning_rate": 0.000254156301059646,
      "loss": 4.7815,
      "step": 20500
    },
    {
      "epoch": 0.4199004771363627,
      "grad_norm": 0.27729693055152893,
      "learning_rate": 0.0002541117510367437,
      "loss": 4.8546,
      "step": 20505
    },
    {
      "epoch": 0.4200028669137673,
      "grad_norm": 0.24958054721355438,
      "learning_rate": 0.00025406718328655777,
      "loss": 4.8021,
      "step": 20510
    },
    {
      "epoch": 0.42010525669117194,
      "grad_norm": 0.25474047660827637,
      "learning_rate": 0.00025402259781667676,
      "loss": 4.765,
      "step": 20515
    },
    {
      "epoch": 0.42020764646857656,
      "grad_norm": 0.24423550069332123,
      "learning_rate": 0.0002539779946346924,
      "loss": 4.7385,
      "step": 20520
    },
    {
      "epoch": 0.4203100362459812,
      "grad_norm": 0.24203768372535706,
      "learning_rate": 0.0002539333737481992,
      "loss": 4.7787,
      "step": 20525
    },
    {
      "epoch": 0.4204124260233858,
      "grad_norm": 0.25625908374786377,
      "learning_rate": 0.00025388873516479495,
      "loss": 4.7171,
      "step": 20530
    },
    {
      "epoch": 0.4205148158007905,
      "grad_norm": 0.24318137764930725,
      "learning_rate": 0.0002538440788920803,
      "loss": 4.7654,
      "step": 20535
    },
    {
      "epoch": 0.4206172055781951,
      "grad_norm": 0.2454153299331665,
      "learning_rate": 0.00025379940493765895,
      "loss": 4.7934,
      "step": 20540
    },
    {
      "epoch": 0.4207195953555997,
      "grad_norm": 0.2600332200527191,
      "learning_rate": 0.00025375471330913764,
      "loss": 4.8145,
      "step": 20545
    },
    {
      "epoch": 0.42082198513300434,
      "grad_norm": 0.23698943853378296,
      "learning_rate": 0.0002537100040141261,
      "loss": 4.7997,
      "step": 20550
    },
    {
      "epoch": 0.42092437491040896,
      "grad_norm": 0.2716785669326782,
      "learning_rate": 0.000253665277060237,
      "loss": 4.7199,
      "step": 20555
    },
    {
      "epoch": 0.4210267646878136,
      "grad_norm": 0.2378966361284256,
      "learning_rate": 0.00025362053245508607,
      "loss": 4.7607,
      "step": 20560
    },
    {
      "epoch": 0.4211291544652182,
      "grad_norm": 0.2592945396900177,
      "learning_rate": 0.00025357577020629205,
      "loss": 4.7984,
      "step": 20565
    },
    {
      "epoch": 0.4212315442426228,
      "grad_norm": 0.2741259038448334,
      "learning_rate": 0.0002535309903214768,
      "loss": 4.731,
      "step": 20570
    },
    {
      "epoch": 0.42133393402002745,
      "grad_norm": 0.2415311634540558,
      "learning_rate": 0.0002534861928082649,
      "loss": 4.7383,
      "step": 20575
    },
    {
      "epoch": 0.42143632379743207,
      "grad_norm": 0.2703542709350586,
      "learning_rate": 0.00025344137767428424,
      "loss": 4.7785,
      "step": 20580
    },
    {
      "epoch": 0.4215387135748367,
      "grad_norm": 0.23549619317054749,
      "learning_rate": 0.00025339654492716543,
      "loss": 4.7548,
      "step": 20585
    },
    {
      "epoch": 0.4216411033522413,
      "grad_norm": 0.2453819215297699,
      "learning_rate": 0.0002533516945745423,
      "loss": 4.7905,
      "step": 20590
    },
    {
      "epoch": 0.42174349312964593,
      "grad_norm": 0.26156413555145264,
      "learning_rate": 0.00025330682662405167,
      "loss": 4.7545,
      "step": 20595
    },
    {
      "epoch": 0.42184588290705055,
      "grad_norm": 0.28923532366752625,
      "learning_rate": 0.00025326194108333313,
      "loss": 4.7925,
      "step": 20600
    },
    {
      "epoch": 0.4219482726844552,
      "grad_norm": 0.24263493716716766,
      "learning_rate": 0.0002532170379600295,
      "loss": 4.7576,
      "step": 20605
    },
    {
      "epoch": 0.4220506624618598,
      "grad_norm": 0.2529463469982147,
      "learning_rate": 0.0002531721172617865,
      "loss": 4.7117,
      "step": 20610
    },
    {
      "epoch": 0.4221530522392644,
      "grad_norm": 0.2368769347667694,
      "learning_rate": 0.0002531271789962528,
      "loss": 4.7411,
      "step": 20615
    },
    {
      "epoch": 0.42225544201666904,
      "grad_norm": 0.253793329000473,
      "learning_rate": 0.00025308222317108026,
      "loss": 4.768,
      "step": 20620
    },
    {
      "epoch": 0.42235783179407366,
      "grad_norm": 0.2393905222415924,
      "learning_rate": 0.0002530372497939234,
      "loss": 4.7746,
      "step": 20625
    },
    {
      "epoch": 0.4224602215714783,
      "grad_norm": 0.24153396487236023,
      "learning_rate": 0.00025299225887244004,
      "loss": 4.7652,
      "step": 20630
    },
    {
      "epoch": 0.42256261134888295,
      "grad_norm": 0.24952054023742676,
      "learning_rate": 0.00025294725041429087,
      "loss": 4.699,
      "step": 20635
    },
    {
      "epoch": 0.4226650011262876,
      "grad_norm": 0.24828636646270752,
      "learning_rate": 0.0002529022244271394,
      "loss": 4.7754,
      "step": 20640
    },
    {
      "epoch": 0.4227673909036922,
      "grad_norm": 0.24602501094341278,
      "learning_rate": 0.0002528571809186525,
      "loss": 4.7367,
      "step": 20645
    },
    {
      "epoch": 0.4228697806810968,
      "grad_norm": 0.31363430619239807,
      "learning_rate": 0.00025281211989649957,
      "loss": 4.7321,
      "step": 20650
    },
    {
      "epoch": 0.42297217045850144,
      "grad_norm": 0.25800901651382446,
      "learning_rate": 0.00025276704136835345,
      "loss": 4.7531,
      "step": 20655
    },
    {
      "epoch": 0.42307456023590606,
      "grad_norm": 0.24754787981510162,
      "learning_rate": 0.0002527219453418896,
      "loss": 4.7282,
      "step": 20660
    },
    {
      "epoch": 0.4231769500133107,
      "grad_norm": 0.24348415434360504,
      "learning_rate": 0.00025267683182478664,
      "loss": 4.7162,
      "step": 20665
    },
    {
      "epoch": 0.4232793397907153,
      "grad_norm": 0.27512019872665405,
      "learning_rate": 0.00025263170082472607,
      "loss": 4.7666,
      "step": 20670
    },
    {
      "epoch": 0.4233817295681199,
      "grad_norm": 0.3168177902698517,
      "learning_rate": 0.0002525865523493925,
      "loss": 4.7289,
      "step": 20675
    },
    {
      "epoch": 0.42348411934552455,
      "grad_norm": 0.24315686523914337,
      "learning_rate": 0.0002525413864064734,
      "loss": 4.7274,
      "step": 20680
    },
    {
      "epoch": 0.42358650912292917,
      "grad_norm": 0.2777538001537323,
      "learning_rate": 0.0002524962030036593,
      "loss": 4.7145,
      "step": 20685
    },
    {
      "epoch": 0.4236888989003338,
      "grad_norm": 0.25496119260787964,
      "learning_rate": 0.00025245100214864357,
      "loss": 4.8036,
      "step": 20690
    },
    {
      "epoch": 0.4237912886777384,
      "grad_norm": 0.25127914547920227,
      "learning_rate": 0.0002524057838491227,
      "loss": 4.7427,
      "step": 20695
    },
    {
      "epoch": 0.42389367845514303,
      "grad_norm": 0.26537182927131653,
      "learning_rate": 0.000252360548112796,
      "loss": 4.7496,
      "step": 20700
    },
    {
      "epoch": 0.42399606823254765,
      "grad_norm": 0.2693651020526886,
      "learning_rate": 0.00025231529494736593,
      "loss": 4.7514,
      "step": 20705
    },
    {
      "epoch": 0.4240984580099523,
      "grad_norm": 0.2502761483192444,
      "learning_rate": 0.0002522700243605378,
      "loss": 4.7721,
      "step": 20710
    },
    {
      "epoch": 0.4242008477873569,
      "grad_norm": 0.263739675283432,
      "learning_rate": 0.00025222473636001995,
      "loss": 4.7405,
      "step": 20715
    },
    {
      "epoch": 0.4243032375647615,
      "grad_norm": 0.2528032958507538,
      "learning_rate": 0.0002521794309535235,
      "loss": 4.7784,
      "step": 20720
    },
    {
      "epoch": 0.42440562734216614,
      "grad_norm": 0.25531187653541565,
      "learning_rate": 0.00025213410814876284,
      "loss": 4.7421,
      "step": 20725
    },
    {
      "epoch": 0.42450801711957076,
      "grad_norm": 0.23182035982608795,
      "learning_rate": 0.0002520887679534551,
      "loss": 4.7566,
      "step": 20730
    },
    {
      "epoch": 0.4246104068969754,
      "grad_norm": 0.26235559582710266,
      "learning_rate": 0.0002520434103753204,
      "loss": 4.7696,
      "step": 20735
    },
    {
      "epoch": 0.42471279667438006,
      "grad_norm": 0.2579345405101776,
      "learning_rate": 0.0002519980354220819,
      "loss": 4.7361,
      "step": 20740
    },
    {
      "epoch": 0.4248151864517847,
      "grad_norm": 0.24372094869613647,
      "learning_rate": 0.00025195264310146564,
      "loss": 4.7092,
      "step": 20745
    },
    {
      "epoch": 0.4249175762291893,
      "grad_norm": 0.26634448766708374,
      "learning_rate": 0.0002519072334212006,
      "loss": 4.807,
      "step": 20750
    },
    {
      "epoch": 0.4250199660065939,
      "grad_norm": 0.2433907836675644,
      "learning_rate": 0.00025186180638901885,
      "loss": 4.737,
      "step": 20755
    },
    {
      "epoch": 0.42512235578399854,
      "grad_norm": 0.24627669155597687,
      "learning_rate": 0.00025181636201265534,
      "loss": 4.7326,
      "step": 20760
    },
    {
      "epoch": 0.42522474556140316,
      "grad_norm": 0.2820776700973511,
      "learning_rate": 0.00025177090029984787,
      "loss": 4.8008,
      "step": 20765
    },
    {
      "epoch": 0.4253271353388078,
      "grad_norm": 0.23752164840698242,
      "learning_rate": 0.00025172542125833736,
      "loss": 4.7742,
      "step": 20770
    },
    {
      "epoch": 0.4254295251162124,
      "grad_norm": 0.24931824207305908,
      "learning_rate": 0.0002516799248958675,
      "loss": 4.7357,
      "step": 20775
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 0.28673723340034485,
      "learning_rate": 0.00025163441122018517,
      "loss": 4.7612,
      "step": 20780
    },
    {
      "epoch": 0.42563430467102165,
      "grad_norm": 0.2612729072570801,
      "learning_rate": 0.00025158888023903993,
      "loss": 4.736,
      "step": 20785
    },
    {
      "epoch": 0.42573669444842627,
      "grad_norm": 0.8941785097122192,
      "learning_rate": 0.00025154333196018456,
      "loss": 4.7835,
      "step": 20790
    },
    {
      "epoch": 0.4258390842258309,
      "grad_norm": 0.2597096264362335,
      "learning_rate": 0.0002514977663913745,
      "loss": 4.7651,
      "step": 20795
    },
    {
      "epoch": 0.4259414740032355,
      "grad_norm": 0.26881104707717896,
      "learning_rate": 0.0002514521835403684,
      "loss": 4.7823,
      "step": 20800
    },
    {
      "epoch": 0.42604386378064013,
      "grad_norm": 0.23725557327270508,
      "learning_rate": 0.00025140658341492767,
      "loss": 4.7017,
      "step": 20805
    },
    {
      "epoch": 0.42614625355804475,
      "grad_norm": 0.23989666998386383,
      "learning_rate": 0.00025136096602281665,
      "loss": 4.7279,
      "step": 20810
    },
    {
      "epoch": 0.4262486433354494,
      "grad_norm": 0.33615273237228394,
      "learning_rate": 0.0002513153313718028,
      "loss": 4.7719,
      "step": 20815
    },
    {
      "epoch": 0.426351033112854,
      "grad_norm": 0.2352549284696579,
      "learning_rate": 0.00025126967946965636,
      "loss": 4.7529,
      "step": 20820
    },
    {
      "epoch": 0.4264534228902586,
      "grad_norm": 0.2769213616847992,
      "learning_rate": 0.00025122401032415053,
      "loss": 4.7272,
      "step": 20825
    },
    {
      "epoch": 0.42655581266766324,
      "grad_norm": 0.2510347366333008,
      "learning_rate": 0.00025117832394306157,
      "loss": 4.7821,
      "step": 20830
    },
    {
      "epoch": 0.42665820244506786,
      "grad_norm": 0.2542051076889038,
      "learning_rate": 0.00025113262033416847,
      "loss": 4.7858,
      "step": 20835
    },
    {
      "epoch": 0.42676059222247253,
      "grad_norm": 0.24848873913288116,
      "learning_rate": 0.00025108689950525335,
      "loss": 4.8271,
      "step": 20840
    },
    {
      "epoch": 0.42686298199987716,
      "grad_norm": 0.3308088481426239,
      "learning_rate": 0.0002510411614641011,
      "loss": 4.7345,
      "step": 20845
    },
    {
      "epoch": 0.4269653717772818,
      "grad_norm": 0.24253885447978973,
      "learning_rate": 0.0002509954062184996,
      "loss": 4.7771,
      "step": 20850
    },
    {
      "epoch": 0.4270677615546864,
      "grad_norm": 0.24381297826766968,
      "learning_rate": 0.0002509496337762398,
      "loss": 4.7391,
      "step": 20855
    },
    {
      "epoch": 0.427170151332091,
      "grad_norm": 0.26204216480255127,
      "learning_rate": 0.0002509038441451153,
      "loss": 4.7607,
      "step": 20860
    },
    {
      "epoch": 0.42727254110949564,
      "grad_norm": 0.2552087604999542,
      "learning_rate": 0.00025085803733292285,
      "loss": 4.757,
      "step": 20865
    },
    {
      "epoch": 0.42737493088690026,
      "grad_norm": 0.2958095371723175,
      "learning_rate": 0.0002508122133474621,
      "loss": 4.7253,
      "step": 20870
    },
    {
      "epoch": 0.4274773206643049,
      "grad_norm": 0.2517816126346588,
      "learning_rate": 0.0002507663721965355,
      "loss": 4.7462,
      "step": 20875
    },
    {
      "epoch": 0.4275797104417095,
      "grad_norm": 0.22977468371391296,
      "learning_rate": 0.0002507205138879485,
      "loss": 4.7558,
      "step": 20880
    },
    {
      "epoch": 0.4276821002191141,
      "grad_norm": 0.25457119941711426,
      "learning_rate": 0.0002506746384295095,
      "loss": 4.7889,
      "step": 20885
    },
    {
      "epoch": 0.42778448999651875,
      "grad_norm": 0.261680543422699,
      "learning_rate": 0.0002506287458290298,
      "loss": 4.7809,
      "step": 20890
    },
    {
      "epoch": 0.42788687977392337,
      "grad_norm": 0.2524440884590149,
      "learning_rate": 0.00025058283609432356,
      "loss": 4.7378,
      "step": 20895
    },
    {
      "epoch": 0.427989269551328,
      "grad_norm": 0.25424009561538696,
      "learning_rate": 0.00025053690923320797,
      "loss": 4.7405,
      "step": 20900
    },
    {
      "epoch": 0.4280916593287326,
      "grad_norm": 0.24107684195041656,
      "learning_rate": 0.00025049096525350307,
      "loss": 4.7329,
      "step": 20905
    },
    {
      "epoch": 0.42819404910613723,
      "grad_norm": 0.2617661654949188,
      "learning_rate": 0.0002504450041630318,
      "loss": 4.7606,
      "step": 20910
    },
    {
      "epoch": 0.42829643888354185,
      "grad_norm": 0.2651958167552948,
      "learning_rate": 0.00025039902596962,
      "loss": 4.7175,
      "step": 20915
    },
    {
      "epoch": 0.4283988286609465,
      "grad_norm": 0.2467733472585678,
      "learning_rate": 0.00025035303068109653,
      "loss": 4.7578,
      "step": 20920
    },
    {
      "epoch": 0.4285012184383511,
      "grad_norm": 0.278435617685318,
      "learning_rate": 0.00025030701830529306,
      "loss": 4.7412,
      "step": 20925
    },
    {
      "epoch": 0.4286036082157557,
      "grad_norm": 0.27353397011756897,
      "learning_rate": 0.0002502609888500441,
      "loss": 4.7226,
      "step": 20930
    },
    {
      "epoch": 0.42870599799316034,
      "grad_norm": 0.2490510493516922,
      "learning_rate": 0.00025021494232318727,
      "loss": 4.7062,
      "step": 20935
    },
    {
      "epoch": 0.428808387770565,
      "grad_norm": 0.7028024792671204,
      "learning_rate": 0.00025016887873256295,
      "loss": 4.6703,
      "step": 20940
    },
    {
      "epoch": 0.42891077754796963,
      "grad_norm": 0.2787671983242035,
      "learning_rate": 0.0002501227980860145,
      "loss": 4.7194,
      "step": 20945
    },
    {
      "epoch": 0.42901316732537426,
      "grad_norm": 0.24122415482997894,
      "learning_rate": 0.0002500767003913881,
      "loss": 4.7723,
      "step": 20950
    },
    {
      "epoch": 0.4291155571027789,
      "grad_norm": 0.24152274429798126,
      "learning_rate": 0.00025003058565653297,
      "loss": 4.7381,
      "step": 20955
    },
    {
      "epoch": 0.4292179468801835,
      "grad_norm": 0.23557929694652557,
      "learning_rate": 0.000249984453889301,
      "loss": 4.7286,
      "step": 20960
    },
    {
      "epoch": 0.4293203366575881,
      "grad_norm": 0.30461108684539795,
      "learning_rate": 0.0002499383050975472,
      "loss": 4.7933,
      "step": 20965
    },
    {
      "epoch": 0.42942272643499274,
      "grad_norm": 0.24944448471069336,
      "learning_rate": 0.00024989213928912953,
      "loss": 4.7598,
      "step": 20970
    },
    {
      "epoch": 0.42952511621239736,
      "grad_norm": 0.23622015118598938,
      "learning_rate": 0.00024984595647190846,
      "loss": 4.7701,
      "step": 20975
    },
    {
      "epoch": 0.429627505989802,
      "grad_norm": 0.24173890054225922,
      "learning_rate": 0.00024979975665374783,
      "loss": 4.7568,
      "step": 20980
    },
    {
      "epoch": 0.4297298957672066,
      "grad_norm": 0.26110169291496277,
      "learning_rate": 0.0002497535398425141,
      "loss": 4.6862,
      "step": 20985
    },
    {
      "epoch": 0.4298322855446112,
      "grad_norm": 0.2331186681985855,
      "learning_rate": 0.00024970730604607656,
      "loss": 4.7182,
      "step": 20990
    },
    {
      "epoch": 0.42993467532201585,
      "grad_norm": 0.24164794385433197,
      "learning_rate": 0.0002496610552723077,
      "loss": 4.7811,
      "step": 20995
    },
    {
      "epoch": 0.43003706509942047,
      "grad_norm": 0.28270652890205383,
      "learning_rate": 0.0002496147875290827,
      "loss": 4.8024,
      "step": 21000
    },
    {
      "epoch": 0.43003706509942047,
      "eval_loss": 4.769644737243652,
      "eval_runtime": 17.2364,
      "eval_samples_per_second": 33.766,
      "eval_steps_per_second": 4.235,
      "step": 21000
    },
    {
      "epoch": 0.4301394548768251,
      "grad_norm": 0.24345621466636658,
      "learning_rate": 0.00024956850282427953,
      "loss": 4.7666,
      "step": 21005
    },
    {
      "epoch": 0.4302418446542297,
      "grad_norm": 0.2555341422557831,
      "learning_rate": 0.00024952220116577925,
      "loss": 4.7304,
      "step": 21010
    },
    {
      "epoch": 0.43034423443163433,
      "grad_norm": 0.25171372294425964,
      "learning_rate": 0.0002494758825614657,
      "loss": 4.8079,
      "step": 21015
    },
    {
      "epoch": 0.43044662420903895,
      "grad_norm": 0.25784316658973694,
      "learning_rate": 0.00024942954701922556,
      "loss": 4.815,
      "step": 21020
    },
    {
      "epoch": 0.4305490139864436,
      "grad_norm": 0.4336875379085541,
      "learning_rate": 0.0002493831945469486,
      "loss": 4.7852,
      "step": 21025
    },
    {
      "epoch": 0.4306514037638482,
      "grad_norm": 0.24383743107318878,
      "learning_rate": 0.00024933682515252716,
      "loss": 4.7327,
      "step": 21030
    },
    {
      "epoch": 0.4307537935412528,
      "grad_norm": 0.2472480684518814,
      "learning_rate": 0.0002492904388438568,
      "loss": 4.749,
      "step": 21035
    },
    {
      "epoch": 0.4308561833186575,
      "grad_norm": 0.25940173864364624,
      "learning_rate": 0.00024924403562883574,
      "loss": 4.7539,
      "step": 21040
    },
    {
      "epoch": 0.4309585730960621,
      "grad_norm": 0.27517426013946533,
      "learning_rate": 0.0002491976155153651,
      "loss": 4.7581,
      "step": 21045
    },
    {
      "epoch": 0.43106096287346674,
      "grad_norm": 0.272371768951416,
      "learning_rate": 0.00024915117851134896,
      "loss": 4.7662,
      "step": 21050
    },
    {
      "epoch": 0.43116335265087136,
      "grad_norm": 0.27990251779556274,
      "learning_rate": 0.00024910472462469417,
      "loss": 4.7864,
      "step": 21055
    },
    {
      "epoch": 0.431265742428276,
      "grad_norm": 0.320916086435318,
      "learning_rate": 0.00024905825386331055,
      "loss": 4.7337,
      "step": 21060
    },
    {
      "epoch": 0.4313681322056806,
      "grad_norm": 0.23804743587970734,
      "learning_rate": 0.0002490117662351108,
      "loss": 4.7596,
      "step": 21065
    },
    {
      "epoch": 0.4314705219830852,
      "grad_norm": 0.24924394488334656,
      "learning_rate": 0.00024896526174801034,
      "loss": 4.72,
      "step": 21070
    },
    {
      "epoch": 0.43157291176048984,
      "grad_norm": 0.2520267963409424,
      "learning_rate": 0.00024891874040992764,
      "loss": 4.6497,
      "step": 21075
    },
    {
      "epoch": 0.43167530153789446,
      "grad_norm": 0.2740488052368164,
      "learning_rate": 0.00024887220222878393,
      "loss": 4.7016,
      "step": 21080
    },
    {
      "epoch": 0.4317776913152991,
      "grad_norm": 0.2445664405822754,
      "learning_rate": 0.00024882564721250336,
      "loss": 4.7656,
      "step": 21085
    },
    {
      "epoch": 0.4318800810927037,
      "grad_norm": 0.28102514147758484,
      "learning_rate": 0.0002487790753690129,
      "loss": 4.749,
      "step": 21090
    },
    {
      "epoch": 0.4319824708701083,
      "grad_norm": 0.282574862241745,
      "learning_rate": 0.00024873248670624257,
      "loss": 4.7722,
      "step": 21095
    },
    {
      "epoch": 0.43208486064751295,
      "grad_norm": 0.2809067964553833,
      "learning_rate": 0.0002486858812321249,
      "loss": 4.7505,
      "step": 21100
    },
    {
      "epoch": 0.43218725042491757,
      "grad_norm": 0.27816784381866455,
      "learning_rate": 0.00024863925895459555,
      "loss": 4.7697,
      "step": 21105
    },
    {
      "epoch": 0.4322896402023222,
      "grad_norm": 0.25122854113578796,
      "learning_rate": 0.00024859261988159306,
      "loss": 4.742,
      "step": 21110
    },
    {
      "epoch": 0.4323920299797268,
      "grad_norm": 0.23729555308818817,
      "learning_rate": 0.0002485459640210586,
      "loss": 4.7465,
      "step": 21115
    },
    {
      "epoch": 0.43249441975713143,
      "grad_norm": 0.2549649178981781,
      "learning_rate": 0.0002484992913809364,
      "loss": 4.7455,
      "step": 21120
    },
    {
      "epoch": 0.43259680953453605,
      "grad_norm": 0.24888847768306732,
      "learning_rate": 0.0002484526019691736,
      "loss": 4.7683,
      "step": 21125
    },
    {
      "epoch": 0.4326991993119407,
      "grad_norm": 0.266985148191452,
      "learning_rate": 0.0002484058957937199,
      "loss": 4.7236,
      "step": 21130
    },
    {
      "epoch": 0.4328015890893453,
      "grad_norm": 0.2714773118495941,
      "learning_rate": 0.00024835917286252814,
      "loss": 4.767,
      "step": 21135
    },
    {
      "epoch": 0.4329039788667499,
      "grad_norm": 0.23292596638202667,
      "learning_rate": 0.00024831243318355396,
      "loss": 4.8115,
      "step": 21140
    },
    {
      "epoch": 0.4330063686441546,
      "grad_norm": 0.24832096695899963,
      "learning_rate": 0.0002482656767647557,
      "loss": 4.6607,
      "step": 21145
    },
    {
      "epoch": 0.4331087584215592,
      "grad_norm": 0.2521432042121887,
      "learning_rate": 0.0002482189036140947,
      "loss": 4.7279,
      "step": 21150
    },
    {
      "epoch": 0.43321114819896384,
      "grad_norm": 0.2571776509284973,
      "learning_rate": 0.0002481721137395351,
      "loss": 4.734,
      "step": 21155
    },
    {
      "epoch": 0.43331353797636846,
      "grad_norm": 0.2571667730808258,
      "learning_rate": 0.0002481253071490439,
      "loss": 4.7468,
      "step": 21160
    },
    {
      "epoch": 0.4334159277537731,
      "grad_norm": 0.25185245275497437,
      "learning_rate": 0.00024807848385059093,
      "loss": 4.72,
      "step": 21165
    },
    {
      "epoch": 0.4335183175311777,
      "grad_norm": 0.5772988796234131,
      "learning_rate": 0.00024803164385214885,
      "loss": 4.7641,
      "step": 21170
    },
    {
      "epoch": 0.4336207073085823,
      "grad_norm": 0.2447091042995453,
      "learning_rate": 0.0002479847871616932,
      "loss": 4.729,
      "step": 21175
    },
    {
      "epoch": 0.43372309708598694,
      "grad_norm": 0.29918625950813293,
      "learning_rate": 0.0002479379137872024,
      "loss": 4.7834,
      "step": 21180
    },
    {
      "epoch": 0.43382548686339156,
      "grad_norm": 0.24188275635242462,
      "learning_rate": 0.00024789102373665763,
      "loss": 4.7368,
      "step": 21185
    },
    {
      "epoch": 0.4339278766407962,
      "grad_norm": 0.24988071620464325,
      "learning_rate": 0.0002478441170180429,
      "loss": 4.7412,
      "step": 21190
    },
    {
      "epoch": 0.4340302664182008,
      "grad_norm": 0.2506921589374542,
      "learning_rate": 0.00024779719363934513,
      "loss": 4.7174,
      "step": 21195
    },
    {
      "epoch": 0.4341326561956054,
      "grad_norm": 0.25739920139312744,
      "learning_rate": 0.000247750253608554,
      "loss": 4.7428,
      "step": 21200
    },
    {
      "epoch": 0.43423504597301005,
      "grad_norm": 0.2615566551685333,
      "learning_rate": 0.00024770329693366206,
      "loss": 4.7792,
      "step": 21205
    },
    {
      "epoch": 0.43433743575041467,
      "grad_norm": 0.2501102685928345,
      "learning_rate": 0.00024765632362266487,
      "loss": 4.7425,
      "step": 21210
    },
    {
      "epoch": 0.4344398255278193,
      "grad_norm": 0.24423962831497192,
      "learning_rate": 0.0002476093336835604,
      "loss": 4.7163,
      "step": 21215
    },
    {
      "epoch": 0.4345422153052239,
      "grad_norm": 0.23754578828811646,
      "learning_rate": 0.0002475623271243499,
      "loss": 4.7604,
      "step": 21220
    },
    {
      "epoch": 0.43464460508262853,
      "grad_norm": 0.24553988873958588,
      "learning_rate": 0.0002475153039530372,
      "loss": 4.7635,
      "step": 21225
    },
    {
      "epoch": 0.43474699486003315,
      "grad_norm": 0.2519501745700836,
      "learning_rate": 0.000247468264177629,
      "loss": 4.7251,
      "step": 21230
    },
    {
      "epoch": 0.4348493846374378,
      "grad_norm": 0.2671903073787689,
      "learning_rate": 0.00024742120780613477,
      "loss": 4.7707,
      "step": 21235
    },
    {
      "epoch": 0.4349517744148424,
      "grad_norm": 0.27295219898223877,
      "learning_rate": 0.00024737413484656704,
      "loss": 4.7181,
      "step": 21240
    },
    {
      "epoch": 0.4350541641922471,
      "grad_norm": 0.24619005620479584,
      "learning_rate": 0.0002473270453069409,
      "loss": 4.7404,
      "step": 21245
    },
    {
      "epoch": 0.4351565539696517,
      "grad_norm": 0.26180773973464966,
      "learning_rate": 0.00024727993919527433,
      "loss": 4.7374,
      "step": 21250
    },
    {
      "epoch": 0.4352589437470563,
      "grad_norm": 0.2482661008834839,
      "learning_rate": 0.0002472328165195883,
      "loss": 4.7288,
      "step": 21255
    },
    {
      "epoch": 0.43536133352446094,
      "grad_norm": 0.25037941336631775,
      "learning_rate": 0.00024718567728790634,
      "loss": 4.7517,
      "step": 21260
    },
    {
      "epoch": 0.43546372330186556,
      "grad_norm": 0.23189325630664825,
      "learning_rate": 0.000247138521508255,
      "loss": 4.7501,
      "step": 21265
    },
    {
      "epoch": 0.4355661130792702,
      "grad_norm": 0.28822433948516846,
      "learning_rate": 0.0002470913491886635,
      "loss": 4.7335,
      "step": 21270
    },
    {
      "epoch": 0.4356685028566748,
      "grad_norm": 0.2456277757883072,
      "learning_rate": 0.00024704416033716404,
      "loss": 4.7491,
      "step": 21275
    },
    {
      "epoch": 0.4357708926340794,
      "grad_norm": 0.2418244481086731,
      "learning_rate": 0.0002469969549617915,
      "loss": 4.7966,
      "step": 21280
    },
    {
      "epoch": 0.43587328241148404,
      "grad_norm": 0.2540426552295685,
      "learning_rate": 0.00024694973307058356,
      "loss": 4.7318,
      "step": 21285
    },
    {
      "epoch": 0.43597567218888866,
      "grad_norm": 0.2391122579574585,
      "learning_rate": 0.00024690249467158095,
      "loss": 4.7782,
      "step": 21290
    },
    {
      "epoch": 0.4360780619662933,
      "grad_norm": 0.24399574100971222,
      "learning_rate": 0.00024685523977282684,
      "loss": 4.699,
      "step": 21295
    },
    {
      "epoch": 0.4361804517436979,
      "grad_norm": 0.2370738983154297,
      "learning_rate": 0.00024680796838236744,
      "loss": 4.7699,
      "step": 21300
    },
    {
      "epoch": 0.4362828415211025,
      "grad_norm": 0.24855202436447144,
      "learning_rate": 0.00024676068050825186,
      "loss": 4.7625,
      "step": 21305
    },
    {
      "epoch": 0.43638523129850715,
      "grad_norm": 0.2379562258720398,
      "learning_rate": 0.00024671337615853175,
      "loss": 4.7163,
      "step": 21310
    },
    {
      "epoch": 0.43648762107591177,
      "grad_norm": 0.2485109567642212,
      "learning_rate": 0.0002466660553412617,
      "loss": 4.7674,
      "step": 21315
    },
    {
      "epoch": 0.4365900108533164,
      "grad_norm": 0.2557595670223236,
      "learning_rate": 0.00024661871806449923,
      "loss": 4.7188,
      "step": 21320
    },
    {
      "epoch": 0.436692400630721,
      "grad_norm": 0.245519757270813,
      "learning_rate": 0.0002465713643363044,
      "loss": 4.7351,
      "step": 21325
    },
    {
      "epoch": 0.43679479040812563,
      "grad_norm": 0.2353067696094513,
      "learning_rate": 0.0002465239941647403,
      "loss": 4.7396,
      "step": 21330
    },
    {
      "epoch": 0.43689718018553025,
      "grad_norm": 0.24086587131023407,
      "learning_rate": 0.00024647660755787275,
      "loss": 4.7536,
      "step": 21335
    },
    {
      "epoch": 0.4369995699629349,
      "grad_norm": 0.28357189893722534,
      "learning_rate": 0.0002464292045237702,
      "loss": 4.7301,
      "step": 21340
    },
    {
      "epoch": 0.43710195974033955,
      "grad_norm": 0.2478339523077011,
      "learning_rate": 0.00024638178507050415,
      "loss": 4.8051,
      "step": 21345
    },
    {
      "epoch": 0.4372043495177442,
      "grad_norm": 0.2508557140827179,
      "learning_rate": 0.00024633434920614877,
      "loss": 4.7126,
      "step": 21350
    },
    {
      "epoch": 0.4373067392951488,
      "grad_norm": 0.25761789083480835,
      "learning_rate": 0.00024628689693878113,
      "loss": 4.7555,
      "step": 21355
    },
    {
      "epoch": 0.4374091290725534,
      "grad_norm": 0.2546573579311371,
      "learning_rate": 0.0002462394282764809,
      "loss": 4.7412,
      "step": 21360
    },
    {
      "epoch": 0.43751151884995804,
      "grad_norm": 0.2487485557794571,
      "learning_rate": 0.00024619194322733065,
      "loss": 4.6789,
      "step": 21365
    },
    {
      "epoch": 0.43761390862736266,
      "grad_norm": 0.24232380092144012,
      "learning_rate": 0.0002461444417994158,
      "loss": 4.7147,
      "step": 21370
    },
    {
      "epoch": 0.4377162984047673,
      "grad_norm": 0.25324276089668274,
      "learning_rate": 0.00024609692400082443,
      "loss": 4.7208,
      "step": 21375
    },
    {
      "epoch": 0.4378186881821719,
      "grad_norm": 0.2620543837547302,
      "learning_rate": 0.00024604938983964757,
      "loss": 4.7737,
      "step": 21380
    },
    {
      "epoch": 0.4379210779595765,
      "grad_norm": 0.23999658226966858,
      "learning_rate": 0.00024600183932397884,
      "loss": 4.7306,
      "step": 21385
    },
    {
      "epoch": 0.43802346773698114,
      "grad_norm": 0.25081706047058105,
      "learning_rate": 0.0002459542724619148,
      "loss": 4.7146,
      "step": 21390
    },
    {
      "epoch": 0.43812585751438576,
      "grad_norm": 0.304277241230011,
      "learning_rate": 0.00024590668926155473,
      "loss": 4.7099,
      "step": 21395
    },
    {
      "epoch": 0.4382282472917904,
      "grad_norm": 0.2500305473804474,
      "learning_rate": 0.0002458590897310007,
      "loss": 4.7558,
      "step": 21400
    },
    {
      "epoch": 0.438330637069195,
      "grad_norm": 0.2485106736421585,
      "learning_rate": 0.00024581147387835764,
      "loss": 4.7192,
      "step": 21405
    },
    {
      "epoch": 0.4384330268465996,
      "grad_norm": 0.25544965267181396,
      "learning_rate": 0.000245763841711733,
      "loss": 4.7744,
      "step": 21410
    },
    {
      "epoch": 0.43853541662400425,
      "grad_norm": 0.2420274317264557,
      "learning_rate": 0.00024571619323923737,
      "loss": 4.7154,
      "step": 21415
    },
    {
      "epoch": 0.43863780640140887,
      "grad_norm": 0.2997153103351593,
      "learning_rate": 0.00024566852846898384,
      "loss": 4.7067,
      "step": 21420
    },
    {
      "epoch": 0.4387401961788135,
      "grad_norm": 0.26231518387794495,
      "learning_rate": 0.0002456208474090884,
      "loss": 4.8259,
      "step": 21425
    },
    {
      "epoch": 0.4388425859562181,
      "grad_norm": 0.23175372183322906,
      "learning_rate": 0.00024557315006766984,
      "loss": 4.7194,
      "step": 21430
    },
    {
      "epoch": 0.43894497573362273,
      "grad_norm": 0.2603735029697418,
      "learning_rate": 0.0002455254364528495,
      "loss": 4.7663,
      "step": 21435
    },
    {
      "epoch": 0.43904736551102735,
      "grad_norm": 0.2584611475467682,
      "learning_rate": 0.0002454777065727518,
      "loss": 4.7981,
      "step": 21440
    },
    {
      "epoch": 0.439149755288432,
      "grad_norm": 0.24514956772327423,
      "learning_rate": 0.0002454299604355038,
      "loss": 4.7201,
      "step": 21445
    },
    {
      "epoch": 0.43925214506583665,
      "grad_norm": 0.24925577640533447,
      "learning_rate": 0.0002453821980492352,
      "loss": 4.7701,
      "step": 21450
    },
    {
      "epoch": 0.4393545348432413,
      "grad_norm": 0.2554652988910675,
      "learning_rate": 0.0002453344194220787,
      "loss": 4.7923,
      "step": 21455
    },
    {
      "epoch": 0.4394569246206459,
      "grad_norm": 0.32403281331062317,
      "learning_rate": 0.0002452866245621696,
      "loss": 4.7854,
      "step": 21460
    },
    {
      "epoch": 0.4395593143980505,
      "grad_norm": 0.2463287115097046,
      "learning_rate": 0.000245238813477646,
      "loss": 4.74,
      "step": 21465
    },
    {
      "epoch": 0.43966170417545514,
      "grad_norm": 0.2618178427219391,
      "learning_rate": 0.00024519098617664875,
      "loss": 4.7173,
      "step": 21470
    },
    {
      "epoch": 0.43976409395285976,
      "grad_norm": 0.24402408301830292,
      "learning_rate": 0.00024514314266732155,
      "loss": 4.7305,
      "step": 21475
    },
    {
      "epoch": 0.4398664837302644,
      "grad_norm": 0.34568819403648376,
      "learning_rate": 0.0002450952829578107,
      "loss": 4.7478,
      "step": 21480
    },
    {
      "epoch": 0.439968873507669,
      "grad_norm": 0.26306089758872986,
      "learning_rate": 0.0002450474070562655,
      "loss": 4.6287,
      "step": 21485
    },
    {
      "epoch": 0.4400712632850736,
      "grad_norm": 0.287149041891098,
      "learning_rate": 0.00024499951497083776,
      "loss": 4.7754,
      "step": 21490
    },
    {
      "epoch": 0.44017365306247824,
      "grad_norm": 0.3165466785430908,
      "learning_rate": 0.00024495160670968215,
      "loss": 4.6776,
      "step": 21495
    },
    {
      "epoch": 0.44027604283988286,
      "grad_norm": 0.26258593797683716,
      "learning_rate": 0.00024490368228095605,
      "loss": 4.8052,
      "step": 21500
    },
    {
      "epoch": 0.4403784326172875,
      "grad_norm": 0.26825210452079773,
      "learning_rate": 0.0002448557416928198,
      "loss": 4.7529,
      "step": 21505
    },
    {
      "epoch": 0.4404808223946921,
      "grad_norm": 0.2573893368244171,
      "learning_rate": 0.0002448077849534361,
      "loss": 4.7737,
      "step": 21510
    },
    {
      "epoch": 0.4405832121720967,
      "grad_norm": 0.24566176533699036,
      "learning_rate": 0.00024475981207097084,
      "loss": 4.73,
      "step": 21515
    },
    {
      "epoch": 0.44068560194950135,
      "grad_norm": 0.2359248548746109,
      "learning_rate": 0.00024471182305359224,
      "loss": 4.7148,
      "step": 21520
    },
    {
      "epoch": 0.44078799172690597,
      "grad_norm": 0.23471441864967346,
      "learning_rate": 0.00024466381790947165,
      "loss": 4.7509,
      "step": 21525
    },
    {
      "epoch": 0.4408903815043106,
      "grad_norm": 0.268425852060318,
      "learning_rate": 0.0002446157966467829,
      "loss": 4.7169,
      "step": 21530
    },
    {
      "epoch": 0.4409927712817152,
      "grad_norm": 0.24401435256004333,
      "learning_rate": 0.00024456775927370267,
      "loss": 4.7522,
      "step": 21535
    },
    {
      "epoch": 0.44109516105911983,
      "grad_norm": 0.24866032600402832,
      "learning_rate": 0.0002445197057984103,
      "loss": 4.6797,
      "step": 21540
    },
    {
      "epoch": 0.44119755083652445,
      "grad_norm": 0.2437106817960739,
      "learning_rate": 0.00024447163622908803,
      "loss": 4.794,
      "step": 21545
    },
    {
      "epoch": 0.44129994061392913,
      "grad_norm": 0.24281218647956848,
      "learning_rate": 0.00024442355057392067,
      "loss": 4.7386,
      "step": 21550
    },
    {
      "epoch": 0.44140233039133375,
      "grad_norm": 0.26372960209846497,
      "learning_rate": 0.0002443754488410959,
      "loss": 4.722,
      "step": 21555
    },
    {
      "epoch": 0.4415047201687384,
      "grad_norm": 0.25185683369636536,
      "learning_rate": 0.00024432733103880414,
      "loss": 4.7104,
      "step": 21560
    },
    {
      "epoch": 0.441607109946143,
      "grad_norm": 0.27980077266693115,
      "learning_rate": 0.0002442791971752383,
      "loss": 4.7497,
      "step": 21565
    },
    {
      "epoch": 0.4417094997235476,
      "grad_norm": 0.28467410802841187,
      "learning_rate": 0.0002442310472585944,
      "loss": 4.7019,
      "step": 21570
    },
    {
      "epoch": 0.44181188950095224,
      "grad_norm": 0.2608586251735687,
      "learning_rate": 0.0002441828812970709,
      "loss": 4.6961,
      "step": 21575
    },
    {
      "epoch": 0.44191427927835686,
      "grad_norm": 0.28476840257644653,
      "learning_rate": 0.00024413469929886918,
      "loss": 4.7437,
      "step": 21580
    },
    {
      "epoch": 0.4420166690557615,
      "grad_norm": 0.28951677680015564,
      "learning_rate": 0.00024408650127219324,
      "loss": 4.7665,
      "step": 21585
    },
    {
      "epoch": 0.4421190588331661,
      "grad_norm": 0.2582278251647949,
      "learning_rate": 0.0002440382872252498,
      "loss": 4.7652,
      "step": 21590
    },
    {
      "epoch": 0.4422214486105707,
      "grad_norm": 0.26101550459861755,
      "learning_rate": 0.00024399005716624835,
      "loss": 4.7786,
      "step": 21595
    },
    {
      "epoch": 0.44232383838797534,
      "grad_norm": 0.2525230348110199,
      "learning_rate": 0.00024394181110340117,
      "loss": 4.7128,
      "step": 21600
    },
    {
      "epoch": 0.44242622816537996,
      "grad_norm": 0.2606767416000366,
      "learning_rate": 0.00024389354904492315,
      "loss": 4.7473,
      "step": 21605
    },
    {
      "epoch": 0.4425286179427846,
      "grad_norm": 0.2644956707954407,
      "learning_rate": 0.00024384527099903194,
      "loss": 4.7048,
      "step": 21610
    },
    {
      "epoch": 0.4426310077201892,
      "grad_norm": 0.2370660901069641,
      "learning_rate": 0.00024379697697394796,
      "loss": 4.7694,
      "step": 21615
    },
    {
      "epoch": 0.44273339749759383,
      "grad_norm": 0.25547847151756287,
      "learning_rate": 0.00024374866697789429,
      "loss": 4.7107,
      "step": 21620
    },
    {
      "epoch": 0.44283578727499845,
      "grad_norm": 0.2429160326719284,
      "learning_rate": 0.0002437003410190968,
      "loss": 4.817,
      "step": 21625
    },
    {
      "epoch": 0.44293817705240307,
      "grad_norm": 0.25499096512794495,
      "learning_rate": 0.00024365199910578397,
      "loss": 4.7773,
      "step": 21630
    },
    {
      "epoch": 0.4430405668298077,
      "grad_norm": 0.23434969782829285,
      "learning_rate": 0.00024360364124618704,
      "loss": 4.7555,
      "step": 21635
    },
    {
      "epoch": 0.4431429566072123,
      "grad_norm": 0.24683192372322083,
      "learning_rate": 0.00024355526744854008,
      "loss": 4.6997,
      "step": 21640
    },
    {
      "epoch": 0.44324534638461693,
      "grad_norm": 0.2424478828907013,
      "learning_rate": 0.00024350687772107973,
      "loss": 4.6979,
      "step": 21645
    },
    {
      "epoch": 0.4433477361620216,
      "grad_norm": 0.27291345596313477,
      "learning_rate": 0.00024345847207204538,
      "loss": 4.7471,
      "step": 21650
    },
    {
      "epoch": 0.44345012593942623,
      "grad_norm": 0.25225764513015747,
      "learning_rate": 0.00024341005050967915,
      "loss": 4.7957,
      "step": 21655
    },
    {
      "epoch": 0.44355251571683085,
      "grad_norm": 0.2572467029094696,
      "learning_rate": 0.00024336161304222583,
      "loss": 4.7486,
      "step": 21660
    },
    {
      "epoch": 0.4436549054942355,
      "grad_norm": 0.45135772228240967,
      "learning_rate": 0.00024331315967793306,
      "loss": 4.6839,
      "step": 21665
    },
    {
      "epoch": 0.4437572952716401,
      "grad_norm": 0.23294630646705627,
      "learning_rate": 0.00024326469042505093,
      "loss": 4.7659,
      "step": 21670
    },
    {
      "epoch": 0.4438596850490447,
      "grad_norm": 0.26809754967689514,
      "learning_rate": 0.00024321620529183248,
      "loss": 4.7783,
      "step": 21675
    },
    {
      "epoch": 0.44396207482644934,
      "grad_norm": 0.24899956583976746,
      "learning_rate": 0.0002431677042865333,
      "loss": 4.7864,
      "step": 21680
    },
    {
      "epoch": 0.44406446460385396,
      "grad_norm": 0.27061086893081665,
      "learning_rate": 0.00024311918741741178,
      "loss": 4.7514,
      "step": 21685
    },
    {
      "epoch": 0.4441668543812586,
      "grad_norm": 0.2461511492729187,
      "learning_rate": 0.00024307065469272895,
      "loss": 4.7489,
      "step": 21690
    },
    {
      "epoch": 0.4442692441586632,
      "grad_norm": 0.6099108457565308,
      "learning_rate": 0.00024302210612074852,
      "loss": 4.6654,
      "step": 21695
    },
    {
      "epoch": 0.4443716339360678,
      "grad_norm": 0.25346803665161133,
      "learning_rate": 0.00024297354170973708,
      "loss": 4.7607,
      "step": 21700
    },
    {
      "epoch": 0.44447402371347244,
      "grad_norm": 0.27422893047332764,
      "learning_rate": 0.0002429249614679636,
      "loss": 4.7332,
      "step": 21705
    },
    {
      "epoch": 0.44457641349087706,
      "grad_norm": 0.25380709767341614,
      "learning_rate": 0.00024287636540369997,
      "loss": 4.7358,
      "step": 21710
    },
    {
      "epoch": 0.4446788032682817,
      "grad_norm": 0.2626173496246338,
      "learning_rate": 0.00024282775352522077,
      "loss": 4.7738,
      "step": 21715
    },
    {
      "epoch": 0.4447811930456863,
      "grad_norm": 0.25741249322891235,
      "learning_rate": 0.00024277912584080323,
      "loss": 4.7849,
      "step": 21720
    },
    {
      "epoch": 0.44488358282309093,
      "grad_norm": 0.253892183303833,
      "learning_rate": 0.0002427304823587272,
      "loss": 4.7328,
      "step": 21725
    },
    {
      "epoch": 0.44498597260049555,
      "grad_norm": 0.24379855394363403,
      "learning_rate": 0.00024268182308727533,
      "loss": 4.7645,
      "step": 21730
    },
    {
      "epoch": 0.44508836237790017,
      "grad_norm": 0.3527223467826843,
      "learning_rate": 0.00024263314803473296,
      "loss": 4.7587,
      "step": 21735
    },
    {
      "epoch": 0.4451907521553048,
      "grad_norm": 0.27313509583473206,
      "learning_rate": 0.000242584457209388,
      "loss": 4.775,
      "step": 21740
    },
    {
      "epoch": 0.4452931419327094,
      "grad_norm": 0.2845230996608734,
      "learning_rate": 0.00024253575061953115,
      "loss": 4.764,
      "step": 21745
    },
    {
      "epoch": 0.4453955317101141,
      "grad_norm": 0.25186723470687866,
      "learning_rate": 0.00024248702827345573,
      "loss": 4.7532,
      "step": 21750
    },
    {
      "epoch": 0.4454979214875187,
      "grad_norm": 0.27293211221694946,
      "learning_rate": 0.00024243829017945787,
      "loss": 4.7009,
      "step": 21755
    },
    {
      "epoch": 0.44560031126492333,
      "grad_norm": 0.2619374990463257,
      "learning_rate": 0.0002423895363458362,
      "loss": 4.7033,
      "step": 21760
    },
    {
      "epoch": 0.44570270104232795,
      "grad_norm": 0.25431960821151733,
      "learning_rate": 0.0002423407667808921,
      "loss": 4.7853,
      "step": 21765
    },
    {
      "epoch": 0.4458050908197326,
      "grad_norm": 0.24628354609012604,
      "learning_rate": 0.00024229198149292974,
      "loss": 4.7179,
      "step": 21770
    },
    {
      "epoch": 0.4459074805971372,
      "grad_norm": 0.27475932240486145,
      "learning_rate": 0.0002422431804902558,
      "loss": 4.7177,
      "step": 21775
    },
    {
      "epoch": 0.4460098703745418,
      "grad_norm": 0.2529583275318146,
      "learning_rate": 0.00024219436378117976,
      "loss": 4.724,
      "step": 21780
    },
    {
      "epoch": 0.44611226015194644,
      "grad_norm": 0.2600528299808502,
      "learning_rate": 0.00024214553137401364,
      "loss": 4.7666,
      "step": 21785
    },
    {
      "epoch": 0.44621464992935106,
      "grad_norm": 0.2494192272424698,
      "learning_rate": 0.00024209668327707229,
      "loss": 4.7305,
      "step": 21790
    },
    {
      "epoch": 0.4463170397067557,
      "grad_norm": 0.2513991594314575,
      "learning_rate": 0.00024204781949867314,
      "loss": 4.8177,
      "step": 21795
    },
    {
      "epoch": 0.4464194294841603,
      "grad_norm": 0.2659304738044739,
      "learning_rate": 0.00024199894004713632,
      "loss": 4.7779,
      "step": 21800
    },
    {
      "epoch": 0.4465218192615649,
      "grad_norm": 0.2444787472486496,
      "learning_rate": 0.0002419500449307846,
      "loss": 4.7648,
      "step": 21805
    },
    {
      "epoch": 0.44662420903896954,
      "grad_norm": 0.288303405046463,
      "learning_rate": 0.00024190113415794342,
      "loss": 4.7192,
      "step": 21810
    },
    {
      "epoch": 0.44672659881637417,
      "grad_norm": 0.24240879714488983,
      "learning_rate": 0.0002418522077369409,
      "loss": 4.7486,
      "step": 21815
    },
    {
      "epoch": 0.4468289885937788,
      "grad_norm": 0.24538648128509521,
      "learning_rate": 0.00024180326567610792,
      "loss": 4.7531,
      "step": 21820
    },
    {
      "epoch": 0.4469313783711834,
      "grad_norm": 0.24396349489688873,
      "learning_rate": 0.0002417543079837778,
      "loss": 4.7794,
      "step": 21825
    },
    {
      "epoch": 0.44703376814858803,
      "grad_norm": 0.2690068483352661,
      "learning_rate": 0.0002417053346682867,
      "loss": 4.7016,
      "step": 21830
    },
    {
      "epoch": 0.44713615792599265,
      "grad_norm": 0.2747132480144501,
      "learning_rate": 0.00024165634573797336,
      "loss": 4.7161,
      "step": 21835
    },
    {
      "epoch": 0.44723854770339727,
      "grad_norm": 0.26216763257980347,
      "learning_rate": 0.0002416073412011793,
      "loss": 4.7635,
      "step": 21840
    },
    {
      "epoch": 0.4473409374808019,
      "grad_norm": 0.2414170503616333,
      "learning_rate": 0.00024155832106624846,
      "loss": 4.7304,
      "step": 21845
    },
    {
      "epoch": 0.4474433272582065,
      "grad_norm": 0.26282960176467896,
      "learning_rate": 0.00024150928534152772,
      "loss": 4.718,
      "step": 21850
    },
    {
      "epoch": 0.4475457170356112,
      "grad_norm": 0.25752347707748413,
      "learning_rate": 0.00024146023403536637,
      "loss": 4.7169,
      "step": 21855
    },
    {
      "epoch": 0.4476481068130158,
      "grad_norm": 0.26910948753356934,
      "learning_rate": 0.00024141116715611655,
      "loss": 4.7501,
      "step": 21860
    },
    {
      "epoch": 0.44775049659042043,
      "grad_norm": 0.26379042863845825,
      "learning_rate": 0.00024136208471213294,
      "loss": 4.7613,
      "step": 21865
    },
    {
      "epoch": 0.44785288636782505,
      "grad_norm": 0.25984200835227966,
      "learning_rate": 0.00024131298671177282,
      "loss": 4.7327,
      "step": 21870
    },
    {
      "epoch": 0.4479552761452297,
      "grad_norm": 0.24715788662433624,
      "learning_rate": 0.00024126387316339624,
      "loss": 4.7458,
      "step": 21875
    },
    {
      "epoch": 0.4480576659226343,
      "grad_norm": 0.24957576394081116,
      "learning_rate": 0.00024121474407536584,
      "loss": 4.7534,
      "step": 21880
    },
    {
      "epoch": 0.4481600557000389,
      "grad_norm": 0.23778215050697327,
      "learning_rate": 0.0002411655994560469,
      "loss": 4.7317,
      "step": 21885
    },
    {
      "epoch": 0.44826244547744354,
      "grad_norm": 0.24345988035202026,
      "learning_rate": 0.00024111643931380743,
      "loss": 4.7047,
      "step": 21890
    },
    {
      "epoch": 0.44836483525484816,
      "grad_norm": 0.24695883691310883,
      "learning_rate": 0.00024106726365701793,
      "loss": 4.7261,
      "step": 21895
    },
    {
      "epoch": 0.4484672250322528,
      "grad_norm": 0.2646474838256836,
      "learning_rate": 0.00024101807249405163,
      "loss": 4.7856,
      "step": 21900
    },
    {
      "epoch": 0.4485696148096574,
      "grad_norm": 0.49546247720718384,
      "learning_rate": 0.0002409688658332844,
      "loss": 4.8024,
      "step": 21905
    },
    {
      "epoch": 0.448672004587062,
      "grad_norm": 0.3293178081512451,
      "learning_rate": 0.00024091964368309478,
      "loss": 4.747,
      "step": 21910
    },
    {
      "epoch": 0.44877439436446664,
      "grad_norm": 0.26098817586898804,
      "learning_rate": 0.00024087040605186383,
      "loss": 4.7359,
      "step": 21915
    },
    {
      "epoch": 0.44887678414187127,
      "grad_norm": 0.24753005802631378,
      "learning_rate": 0.00024082115294797541,
      "loss": 4.7252,
      "step": 21920
    },
    {
      "epoch": 0.4489791739192759,
      "grad_norm": 0.24103949964046478,
      "learning_rate": 0.00024077188437981588,
      "loss": 4.7508,
      "step": 21925
    },
    {
      "epoch": 0.4490815636966805,
      "grad_norm": 0.2502189576625824,
      "learning_rate": 0.0002407226003557743,
      "loss": 4.8036,
      "step": 21930
    },
    {
      "epoch": 0.44918395347408513,
      "grad_norm": 0.2760987877845764,
      "learning_rate": 0.00024067330088424238,
      "loss": 4.7353,
      "step": 21935
    },
    {
      "epoch": 0.44928634325148975,
      "grad_norm": 0.2822510004043579,
      "learning_rate": 0.00024062398597361433,
      "loss": 4.7491,
      "step": 21940
    },
    {
      "epoch": 0.44938873302889437,
      "grad_norm": 0.24212375283241272,
      "learning_rate": 0.00024057465563228714,
      "loss": 4.7312,
      "step": 21945
    },
    {
      "epoch": 0.449491122806299,
      "grad_norm": 0.25818729400634766,
      "learning_rate": 0.00024052530986866036,
      "loss": 4.7185,
      "step": 21950
    },
    {
      "epoch": 0.44959351258370367,
      "grad_norm": 0.2500952482223511,
      "learning_rate": 0.00024047594869113623,
      "loss": 4.725,
      "step": 21955
    },
    {
      "epoch": 0.4496959023611083,
      "grad_norm": 0.2536247968673706,
      "learning_rate": 0.00024042657210811947,
      "loss": 4.7612,
      "step": 21960
    },
    {
      "epoch": 0.4497982921385129,
      "grad_norm": 0.2786836326122284,
      "learning_rate": 0.00024037718012801757,
      "loss": 4.7726,
      "step": 21965
    },
    {
      "epoch": 0.44990068191591753,
      "grad_norm": 0.2442341446876526,
      "learning_rate": 0.00024032777275924053,
      "loss": 4.7498,
      "step": 21970
    },
    {
      "epoch": 0.45000307169332215,
      "grad_norm": 0.25481003522872925,
      "learning_rate": 0.00024027835001020113,
      "loss": 4.7003,
      "step": 21975
    },
    {
      "epoch": 0.4501054614707268,
      "grad_norm": 0.35789549350738525,
      "learning_rate": 0.00024022891188931462,
      "loss": 4.6755,
      "step": 21980
    },
    {
      "epoch": 0.4502078512481314,
      "grad_norm": 0.2654735743999481,
      "learning_rate": 0.00024017945840499886,
      "loss": 4.7754,
      "step": 21985
    },
    {
      "epoch": 0.450310241025536,
      "grad_norm": 0.24647344648838043,
      "learning_rate": 0.00024012998956567442,
      "loss": 4.7571,
      "step": 21990
    },
    {
      "epoch": 0.45041263080294064,
      "grad_norm": 0.256874680519104,
      "learning_rate": 0.00024008050537976442,
      "loss": 4.7364,
      "step": 21995
    },
    {
      "epoch": 0.45051502058034526,
      "grad_norm": 0.26698172092437744,
      "learning_rate": 0.00024003100585569465,
      "loss": 4.7593,
      "step": 22000
    },
    {
      "epoch": 0.45051502058034526,
      "eval_loss": 4.759491920471191,
      "eval_runtime": 17.2162,
      "eval_samples_per_second": 33.805,
      "eval_steps_per_second": 4.24,
      "step": 22000
    },
    {
      "epoch": 0.4506174103577499,
      "grad_norm": 0.22964538633823395,
      "learning_rate": 0.00023998149100189347,
      "loss": 4.7779,
      "step": 22005
    },
    {
      "epoch": 0.4507198001351545,
      "grad_norm": 0.25614720582962036,
      "learning_rate": 0.00023993196082679183,
      "loss": 4.7512,
      "step": 22010
    },
    {
      "epoch": 0.4508221899125591,
      "grad_norm": 0.2806569039821625,
      "learning_rate": 0.00023988241533882336,
      "loss": 4.7031,
      "step": 22015
    },
    {
      "epoch": 0.45092457968996374,
      "grad_norm": 0.2393021583557129,
      "learning_rate": 0.00023983285454642423,
      "loss": 4.7856,
      "step": 22020
    },
    {
      "epoch": 0.45102696946736837,
      "grad_norm": 0.24871160089969635,
      "learning_rate": 0.0002397832784580332,
      "loss": 4.7386,
      "step": 22025
    },
    {
      "epoch": 0.451129359244773,
      "grad_norm": 0.3182537853717804,
      "learning_rate": 0.00023973368708209172,
      "loss": 4.7502,
      "step": 22030
    },
    {
      "epoch": 0.4512317490221776,
      "grad_norm": 0.2552315592765808,
      "learning_rate": 0.00023968408042704384,
      "loss": 4.7061,
      "step": 22035
    },
    {
      "epoch": 0.45133413879958223,
      "grad_norm": 0.24092653393745422,
      "learning_rate": 0.0002396344585013361,
      "loss": 4.7735,
      "step": 22040
    },
    {
      "epoch": 0.45143652857698685,
      "grad_norm": 0.24381929636001587,
      "learning_rate": 0.00023958482131341774,
      "loss": 4.7371,
      "step": 22045
    },
    {
      "epoch": 0.45153891835439147,
      "grad_norm": 0.25864145159721375,
      "learning_rate": 0.00023953516887174054,
      "loss": 4.7348,
      "step": 22050
    },
    {
      "epoch": 0.45164130813179615,
      "grad_norm": 0.23586764931678772,
      "learning_rate": 0.0002394855011847589,
      "loss": 4.7175,
      "step": 22055
    },
    {
      "epoch": 0.45174369790920077,
      "grad_norm": 0.25393882393836975,
      "learning_rate": 0.0002394358182609299,
      "loss": 4.7234,
      "step": 22060
    },
    {
      "epoch": 0.4518460876866054,
      "grad_norm": 0.23844827711582184,
      "learning_rate": 0.00023938612010871303,
      "loss": 4.7125,
      "step": 22065
    },
    {
      "epoch": 0.45194847746401,
      "grad_norm": 0.2636282742023468,
      "learning_rate": 0.00023933640673657052,
      "loss": 4.74,
      "step": 22070
    },
    {
      "epoch": 0.45205086724141463,
      "grad_norm": 0.23698267340660095,
      "learning_rate": 0.00023928667815296723,
      "loss": 4.7738,
      "step": 22075
    },
    {
      "epoch": 0.45215325701881925,
      "grad_norm": 0.26341021060943604,
      "learning_rate": 0.0002392369343663704,
      "loss": 4.7323,
      "step": 22080
    },
    {
      "epoch": 0.4522556467962239,
      "grad_norm": 0.23851220309734344,
      "learning_rate": 0.00023918717538525002,
      "loss": 4.7549,
      "step": 22085
    },
    {
      "epoch": 0.4523580365736285,
      "grad_norm": 0.2447780817747116,
      "learning_rate": 0.0002391374012180787,
      "loss": 4.7244,
      "step": 22090
    },
    {
      "epoch": 0.4524604263510331,
      "grad_norm": 0.24237054586410522,
      "learning_rate": 0.0002390876118733315,
      "loss": 4.7143,
      "step": 22095
    },
    {
      "epoch": 0.45256281612843774,
      "grad_norm": 0.24432867765426636,
      "learning_rate": 0.0002390378073594862,
      "loss": 4.7231,
      "step": 22100
    },
    {
      "epoch": 0.45266520590584236,
      "grad_norm": 0.2664358913898468,
      "learning_rate": 0.0002389879876850231,
      "loss": 4.7944,
      "step": 22105
    },
    {
      "epoch": 0.452767595683247,
      "grad_norm": 0.2870197594165802,
      "learning_rate": 0.00023893815285842498,
      "loss": 4.7442,
      "step": 22110
    },
    {
      "epoch": 0.4528699854606516,
      "grad_norm": 0.23302584886550903,
      "learning_rate": 0.00023888830288817738,
      "loss": 4.7641,
      "step": 22115
    },
    {
      "epoch": 0.4529723752380562,
      "grad_norm": 0.2441381812095642,
      "learning_rate": 0.00023883843778276838,
      "loss": 4.771,
      "step": 22120
    },
    {
      "epoch": 0.45307476501546085,
      "grad_norm": 0.23622973263263702,
      "learning_rate": 0.00023878855755068846,
      "loss": 4.7065,
      "step": 22125
    },
    {
      "epoch": 0.45317715479286547,
      "grad_norm": 0.24003490805625916,
      "learning_rate": 0.00023873866220043094,
      "loss": 4.71,
      "step": 22130
    },
    {
      "epoch": 0.4532795445702701,
      "grad_norm": 0.234433114528656,
      "learning_rate": 0.00023868875174049158,
      "loss": 4.7812,
      "step": 22135
    },
    {
      "epoch": 0.4533819343476747,
      "grad_norm": 0.24909816682338715,
      "learning_rate": 0.00023863882617936857,
      "loss": 4.7732,
      "step": 22140
    },
    {
      "epoch": 0.45348432412507933,
      "grad_norm": 0.25630414485931396,
      "learning_rate": 0.000238588885525563,
      "loss": 4.703,
      "step": 22145
    },
    {
      "epoch": 0.45358671390248395,
      "grad_norm": 0.27781715989112854,
      "learning_rate": 0.0002385389297875783,
      "loss": 4.7235,
      "step": 22150
    },
    {
      "epoch": 0.4536891036798886,
      "grad_norm": 0.25707268714904785,
      "learning_rate": 0.00023848895897392045,
      "loss": 4.7447,
      "step": 22155
    },
    {
      "epoch": 0.45379149345729325,
      "grad_norm": 0.25520217418670654,
      "learning_rate": 0.00023843897309309812,
      "loss": 4.7422,
      "step": 22160
    },
    {
      "epoch": 0.45389388323469787,
      "grad_norm": 0.25873681902885437,
      "learning_rate": 0.00023838897215362246,
      "loss": 4.7242,
      "step": 22165
    },
    {
      "epoch": 0.4539962730121025,
      "grad_norm": 0.25034627318382263,
      "learning_rate": 0.00023833895616400726,
      "loss": 4.6886,
      "step": 22170
    },
    {
      "epoch": 0.4540986627895071,
      "grad_norm": 0.24327032268047333,
      "learning_rate": 0.0002382889251327688,
      "loss": 4.7031,
      "step": 22175
    },
    {
      "epoch": 0.45420105256691173,
      "grad_norm": 0.3189115524291992,
      "learning_rate": 0.00023823887906842593,
      "loss": 4.744,
      "step": 22180
    },
    {
      "epoch": 0.45430344234431635,
      "grad_norm": 0.25070321559906006,
      "learning_rate": 0.00023818881797950012,
      "loss": 4.778,
      "step": 22185
    },
    {
      "epoch": 0.454405832121721,
      "grad_norm": 0.27552253007888794,
      "learning_rate": 0.00023813874187451531,
      "loss": 4.7648,
      "step": 22190
    },
    {
      "epoch": 0.4545082218991256,
      "grad_norm": 0.24581551551818848,
      "learning_rate": 0.00023808865076199807,
      "loss": 4.7241,
      "step": 22195
    },
    {
      "epoch": 0.4546106116765302,
      "grad_norm": 0.25316470861434937,
      "learning_rate": 0.00023803854465047748,
      "loss": 4.7666,
      "step": 22200
    },
    {
      "epoch": 0.45471300145393484,
      "grad_norm": 0.24150055646896362,
      "learning_rate": 0.00023798842354848522,
      "loss": 4.6931,
      "step": 22205
    },
    {
      "epoch": 0.45481539123133946,
      "grad_norm": 0.2623421251773834,
      "learning_rate": 0.0002379382874645555,
      "loss": 4.7356,
      "step": 22210
    },
    {
      "epoch": 0.4549177810087441,
      "grad_norm": 0.2407127320766449,
      "learning_rate": 0.00023788813640722507,
      "loss": 4.7344,
      "step": 22215
    },
    {
      "epoch": 0.4550201707861487,
      "grad_norm": 0.2920372188091278,
      "learning_rate": 0.00023783797038503314,
      "loss": 4.753,
      "step": 22220
    },
    {
      "epoch": 0.4551225605635533,
      "grad_norm": 0.24697765707969666,
      "learning_rate": 0.0002377877894065217,
      "loss": 4.7418,
      "step": 22225
    },
    {
      "epoch": 0.45522495034095795,
      "grad_norm": 0.28075122833251953,
      "learning_rate": 0.0002377375934802351,
      "loss": 4.6846,
      "step": 22230
    },
    {
      "epoch": 0.45532734011836257,
      "grad_norm": 0.25362035632133484,
      "learning_rate": 0.00023768738261472028,
      "loss": 4.7605,
      "step": 22235
    },
    {
      "epoch": 0.4554297298957672,
      "grad_norm": 0.27194178104400635,
      "learning_rate": 0.00023763715681852674,
      "loss": 4.7013,
      "step": 22240
    },
    {
      "epoch": 0.4555321196731718,
      "grad_norm": 0.3001675009727478,
      "learning_rate": 0.00023758691610020646,
      "loss": 4.7209,
      "step": 22245
    },
    {
      "epoch": 0.45563450945057643,
      "grad_norm": 0.2572694420814514,
      "learning_rate": 0.00023753666046831407,
      "loss": 4.7078,
      "step": 22250
    },
    {
      "epoch": 0.45573689922798105,
      "grad_norm": 0.23649387061595917,
      "learning_rate": 0.00023748638993140669,
      "loss": 4.7168,
      "step": 22255
    },
    {
      "epoch": 0.45583928900538573,
      "grad_norm": 0.25332945585250854,
      "learning_rate": 0.00023743610449804392,
      "loss": 4.7397,
      "step": 22260
    },
    {
      "epoch": 0.45594167878279035,
      "grad_norm": 0.25807368755340576,
      "learning_rate": 0.00023738580417678794,
      "loss": 4.7024,
      "step": 22265
    },
    {
      "epoch": 0.45604406856019497,
      "grad_norm": 0.259543240070343,
      "learning_rate": 0.00023733548897620353,
      "loss": 4.7819,
      "step": 22270
    },
    {
      "epoch": 0.4561464583375996,
      "grad_norm": 0.26395493745803833,
      "learning_rate": 0.00023728515890485794,
      "loss": 4.7047,
      "step": 22275
    },
    {
      "epoch": 0.4562488481150042,
      "grad_norm": 0.9881982207298279,
      "learning_rate": 0.00023723481397132086,
      "loss": 4.6476,
      "step": 22280
    },
    {
      "epoch": 0.45635123789240883,
      "grad_norm": 0.24993281066417694,
      "learning_rate": 0.00023718445418416474,
      "loss": 4.712,
      "step": 22285
    },
    {
      "epoch": 0.45645362766981346,
      "grad_norm": 0.27237850427627563,
      "learning_rate": 0.00023713407955196436,
      "loss": 4.6818,
      "step": 22290
    },
    {
      "epoch": 0.4565560174472181,
      "grad_norm": 0.3099258542060852,
      "learning_rate": 0.00023708369008329703,
      "loss": 4.7796,
      "step": 22295
    },
    {
      "epoch": 0.4566584072246227,
      "grad_norm": 0.2862955331802368,
      "learning_rate": 0.00023703328578674281,
      "loss": 4.7674,
      "step": 22300
    },
    {
      "epoch": 0.4567607970020273,
      "grad_norm": 0.2516242563724518,
      "learning_rate": 0.00023698286667088393,
      "loss": 4.7262,
      "step": 22305
    },
    {
      "epoch": 0.45686318677943194,
      "grad_norm": 0.24596908688545227,
      "learning_rate": 0.00023693243274430553,
      "loss": 4.744,
      "step": 22310
    },
    {
      "epoch": 0.45696557655683656,
      "grad_norm": 0.2882581651210785,
      "learning_rate": 0.00023688198401559496,
      "loss": 4.7209,
      "step": 22315
    },
    {
      "epoch": 0.4570679663342412,
      "grad_norm": 0.2525341808795929,
      "learning_rate": 0.00023683152049334218,
      "loss": 4.7341,
      "step": 22320
    },
    {
      "epoch": 0.4571703561116458,
      "grad_norm": 0.24643680453300476,
      "learning_rate": 0.0002367810421861398,
      "loss": 4.7168,
      "step": 22325
    },
    {
      "epoch": 0.4572727458890504,
      "grad_norm": 0.25516244769096375,
      "learning_rate": 0.0002367305491025828,
      "loss": 4.7043,
      "step": 22330
    },
    {
      "epoch": 0.45737513566645505,
      "grad_norm": 0.24852685630321503,
      "learning_rate": 0.00023668004125126875,
      "loss": 4.7319,
      "step": 22335
    },
    {
      "epoch": 0.45747752544385967,
      "grad_norm": 0.26832687854766846,
      "learning_rate": 0.00023662951864079765,
      "loss": 4.7021,
      "step": 22340
    },
    {
      "epoch": 0.4575799152212643,
      "grad_norm": 0.25500577688217163,
      "learning_rate": 0.00023657898127977218,
      "loss": 4.7295,
      "step": 22345
    },
    {
      "epoch": 0.4576823049986689,
      "grad_norm": 0.253725528717041,
      "learning_rate": 0.00023652842917679727,
      "loss": 4.7624,
      "step": 22350
    },
    {
      "epoch": 0.45778469477607353,
      "grad_norm": 0.24359218776226044,
      "learning_rate": 0.00023647786234048062,
      "loss": 4.7484,
      "step": 22355
    },
    {
      "epoch": 0.4578870845534782,
      "grad_norm": 0.2668800354003906,
      "learning_rate": 0.00023642728077943228,
      "loss": 4.7343,
      "step": 22360
    },
    {
      "epoch": 0.45798947433088283,
      "grad_norm": 0.2451476752758026,
      "learning_rate": 0.0002363766845022649,
      "loss": 4.7147,
      "step": 22365
    },
    {
      "epoch": 0.45809186410828745,
      "grad_norm": 0.26905909180641174,
      "learning_rate": 0.0002363260735175936,
      "loss": 4.7641,
      "step": 22370
    },
    {
      "epoch": 0.45819425388569207,
      "grad_norm": 0.2415500432252884,
      "learning_rate": 0.0002362754478340359,
      "loss": 4.7479,
      "step": 22375
    },
    {
      "epoch": 0.4582966436630967,
      "grad_norm": 0.2391519844532013,
      "learning_rate": 0.000236224807460212,
      "loss": 4.7368,
      "step": 22380
    },
    {
      "epoch": 0.4583990334405013,
      "grad_norm": 0.2539454996585846,
      "learning_rate": 0.00023617415240474452,
      "loss": 4.6922,
      "step": 22385
    },
    {
      "epoch": 0.45850142321790593,
      "grad_norm": 0.5625364780426025,
      "learning_rate": 0.00023612348267625853,
      "loss": 4.7439,
      "step": 22390
    },
    {
      "epoch": 0.45860381299531056,
      "grad_norm": 0.23095041513442993,
      "learning_rate": 0.00023607279828338177,
      "loss": 4.6718,
      "step": 22395
    },
    {
      "epoch": 0.4587062027727152,
      "grad_norm": 0.26877298951148987,
      "learning_rate": 0.00023602209923474427,
      "loss": 4.7454,
      "step": 22400
    },
    {
      "epoch": 0.4588085925501198,
      "grad_norm": 0.2572716474533081,
      "learning_rate": 0.0002359713855389786,
      "loss": 4.656,
      "step": 22405
    },
    {
      "epoch": 0.4589109823275244,
      "grad_norm": 0.2368837147951126,
      "learning_rate": 0.00023592065720471996,
      "loss": 4.7213,
      "step": 22410
    },
    {
      "epoch": 0.45901337210492904,
      "grad_norm": 0.26700299978256226,
      "learning_rate": 0.00023586991424060592,
      "loss": 4.7123,
      "step": 22415
    },
    {
      "epoch": 0.45911576188233366,
      "grad_norm": 0.260018914937973,
      "learning_rate": 0.00023581915665527652,
      "loss": 4.7315,
      "step": 22420
    },
    {
      "epoch": 0.4592181516597383,
      "grad_norm": 0.26031601428985596,
      "learning_rate": 0.0002357683844573744,
      "loss": 4.7261,
      "step": 22425
    },
    {
      "epoch": 0.4593205414371429,
      "grad_norm": 0.24976696074008942,
      "learning_rate": 0.00023571759765554464,
      "loss": 4.7519,
      "step": 22430
    },
    {
      "epoch": 0.4594229312145475,
      "grad_norm": 0.25443023443222046,
      "learning_rate": 0.0002356667962584347,
      "loss": 4.7378,
      "step": 22435
    },
    {
      "epoch": 0.45952532099195215,
      "grad_norm": 0.24392306804656982,
      "learning_rate": 0.00023561598027469472,
      "loss": 4.7262,
      "step": 22440
    },
    {
      "epoch": 0.45962771076935677,
      "grad_norm": 0.2588663399219513,
      "learning_rate": 0.00023556514971297712,
      "loss": 4.7743,
      "step": 22445
    },
    {
      "epoch": 0.4597301005467614,
      "grad_norm": 0.24382399022579193,
      "learning_rate": 0.00023551430458193705,
      "loss": 4.7749,
      "step": 22450
    },
    {
      "epoch": 0.459832490324166,
      "grad_norm": 0.25846609473228455,
      "learning_rate": 0.0002354634448902319,
      "loss": 4.7296,
      "step": 22455
    },
    {
      "epoch": 0.4599348801015707,
      "grad_norm": 0.2652266323566437,
      "learning_rate": 0.0002354125706465216,
      "loss": 4.7466,
      "step": 22460
    },
    {
      "epoch": 0.4600372698789753,
      "grad_norm": 0.24637307226657867,
      "learning_rate": 0.0002353616818594687,
      "loss": 4.7446,
      "step": 22465
    },
    {
      "epoch": 0.46013965965637993,
      "grad_norm": 0.2609288692474365,
      "learning_rate": 0.00023531077853773804,
      "loss": 4.7121,
      "step": 22470
    },
    {
      "epoch": 0.46024204943378455,
      "grad_norm": 0.5267515778541565,
      "learning_rate": 0.00023525986068999706,
      "loss": 4.7199,
      "step": 22475
    },
    {
      "epoch": 0.46034443921118917,
      "grad_norm": 0.23161795735359192,
      "learning_rate": 0.00023520892832491557,
      "loss": 4.7501,
      "step": 22480
    },
    {
      "epoch": 0.4604468289885938,
      "grad_norm": 0.2667028605937958,
      "learning_rate": 0.00023515798145116597,
      "loss": 4.76,
      "step": 22485
    },
    {
      "epoch": 0.4605492187659984,
      "grad_norm": 0.2522803246974945,
      "learning_rate": 0.00023510702007742304,
      "loss": 4.7654,
      "step": 22490
    },
    {
      "epoch": 0.46065160854340303,
      "grad_norm": 0.28176870942115784,
      "learning_rate": 0.00023505604421236408,
      "loss": 4.7748,
      "step": 22495
    },
    {
      "epoch": 0.46075399832080766,
      "grad_norm": 0.27334338426589966,
      "learning_rate": 0.00023500505386466885,
      "loss": 4.7205,
      "step": 22500
    },
    {
      "epoch": 0.4608563880982123,
      "grad_norm": 0.2692910134792328,
      "learning_rate": 0.00023495404904301946,
      "loss": 4.7661,
      "step": 22505
    },
    {
      "epoch": 0.4609587778756169,
      "grad_norm": 0.24851375818252563,
      "learning_rate": 0.0002349030297561007,
      "loss": 4.7542,
      "step": 22510
    },
    {
      "epoch": 0.4610611676530215,
      "grad_norm": 0.25268369913101196,
      "learning_rate": 0.00023485199601259966,
      "loss": 4.7673,
      "step": 22515
    },
    {
      "epoch": 0.46116355743042614,
      "grad_norm": 0.24291718006134033,
      "learning_rate": 0.00023480094782120598,
      "loss": 4.7623,
      "step": 22520
    },
    {
      "epoch": 0.46126594720783076,
      "grad_norm": 0.2530502378940582,
      "learning_rate": 0.0002347498851906117,
      "loss": 4.7235,
      "step": 22525
    },
    {
      "epoch": 0.4613683369852354,
      "grad_norm": 0.2749159634113312,
      "learning_rate": 0.00023469880812951132,
      "loss": 4.7397,
      "step": 22530
    },
    {
      "epoch": 0.46147072676264,
      "grad_norm": 0.27258047461509705,
      "learning_rate": 0.00023464771664660183,
      "loss": 4.7494,
      "step": 22535
    },
    {
      "epoch": 0.4615731165400446,
      "grad_norm": 0.2481284886598587,
      "learning_rate": 0.00023459661075058268,
      "loss": 4.7916,
      "step": 22540
    },
    {
      "epoch": 0.46167550631744925,
      "grad_norm": 0.25946375727653503,
      "learning_rate": 0.0002345454904501557,
      "loss": 4.739,
      "step": 22545
    },
    {
      "epoch": 0.46177789609485387,
      "grad_norm": 0.25661179423332214,
      "learning_rate": 0.00023449435575402533,
      "loss": 4.695,
      "step": 22550
    },
    {
      "epoch": 0.4618802858722585,
      "grad_norm": 0.24974200129508972,
      "learning_rate": 0.00023444320667089828,
      "loss": 4.7466,
      "step": 22555
    },
    {
      "epoch": 0.4619826756496631,
      "grad_norm": 0.2404831200838089,
      "learning_rate": 0.0002343920432094838,
      "loss": 4.6757,
      "step": 22560
    },
    {
      "epoch": 0.4620850654270678,
      "grad_norm": 0.2554483115673065,
      "learning_rate": 0.00023434086537849365,
      "loss": 4.7205,
      "step": 22565
    },
    {
      "epoch": 0.4621874552044724,
      "grad_norm": 0.24611206352710724,
      "learning_rate": 0.00023428967318664187,
      "loss": 4.7079,
      "step": 22570
    },
    {
      "epoch": 0.46228984498187703,
      "grad_norm": 0.24876458942890167,
      "learning_rate": 0.00023423846664264507,
      "loss": 4.7568,
      "step": 22575
    },
    {
      "epoch": 0.46239223475928165,
      "grad_norm": 0.26299282908439636,
      "learning_rate": 0.00023418724575522236,
      "loss": 4.6751,
      "step": 22580
    },
    {
      "epoch": 0.46249462453668627,
      "grad_norm": 0.2555292844772339,
      "learning_rate": 0.00023413601053309506,
      "loss": 4.6836,
      "step": 22585
    },
    {
      "epoch": 0.4625970143140909,
      "grad_norm": 0.2545802891254425,
      "learning_rate": 0.00023408476098498717,
      "loss": 4.6831,
      "step": 22590
    },
    {
      "epoch": 0.4626994040914955,
      "grad_norm": 0.26119187474250793,
      "learning_rate": 0.000234033497119625,
      "loss": 4.7271,
      "step": 22595
    },
    {
      "epoch": 0.46280179386890014,
      "grad_norm": 0.2618061900138855,
      "learning_rate": 0.00023398221894573736,
      "loss": 4.752,
      "step": 22600
    },
    {
      "epoch": 0.46290418364630476,
      "grad_norm": 0.2994387447834015,
      "learning_rate": 0.0002339309264720555,
      "loss": 4.7242,
      "step": 22605
    },
    {
      "epoch": 0.4630065734237094,
      "grad_norm": 0.2792063355445862,
      "learning_rate": 0.00023387961970731301,
      "loss": 4.7183,
      "step": 22610
    },
    {
      "epoch": 0.463108963201114,
      "grad_norm": 1.7313650846481323,
      "learning_rate": 0.00023382829866024603,
      "loss": 4.6868,
      "step": 22615
    },
    {
      "epoch": 0.4632113529785186,
      "grad_norm": 0.25411492586135864,
      "learning_rate": 0.00023377696333959304,
      "loss": 4.7746,
      "step": 22620
    },
    {
      "epoch": 0.46331374275592324,
      "grad_norm": 0.23991341888904572,
      "learning_rate": 0.00023372561375409508,
      "loss": 4.7603,
      "step": 22625
    },
    {
      "epoch": 0.46341613253332786,
      "grad_norm": 0.2458183765411377,
      "learning_rate": 0.00023367424991249536,
      "loss": 4.7253,
      "step": 22630
    },
    {
      "epoch": 0.4635185223107325,
      "grad_norm": 0.26310741901397705,
      "learning_rate": 0.00023362287182353983,
      "loss": 4.7386,
      "step": 22635
    },
    {
      "epoch": 0.4636209120881371,
      "grad_norm": 0.2550501525402069,
      "learning_rate": 0.0002335714794959767,
      "loss": 4.7315,
      "step": 22640
    },
    {
      "epoch": 0.4637233018655417,
      "grad_norm": 0.25134924054145813,
      "learning_rate": 0.0002335200729385566,
      "loss": 4.7268,
      "step": 22645
    },
    {
      "epoch": 0.46382569164294635,
      "grad_norm": 0.24425670504570007,
      "learning_rate": 0.00023346865216003268,
      "loss": 4.7531,
      "step": 22650
    },
    {
      "epoch": 0.46392808142035097,
      "grad_norm": 0.25438258051872253,
      "learning_rate": 0.0002334172171691603,
      "loss": 4.7171,
      "step": 22655
    },
    {
      "epoch": 0.4640304711977556,
      "grad_norm": 0.2890762984752655,
      "learning_rate": 0.00023336576797469753,
      "loss": 4.6838,
      "step": 22660
    },
    {
      "epoch": 0.46413286097516027,
      "grad_norm": 0.23798921704292297,
      "learning_rate": 0.00023331430458540465,
      "loss": 4.7094,
      "step": 22665
    },
    {
      "epoch": 0.4642352507525649,
      "grad_norm": 0.3062776029109955,
      "learning_rate": 0.0002332628270100444,
      "loss": 4.7395,
      "step": 22670
    },
    {
      "epoch": 0.4643376405299695,
      "grad_norm": 0.24925167858600616,
      "learning_rate": 0.00023321133525738198,
      "loss": 4.7799,
      "step": 22675
    },
    {
      "epoch": 0.46444003030737413,
      "grad_norm": 0.2531532943248749,
      "learning_rate": 0.00023315982933618497,
      "loss": 4.7545,
      "step": 22680
    },
    {
      "epoch": 0.46454242008477875,
      "grad_norm": 0.24798624217510223,
      "learning_rate": 0.00023310830925522336,
      "loss": 4.7217,
      "step": 22685
    },
    {
      "epoch": 0.46464480986218337,
      "grad_norm": 0.2451784610748291,
      "learning_rate": 0.0002330567750232696,
      "loss": 4.7494,
      "step": 22690
    },
    {
      "epoch": 0.464747199639588,
      "grad_norm": 0.26014596223831177,
      "learning_rate": 0.0002330052266490985,
      "loss": 4.7399,
      "step": 22695
    },
    {
      "epoch": 0.4648495894169926,
      "grad_norm": 0.24459093809127808,
      "learning_rate": 0.00023295366414148724,
      "loss": 4.7729,
      "step": 22700
    },
    {
      "epoch": 0.46495197919439724,
      "grad_norm": 0.2558246850967407,
      "learning_rate": 0.00023290208750921545,
      "loss": 4.7228,
      "step": 22705
    },
    {
      "epoch": 0.46505436897180186,
      "grad_norm": 0.24167120456695557,
      "learning_rate": 0.0002328504967610653,
      "loss": 4.6931,
      "step": 22710
    },
    {
      "epoch": 0.4651567587492065,
      "grad_norm": 0.23493839800357819,
      "learning_rate": 0.00023279889190582108,
      "loss": 4.704,
      "step": 22715
    },
    {
      "epoch": 0.4652591485266111,
      "grad_norm": 0.25040683150291443,
      "learning_rate": 0.00023274727295226973,
      "loss": 4.7672,
      "step": 22720
    },
    {
      "epoch": 0.4653615383040157,
      "grad_norm": 0.25645115971565247,
      "learning_rate": 0.00023269563990920044,
      "loss": 4.6998,
      "step": 22725
    },
    {
      "epoch": 0.46546392808142034,
      "grad_norm": 0.27187323570251465,
      "learning_rate": 0.00023264399278540487,
      "loss": 4.7217,
      "step": 22730
    },
    {
      "epoch": 0.46556631785882496,
      "grad_norm": 0.25805917382240295,
      "learning_rate": 0.00023259233158967713,
      "loss": 4.7406,
      "step": 22735
    },
    {
      "epoch": 0.4656687076362296,
      "grad_norm": 0.24273861944675446,
      "learning_rate": 0.0002325406563308135,
      "loss": 4.7488,
      "step": 22740
    },
    {
      "epoch": 0.4657710974136342,
      "grad_norm": 0.24475401639938354,
      "learning_rate": 0.00023248896701761303,
      "loss": 4.7654,
      "step": 22745
    },
    {
      "epoch": 0.4658734871910388,
      "grad_norm": 0.23977327346801758,
      "learning_rate": 0.00023243726365887677,
      "loss": 4.7662,
      "step": 22750
    },
    {
      "epoch": 0.46597587696844345,
      "grad_norm": 0.2601017355918884,
      "learning_rate": 0.00023238554626340841,
      "loss": 4.7241,
      "step": 22755
    },
    {
      "epoch": 0.46607826674584807,
      "grad_norm": 0.2603398859500885,
      "learning_rate": 0.00023233381484001394,
      "loss": 4.7117,
      "step": 22760
    },
    {
      "epoch": 0.46618065652325275,
      "grad_norm": 0.6911476254463196,
      "learning_rate": 0.0002322820693975018,
      "loss": 4.789,
      "step": 22765
    },
    {
      "epoch": 0.46628304630065737,
      "grad_norm": 0.2640589773654938,
      "learning_rate": 0.0002322303099446827,
      "loss": 4.7677,
      "step": 22770
    },
    {
      "epoch": 0.466385436078062,
      "grad_norm": 0.2652299702167511,
      "learning_rate": 0.00023217853649036988,
      "loss": 4.6717,
      "step": 22775
    },
    {
      "epoch": 0.4664878258554666,
      "grad_norm": 0.2788437604904175,
      "learning_rate": 0.00023212674904337887,
      "loss": 4.7015,
      "step": 22780
    },
    {
      "epoch": 0.46659021563287123,
      "grad_norm": 0.24584704637527466,
      "learning_rate": 0.00023207494761252754,
      "loss": 4.7687,
      "step": 22785
    },
    {
      "epoch": 0.46669260541027585,
      "grad_norm": 0.22849427163600922,
      "learning_rate": 0.00023202313220663633,
      "loss": 4.6997,
      "step": 22790
    },
    {
      "epoch": 0.4667949951876805,
      "grad_norm": 0.2581017017364502,
      "learning_rate": 0.0002319713028345278,
      "loss": 4.763,
      "step": 22795
    },
    {
      "epoch": 0.4668973849650851,
      "grad_norm": 0.25005319714546204,
      "learning_rate": 0.00023191945950502718,
      "loss": 4.7233,
      "step": 22800
    },
    {
      "epoch": 0.4669997747424897,
      "grad_norm": 0.24875850975513458,
      "learning_rate": 0.0002318676022269618,
      "loss": 4.7219,
      "step": 22805
    },
    {
      "epoch": 0.46710216451989434,
      "grad_norm": 0.233003169298172,
      "learning_rate": 0.0002318157310091615,
      "loss": 4.748,
      "step": 22810
    },
    {
      "epoch": 0.46720455429729896,
      "grad_norm": 0.253690242767334,
      "learning_rate": 0.00023176384586045854,
      "loss": 4.7961,
      "step": 22815
    },
    {
      "epoch": 0.4673069440747036,
      "grad_norm": 0.2438765913248062,
      "learning_rate": 0.00023171194678968743,
      "loss": 4.7664,
      "step": 22820
    },
    {
      "epoch": 0.4674093338521082,
      "grad_norm": 0.25183239579200745,
      "learning_rate": 0.00023166003380568515,
      "loss": 4.7098,
      "step": 22825
    },
    {
      "epoch": 0.4675117236295128,
      "grad_norm": 0.24586036801338196,
      "learning_rate": 0.00023160810691729103,
      "loss": 4.7102,
      "step": 22830
    },
    {
      "epoch": 0.46761411340691744,
      "grad_norm": 0.3420575261116028,
      "learning_rate": 0.00023155616613334672,
      "loss": 4.7272,
      "step": 22835
    },
    {
      "epoch": 0.46771650318432206,
      "grad_norm": 0.2641162574291229,
      "learning_rate": 0.00023150421146269617,
      "loss": 4.7353,
      "step": 22840
    },
    {
      "epoch": 0.4678188929617267,
      "grad_norm": 0.25523972511291504,
      "learning_rate": 0.000231452242914186,
      "loss": 4.7731,
      "step": 22845
    },
    {
      "epoch": 0.4679212827391313,
      "grad_norm": 0.25930434465408325,
      "learning_rate": 0.0002314002604966648,
      "loss": 4.7182,
      "step": 22850
    },
    {
      "epoch": 0.4680236725165359,
      "grad_norm": 0.2686781883239746,
      "learning_rate": 0.00023134826421898375,
      "loss": 4.792,
      "step": 22855
    },
    {
      "epoch": 0.46812606229394055,
      "grad_norm": 0.2687591314315796,
      "learning_rate": 0.00023129625408999644,
      "loss": 4.7289,
      "step": 22860
    },
    {
      "epoch": 0.46822845207134517,
      "grad_norm": 0.24362003803253174,
      "learning_rate": 0.0002312442301185586,
      "loss": 4.7025,
      "step": 22865
    },
    {
      "epoch": 0.46833084184874985,
      "grad_norm": 0.25301697850227356,
      "learning_rate": 0.00023119219231352848,
      "loss": 4.7109,
      "step": 22870
    },
    {
      "epoch": 0.46843323162615447,
      "grad_norm": 0.2592464089393616,
      "learning_rate": 0.00023114014068376668,
      "loss": 4.7695,
      "step": 22875
    },
    {
      "epoch": 0.4685356214035591,
      "grad_norm": 0.2599638104438782,
      "learning_rate": 0.0002310880752381361,
      "loss": 4.6963,
      "step": 22880
    },
    {
      "epoch": 0.4686380111809637,
      "grad_norm": 0.25477394461631775,
      "learning_rate": 0.00023103599598550196,
      "loss": 4.7334,
      "step": 22885
    },
    {
      "epoch": 0.46874040095836833,
      "grad_norm": 0.2755983769893646,
      "learning_rate": 0.00023098390293473196,
      "loss": 4.7255,
      "step": 22890
    },
    {
      "epoch": 0.46884279073577295,
      "grad_norm": 0.26748162508010864,
      "learning_rate": 0.000230931796094696,
      "loss": 4.7001,
      "step": 22895
    },
    {
      "epoch": 0.4689451805131776,
      "grad_norm": 0.278424471616745,
      "learning_rate": 0.00023087967547426644,
      "loss": 4.7336,
      "step": 22900
    },
    {
      "epoch": 0.4690475702905822,
      "grad_norm": 0.2655482590198517,
      "learning_rate": 0.00023082754108231797,
      "loss": 4.7248,
      "step": 22905
    },
    {
      "epoch": 0.4691499600679868,
      "grad_norm": 0.2636497914791107,
      "learning_rate": 0.00023077539292772753,
      "loss": 4.7207,
      "step": 22910
    },
    {
      "epoch": 0.46925234984539144,
      "grad_norm": 0.27643537521362305,
      "learning_rate": 0.00023072323101937456,
      "loss": 4.772,
      "step": 22915
    },
    {
      "epoch": 0.46935473962279606,
      "grad_norm": 0.28624022006988525,
      "learning_rate": 0.0002306710553661407,
      "loss": 4.7101,
      "step": 22920
    },
    {
      "epoch": 0.4694571294002007,
      "grad_norm": 0.25350862741470337,
      "learning_rate": 0.00023061886597691,
      "loss": 4.715,
      "step": 22925
    },
    {
      "epoch": 0.4695595191776053,
      "grad_norm": 0.2603566646575928,
      "learning_rate": 0.00023056666286056884,
      "loss": 4.7286,
      "step": 22930
    },
    {
      "epoch": 0.4696619089550099,
      "grad_norm": 0.2457352727651596,
      "learning_rate": 0.0002305144460260059,
      "loss": 4.7614,
      "step": 22935
    },
    {
      "epoch": 0.46976429873241454,
      "grad_norm": 0.24755015969276428,
      "learning_rate": 0.0002304622154821123,
      "loss": 4.7627,
      "step": 22940
    },
    {
      "epoch": 0.46986668850981916,
      "grad_norm": 0.24351108074188232,
      "learning_rate": 0.0002304099712377813,
      "loss": 4.7481,
      "step": 22945
    },
    {
      "epoch": 0.4699690782872238,
      "grad_norm": 0.2639317810535431,
      "learning_rate": 0.00023035771330190877,
      "loss": 4.6929,
      "step": 22950
    },
    {
      "epoch": 0.4700714680646284,
      "grad_norm": 0.25225552916526794,
      "learning_rate": 0.00023030544168339264,
      "loss": 4.7569,
      "step": 22955
    },
    {
      "epoch": 0.470173857842033,
      "grad_norm": 0.27675190567970276,
      "learning_rate": 0.0002302531563911334,
      "loss": 4.7458,
      "step": 22960
    },
    {
      "epoch": 0.47027624761943765,
      "grad_norm": 0.2610032856464386,
      "learning_rate": 0.0002302008574340336,
      "loss": 4.7324,
      "step": 22965
    },
    {
      "epoch": 0.4703786373968423,
      "grad_norm": 0.24219560623168945,
      "learning_rate": 0.00023014854482099838,
      "loss": 4.7042,
      "step": 22970
    },
    {
      "epoch": 0.47048102717424695,
      "grad_norm": 0.25723206996917725,
      "learning_rate": 0.00023009621856093508,
      "loss": 4.762,
      "step": 22975
    },
    {
      "epoch": 0.47058341695165157,
      "grad_norm": 0.2543880343437195,
      "learning_rate": 0.00023004387866275338,
      "loss": 4.742,
      "step": 22980
    },
    {
      "epoch": 0.4706858067290562,
      "grad_norm": 0.2550751268863678,
      "learning_rate": 0.0002299915251353653,
      "loss": 4.7214,
      "step": 22985
    },
    {
      "epoch": 0.4707881965064608,
      "grad_norm": 0.23944367468357086,
      "learning_rate": 0.0002299391579876851,
      "loss": 4.7017,
      "step": 22990
    },
    {
      "epoch": 0.47089058628386543,
      "grad_norm": 0.5354617834091187,
      "learning_rate": 0.00022988677722862944,
      "loss": 4.6972,
      "step": 22995
    },
    {
      "epoch": 0.47099297606127005,
      "grad_norm": 0.24720068275928497,
      "learning_rate": 0.00022983438286711735,
      "loss": 4.7146,
      "step": 23000
    },
    {
      "epoch": 0.47099297606127005,
      "eval_loss": 4.7486653327941895,
      "eval_runtime": 31.0651,
      "eval_samples_per_second": 18.735,
      "eval_steps_per_second": 2.35,
      "step": 23000
    },
    {
      "epoch": 0.4710953658386747,
      "grad_norm": 0.24126434326171875,
      "learning_rate": 0.00022978197491207,
      "loss": 4.7339,
      "step": 23005
    },
    {
      "epoch": 0.4711977556160793,
      "grad_norm": 0.2385326325893402,
      "learning_rate": 0.00022972955337241112,
      "loss": 4.7227,
      "step": 23010
    },
    {
      "epoch": 0.4713001453934839,
      "grad_norm": 0.2848736047744751,
      "learning_rate": 0.00022967711825706648,
      "loss": 4.7474,
      "step": 23015
    },
    {
      "epoch": 0.47140253517088854,
      "grad_norm": 0.26521795988082886,
      "learning_rate": 0.00022962466957496437,
      "loss": 4.738,
      "step": 23020
    },
    {
      "epoch": 0.47150492494829316,
      "grad_norm": 0.24742761254310608,
      "learning_rate": 0.0002295722073350353,
      "loss": 4.6882,
      "step": 23025
    },
    {
      "epoch": 0.4716073147256978,
      "grad_norm": 0.2614652216434479,
      "learning_rate": 0.00022951973154621208,
      "loss": 4.7519,
      "step": 23030
    },
    {
      "epoch": 0.4717097045031024,
      "grad_norm": 0.24566341936588287,
      "learning_rate": 0.0002294672422174299,
      "loss": 4.6895,
      "step": 23035
    },
    {
      "epoch": 0.471812094280507,
      "grad_norm": 0.25319018959999084,
      "learning_rate": 0.00022941473935762612,
      "loss": 4.7537,
      "step": 23040
    },
    {
      "epoch": 0.47191448405791164,
      "grad_norm": 0.24296757578849792,
      "learning_rate": 0.0002293622229757406,
      "loss": 4.7501,
      "step": 23045
    },
    {
      "epoch": 0.47201687383531626,
      "grad_norm": 0.27099448442459106,
      "learning_rate": 0.00022930969308071532,
      "loss": 4.706,
      "step": 23050
    },
    {
      "epoch": 0.4721192636127209,
      "grad_norm": 0.2669123113155365,
      "learning_rate": 0.0002292571496814947,
      "loss": 4.7285,
      "step": 23055
    },
    {
      "epoch": 0.4722216533901255,
      "grad_norm": 0.26060131192207336,
      "learning_rate": 0.00022920459278702527,
      "loss": 4.7548,
      "step": 23060
    },
    {
      "epoch": 0.4723240431675301,
      "grad_norm": 0.2891134023666382,
      "learning_rate": 0.0002291520224062561,
      "loss": 4.6762,
      "step": 23065
    },
    {
      "epoch": 0.4724264329449348,
      "grad_norm": 0.246364563703537,
      "learning_rate": 0.0002290994385481384,
      "loss": 4.6962,
      "step": 23070
    },
    {
      "epoch": 0.4725288227223394,
      "grad_norm": 0.26051563024520874,
      "learning_rate": 0.0002290468412216257,
      "loss": 4.7511,
      "step": 23075
    },
    {
      "epoch": 0.47263121249974405,
      "grad_norm": 0.2498205602169037,
      "learning_rate": 0.00022899423043567385,
      "loss": 4.7198,
      "step": 23080
    },
    {
      "epoch": 0.47273360227714867,
      "grad_norm": 0.25932613015174866,
      "learning_rate": 0.00022894160619924095,
      "loss": 4.7391,
      "step": 23085
    },
    {
      "epoch": 0.4728359920545533,
      "grad_norm": 0.24442742764949799,
      "learning_rate": 0.00022888896852128748,
      "loss": 4.7416,
      "step": 23090
    },
    {
      "epoch": 0.4729383818319579,
      "grad_norm": 0.23264530301094055,
      "learning_rate": 0.0002288363174107761,
      "loss": 4.7422,
      "step": 23095
    },
    {
      "epoch": 0.47304077160936253,
      "grad_norm": 0.24659916758537292,
      "learning_rate": 0.00022878365287667181,
      "loss": 4.7565,
      "step": 23100
    },
    {
      "epoch": 0.47314316138676715,
      "grad_norm": 0.2867399752140045,
      "learning_rate": 0.00022873097492794188,
      "loss": 4.7029,
      "step": 23105
    },
    {
      "epoch": 0.4732455511641718,
      "grad_norm": 3.2085673809051514,
      "learning_rate": 0.00022867828357355587,
      "loss": 4.7253,
      "step": 23110
    },
    {
      "epoch": 0.4733479409415764,
      "grad_norm": 0.26261451840400696,
      "learning_rate": 0.00022862557882248562,
      "loss": 4.732,
      "step": 23115
    },
    {
      "epoch": 0.473450330718981,
      "grad_norm": 0.24705669283866882,
      "learning_rate": 0.00022857286068370532,
      "loss": 4.7413,
      "step": 23120
    },
    {
      "epoch": 0.47355272049638564,
      "grad_norm": 0.24625122547149658,
      "learning_rate": 0.00022852012916619135,
      "loss": 4.6805,
      "step": 23125
    },
    {
      "epoch": 0.47365511027379026,
      "grad_norm": 0.24632865190505981,
      "learning_rate": 0.00022846738427892232,
      "loss": 4.7188,
      "step": 23130
    },
    {
      "epoch": 0.4737575000511949,
      "grad_norm": 0.2599443197250366,
      "learning_rate": 0.00022841462603087923,
      "loss": 4.7463,
      "step": 23135
    },
    {
      "epoch": 0.4738598898285995,
      "grad_norm": 0.2681810259819031,
      "learning_rate": 0.0002283618544310454,
      "loss": 4.6996,
      "step": 23140
    },
    {
      "epoch": 0.4739622796060041,
      "grad_norm": 0.24931111931800842,
      "learning_rate": 0.00022830906948840624,
      "loss": 4.754,
      "step": 23145
    },
    {
      "epoch": 0.47406466938340874,
      "grad_norm": 0.2588179409503937,
      "learning_rate": 0.00022825627121194955,
      "loss": 4.7328,
      "step": 23150
    },
    {
      "epoch": 0.47416705916081336,
      "grad_norm": 0.2509627938270569,
      "learning_rate": 0.00022820345961066543,
      "loss": 4.7174,
      "step": 23155
    },
    {
      "epoch": 0.474269448938218,
      "grad_norm": 0.2449938803911209,
      "learning_rate": 0.0002281506346935461,
      "loss": 4.7319,
      "step": 23160
    },
    {
      "epoch": 0.4743718387156226,
      "grad_norm": 0.24512113630771637,
      "learning_rate": 0.0002280977964695863,
      "loss": 4.7549,
      "step": 23165
    },
    {
      "epoch": 0.4744742284930273,
      "grad_norm": 0.2512456476688385,
      "learning_rate": 0.00022804494494778276,
      "loss": 4.7605,
      "step": 23170
    },
    {
      "epoch": 0.4745766182704319,
      "grad_norm": 0.24117611348628998,
      "learning_rate": 0.00022799208013713466,
      "loss": 4.8048,
      "step": 23175
    },
    {
      "epoch": 0.4746790080478365,
      "grad_norm": 0.2468961775302887,
      "learning_rate": 0.0002279392020466433,
      "loss": 4.7062,
      "step": 23180
    },
    {
      "epoch": 0.47478139782524115,
      "grad_norm": 0.24405434727668762,
      "learning_rate": 0.00022788631068531244,
      "loss": 4.742,
      "step": 23185
    },
    {
      "epoch": 0.47488378760264577,
      "grad_norm": 0.34417128562927246,
      "learning_rate": 0.00022783340606214787,
      "loss": 4.761,
      "step": 23190
    },
    {
      "epoch": 0.4749861773800504,
      "grad_norm": 0.26643431186676025,
      "learning_rate": 0.00022778048818615783,
      "loss": 4.7816,
      "step": 23195
    },
    {
      "epoch": 0.475088567157455,
      "grad_norm": 0.26170021295547485,
      "learning_rate": 0.00022772755706635268,
      "loss": 4.7632,
      "step": 23200
    },
    {
      "epoch": 0.47519095693485963,
      "grad_norm": 0.2950766980648041,
      "learning_rate": 0.0002276746127117451,
      "loss": 4.7482,
      "step": 23205
    },
    {
      "epoch": 0.47529334671226425,
      "grad_norm": 0.2571079730987549,
      "learning_rate": 0.00022762165513135006,
      "loss": 4.7225,
      "step": 23210
    },
    {
      "epoch": 0.4753957364896689,
      "grad_norm": 0.26502344012260437,
      "learning_rate": 0.00022756868433418474,
      "loss": 4.7469,
      "step": 23215
    },
    {
      "epoch": 0.4754981262670735,
      "grad_norm": 0.26222342252731323,
      "learning_rate": 0.0002275157003292685,
      "loss": 4.7223,
      "step": 23220
    },
    {
      "epoch": 0.4756005160444781,
      "grad_norm": 0.2550445795059204,
      "learning_rate": 0.00022746270312562304,
      "loss": 4.6959,
      "step": 23225
    },
    {
      "epoch": 0.47570290582188274,
      "grad_norm": 0.25171804428100586,
      "learning_rate": 0.00022740969273227232,
      "loss": 4.7106,
      "step": 23230
    },
    {
      "epoch": 0.47580529559928736,
      "grad_norm": 0.23876318335533142,
      "learning_rate": 0.00022735666915824244,
      "loss": 4.678,
      "step": 23235
    },
    {
      "epoch": 0.475907685376692,
      "grad_norm": 0.29059797525405884,
      "learning_rate": 0.00022730363241256192,
      "loss": 4.708,
      "step": 23240
    },
    {
      "epoch": 0.4760100751540966,
      "grad_norm": 0.24677996337413788,
      "learning_rate": 0.0002272505825042613,
      "loss": 4.7111,
      "step": 23245
    },
    {
      "epoch": 0.4761124649315012,
      "grad_norm": 0.25260820984840393,
      "learning_rate": 0.00022719751944237355,
      "loss": 4.7577,
      "step": 23250
    },
    {
      "epoch": 0.47621485470890584,
      "grad_norm": 0.24880164861679077,
      "learning_rate": 0.0002271444432359338,
      "loss": 4.708,
      "step": 23255
    },
    {
      "epoch": 0.47631724448631046,
      "grad_norm": 0.2399342954158783,
      "learning_rate": 0.0002270913538939794,
      "loss": 4.7527,
      "step": 23260
    },
    {
      "epoch": 0.4764196342637151,
      "grad_norm": 0.25597113370895386,
      "learning_rate": 0.00022703825142555,
      "loss": 4.7383,
      "step": 23265
    },
    {
      "epoch": 0.4765220240411197,
      "grad_norm": 0.25812873244285583,
      "learning_rate": 0.00022698513583968737,
      "loss": 4.6974,
      "step": 23270
    },
    {
      "epoch": 0.4766244138185244,
      "grad_norm": 0.24173590540885925,
      "learning_rate": 0.0002269320071454356,
      "loss": 4.7384,
      "step": 23275
    },
    {
      "epoch": 0.476726803595929,
      "grad_norm": 0.25427424907684326,
      "learning_rate": 0.00022687886535184117,
      "loss": 4.7333,
      "step": 23280
    },
    {
      "epoch": 0.4768291933733336,
      "grad_norm": 0.24102525413036346,
      "learning_rate": 0.00022682571046795242,
      "loss": 4.751,
      "step": 23285
    },
    {
      "epoch": 0.47693158315073825,
      "grad_norm": 0.2437005639076233,
      "learning_rate": 0.00022677254250282026,
      "loss": 4.6898,
      "step": 23290
    },
    {
      "epoch": 0.47703397292814287,
      "grad_norm": 0.255088746547699,
      "learning_rate": 0.00022671936146549757,
      "loss": 4.7477,
      "step": 23295
    },
    {
      "epoch": 0.4771363627055475,
      "grad_norm": 0.2773576080799103,
      "learning_rate": 0.00022666616736503962,
      "loss": 4.7439,
      "step": 23300
    },
    {
      "epoch": 0.4772387524829521,
      "grad_norm": 0.2571176588535309,
      "learning_rate": 0.00022661296021050391,
      "loss": 4.7332,
      "step": 23305
    },
    {
      "epoch": 0.47734114226035673,
      "grad_norm": 0.2412247359752655,
      "learning_rate": 0.00022655974001095012,
      "loss": 4.7016,
      "step": 23310
    },
    {
      "epoch": 0.47744353203776135,
      "grad_norm": 0.2535102069377899,
      "learning_rate": 0.00022650650677544004,
      "loss": 4.7533,
      "step": 23315
    },
    {
      "epoch": 0.477545921815166,
      "grad_norm": 0.27320313453674316,
      "learning_rate": 0.00022645326051303784,
      "loss": 4.6689,
      "step": 23320
    },
    {
      "epoch": 0.4776483115925706,
      "grad_norm": 0.25510895252227783,
      "learning_rate": 0.0002264000012328099,
      "loss": 4.7309,
      "step": 23325
    },
    {
      "epoch": 0.4777507013699752,
      "grad_norm": 0.2580207586288452,
      "learning_rate": 0.00022634672894382464,
      "loss": 4.7344,
      "step": 23330
    },
    {
      "epoch": 0.47785309114737984,
      "grad_norm": 0.2654505968093872,
      "learning_rate": 0.00022629344365515295,
      "loss": 4.6744,
      "step": 23335
    },
    {
      "epoch": 0.47795548092478446,
      "grad_norm": 0.24057354032993317,
      "learning_rate": 0.00022624014537586776,
      "loss": 4.7311,
      "step": 23340
    },
    {
      "epoch": 0.4780578707021891,
      "grad_norm": 0.2518256604671478,
      "learning_rate": 0.00022618683411504423,
      "loss": 4.7433,
      "step": 23345
    },
    {
      "epoch": 0.4781602604795937,
      "grad_norm": 0.2699842154979706,
      "learning_rate": 0.00022613350988175983,
      "loss": 4.6862,
      "step": 23350
    },
    {
      "epoch": 0.4782626502569983,
      "grad_norm": 0.24945512413978577,
      "learning_rate": 0.0002260801726850941,
      "loss": 4.7247,
      "step": 23355
    },
    {
      "epoch": 0.47836504003440294,
      "grad_norm": 0.25566694140434265,
      "learning_rate": 0.00022602682253412886,
      "loss": 4.7578,
      "step": 23360
    },
    {
      "epoch": 0.47846742981180757,
      "grad_norm": 0.2524034082889557,
      "learning_rate": 0.0002259734594379482,
      "loss": 4.7608,
      "step": 23365
    },
    {
      "epoch": 0.4785698195892122,
      "grad_norm": 0.24215050041675568,
      "learning_rate": 0.0002259200834056382,
      "loss": 4.7336,
      "step": 23370
    },
    {
      "epoch": 0.47867220936661686,
      "grad_norm": 0.26179876923561096,
      "learning_rate": 0.00022586669444628747,
      "loss": 4.7207,
      "step": 23375
    },
    {
      "epoch": 0.4787745991440215,
      "grad_norm": 0.2582859694957733,
      "learning_rate": 0.0002258132925689865,
      "loss": 4.6859,
      "step": 23380
    },
    {
      "epoch": 0.4788769889214261,
      "grad_norm": 0.2548452913761139,
      "learning_rate": 0.00022575987778282819,
      "loss": 4.7869,
      "step": 23385
    },
    {
      "epoch": 0.4789793786988307,
      "grad_norm": 0.235408753156662,
      "learning_rate": 0.0002257064500969075,
      "loss": 4.7492,
      "step": 23390
    },
    {
      "epoch": 0.47908176847623535,
      "grad_norm": 0.24802841246128082,
      "learning_rate": 0.00022565300952032174,
      "loss": 4.6972,
      "step": 23395
    },
    {
      "epoch": 0.47918415825363997,
      "grad_norm": 0.4069477319717407,
      "learning_rate": 0.0002255995560621702,
      "loss": 4.8212,
      "step": 23400
    },
    {
      "epoch": 0.4792865480310446,
      "grad_norm": 0.2730862498283386,
      "learning_rate": 0.00022554608973155468,
      "loss": 4.7312,
      "step": 23405
    },
    {
      "epoch": 0.4793889378084492,
      "grad_norm": 0.9719353914260864,
      "learning_rate": 0.00022549261053757884,
      "loss": 4.6475,
      "step": 23410
    },
    {
      "epoch": 0.47949132758585383,
      "grad_norm": 0.25093239545822144,
      "learning_rate": 0.00022543911848934867,
      "loss": 4.6942,
      "step": 23415
    },
    {
      "epoch": 0.47959371736325845,
      "grad_norm": 0.2552156448364258,
      "learning_rate": 0.0002253856135959724,
      "loss": 4.716,
      "step": 23420
    },
    {
      "epoch": 0.4796961071406631,
      "grad_norm": 0.2522520422935486,
      "learning_rate": 0.00022533209586656037,
      "loss": 4.7008,
      "step": 23425
    },
    {
      "epoch": 0.4797984969180677,
      "grad_norm": 0.2818015515804291,
      "learning_rate": 0.00022527856531022523,
      "loss": 4.7332,
      "step": 23430
    },
    {
      "epoch": 0.4799008866954723,
      "grad_norm": 0.28834909200668335,
      "learning_rate": 0.00022522502193608154,
      "loss": 4.7325,
      "step": 23435
    },
    {
      "epoch": 0.48000327647287694,
      "grad_norm": 0.3477283716201782,
      "learning_rate": 0.00022517146575324638,
      "loss": 4.729,
      "step": 23440
    },
    {
      "epoch": 0.48010566625028156,
      "grad_norm": 0.24907995760440826,
      "learning_rate": 0.00022511789677083876,
      "loss": 4.7405,
      "step": 23445
    },
    {
      "epoch": 0.4802080560276862,
      "grad_norm": 0.2532285153865814,
      "learning_rate": 0.00022506431499798004,
      "loss": 4.7097,
      "step": 23450
    },
    {
      "epoch": 0.4803104458050908,
      "grad_norm": 0.2618994116783142,
      "learning_rate": 0.00022501072044379361,
      "loss": 4.7868,
      "step": 23455
    },
    {
      "epoch": 0.4804128355824954,
      "grad_norm": 0.259520024061203,
      "learning_rate": 0.00022495711311740509,
      "loss": 4.7504,
      "step": 23460
    },
    {
      "epoch": 0.48051522535990004,
      "grad_norm": 0.2821270823478699,
      "learning_rate": 0.00022490349302794235,
      "loss": 4.7465,
      "step": 23465
    },
    {
      "epoch": 0.48061761513730467,
      "grad_norm": 0.2771863639354706,
      "learning_rate": 0.00022484986018453534,
      "loss": 4.7235,
      "step": 23470
    },
    {
      "epoch": 0.48072000491470934,
      "grad_norm": 0.2585529386997223,
      "learning_rate": 0.00022479621459631623,
      "loss": 4.6119,
      "step": 23475
    },
    {
      "epoch": 0.48082239469211396,
      "grad_norm": 0.25866615772247314,
      "learning_rate": 0.00022474255627241935,
      "loss": 4.7308,
      "step": 23480
    },
    {
      "epoch": 0.4809247844695186,
      "grad_norm": 0.26243090629577637,
      "learning_rate": 0.00022468888522198115,
      "loss": 4.7026,
      "step": 23485
    },
    {
      "epoch": 0.4810271742469232,
      "grad_norm": 0.2763725817203522,
      "learning_rate": 0.00022463520145414036,
      "loss": 4.7614,
      "step": 23490
    },
    {
      "epoch": 0.4811295640243278,
      "grad_norm": 0.26582959294319153,
      "learning_rate": 0.00022458150497803775,
      "loss": 4.7269,
      "step": 23495
    },
    {
      "epoch": 0.48123195380173245,
      "grad_norm": 0.5712265372276306,
      "learning_rate": 0.00022452779580281636,
      "loss": 4.6891,
      "step": 23500
    },
    {
      "epoch": 0.48133434357913707,
      "grad_norm": 0.2590843141078949,
      "learning_rate": 0.00022447407393762125,
      "loss": 4.7279,
      "step": 23505
    },
    {
      "epoch": 0.4814367333565417,
      "grad_norm": 0.2678948640823364,
      "learning_rate": 0.00022442033939159984,
      "loss": 4.6712,
      "step": 23510
    },
    {
      "epoch": 0.4815391231339463,
      "grad_norm": 0.24214579164981842,
      "learning_rate": 0.00022436659217390153,
      "loss": 4.7322,
      "step": 23515
    },
    {
      "epoch": 0.48164151291135093,
      "grad_norm": 0.2527870237827301,
      "learning_rate": 0.00022431283229367802,
      "loss": 4.7619,
      "step": 23520
    },
    {
      "epoch": 0.48174390268875555,
      "grad_norm": 0.28357213735580444,
      "learning_rate": 0.000224259059760083,
      "loss": 4.6889,
      "step": 23525
    },
    {
      "epoch": 0.4818462924661602,
      "grad_norm": 0.2707435190677643,
      "learning_rate": 0.0002242052745822725,
      "loss": 4.7336,
      "step": 23530
    },
    {
      "epoch": 0.4819486822435648,
      "grad_norm": 0.29873546957969666,
      "learning_rate": 0.00022415147676940454,
      "loss": 4.7022,
      "step": 23535
    },
    {
      "epoch": 0.4820510720209694,
      "grad_norm": 0.25273212790489197,
      "learning_rate": 0.00022409766633063938,
      "loss": 4.7368,
      "step": 23540
    },
    {
      "epoch": 0.48215346179837404,
      "grad_norm": 0.25742873549461365,
      "learning_rate": 0.0002240438432751395,
      "loss": 4.689,
      "step": 23545
    },
    {
      "epoch": 0.48225585157577866,
      "grad_norm": 0.24842469394207,
      "learning_rate": 0.00022399000761206934,
      "loss": 4.6942,
      "step": 23550
    },
    {
      "epoch": 0.4823582413531833,
      "grad_norm": 0.26437339186668396,
      "learning_rate": 0.0002239361593505956,
      "loss": 4.6946,
      "step": 23555
    },
    {
      "epoch": 0.4824606311305879,
      "grad_norm": 0.27191320061683655,
      "learning_rate": 0.00022388229849988717,
      "loss": 4.7427,
      "step": 23560
    },
    {
      "epoch": 0.4825630209079925,
      "grad_norm": 0.2621234953403473,
      "learning_rate": 0.00022382842506911497,
      "loss": 4.7201,
      "step": 23565
    },
    {
      "epoch": 0.48266541068539714,
      "grad_norm": 0.2691076397895813,
      "learning_rate": 0.00022377453906745215,
      "loss": 4.742,
      "step": 23570
    },
    {
      "epoch": 0.4827678004628018,
      "grad_norm": 0.253271222114563,
      "learning_rate": 0.00022372064050407397,
      "loss": 4.6797,
      "step": 23575
    },
    {
      "epoch": 0.48287019024020644,
      "grad_norm": 0.24928729236125946,
      "learning_rate": 0.00022366672938815783,
      "loss": 4.7364,
      "step": 23580
    },
    {
      "epoch": 0.48297258001761106,
      "grad_norm": 0.2914057672023773,
      "learning_rate": 0.00022361280572888328,
      "loss": 4.7112,
      "step": 23585
    },
    {
      "epoch": 0.4830749697950157,
      "grad_norm": 0.46790847182273865,
      "learning_rate": 0.000223558869535432,
      "loss": 4.8043,
      "step": 23590
    },
    {
      "epoch": 0.4831773595724203,
      "grad_norm": 0.26335951685905457,
      "learning_rate": 0.0002235049208169877,
      "loss": 4.7362,
      "step": 23595
    },
    {
      "epoch": 0.4832797493498249,
      "grad_norm": 0.2944200932979584,
      "learning_rate": 0.00022345095958273641,
      "loss": 4.7088,
      "step": 23600
    },
    {
      "epoch": 0.48338213912722955,
      "grad_norm": 0.24995774030685425,
      "learning_rate": 0.00022339698584186623,
      "loss": 4.7012,
      "step": 23605
    },
    {
      "epoch": 0.48348452890463417,
      "grad_norm": 0.31491002440452576,
      "learning_rate": 0.00022334299960356726,
      "loss": 4.6802,
      "step": 23610
    },
    {
      "epoch": 0.4835869186820388,
      "grad_norm": 0.24595631659030914,
      "learning_rate": 0.00022328900087703195,
      "loss": 4.7284,
      "step": 23615
    },
    {
      "epoch": 0.4836893084594434,
      "grad_norm": 0.2661062777042389,
      "learning_rate": 0.00022323498967145462,
      "loss": 4.7217,
      "step": 23620
    },
    {
      "epoch": 0.48379169823684803,
      "grad_norm": 0.25394463539123535,
      "learning_rate": 0.0002231809659960319,
      "loss": 4.7482,
      "step": 23625
    },
    {
      "epoch": 0.48389408801425265,
      "grad_norm": 0.27144917845726013,
      "learning_rate": 0.00022312692985996255,
      "loss": 4.6937,
      "step": 23630
    },
    {
      "epoch": 0.4839964777916573,
      "grad_norm": 0.2578558027744293,
      "learning_rate": 0.00022307288127244732,
      "loss": 4.6935,
      "step": 23635
    },
    {
      "epoch": 0.4840988675690619,
      "grad_norm": 0.24560491740703583,
      "learning_rate": 0.0002230188202426892,
      "loss": 4.7088,
      "step": 23640
    },
    {
      "epoch": 0.4842012573464665,
      "grad_norm": 0.23649932444095612,
      "learning_rate": 0.00022296474677989324,
      "loss": 4.6857,
      "step": 23645
    },
    {
      "epoch": 0.48430364712387114,
      "grad_norm": 0.25463882088661194,
      "learning_rate": 0.00022291066089326662,
      "loss": 4.6973,
      "step": 23650
    },
    {
      "epoch": 0.48440603690127576,
      "grad_norm": 0.2707253694534302,
      "learning_rate": 0.00022285656259201866,
      "loss": 4.7359,
      "step": 23655
    },
    {
      "epoch": 0.4845084266786804,
      "grad_norm": 0.24852105975151062,
      "learning_rate": 0.0002228024518853607,
      "loss": 4.7687,
      "step": 23660
    },
    {
      "epoch": 0.484610816456085,
      "grad_norm": 0.25112882256507874,
      "learning_rate": 0.0002227483287825063,
      "loss": 4.7106,
      "step": 23665
    },
    {
      "epoch": 0.4847132062334896,
      "grad_norm": 0.25372880697250366,
      "learning_rate": 0.00022269419329267113,
      "loss": 4.7425,
      "step": 23670
    },
    {
      "epoch": 0.48481559601089425,
      "grad_norm": 0.2551497220993042,
      "learning_rate": 0.00022264004542507288,
      "loss": 4.7361,
      "step": 23675
    },
    {
      "epoch": 0.4849179857882989,
      "grad_norm": 0.25463464856147766,
      "learning_rate": 0.0002225858851889314,
      "loss": 4.7486,
      "step": 23680
    },
    {
      "epoch": 0.48502037556570354,
      "grad_norm": 0.25041845440864563,
      "learning_rate": 0.0002225317125934687,
      "loss": 4.7456,
      "step": 23685
    },
    {
      "epoch": 0.48512276534310816,
      "grad_norm": 0.26744791865348816,
      "learning_rate": 0.00022247752764790877,
      "loss": 4.7191,
      "step": 23690
    },
    {
      "epoch": 0.4852251551205128,
      "grad_norm": 0.27257540822029114,
      "learning_rate": 0.00022242333036147775,
      "loss": 4.6903,
      "step": 23695
    },
    {
      "epoch": 0.4853275448979174,
      "grad_norm": 0.24697956442832947,
      "learning_rate": 0.00022236912074340403,
      "loss": 4.7408,
      "step": 23700
    },
    {
      "epoch": 0.48542993467532203,
      "grad_norm": 0.28456252813339233,
      "learning_rate": 0.00022231489880291787,
      "loss": 4.6968,
      "step": 23705
    },
    {
      "epoch": 0.48553232445272665,
      "grad_norm": 0.2713773250579834,
      "learning_rate": 0.00022226066454925179,
      "loss": 4.7041,
      "step": 23710
    },
    {
      "epoch": 0.48563471423013127,
      "grad_norm": 0.25253403186798096,
      "learning_rate": 0.00022220641799164032,
      "loss": 4.7425,
      "step": 23715
    },
    {
      "epoch": 0.4857371040075359,
      "grad_norm": 0.26198431849479675,
      "learning_rate": 0.00022215215913932003,
      "loss": 4.7672,
      "step": 23720
    },
    {
      "epoch": 0.4858394937849405,
      "grad_norm": 0.2633711099624634,
      "learning_rate": 0.00022209788800152982,
      "loss": 4.7149,
      "step": 23725
    },
    {
      "epoch": 0.48594188356234513,
      "grad_norm": 0.2900274991989136,
      "learning_rate": 0.00022204360458751046,
      "loss": 4.6937,
      "step": 23730
    },
    {
      "epoch": 0.48604427333974976,
      "grad_norm": 0.23595717549324036,
      "learning_rate": 0.00022198930890650486,
      "loss": 4.7202,
      "step": 23735
    },
    {
      "epoch": 0.4861466631171544,
      "grad_norm": 0.24150396883487701,
      "learning_rate": 0.00022193500096775803,
      "loss": 4.7367,
      "step": 23740
    },
    {
      "epoch": 0.486249052894559,
      "grad_norm": 0.2442939579486847,
      "learning_rate": 0.00022188068078051712,
      "loss": 4.711,
      "step": 23745
    },
    {
      "epoch": 0.4863514426719636,
      "grad_norm": 0.272164523601532,
      "learning_rate": 0.00022182634835403125,
      "loss": 4.7202,
      "step": 23750
    },
    {
      "epoch": 0.48645383244936824,
      "grad_norm": 0.2705736756324768,
      "learning_rate": 0.0002217720036975518,
      "loss": 4.7096,
      "step": 23755
    },
    {
      "epoch": 0.48655622222677286,
      "grad_norm": 0.2533026337623596,
      "learning_rate": 0.0002217176468203321,
      "loss": 4.7322,
      "step": 23760
    },
    {
      "epoch": 0.4866586120041775,
      "grad_norm": 0.24849478900432587,
      "learning_rate": 0.00022166327773162742,
      "loss": 4.7143,
      "step": 23765
    },
    {
      "epoch": 0.4867610017815821,
      "grad_norm": 0.23685523867607117,
      "learning_rate": 0.00022160889644069553,
      "loss": 4.7141,
      "step": 23770
    },
    {
      "epoch": 0.4868633915589867,
      "grad_norm": 0.27315205335617065,
      "learning_rate": 0.00022155450295679584,
      "loss": 4.6996,
      "step": 23775
    },
    {
      "epoch": 0.4869657813363914,
      "grad_norm": 0.2574215531349182,
      "learning_rate": 0.00022150009728919015,
      "loss": 4.763,
      "step": 23780
    },
    {
      "epoch": 0.487068171113796,
      "grad_norm": 0.2711694836616516,
      "learning_rate": 0.0002214456794471421,
      "loss": 4.714,
      "step": 23785
    },
    {
      "epoch": 0.48717056089120064,
      "grad_norm": 0.2520192563533783,
      "learning_rate": 0.00022139124943991746,
      "loss": 4.7763,
      "step": 23790
    },
    {
      "epoch": 0.48727295066860526,
      "grad_norm": 0.3664661943912506,
      "learning_rate": 0.00022133680727678436,
      "loss": 4.7257,
      "step": 23795
    },
    {
      "epoch": 0.4873753404460099,
      "grad_norm": 0.26421457529067993,
      "learning_rate": 0.00022128235296701252,
      "loss": 4.7375,
      "step": 23800
    },
    {
      "epoch": 0.4874777302234145,
      "grad_norm": 0.2750388979911804,
      "learning_rate": 0.00022122788651987406,
      "loss": 4.7304,
      "step": 23805
    },
    {
      "epoch": 0.48758012000081913,
      "grad_norm": 0.25376227498054504,
      "learning_rate": 0.0002211734079446431,
      "loss": 4.7056,
      "step": 23810
    },
    {
      "epoch": 0.48768250977822375,
      "grad_norm": 0.26880383491516113,
      "learning_rate": 0.00022111891725059573,
      "loss": 4.7369,
      "step": 23815
    },
    {
      "epoch": 0.48778489955562837,
      "grad_norm": 0.28430506587028503,
      "learning_rate": 0.00022106441444701022,
      "loss": 4.748,
      "step": 23820
    },
    {
      "epoch": 0.487887289333033,
      "grad_norm": 0.27597060799598694,
      "learning_rate": 0.00022100989954316683,
      "loss": 4.7576,
      "step": 23825
    },
    {
      "epoch": 0.4879896791104376,
      "grad_norm": 0.2408474236726761,
      "learning_rate": 0.00022095537254834797,
      "loss": 4.7158,
      "step": 23830
    },
    {
      "epoch": 0.48809206888784223,
      "grad_norm": 0.249201238155365,
      "learning_rate": 0.0002209008334718379,
      "loss": 4.7159,
      "step": 23835
    },
    {
      "epoch": 0.48819445866524686,
      "grad_norm": 0.248459592461586,
      "learning_rate": 0.00022084628232292325,
      "loss": 4.6668,
      "step": 23840
    },
    {
      "epoch": 0.4882968484426515,
      "grad_norm": 0.24212941527366638,
      "learning_rate": 0.0002207917191108924,
      "loss": 4.765,
      "step": 23845
    },
    {
      "epoch": 0.4883992382200561,
      "grad_norm": 0.23875699937343597,
      "learning_rate": 0.00022073714384503605,
      "loss": 4.75,
      "step": 23850
    },
    {
      "epoch": 0.4885016279974607,
      "grad_norm": 0.251152366399765,
      "learning_rate": 0.00022068255653464675,
      "loss": 4.7478,
      "step": 23855
    },
    {
      "epoch": 0.48860401777486534,
      "grad_norm": 0.2696307897567749,
      "learning_rate": 0.0002206279571890191,
      "loss": 4.6872,
      "step": 23860
    },
    {
      "epoch": 0.48870640755226996,
      "grad_norm": 0.28312090039253235,
      "learning_rate": 0.00022057334581744993,
      "loss": 4.7609,
      "step": 23865
    },
    {
      "epoch": 0.4888087973296746,
      "grad_norm": 0.24442794919013977,
      "learning_rate": 0.00022051872242923802,
      "loss": 4.7271,
      "step": 23870
    },
    {
      "epoch": 0.4889111871070792,
      "grad_norm": 0.25698035955429077,
      "learning_rate": 0.0002204640870336841,
      "loss": 4.7861,
      "step": 23875
    },
    {
      "epoch": 0.4890135768844839,
      "grad_norm": 0.2503564953804016,
      "learning_rate": 0.00022040943964009107,
      "loss": 4.7242,
      "step": 23880
    },
    {
      "epoch": 0.4891159666618885,
      "grad_norm": 0.28338170051574707,
      "learning_rate": 0.00022035478025776386,
      "loss": 4.6962,
      "step": 23885
    },
    {
      "epoch": 0.4892183564392931,
      "grad_norm": 0.2796652019023895,
      "learning_rate": 0.00022030010889600939,
      "loss": 4.7119,
      "step": 23890
    },
    {
      "epoch": 0.48932074621669774,
      "grad_norm": 0.2485644519329071,
      "learning_rate": 0.00022024542556413668,
      "loss": 4.7181,
      "step": 23895
    },
    {
      "epoch": 0.48942313599410237,
      "grad_norm": 0.35051485896110535,
      "learning_rate": 0.00022019073027145669,
      "loss": 4.7348,
      "step": 23900
    },
    {
      "epoch": 0.489525525771507,
      "grad_norm": 0.25420159101486206,
      "learning_rate": 0.0002201360230272825,
      "loss": 4.7435,
      "step": 23905
    },
    {
      "epoch": 0.4896279155489116,
      "grad_norm": 0.25134554505348206,
      "learning_rate": 0.0002200813038409292,
      "loss": 4.7209,
      "step": 23910
    },
    {
      "epoch": 0.48973030532631623,
      "grad_norm": 0.2469230741262436,
      "learning_rate": 0.00022002657272171398,
      "loss": 4.7314,
      "step": 23915
    },
    {
      "epoch": 0.48983269510372085,
      "grad_norm": 0.2539413869380951,
      "learning_rate": 0.00021997182967895598,
      "loss": 4.7025,
      "step": 23920
    },
    {
      "epoch": 0.48993508488112547,
      "grad_norm": 0.2581627070903778,
      "learning_rate": 0.00021991707472197633,
      "loss": 4.7102,
      "step": 23925
    },
    {
      "epoch": 0.4900374746585301,
      "grad_norm": 0.2602821886539459,
      "learning_rate": 0.00021986230786009828,
      "loss": 4.6872,
      "step": 23930
    },
    {
      "epoch": 0.4901398644359347,
      "grad_norm": 0.251945823431015,
      "learning_rate": 0.0002198075291026471,
      "loss": 4.7519,
      "step": 23935
    },
    {
      "epoch": 0.49024225421333933,
      "grad_norm": 0.25386083126068115,
      "learning_rate": 0.00021975273845895003,
      "loss": 4.7025,
      "step": 23940
    },
    {
      "epoch": 0.49034464399074396,
      "grad_norm": 0.25990474224090576,
      "learning_rate": 0.0002196979359383364,
      "loss": 4.6574,
      "step": 23945
    },
    {
      "epoch": 0.4904470337681486,
      "grad_norm": 0.2560100853443146,
      "learning_rate": 0.0002196431215501375,
      "loss": 4.782,
      "step": 23950
    },
    {
      "epoch": 0.4905494235455532,
      "grad_norm": 0.2552954852581024,
      "learning_rate": 0.00021958829530368667,
      "loss": 4.7536,
      "step": 23955
    },
    {
      "epoch": 0.4906518133229578,
      "grad_norm": 0.24191182851791382,
      "learning_rate": 0.00021953345720831925,
      "loss": 4.7152,
      "step": 23960
    },
    {
      "epoch": 0.49075420310036244,
      "grad_norm": 0.2537633180618286,
      "learning_rate": 0.00021947860727337268,
      "loss": 4.6867,
      "step": 23965
    },
    {
      "epoch": 0.49085659287776706,
      "grad_norm": 0.24744947254657745,
      "learning_rate": 0.00021942374550818627,
      "loss": 4.7524,
      "step": 23970
    },
    {
      "epoch": 0.4909589826551717,
      "grad_norm": 0.27883878350257874,
      "learning_rate": 0.00021936887192210147,
      "loss": 4.7562,
      "step": 23975
    },
    {
      "epoch": 0.4910613724325763,
      "grad_norm": 0.2712114155292511,
      "learning_rate": 0.00021931398652446168,
      "loss": 4.7652,
      "step": 23980
    },
    {
      "epoch": 0.491163762209981,
      "grad_norm": 0.2643401026725769,
      "learning_rate": 0.00021925908932461237,
      "loss": 4.6906,
      "step": 23985
    },
    {
      "epoch": 0.4912661519873856,
      "grad_norm": 0.25378501415252686,
      "learning_rate": 0.00021920418033190097,
      "loss": 4.7286,
      "step": 23990
    },
    {
      "epoch": 0.4913685417647902,
      "grad_norm": 0.2989406883716583,
      "learning_rate": 0.0002191492595556769,
      "loss": 4.7293,
      "step": 23995
    },
    {
      "epoch": 0.49147093154219484,
      "grad_norm": 0.27138394117355347,
      "learning_rate": 0.00021909432700529158,
      "loss": 4.6571,
      "step": 24000
    },
    {
      "epoch": 0.49147093154219484,
      "eval_loss": 4.740637302398682,
      "eval_runtime": 17.1233,
      "eval_samples_per_second": 33.989,
      "eval_steps_per_second": 4.263,
      "step": 24000
    },
    {
      "epoch": 0.49157332131959947,
      "grad_norm": 0.24590237438678741,
      "learning_rate": 0.0002190393826900985,
      "loss": 4.722,
      "step": 24005
    },
    {
      "epoch": 0.4916757110970041,
      "grad_norm": 0.24786077439785004,
      "learning_rate": 0.0002189844266194532,
      "loss": 4.7176,
      "step": 24010
    },
    {
      "epoch": 0.4917781008744087,
      "grad_norm": 0.2574801445007324,
      "learning_rate": 0.00021892945880271303,
      "loss": 4.7583,
      "step": 24015
    },
    {
      "epoch": 0.49188049065181333,
      "grad_norm": 0.25858595967292786,
      "learning_rate": 0.00021887447924923752,
      "loss": 4.6977,
      "step": 24020
    },
    {
      "epoch": 0.49198288042921795,
      "grad_norm": 0.6016584038734436,
      "learning_rate": 0.0002188194879683881,
      "loss": 4.6965,
      "step": 24025
    },
    {
      "epoch": 0.49208527020662257,
      "grad_norm": 0.24698983132839203,
      "learning_rate": 0.0002187644849695282,
      "loss": 4.7379,
      "step": 24030
    },
    {
      "epoch": 0.4921876599840272,
      "grad_norm": 0.2544917166233063,
      "learning_rate": 0.00021870947026202331,
      "loss": 4.732,
      "step": 24035
    },
    {
      "epoch": 0.4922900497614318,
      "grad_norm": 0.24248680472373962,
      "learning_rate": 0.00021865444385524093,
      "loss": 4.6828,
      "step": 24040
    },
    {
      "epoch": 0.49239243953883644,
      "grad_norm": 0.2829208970069885,
      "learning_rate": 0.00021859940575855044,
      "loss": 4.711,
      "step": 24045
    },
    {
      "epoch": 0.49249482931624106,
      "grad_norm": 0.2585568130016327,
      "learning_rate": 0.00021854435598132324,
      "loss": 4.6618,
      "step": 24050
    },
    {
      "epoch": 0.4925972190936457,
      "grad_norm": 0.2694754898548126,
      "learning_rate": 0.0002184892945329328,
      "loss": 4.7447,
      "step": 24055
    },
    {
      "epoch": 0.4926996088710503,
      "grad_norm": 0.3631758689880371,
      "learning_rate": 0.0002184342214227545,
      "loss": 4.7033,
      "step": 24060
    },
    {
      "epoch": 0.4928019986484549,
      "grad_norm": 0.24882563948631287,
      "learning_rate": 0.00021837913666016576,
      "loss": 4.7405,
      "step": 24065
    },
    {
      "epoch": 0.49290438842585954,
      "grad_norm": 0.24229872226715088,
      "learning_rate": 0.00021832404025454592,
      "loss": 4.7549,
      "step": 24070
    },
    {
      "epoch": 0.49300677820326416,
      "grad_norm": 0.2614671289920807,
      "learning_rate": 0.00021826893221527636,
      "loss": 4.7364,
      "step": 24075
    },
    {
      "epoch": 0.4931091679806688,
      "grad_norm": 0.26160407066345215,
      "learning_rate": 0.00021821381255174042,
      "loss": 4.7489,
      "step": 24080
    },
    {
      "epoch": 0.49321155775807346,
      "grad_norm": 0.2651068866252899,
      "learning_rate": 0.0002181586812733234,
      "loss": 4.711,
      "step": 24085
    },
    {
      "epoch": 0.4933139475354781,
      "grad_norm": 0.2535269856452942,
      "learning_rate": 0.00021810353838941262,
      "loss": 4.7428,
      "step": 24090
    },
    {
      "epoch": 0.4934163373128827,
      "grad_norm": 0.3206854462623596,
      "learning_rate": 0.00021804838390939734,
      "loss": 4.7096,
      "step": 24095
    },
    {
      "epoch": 0.4935187270902873,
      "grad_norm": 0.2458593249320984,
      "learning_rate": 0.00021799321784266877,
      "loss": 4.713,
      "step": 24100
    },
    {
      "epoch": 0.49362111686769194,
      "grad_norm": 0.24413485825061798,
      "learning_rate": 0.00021793804019862024,
      "loss": 4.7451,
      "step": 24105
    },
    {
      "epoch": 0.49372350664509657,
      "grad_norm": 0.2457171082496643,
      "learning_rate": 0.00021788285098664682,
      "loss": 4.6979,
      "step": 24110
    },
    {
      "epoch": 0.4938258964225012,
      "grad_norm": 0.29445162415504456,
      "learning_rate": 0.0002178276502161457,
      "loss": 4.6508,
      "step": 24115
    },
    {
      "epoch": 0.4939282861999058,
      "grad_norm": 0.24881939589977264,
      "learning_rate": 0.00021777243789651603,
      "loss": 4.7094,
      "step": 24120
    },
    {
      "epoch": 0.49403067597731043,
      "grad_norm": 0.2543780207633972,
      "learning_rate": 0.00021771721403715896,
      "loss": 4.7056,
      "step": 24125
    },
    {
      "epoch": 0.49413306575471505,
      "grad_norm": 0.26519331336021423,
      "learning_rate": 0.00021766197864747742,
      "loss": 4.7269,
      "step": 24130
    },
    {
      "epoch": 0.49423545553211967,
      "grad_norm": 0.28295549750328064,
      "learning_rate": 0.00021760673173687657,
      "loss": 4.7403,
      "step": 24135
    },
    {
      "epoch": 0.4943378453095243,
      "grad_norm": 0.2623783349990845,
      "learning_rate": 0.00021755147331476327,
      "loss": 4.7187,
      "step": 24140
    },
    {
      "epoch": 0.4944402350869289,
      "grad_norm": 0.26957452297210693,
      "learning_rate": 0.00021749620339054653,
      "loss": 4.7374,
      "step": 24145
    },
    {
      "epoch": 0.49454262486433354,
      "grad_norm": 0.251097708940506,
      "learning_rate": 0.00021744092197363725,
      "loss": 4.7435,
      "step": 24150
    },
    {
      "epoch": 0.49464501464173816,
      "grad_norm": 0.24897147715091705,
      "learning_rate": 0.0002173856290734483,
      "loss": 4.7092,
      "step": 24155
    },
    {
      "epoch": 0.4947474044191428,
      "grad_norm": 0.27389004826545715,
      "learning_rate": 0.00021733032469939447,
      "loss": 4.763,
      "step": 24160
    },
    {
      "epoch": 0.4948497941965474,
      "grad_norm": 0.2514142096042633,
      "learning_rate": 0.00021727500886089255,
      "loss": 4.749,
      "step": 24165
    },
    {
      "epoch": 0.494952183973952,
      "grad_norm": 0.24971139430999756,
      "learning_rate": 0.00021721968156736122,
      "loss": 4.736,
      "step": 24170
    },
    {
      "epoch": 0.49505457375135664,
      "grad_norm": 0.2452549934387207,
      "learning_rate": 0.0002171643428282212,
      "loss": 4.732,
      "step": 24175
    },
    {
      "epoch": 0.49515696352876126,
      "grad_norm": 0.2623090445995331,
      "learning_rate": 0.00021710899265289514,
      "loss": 4.6775,
      "step": 24180
    },
    {
      "epoch": 0.49525935330616594,
      "grad_norm": 0.24953871965408325,
      "learning_rate": 0.0002170536310508075,
      "loss": 4.7024,
      "step": 24185
    },
    {
      "epoch": 0.49536174308357056,
      "grad_norm": 0.2427460104227066,
      "learning_rate": 0.0002169982580313849,
      "loss": 4.725,
      "step": 24190
    },
    {
      "epoch": 0.4954641328609752,
      "grad_norm": 0.24066436290740967,
      "learning_rate": 0.00021694287360405574,
      "loss": 4.7284,
      "step": 24195
    },
    {
      "epoch": 0.4955665226383798,
      "grad_norm": 0.25126057863235474,
      "learning_rate": 0.00021688747777825042,
      "loss": 4.7395,
      "step": 24200
    },
    {
      "epoch": 0.4956689124157844,
      "grad_norm": 0.3491162061691284,
      "learning_rate": 0.00021683207056340134,
      "loss": 4.7523,
      "step": 24205
    },
    {
      "epoch": 0.49577130219318905,
      "grad_norm": 0.27271589636802673,
      "learning_rate": 0.00021677665196894269,
      "loss": 4.7545,
      "step": 24210
    },
    {
      "epoch": 0.49587369197059367,
      "grad_norm": 0.2561257779598236,
      "learning_rate": 0.0002167212220043107,
      "loss": 4.7135,
      "step": 24215
    },
    {
      "epoch": 0.4959760817479983,
      "grad_norm": 0.2725783586502075,
      "learning_rate": 0.00021666578067894364,
      "loss": 4.7224,
      "step": 24220
    },
    {
      "epoch": 0.4960784715254029,
      "grad_norm": 0.2379966825246811,
      "learning_rate": 0.00021661032800228145,
      "loss": 4.6912,
      "step": 24225
    },
    {
      "epoch": 0.49618086130280753,
      "grad_norm": 0.2789928913116455,
      "learning_rate": 0.00021655486398376624,
      "loss": 4.7488,
      "step": 24230
    },
    {
      "epoch": 0.49628325108021215,
      "grad_norm": 0.26592832803726196,
      "learning_rate": 0.0002164993886328419,
      "loss": 4.7053,
      "step": 24235
    },
    {
      "epoch": 0.4963856408576168,
      "grad_norm": 0.268175333738327,
      "learning_rate": 0.0002164439019589544,
      "loss": 4.6547,
      "step": 24240
    },
    {
      "epoch": 0.4964880306350214,
      "grad_norm": 0.2482205480337143,
      "learning_rate": 0.00021638840397155147,
      "loss": 4.738,
      "step": 24245
    },
    {
      "epoch": 0.496590420412426,
      "grad_norm": 0.25238728523254395,
      "learning_rate": 0.0002163328946800829,
      "loss": 4.7493,
      "step": 24250
    },
    {
      "epoch": 0.49669281018983064,
      "grad_norm": 0.24435508251190186,
      "learning_rate": 0.0002162773740940003,
      "loss": 4.681,
      "step": 24255
    },
    {
      "epoch": 0.49679519996723526,
      "grad_norm": 0.24986614286899567,
      "learning_rate": 0.00021622184222275725,
      "loss": 4.7567,
      "step": 24260
    },
    {
      "epoch": 0.4968975897446399,
      "grad_norm": 0.25959527492523193,
      "learning_rate": 0.00021616629907580934,
      "loss": 4.6976,
      "step": 24265
    },
    {
      "epoch": 0.4969999795220445,
      "grad_norm": 0.24627871811389923,
      "learning_rate": 0.00021611074466261392,
      "loss": 4.7185,
      "step": 24270
    },
    {
      "epoch": 0.4971023692994491,
      "grad_norm": 0.2620392441749573,
      "learning_rate": 0.0002160551789926304,
      "loss": 4.7508,
      "step": 24275
    },
    {
      "epoch": 0.49720475907685374,
      "grad_norm": 0.26908978819847107,
      "learning_rate": 0.0002159996020753199,
      "loss": 4.7035,
      "step": 24280
    },
    {
      "epoch": 0.4973071488542584,
      "grad_norm": 0.25050944089889526,
      "learning_rate": 0.00021594401392014572,
      "loss": 4.7076,
      "step": 24285
    },
    {
      "epoch": 0.49740953863166304,
      "grad_norm": 0.26778820157051086,
      "learning_rate": 0.00021588841453657295,
      "loss": 4.7205,
      "step": 24290
    },
    {
      "epoch": 0.49751192840906766,
      "grad_norm": 0.24914218485355377,
      "learning_rate": 0.00021583280393406852,
      "loss": 4.6853,
      "step": 24295
    },
    {
      "epoch": 0.4976143181864723,
      "grad_norm": 0.2422572374343872,
      "learning_rate": 0.00021577718212210145,
      "loss": 4.7562,
      "step": 24300
    },
    {
      "epoch": 0.4977167079638769,
      "grad_norm": 0.26418182253837585,
      "learning_rate": 0.0002157215491101424,
      "loss": 4.7084,
      "step": 24305
    },
    {
      "epoch": 0.4978190977412815,
      "grad_norm": 0.26421383023262024,
      "learning_rate": 0.00021566590490766424,
      "loss": 4.7017,
      "step": 24310
    },
    {
      "epoch": 0.49792148751868615,
      "grad_norm": 0.2425071895122528,
      "learning_rate": 0.00021561024952414156,
      "loss": 4.767,
      "step": 24315
    },
    {
      "epoch": 0.49802387729609077,
      "grad_norm": 0.2629650831222534,
      "learning_rate": 0.00021555458296905087,
      "loss": 4.6466,
      "step": 24320
    },
    {
      "epoch": 0.4981262670734954,
      "grad_norm": 0.28130167722702026,
      "learning_rate": 0.00021549890525187062,
      "loss": 4.6324,
      "step": 24325
    },
    {
      "epoch": 0.4982286568509,
      "grad_norm": 0.2503277659416199,
      "learning_rate": 0.00021544321638208115,
      "loss": 4.7381,
      "step": 24330
    },
    {
      "epoch": 0.49833104662830463,
      "grad_norm": 0.2546005845069885,
      "learning_rate": 0.0002153875163691647,
      "loss": 4.6521,
      "step": 24335
    },
    {
      "epoch": 0.49843343640570925,
      "grad_norm": 3.1699318885803223,
      "learning_rate": 0.0002153318052226054,
      "loss": 4.7349,
      "step": 24340
    },
    {
      "epoch": 0.4985358261831139,
      "grad_norm": 0.25961750745773315,
      "learning_rate": 0.0002152760829518893,
      "loss": 4.6934,
      "step": 24345
    },
    {
      "epoch": 0.4986382159605185,
      "grad_norm": 0.2727004587650299,
      "learning_rate": 0.00021522034956650434,
      "loss": 4.7158,
      "step": 24350
    },
    {
      "epoch": 0.4987406057379231,
      "grad_norm": 0.25228407979011536,
      "learning_rate": 0.00021516460507594025,
      "loss": 4.7123,
      "step": 24355
    },
    {
      "epoch": 0.49884299551532774,
      "grad_norm": 0.2500511109828949,
      "learning_rate": 0.00021510884948968884,
      "loss": 4.708,
      "step": 24360
    },
    {
      "epoch": 0.49894538529273236,
      "grad_norm": 0.258484423160553,
      "learning_rate": 0.00021505308281724367,
      "loss": 4.7536,
      "step": 24365
    },
    {
      "epoch": 0.499047775070137,
      "grad_norm": 0.24826906621456146,
      "learning_rate": 0.0002149973050681002,
      "loss": 4.7293,
      "step": 24370
    },
    {
      "epoch": 0.4991501648475416,
      "grad_norm": 0.23662418127059937,
      "learning_rate": 0.00021494151625175585,
      "loss": 4.7134,
      "step": 24375
    },
    {
      "epoch": 0.4992525546249462,
      "grad_norm": 0.25076043605804443,
      "learning_rate": 0.0002148857163777098,
      "loss": 4.7352,
      "step": 24380
    },
    {
      "epoch": 0.49935494440235084,
      "grad_norm": 0.2418327033519745,
      "learning_rate": 0.0002148299054554633,
      "loss": 4.7292,
      "step": 24385
    },
    {
      "epoch": 0.4994573341797555,
      "grad_norm": 0.25087326765060425,
      "learning_rate": 0.0002147740834945193,
      "loss": 4.7174,
      "step": 24390
    },
    {
      "epoch": 0.49955972395716014,
      "grad_norm": 0.282532274723053,
      "learning_rate": 0.00021471825050438267,
      "loss": 4.6723,
      "step": 24395
    },
    {
      "epoch": 0.49966211373456476,
      "grad_norm": 0.252346009016037,
      "learning_rate": 0.00021466240649456022,
      "loss": 4.733,
      "step": 24400
    },
    {
      "epoch": 0.4997645035119694,
      "grad_norm": 0.2513236701488495,
      "learning_rate": 0.00021460655147456065,
      "loss": 4.7049,
      "step": 24405
    },
    {
      "epoch": 0.499866893289374,
      "grad_norm": 0.2769632339477539,
      "learning_rate": 0.00021455068545389442,
      "loss": 4.6374,
      "step": 24410
    },
    {
      "epoch": 0.4999692830667786,
      "grad_norm": 0.24667078256607056,
      "learning_rate": 0.000214494808442074,
      "loss": 4.6938,
      "step": 24415
    },
    {
      "epoch": 0.5000716728441832,
      "grad_norm": 0.2368348091840744,
      "learning_rate": 0.00021443892044861355,
      "loss": 4.6575,
      "step": 24420
    },
    {
      "epoch": 0.5001740626215878,
      "grad_norm": 0.24722762405872345,
      "learning_rate": 0.00021438302148302931,
      "loss": 4.6666,
      "step": 24425
    },
    {
      "epoch": 0.5002764523989924,
      "grad_norm": 0.26125872135162354,
      "learning_rate": 0.0002143271115548393,
      "loss": 4.7624,
      "step": 24430
    },
    {
      "epoch": 0.500378842176397,
      "grad_norm": 0.24780777096748352,
      "learning_rate": 0.00021427119067356337,
      "loss": 4.7497,
      "step": 24435
    },
    {
      "epoch": 0.5004812319538018,
      "grad_norm": 0.25851914286613464,
      "learning_rate": 0.00021421525884872327,
      "loss": 4.6976,
      "step": 24440
    },
    {
      "epoch": 0.5005836217312064,
      "grad_norm": 0.2696174085140228,
      "learning_rate": 0.00021415931608984254,
      "loss": 4.6988,
      "step": 24445
    },
    {
      "epoch": 0.500686011508611,
      "grad_norm": 0.2505180835723877,
      "learning_rate": 0.00021410336240644672,
      "loss": 4.7444,
      "step": 24450
    },
    {
      "epoch": 0.5007884012860156,
      "grad_norm": 0.2612415552139282,
      "learning_rate": 0.00021404739780806313,
      "loss": 4.6558,
      "step": 24455
    },
    {
      "epoch": 0.5008907910634203,
      "grad_norm": 0.26742544770240784,
      "learning_rate": 0.00021399142230422096,
      "loss": 4.7245,
      "step": 24460
    },
    {
      "epoch": 0.5009931808408249,
      "grad_norm": 0.27145445346832275,
      "learning_rate": 0.0002139354359044512,
      "loss": 4.7398,
      "step": 24465
    },
    {
      "epoch": 0.5010955706182295,
      "grad_norm": 0.268049418926239,
      "learning_rate": 0.00021387943861828684,
      "loss": 4.7316,
      "step": 24470
    },
    {
      "epoch": 0.5011979603956341,
      "grad_norm": 0.2802181839942932,
      "learning_rate": 0.00021382343045526253,
      "loss": 4.7352,
      "step": 24475
    },
    {
      "epoch": 0.5013003501730388,
      "grad_norm": 0.25395095348358154,
      "learning_rate": 0.00021376741142491493,
      "loss": 4.7198,
      "step": 24480
    },
    {
      "epoch": 0.5014027399504434,
      "grad_norm": 0.27157557010650635,
      "learning_rate": 0.00021371138153678253,
      "loss": 4.7411,
      "step": 24485
    },
    {
      "epoch": 0.501505129727848,
      "grad_norm": 0.252315491437912,
      "learning_rate": 0.0002136553408004056,
      "loss": 4.7233,
      "step": 24490
    },
    {
      "epoch": 0.5016075195052526,
      "grad_norm": 0.2508240044116974,
      "learning_rate": 0.00021359928922532625,
      "loss": 4.7015,
      "step": 24495
    },
    {
      "epoch": 0.5017099092826572,
      "grad_norm": 0.27361515164375305,
      "learning_rate": 0.00021354322682108853,
      "loss": 4.7081,
      "step": 24500
    },
    {
      "epoch": 0.5018122990600619,
      "grad_norm": 0.2511800527572632,
      "learning_rate": 0.00021348715359723825,
      "loss": 4.7356,
      "step": 24505
    },
    {
      "epoch": 0.5019146888374665,
      "grad_norm": 0.25007495284080505,
      "learning_rate": 0.00021343106956332313,
      "loss": 4.7204,
      "step": 24510
    },
    {
      "epoch": 0.5020170786148711,
      "grad_norm": 0.2599777281284332,
      "learning_rate": 0.00021337497472889263,
      "loss": 4.7074,
      "step": 24515
    },
    {
      "epoch": 0.5021194683922757,
      "grad_norm": 0.26869091391563416,
      "learning_rate": 0.00021331886910349815,
      "loss": 4.7148,
      "step": 24520
    },
    {
      "epoch": 0.5022218581696803,
      "grad_norm": 0.27923208475112915,
      "learning_rate": 0.00021326275269669296,
      "loss": 4.655,
      "step": 24525
    },
    {
      "epoch": 0.502324247947085,
      "grad_norm": 0.2597445547580719,
      "learning_rate": 0.000213206625518032,
      "loss": 4.7167,
      "step": 24530
    },
    {
      "epoch": 0.5024266377244896,
      "grad_norm": 0.3291465938091278,
      "learning_rate": 0.00021315048757707215,
      "loss": 4.7631,
      "step": 24535
    },
    {
      "epoch": 0.5025290275018942,
      "grad_norm": 0.25353655219078064,
      "learning_rate": 0.0002130943388833721,
      "loss": 4.7166,
      "step": 24540
    },
    {
      "epoch": 0.5026314172792988,
      "grad_norm": 0.24162156879901886,
      "learning_rate": 0.00021303817944649242,
      "loss": 4.7468,
      "step": 24545
    },
    {
      "epoch": 0.5027338070567035,
      "grad_norm": 0.24550728499889374,
      "learning_rate": 0.00021298200927599546,
      "loss": 4.7383,
      "step": 24550
    },
    {
      "epoch": 0.5028361968341081,
      "grad_norm": 0.2765519320964813,
      "learning_rate": 0.00021292582838144546,
      "loss": 4.7263,
      "step": 24555
    },
    {
      "epoch": 0.5029385866115127,
      "grad_norm": 0.2508819103240967,
      "learning_rate": 0.00021286963677240832,
      "loss": 4.6969,
      "step": 24560
    },
    {
      "epoch": 0.5030409763889173,
      "grad_norm": 0.2590639889240265,
      "learning_rate": 0.00021281343445845197,
      "loss": 4.7415,
      "step": 24565
    },
    {
      "epoch": 0.5031433661663219,
      "grad_norm": 0.26587069034576416,
      "learning_rate": 0.00021275722144914607,
      "loss": 4.7175,
      "step": 24570
    },
    {
      "epoch": 0.5032457559437266,
      "grad_norm": 0.25631362199783325,
      "learning_rate": 0.00021270099775406206,
      "loss": 4.7137,
      "step": 24575
    },
    {
      "epoch": 0.5033481457211312,
      "grad_norm": 0.261626660823822,
      "learning_rate": 0.0002126447633827733,
      "loss": 4.6867,
      "step": 24580
    },
    {
      "epoch": 0.5034505354985358,
      "grad_norm": 0.2893737256526947,
      "learning_rate": 0.00021258851834485485,
      "loss": 4.7523,
      "step": 24585
    },
    {
      "epoch": 0.5035529252759404,
      "grad_norm": 0.27564457058906555,
      "learning_rate": 0.00021253226264988367,
      "loss": 4.7037,
      "step": 24590
    },
    {
      "epoch": 0.503655315053345,
      "grad_norm": 0.26541513204574585,
      "learning_rate": 0.00021247599630743857,
      "loss": 4.645,
      "step": 24595
    },
    {
      "epoch": 0.5037577048307497,
      "grad_norm": 0.24438264966011047,
      "learning_rate": 0.00021241971932710004,
      "loss": 4.7093,
      "step": 24600
    },
    {
      "epoch": 0.5038600946081543,
      "grad_norm": 0.2826894223690033,
      "learning_rate": 0.00021236343171845048,
      "loss": 4.7249,
      "step": 24605
    },
    {
      "epoch": 0.5039624843855589,
      "grad_norm": 0.2549968957901001,
      "learning_rate": 0.00021230713349107406,
      "loss": 4.708,
      "step": 24610
    },
    {
      "epoch": 0.5040648741629635,
      "grad_norm": 0.24997398257255554,
      "learning_rate": 0.00021225082465455684,
      "loss": 4.6947,
      "step": 24615
    },
    {
      "epoch": 0.5041672639403681,
      "grad_norm": 0.2487078458070755,
      "learning_rate": 0.0002121945052184865,
      "loss": 4.7086,
      "step": 24620
    },
    {
      "epoch": 0.5042696537177728,
      "grad_norm": 0.244797483086586,
      "learning_rate": 0.00021213817519245276,
      "loss": 4.6922,
      "step": 24625
    },
    {
      "epoch": 0.5043720434951774,
      "grad_norm": 0.24336354434490204,
      "learning_rate": 0.00021208183458604698,
      "loss": 4.7448,
      "step": 24630
    },
    {
      "epoch": 0.504474433272582,
      "grad_norm": 0.2476019561290741,
      "learning_rate": 0.00021202548340886235,
      "loss": 4.6856,
      "step": 24635
    },
    {
      "epoch": 0.5045768230499867,
      "grad_norm": 0.2573014199733734,
      "learning_rate": 0.00021196912167049397,
      "loss": 4.7023,
      "step": 24640
    },
    {
      "epoch": 0.5046792128273914,
      "grad_norm": 0.26645344495773315,
      "learning_rate": 0.00021191274938053853,
      "loss": 4.705,
      "step": 24645
    },
    {
      "epoch": 0.504781602604796,
      "grad_norm": 0.25060755014419556,
      "learning_rate": 0.0002118563665485947,
      "loss": 4.6944,
      "step": 24650
    },
    {
      "epoch": 0.5048839923822006,
      "grad_norm": 0.245570108294487,
      "learning_rate": 0.00021179997318426287,
      "loss": 4.7521,
      "step": 24655
    },
    {
      "epoch": 0.5049863821596052,
      "grad_norm": 0.2431022971868515,
      "learning_rate": 0.00021174356929714523,
      "loss": 4.7062,
      "step": 24660
    },
    {
      "epoch": 0.5050887719370099,
      "grad_norm": 0.25514674186706543,
      "learning_rate": 0.00021168715489684582,
      "loss": 4.6733,
      "step": 24665
    },
    {
      "epoch": 0.5051911617144145,
      "grad_norm": 0.25629234313964844,
      "learning_rate": 0.00021163072999297032,
      "loss": 4.7279,
      "step": 24670
    },
    {
      "epoch": 0.5052935514918191,
      "grad_norm": 0.2432234287261963,
      "learning_rate": 0.00021157429459512636,
      "loss": 4.7288,
      "step": 24675
    },
    {
      "epoch": 0.5053959412692237,
      "grad_norm": 0.2488624006509781,
      "learning_rate": 0.00021151784871292322,
      "loss": 4.7273,
      "step": 24680
    },
    {
      "epoch": 0.5054983310466283,
      "grad_norm": 0.2719564139842987,
      "learning_rate": 0.00021146139235597212,
      "loss": 4.6633,
      "step": 24685
    },
    {
      "epoch": 0.505600720824033,
      "grad_norm": 0.26692262291908264,
      "learning_rate": 0.0002114049255338859,
      "loss": 4.7522,
      "step": 24690
    },
    {
      "epoch": 0.5057031106014376,
      "grad_norm": 0.26064640283584595,
      "learning_rate": 0.00021134844825627935,
      "loss": 4.7518,
      "step": 24695
    },
    {
      "epoch": 0.5058055003788422,
      "grad_norm": 0.2623884677886963,
      "learning_rate": 0.00021129196053276892,
      "loss": 4.6863,
      "step": 24700
    },
    {
      "epoch": 0.5059078901562468,
      "grad_norm": 0.2715488374233246,
      "learning_rate": 0.0002112354623729728,
      "loss": 4.7502,
      "step": 24705
    },
    {
      "epoch": 0.5060102799336514,
      "grad_norm": 0.2498241662979126,
      "learning_rate": 0.00021117895378651113,
      "loss": 4.6931,
      "step": 24710
    },
    {
      "epoch": 0.5061126697110561,
      "grad_norm": 0.26348575949668884,
      "learning_rate": 0.00021112243478300566,
      "loss": 4.6818,
      "step": 24715
    },
    {
      "epoch": 0.5062150594884607,
      "grad_norm": 0.2448241412639618,
      "learning_rate": 0.00021106590537207999,
      "loss": 4.6994,
      "step": 24720
    },
    {
      "epoch": 0.5063174492658653,
      "grad_norm": 0.2700660228729248,
      "learning_rate": 0.00021100936556335946,
      "loss": 4.729,
      "step": 24725
    },
    {
      "epoch": 0.5064198390432699,
      "grad_norm": 0.25830090045928955,
      "learning_rate": 0.00021095281536647126,
      "loss": 4.6994,
      "step": 24730
    },
    {
      "epoch": 0.5065222288206745,
      "grad_norm": 0.2823108434677124,
      "learning_rate": 0.00021089625479104424,
      "loss": 4.6936,
      "step": 24735
    },
    {
      "epoch": 0.5066246185980792,
      "grad_norm": 0.27551400661468506,
      "learning_rate": 0.00021083968384670907,
      "loss": 4.6821,
      "step": 24740
    },
    {
      "epoch": 0.5067270083754838,
      "grad_norm": 0.25167158246040344,
      "learning_rate": 0.00021078310254309815,
      "loss": 4.6748,
      "step": 24745
    },
    {
      "epoch": 0.5068293981528884,
      "grad_norm": 0.28884032368659973,
      "learning_rate": 0.0002107265108898457,
      "loss": 4.7889,
      "step": 24750
    },
    {
      "epoch": 0.506931787930293,
      "grad_norm": 0.26263895630836487,
      "learning_rate": 0.0002106699088965877,
      "loss": 4.7195,
      "step": 24755
    },
    {
      "epoch": 0.5070341777076977,
      "grad_norm": 0.26021355390548706,
      "learning_rate": 0.00021061329657296184,
      "loss": 4.7819,
      "step": 24760
    },
    {
      "epoch": 0.5071365674851023,
      "grad_norm": 0.2741910517215729,
      "learning_rate": 0.00021055667392860764,
      "loss": 4.7208,
      "step": 24765
    },
    {
      "epoch": 0.5072389572625069,
      "grad_norm": 0.3258890211582184,
      "learning_rate": 0.00021050004097316623,
      "loss": 4.6659,
      "step": 24770
    },
    {
      "epoch": 0.5073413470399115,
      "grad_norm": 0.2708551287651062,
      "learning_rate": 0.00021044339771628072,
      "loss": 4.7837,
      "step": 24775
    },
    {
      "epoch": 0.5074437368173161,
      "grad_norm": 0.26590174436569214,
      "learning_rate": 0.0002103867441675958,
      "loss": 4.7363,
      "step": 24780
    },
    {
      "epoch": 0.5075461265947208,
      "grad_norm": 0.25326523184776306,
      "learning_rate": 0.00021033008033675794,
      "loss": 4.726,
      "step": 24785
    },
    {
      "epoch": 0.5076485163721254,
      "grad_norm": 0.2555302381515503,
      "learning_rate": 0.00021027340623341542,
      "loss": 4.6985,
      "step": 24790
    },
    {
      "epoch": 0.50775090614953,
      "grad_norm": 0.25960206985473633,
      "learning_rate": 0.00021021672186721826,
      "loss": 4.7171,
      "step": 24795
    },
    {
      "epoch": 0.5078532959269346,
      "grad_norm": 0.27280300855636597,
      "learning_rate": 0.00021016002724781808,
      "loss": 4.6646,
      "step": 24800
    },
    {
      "epoch": 0.5079556857043392,
      "grad_norm": 0.28094059228897095,
      "learning_rate": 0.00021010332238486853,
      "loss": 4.7707,
      "step": 24805
    },
    {
      "epoch": 0.5080580754817439,
      "grad_norm": 0.26008355617523193,
      "learning_rate": 0.0002100466072880248,
      "loss": 4.7649,
      "step": 24810
    },
    {
      "epoch": 0.5081604652591485,
      "grad_norm": 0.2612993121147156,
      "learning_rate": 0.00020998988196694378,
      "loss": 4.6929,
      "step": 24815
    },
    {
      "epoch": 0.5082628550365531,
      "grad_norm": 0.2467452585697174,
      "learning_rate": 0.00020993314643128424,
      "loss": 4.7253,
      "step": 24820
    },
    {
      "epoch": 0.5083652448139577,
      "grad_norm": 0.28310737013816833,
      "learning_rate": 0.00020987640069070667,
      "loss": 4.6759,
      "step": 24825
    },
    {
      "epoch": 0.5084676345913623,
      "grad_norm": 0.270855575799942,
      "learning_rate": 0.0002098196447548732,
      "loss": 4.699,
      "step": 24830
    },
    {
      "epoch": 0.508570024368767,
      "grad_norm": 0.2647227644920349,
      "learning_rate": 0.0002097628786334478,
      "loss": 4.7276,
      "step": 24835
    },
    {
      "epoch": 0.5086724141461716,
      "grad_norm": 0.28125500679016113,
      "learning_rate": 0.0002097061023360961,
      "loss": 4.7177,
      "step": 24840
    },
    {
      "epoch": 0.5087748039235763,
      "grad_norm": 0.24579207599163055,
      "learning_rate": 0.00020964931587248557,
      "loss": 4.7113,
      "step": 24845
    },
    {
      "epoch": 0.508877193700981,
      "grad_norm": 0.2567944824695587,
      "learning_rate": 0.00020959251925228528,
      "loss": 4.7238,
      "step": 24850
    },
    {
      "epoch": 0.5089795834783856,
      "grad_norm": 0.2535143196582794,
      "learning_rate": 0.00020953571248516606,
      "loss": 4.7567,
      "step": 24855
    },
    {
      "epoch": 0.5090819732557902,
      "grad_norm": 0.2610457241535187,
      "learning_rate": 0.00020947889558080052,
      "loss": 4.7,
      "step": 24860
    },
    {
      "epoch": 0.5091843630331948,
      "grad_norm": 0.24878627061843872,
      "learning_rate": 0.00020942206854886303,
      "loss": 4.7271,
      "step": 24865
    },
    {
      "epoch": 0.5092867528105994,
      "grad_norm": 0.2622509002685547,
      "learning_rate": 0.00020936523139902948,
      "loss": 4.6982,
      "step": 24870
    },
    {
      "epoch": 0.509389142588004,
      "grad_norm": 0.26678067445755005,
      "learning_rate": 0.0002093083841409778,
      "loss": 4.6707,
      "step": 24875
    },
    {
      "epoch": 0.5094915323654087,
      "grad_norm": 0.24062874913215637,
      "learning_rate": 0.00020925152678438736,
      "loss": 4.6962,
      "step": 24880
    },
    {
      "epoch": 0.5095939221428133,
      "grad_norm": 0.24843214452266693,
      "learning_rate": 0.0002091946593389394,
      "loss": 4.7047,
      "step": 24885
    },
    {
      "epoch": 0.5096963119202179,
      "grad_norm": 0.25527897477149963,
      "learning_rate": 0.0002091377818143168,
      "loss": 4.709,
      "step": 24890
    },
    {
      "epoch": 0.5097987016976225,
      "grad_norm": 0.2503810524940491,
      "learning_rate": 0.00020908089422020426,
      "loss": 4.7533,
      "step": 24895
    },
    {
      "epoch": 0.5099010914750272,
      "grad_norm": 0.23670713603496552,
      "learning_rate": 0.000209023996566288,
      "loss": 4.7076,
      "step": 24900
    },
    {
      "epoch": 0.5100034812524318,
      "grad_norm": 0.24563787877559662,
      "learning_rate": 0.00020896708886225624,
      "loss": 4.6996,
      "step": 24905
    },
    {
      "epoch": 0.5101058710298364,
      "grad_norm": 0.27107053995132446,
      "learning_rate": 0.00020891017111779862,
      "loss": 4.6914,
      "step": 24910
    },
    {
      "epoch": 0.510208260807241,
      "grad_norm": 0.25058692693710327,
      "learning_rate": 0.00020885324334260667,
      "loss": 4.6982,
      "step": 24915
    },
    {
      "epoch": 0.5103106505846456,
      "grad_norm": 0.2572046220302582,
      "learning_rate": 0.00020879630554637365,
      "loss": 4.6911,
      "step": 24920
    },
    {
      "epoch": 0.5104130403620503,
      "grad_norm": 0.26619139313697815,
      "learning_rate": 0.0002087393577387943,
      "loss": 4.7387,
      "step": 24925
    },
    {
      "epoch": 0.5105154301394549,
      "grad_norm": 0.24175913631916046,
      "learning_rate": 0.00020868239992956537,
      "loss": 4.7599,
      "step": 24930
    },
    {
      "epoch": 0.5106178199168595,
      "grad_norm": 0.24409683048725128,
      "learning_rate": 0.0002086254321283851,
      "loss": 4.6678,
      "step": 24935
    },
    {
      "epoch": 0.5107202096942641,
      "grad_norm": 0.26042482256889343,
      "learning_rate": 0.0002085684543449534,
      "loss": 4.7323,
      "step": 24940
    },
    {
      "epoch": 0.5108225994716687,
      "grad_norm": 0.284839928150177,
      "learning_rate": 0.00020851146658897214,
      "loss": 4.763,
      "step": 24945
    },
    {
      "epoch": 0.5109249892490734,
      "grad_norm": 0.284509539604187,
      "learning_rate": 0.00020845446887014465,
      "loss": 4.6373,
      "step": 24950
    },
    {
      "epoch": 0.511027379026478,
      "grad_norm": 0.2701762616634369,
      "learning_rate": 0.00020839746119817598,
      "loss": 4.7613,
      "step": 24955
    },
    {
      "epoch": 0.5111297688038826,
      "grad_norm": 0.26169511675834656,
      "learning_rate": 0.00020834044358277297,
      "loss": 4.7125,
      "step": 24960
    },
    {
      "epoch": 0.5112321585812872,
      "grad_norm": 0.34681040048599243,
      "learning_rate": 0.00020828341603364414,
      "loss": 4.7033,
      "step": 24965
    },
    {
      "epoch": 0.5113345483586919,
      "grad_norm": 0.26275140047073364,
      "learning_rate": 0.0002082263785604996,
      "loss": 4.7016,
      "step": 24970
    },
    {
      "epoch": 0.5114369381360965,
      "grad_norm": 0.4569779336452484,
      "learning_rate": 0.00020816933117305124,
      "loss": 4.6567,
      "step": 24975
    },
    {
      "epoch": 0.5115393279135011,
      "grad_norm": 0.27598029375076294,
      "learning_rate": 0.00020811227388101264,
      "loss": 4.6915,
      "step": 24980
    },
    {
      "epoch": 0.5116417176909057,
      "grad_norm": 0.2594294250011444,
      "learning_rate": 0.000208055206694099,
      "loss": 4.7079,
      "step": 24985
    },
    {
      "epoch": 0.5117441074683103,
      "grad_norm": 0.27375197410583496,
      "learning_rate": 0.0002079981296220273,
      "loss": 4.6944,
      "step": 24990
    },
    {
      "epoch": 0.511846497245715,
      "grad_norm": 0.26600924134254456,
      "learning_rate": 0.00020794104267451605,
      "loss": 4.7308,
      "step": 24995
    },
    {
      "epoch": 0.5119488870231196,
      "grad_norm": 0.27649980783462524,
      "learning_rate": 0.0002078839458612857,
      "loss": 4.7236,
      "step": 25000
    },
    {
      "epoch": 0.5119488870231196,
      "eval_loss": 4.726166725158691,
      "eval_runtime": 17.1335,
      "eval_samples_per_second": 33.969,
      "eval_steps_per_second": 4.261,
      "step": 25000
    },
    {
      "epoch": 0.5120512768005242,
      "grad_norm": 0.27099522948265076,
      "learning_rate": 0.0002078268391920581,
      "loss": 4.74,
      "step": 25005
    },
    {
      "epoch": 0.5121536665779288,
      "grad_norm": 0.2632291913032532,
      "learning_rate": 0.00020776972267655687,
      "loss": 4.7022,
      "step": 25010
    },
    {
      "epoch": 0.5122560563553334,
      "grad_norm": 0.3414386212825775,
      "learning_rate": 0.00020771259632450747,
      "loss": 4.6967,
      "step": 25015
    },
    {
      "epoch": 0.5123584461327381,
      "grad_norm": 0.2538946568965912,
      "learning_rate": 0.0002076554601456368,
      "loss": 4.6859,
      "step": 25020
    },
    {
      "epoch": 0.5124608359101427,
      "grad_norm": 0.2705353796482086,
      "learning_rate": 0.00020759831414967356,
      "loss": 4.6758,
      "step": 25025
    },
    {
      "epoch": 0.5125632256875473,
      "grad_norm": 0.2563190162181854,
      "learning_rate": 0.00020754115834634808,
      "loss": 4.6858,
      "step": 25030
    },
    {
      "epoch": 0.5126656154649519,
      "grad_norm": 0.2726873457431793,
      "learning_rate": 0.00020748399274539238,
      "loss": 4.6825,
      "step": 25035
    },
    {
      "epoch": 0.5127680052423566,
      "grad_norm": 0.24901781976222992,
      "learning_rate": 0.00020742681735654015,
      "loss": 4.6136,
      "step": 25040
    },
    {
      "epoch": 0.5128703950197612,
      "grad_norm": 0.28644707798957825,
      "learning_rate": 0.00020736963218952677,
      "loss": 4.7109,
      "step": 25045
    },
    {
      "epoch": 0.5129727847971659,
      "grad_norm": 0.2551325559616089,
      "learning_rate": 0.00020731243725408922,
      "loss": 4.6807,
      "step": 25050
    },
    {
      "epoch": 0.5130751745745705,
      "grad_norm": 0.27223217487335205,
      "learning_rate": 0.00020725523255996615,
      "loss": 4.7506,
      "step": 25055
    },
    {
      "epoch": 0.5131775643519751,
      "grad_norm": 0.25584861636161804,
      "learning_rate": 0.00020719801811689797,
      "loss": 4.7083,
      "step": 25060
    },
    {
      "epoch": 0.5132799541293798,
      "grad_norm": 0.2572309076786041,
      "learning_rate": 0.00020714079393462667,
      "loss": 4.7186,
      "step": 25065
    },
    {
      "epoch": 0.5133823439067844,
      "grad_norm": 0.2758883237838745,
      "learning_rate": 0.00020708356002289584,
      "loss": 4.6903,
      "step": 25070
    },
    {
      "epoch": 0.513484733684189,
      "grad_norm": 0.2584627866744995,
      "learning_rate": 0.00020702631639145087,
      "loss": 4.7015,
      "step": 25075
    },
    {
      "epoch": 0.5135871234615936,
      "grad_norm": 0.2779619097709656,
      "learning_rate": 0.00020696906305003862,
      "loss": 4.6788,
      "step": 25080
    },
    {
      "epoch": 0.5136895132389983,
      "grad_norm": 0.2768387198448181,
      "learning_rate": 0.0002069118000084079,
      "loss": 4.7122,
      "step": 25085
    },
    {
      "epoch": 0.5137919030164029,
      "grad_norm": 0.2851027846336365,
      "learning_rate": 0.00020685452727630882,
      "loss": 4.683,
      "step": 25090
    },
    {
      "epoch": 0.5138942927938075,
      "grad_norm": 0.25141647458076477,
      "learning_rate": 0.00020679724486349335,
      "loss": 4.6939,
      "step": 25095
    },
    {
      "epoch": 0.5139966825712121,
      "grad_norm": 0.2560679614543915,
      "learning_rate": 0.00020673995277971512,
      "loss": 4.7098,
      "step": 25100
    },
    {
      "epoch": 0.5140990723486167,
      "grad_norm": 0.24338766932487488,
      "learning_rate": 0.00020668265103472924,
      "loss": 4.7023,
      "step": 25105
    },
    {
      "epoch": 0.5142014621260214,
      "grad_norm": 0.26247096061706543,
      "learning_rate": 0.00020662533963829266,
      "loss": 4.7534,
      "step": 25110
    },
    {
      "epoch": 0.514303851903426,
      "grad_norm": 0.23718619346618652,
      "learning_rate": 0.0002065680186001639,
      "loss": 4.7087,
      "step": 25115
    },
    {
      "epoch": 0.5144062416808306,
      "grad_norm": 0.2554474472999573,
      "learning_rate": 0.00020651068793010305,
      "loss": 4.6975,
      "step": 25120
    },
    {
      "epoch": 0.5145086314582352,
      "grad_norm": 0.3297739326953888,
      "learning_rate": 0.00020645334763787194,
      "loss": 4.701,
      "step": 25125
    },
    {
      "epoch": 0.5146110212356398,
      "grad_norm": 0.28824537992477417,
      "learning_rate": 0.00020639599773323395,
      "loss": 4.6585,
      "step": 25130
    },
    {
      "epoch": 0.5147134110130445,
      "grad_norm": 0.29799553751945496,
      "learning_rate": 0.0002063386382259542,
      "loss": 4.7184,
      "step": 25135
    },
    {
      "epoch": 0.5148158007904491,
      "grad_norm": 0.2818523645401001,
      "learning_rate": 0.0002062812691257994,
      "loss": 4.7205,
      "step": 25140
    },
    {
      "epoch": 0.5149181905678537,
      "grad_norm": 0.5003705620765686,
      "learning_rate": 0.00020622389044253785,
      "loss": 4.7401,
      "step": 25145
    },
    {
      "epoch": 0.5150205803452583,
      "grad_norm": 0.24921785295009613,
      "learning_rate": 0.00020616650218593946,
      "loss": 4.7274,
      "step": 25150
    },
    {
      "epoch": 0.515122970122663,
      "grad_norm": 0.30534884333610535,
      "learning_rate": 0.00020610910436577595,
      "loss": 4.6842,
      "step": 25155
    },
    {
      "epoch": 0.5152253599000676,
      "grad_norm": 0.26093873381614685,
      "learning_rate": 0.00020605169699182044,
      "loss": 4.7124,
      "step": 25160
    },
    {
      "epoch": 0.5153277496774722,
      "grad_norm": 0.2568979263305664,
      "learning_rate": 0.00020599428007384783,
      "loss": 4.713,
      "step": 25165
    },
    {
      "epoch": 0.5154301394548768,
      "grad_norm": 0.25099727511405945,
      "learning_rate": 0.0002059368536216346,
      "loss": 4.679,
      "step": 25170
    },
    {
      "epoch": 0.5155325292322814,
      "grad_norm": 0.2550967037677765,
      "learning_rate": 0.0002058794176449588,
      "loss": 4.6942,
      "step": 25175
    },
    {
      "epoch": 0.515634919009686,
      "grad_norm": 0.2863609194755554,
      "learning_rate": 0.00020582197215360018,
      "loss": 4.7301,
      "step": 25180
    },
    {
      "epoch": 0.5157373087870907,
      "grad_norm": 0.2509368062019348,
      "learning_rate": 0.00020576451715734014,
      "loss": 4.7251,
      "step": 25185
    },
    {
      "epoch": 0.5158396985644953,
      "grad_norm": 0.26072463393211365,
      "learning_rate": 0.00020570705266596156,
      "loss": 4.7283,
      "step": 25190
    },
    {
      "epoch": 0.5159420883418999,
      "grad_norm": 0.24949073791503906,
      "learning_rate": 0.00020564957868924902,
      "loss": 4.6715,
      "step": 25195
    },
    {
      "epoch": 0.5160444781193045,
      "grad_norm": 0.27192243933677673,
      "learning_rate": 0.00020559209523698873,
      "loss": 4.7054,
      "step": 25200
    },
    {
      "epoch": 0.5161468678967092,
      "grad_norm": 0.2541794180870056,
      "learning_rate": 0.0002055346023189685,
      "loss": 4.7155,
      "step": 25205
    },
    {
      "epoch": 0.5162492576741138,
      "grad_norm": 0.24772867560386658,
      "learning_rate": 0.00020547709994497777,
      "loss": 4.7165,
      "step": 25210
    },
    {
      "epoch": 0.5163516474515184,
      "grad_norm": 0.25157585740089417,
      "learning_rate": 0.00020541958812480751,
      "loss": 4.7328,
      "step": 25215
    },
    {
      "epoch": 0.516454037228923,
      "grad_norm": 0.2582609951496124,
      "learning_rate": 0.0002053620668682504,
      "loss": 4.6885,
      "step": 25220
    },
    {
      "epoch": 0.5165564270063276,
      "grad_norm": 0.2692011594772339,
      "learning_rate": 0.00020530453618510064,
      "loss": 4.7216,
      "step": 25225
    },
    {
      "epoch": 0.5166588167837323,
      "grad_norm": 0.2550601363182068,
      "learning_rate": 0.00020524699608515416,
      "loss": 4.7391,
      "step": 25230
    },
    {
      "epoch": 0.5167612065611369,
      "grad_norm": 0.2598309814929962,
      "learning_rate": 0.00020518944657820827,
      "loss": 4.7313,
      "step": 25235
    },
    {
      "epoch": 0.5168635963385415,
      "grad_norm": 0.2678385078907013,
      "learning_rate": 0.00020513188767406216,
      "loss": 4.7061,
      "step": 25240
    },
    {
      "epoch": 0.5169659861159461,
      "grad_norm": 0.251507431268692,
      "learning_rate": 0.00020507431938251641,
      "loss": 4.729,
      "step": 25245
    },
    {
      "epoch": 0.5170683758933509,
      "grad_norm": 0.2598932981491089,
      "learning_rate": 0.0002050167417133733,
      "loss": 4.7031,
      "step": 25250
    },
    {
      "epoch": 0.5171707656707555,
      "grad_norm": 0.25043338537216187,
      "learning_rate": 0.00020495915467643664,
      "loss": 4.7563,
      "step": 25255
    },
    {
      "epoch": 0.5172731554481601,
      "grad_norm": 0.25743046402931213,
      "learning_rate": 0.00020490155828151185,
      "loss": 4.6892,
      "step": 25260
    },
    {
      "epoch": 0.5173755452255647,
      "grad_norm": 0.2507784962654114,
      "learning_rate": 0.0002048439525384061,
      "loss": 4.6957,
      "step": 25265
    },
    {
      "epoch": 0.5174779350029693,
      "grad_norm": 0.2602917551994324,
      "learning_rate": 0.00020478633745692784,
      "loss": 4.6799,
      "step": 25270
    },
    {
      "epoch": 0.517580324780374,
      "grad_norm": 0.32359644770622253,
      "learning_rate": 0.00020472871304688743,
      "loss": 4.7462,
      "step": 25275
    },
    {
      "epoch": 0.5176827145577786,
      "grad_norm": 0.27157193422317505,
      "learning_rate": 0.00020467107931809663,
      "loss": 4.7443,
      "step": 25280
    },
    {
      "epoch": 0.5177851043351832,
      "grad_norm": 0.2694609463214874,
      "learning_rate": 0.00020461343628036882,
      "loss": 4.7407,
      "step": 25285
    },
    {
      "epoch": 0.5178874941125878,
      "grad_norm": 0.25350046157836914,
      "learning_rate": 0.00020455578394351894,
      "loss": 4.6895,
      "step": 25290
    },
    {
      "epoch": 0.5179898838899925,
      "grad_norm": 0.2541326880455017,
      "learning_rate": 0.0002044981223173636,
      "loss": 4.7259,
      "step": 25295
    },
    {
      "epoch": 0.5180922736673971,
      "grad_norm": 0.2908807098865509,
      "learning_rate": 0.00020444045141172095,
      "loss": 4.7015,
      "step": 25300
    },
    {
      "epoch": 0.5181946634448017,
      "grad_norm": 0.24267639219760895,
      "learning_rate": 0.00020438277123641067,
      "loss": 4.6619,
      "step": 25305
    },
    {
      "epoch": 0.5182970532222063,
      "grad_norm": 0.2692250907421112,
      "learning_rate": 0.00020432508180125413,
      "loss": 4.6891,
      "step": 25310
    },
    {
      "epoch": 0.5183994429996109,
      "grad_norm": 0.6035641431808472,
      "learning_rate": 0.0002042673831160741,
      "loss": 4.7316,
      "step": 25315
    },
    {
      "epoch": 0.5185018327770156,
      "grad_norm": 0.2575273811817169,
      "learning_rate": 0.00020420967519069517,
      "loss": 4.7513,
      "step": 25320
    },
    {
      "epoch": 0.5186042225544202,
      "grad_norm": 0.25130707025527954,
      "learning_rate": 0.00020415195803494325,
      "loss": 4.7517,
      "step": 25325
    },
    {
      "epoch": 0.5187066123318248,
      "grad_norm": 0.2885972857475281,
      "learning_rate": 0.00020409423165864598,
      "loss": 4.6608,
      "step": 25330
    },
    {
      "epoch": 0.5188090021092294,
      "grad_norm": 0.25472113490104675,
      "learning_rate": 0.00020403649607163253,
      "loss": 4.7282,
      "step": 25335
    },
    {
      "epoch": 0.518911391886634,
      "grad_norm": 0.3163636326789856,
      "learning_rate": 0.0002039787512837336,
      "loss": 4.7619,
      "step": 25340
    },
    {
      "epoch": 0.5190137816640387,
      "grad_norm": 0.24546130001544952,
      "learning_rate": 0.00020392099730478158,
      "loss": 4.6859,
      "step": 25345
    },
    {
      "epoch": 0.5191161714414433,
      "grad_norm": 0.2689363658428192,
      "learning_rate": 0.00020386323414461026,
      "loss": 4.8244,
      "step": 25350
    },
    {
      "epoch": 0.5192185612188479,
      "grad_norm": 0.25374606251716614,
      "learning_rate": 0.00020380546181305513,
      "loss": 4.6985,
      "step": 25355
    },
    {
      "epoch": 0.5193209509962525,
      "grad_norm": 0.242715984582901,
      "learning_rate": 0.00020374768031995307,
      "loss": 4.7155,
      "step": 25360
    },
    {
      "epoch": 0.5194233407736571,
      "grad_norm": 0.2540392577648163,
      "learning_rate": 0.00020368988967514275,
      "loss": 4.6721,
      "step": 25365
    },
    {
      "epoch": 0.5195257305510618,
      "grad_norm": 0.2697451114654541,
      "learning_rate": 0.00020363208988846424,
      "loss": 4.6991,
      "step": 25370
    },
    {
      "epoch": 0.5196281203284664,
      "grad_norm": 0.24948923289775848,
      "learning_rate": 0.00020357428096975922,
      "loss": 4.7085,
      "step": 25375
    },
    {
      "epoch": 0.519730510105871,
      "grad_norm": 0.25771522521972656,
      "learning_rate": 0.0002035164629288709,
      "loss": 4.7562,
      "step": 25380
    },
    {
      "epoch": 0.5198328998832756,
      "grad_norm": 0.2981756627559662,
      "learning_rate": 0.00020345863577564402,
      "loss": 4.6578,
      "step": 25385
    },
    {
      "epoch": 0.5199352896606803,
      "grad_norm": 0.25592824816703796,
      "learning_rate": 0.00020340079951992497,
      "loss": 4.7013,
      "step": 25390
    },
    {
      "epoch": 0.5200376794380849,
      "grad_norm": 0.2738116979598999,
      "learning_rate": 0.0002033429541715616,
      "loss": 4.7512,
      "step": 25395
    },
    {
      "epoch": 0.5201400692154895,
      "grad_norm": 0.2536337375640869,
      "learning_rate": 0.00020328509974040332,
      "loss": 4.7228,
      "step": 25400
    },
    {
      "epoch": 0.5202424589928941,
      "grad_norm": 0.24680015444755554,
      "learning_rate": 0.0002032272362363011,
      "loss": 4.6811,
      "step": 25405
    },
    {
      "epoch": 0.5203448487702987,
      "grad_norm": 0.2532963454723358,
      "learning_rate": 0.0002031693636691075,
      "loss": 4.6936,
      "step": 25410
    },
    {
      "epoch": 0.5204472385477034,
      "grad_norm": 0.2937077283859253,
      "learning_rate": 0.00020311148204867652,
      "loss": 4.6856,
      "step": 25415
    },
    {
      "epoch": 0.520549628325108,
      "grad_norm": 0.2579612731933594,
      "learning_rate": 0.00020305359138486385,
      "loss": 4.7116,
      "step": 25420
    },
    {
      "epoch": 0.5206520181025126,
      "grad_norm": 0.27243170142173767,
      "learning_rate": 0.00020299569168752657,
      "loss": 4.6914,
      "step": 25425
    },
    {
      "epoch": 0.5207544078799172,
      "grad_norm": 0.25620657205581665,
      "learning_rate": 0.00020293778296652328,
      "loss": 4.6985,
      "step": 25430
    },
    {
      "epoch": 0.5208567976573218,
      "grad_norm": 0.2616143226623535,
      "learning_rate": 0.0002028798652317143,
      "loss": 4.6425,
      "step": 25435
    },
    {
      "epoch": 0.5209591874347265,
      "grad_norm": 0.30034929513931274,
      "learning_rate": 0.00020282193849296143,
      "loss": 4.7288,
      "step": 25440
    },
    {
      "epoch": 0.5210615772121311,
      "grad_norm": 0.25712305307388306,
      "learning_rate": 0.00020276400276012785,
      "loss": 4.7359,
      "step": 25445
    },
    {
      "epoch": 0.5211639669895357,
      "grad_norm": 0.2787604331970215,
      "learning_rate": 0.00020270605804307843,
      "loss": 4.6328,
      "step": 25450
    },
    {
      "epoch": 0.5212663567669404,
      "grad_norm": 0.26481419801712036,
      "learning_rate": 0.00020264810435167944,
      "loss": 4.6835,
      "step": 25455
    },
    {
      "epoch": 0.5213687465443451,
      "grad_norm": 0.2781488299369812,
      "learning_rate": 0.0002025901416957988,
      "loss": 4.7679,
      "step": 25460
    },
    {
      "epoch": 0.5214711363217497,
      "grad_norm": 0.29370707273483276,
      "learning_rate": 0.00020253217008530598,
      "loss": 4.7493,
      "step": 25465
    },
    {
      "epoch": 0.5215735260991543,
      "grad_norm": 0.28654757142066956,
      "learning_rate": 0.0002024741895300718,
      "loss": 4.7388,
      "step": 25470
    },
    {
      "epoch": 0.5216759158765589,
      "grad_norm": 0.2706935405731201,
      "learning_rate": 0.00020241620003996877,
      "loss": 4.6992,
      "step": 25475
    },
    {
      "epoch": 0.5217783056539635,
      "grad_norm": 0.27572545409202576,
      "learning_rate": 0.0002023582016248708,
      "loss": 4.6964,
      "step": 25480
    },
    {
      "epoch": 0.5218806954313682,
      "grad_norm": 0.2670837640762329,
      "learning_rate": 0.00020230019429465343,
      "loss": 4.7087,
      "step": 25485
    },
    {
      "epoch": 0.5219830852087728,
      "grad_norm": 0.24532637000083923,
      "learning_rate": 0.00020224217805919363,
      "loss": 4.7042,
      "step": 25490
    },
    {
      "epoch": 0.5220854749861774,
      "grad_norm": 0.2530348300933838,
      "learning_rate": 0.00020218415292837,
      "loss": 4.7387,
      "step": 25495
    },
    {
      "epoch": 0.522187864763582,
      "grad_norm": 0.23660606145858765,
      "learning_rate": 0.00020212611891206247,
      "loss": 4.7322,
      "step": 25500
    },
    {
      "epoch": 0.5222902545409867,
      "grad_norm": 0.28594502806663513,
      "learning_rate": 0.0002020680760201526,
      "loss": 4.7184,
      "step": 25505
    },
    {
      "epoch": 0.5223926443183913,
      "grad_norm": 0.2688608169555664,
      "learning_rate": 0.00020201002426252358,
      "loss": 4.7371,
      "step": 25510
    },
    {
      "epoch": 0.5224950340957959,
      "grad_norm": 0.26218748092651367,
      "learning_rate": 0.00020195196364905982,
      "loss": 4.7225,
      "step": 25515
    },
    {
      "epoch": 0.5225974238732005,
      "grad_norm": 0.24824967980384827,
      "learning_rate": 0.0002018938941896475,
      "loss": 4.6963,
      "step": 25520
    },
    {
      "epoch": 0.5226998136506051,
      "grad_norm": 0.30935096740722656,
      "learning_rate": 0.0002018358158941742,
      "loss": 4.767,
      "step": 25525
    },
    {
      "epoch": 0.5228022034280098,
      "grad_norm": 0.27268844842910767,
      "learning_rate": 0.0002017777287725289,
      "loss": 4.6938,
      "step": 25530
    },
    {
      "epoch": 0.5229045932054144,
      "grad_norm": 0.2782441973686218,
      "learning_rate": 0.00020171963283460237,
      "loss": 4.6773,
      "step": 25535
    },
    {
      "epoch": 0.523006982982819,
      "grad_norm": 0.2511165142059326,
      "learning_rate": 0.00020166152809028654,
      "loss": 4.7178,
      "step": 25540
    },
    {
      "epoch": 0.5231093727602236,
      "grad_norm": 0.28301969170570374,
      "learning_rate": 0.00020160341454947514,
      "loss": 4.6913,
      "step": 25545
    },
    {
      "epoch": 0.5232117625376282,
      "grad_norm": 0.25940707325935364,
      "learning_rate": 0.0002015452922220631,
      "loss": 4.7002,
      "step": 25550
    },
    {
      "epoch": 0.5233141523150329,
      "grad_norm": 0.25021982192993164,
      "learning_rate": 0.00020148716111794717,
      "loss": 4.6538,
      "step": 25555
    },
    {
      "epoch": 0.5234165420924375,
      "grad_norm": 0.2511366009712219,
      "learning_rate": 0.00020142902124702534,
      "loss": 4.7248,
      "step": 25560
    },
    {
      "epoch": 0.5235189318698421,
      "grad_norm": 0.25236549973487854,
      "learning_rate": 0.00020137087261919723,
      "loss": 4.7055,
      "step": 25565
    },
    {
      "epoch": 0.5236213216472467,
      "grad_norm": 0.42735183238983154,
      "learning_rate": 0.00020131271524436388,
      "loss": 4.6687,
      "step": 25570
    },
    {
      "epoch": 0.5237237114246514,
      "grad_norm": 0.26809757947921753,
      "learning_rate": 0.00020125454913242778,
      "loss": 4.7019,
      "step": 25575
    },
    {
      "epoch": 0.523826101202056,
      "grad_norm": 0.3365897238254547,
      "learning_rate": 0.0002011963742932931,
      "loss": 4.6834,
      "step": 25580
    },
    {
      "epoch": 0.5239284909794606,
      "grad_norm": 0.2797791063785553,
      "learning_rate": 0.00020113819073686528,
      "loss": 4.7245,
      "step": 25585
    },
    {
      "epoch": 0.5240308807568652,
      "grad_norm": 0.25630486011505127,
      "learning_rate": 0.0002010799984730514,
      "loss": 4.7393,
      "step": 25590
    },
    {
      "epoch": 0.5241332705342698,
      "grad_norm": 0.26039737462997437,
      "learning_rate": 0.00020102179751175982,
      "loss": 4.7142,
      "step": 25595
    },
    {
      "epoch": 0.5242356603116745,
      "grad_norm": 0.2776634097099304,
      "learning_rate": 0.00020096358786290069,
      "loss": 4.676,
      "step": 25600
    },
    {
      "epoch": 0.5243380500890791,
      "grad_norm": 0.2771133482456207,
      "learning_rate": 0.00020090536953638535,
      "loss": 4.7037,
      "step": 25605
    },
    {
      "epoch": 0.5244404398664837,
      "grad_norm": 0.24834643304347992,
      "learning_rate": 0.00020084714254212677,
      "loss": 4.6868,
      "step": 25610
    },
    {
      "epoch": 0.5245428296438883,
      "grad_norm": 0.2764914929866791,
      "learning_rate": 0.00020078890689003936,
      "loss": 4.7024,
      "step": 25615
    },
    {
      "epoch": 0.5246452194212929,
      "grad_norm": 0.2530945837497711,
      "learning_rate": 0.00020073066259003897,
      "loss": 4.7163,
      "step": 25620
    },
    {
      "epoch": 0.5247476091986976,
      "grad_norm": 0.24554969370365143,
      "learning_rate": 0.000200672409652043,
      "loss": 4.692,
      "step": 25625
    },
    {
      "epoch": 0.5248499989761022,
      "grad_norm": 0.25487956404685974,
      "learning_rate": 0.00020061414808597027,
      "loss": 4.7367,
      "step": 25630
    },
    {
      "epoch": 0.5249523887535068,
      "grad_norm": 0.24713367223739624,
      "learning_rate": 0.00020055587790174107,
      "loss": 4.6828,
      "step": 25635
    },
    {
      "epoch": 0.5250547785309114,
      "grad_norm": 0.27729615569114685,
      "learning_rate": 0.00020049759910927716,
      "loss": 4.6629,
      "step": 25640
    },
    {
      "epoch": 0.525157168308316,
      "grad_norm": 0.25373226404190063,
      "learning_rate": 0.00020043931171850177,
      "loss": 4.6707,
      "step": 25645
    },
    {
      "epoch": 0.5252595580857207,
      "grad_norm": 0.2616504728794098,
      "learning_rate": 0.00020038101573933956,
      "loss": 4.745,
      "step": 25650
    },
    {
      "epoch": 0.5253619478631254,
      "grad_norm": 0.2694254219532013,
      "learning_rate": 0.0002003227111817167,
      "loss": 4.659,
      "step": 25655
    },
    {
      "epoch": 0.52546433764053,
      "grad_norm": 0.23335018754005432,
      "learning_rate": 0.0002002643980555609,
      "loss": 4.6865,
      "step": 25660
    },
    {
      "epoch": 0.5255667274179346,
      "grad_norm": 0.2629484236240387,
      "learning_rate": 0.00020020607637080104,
      "loss": 4.7246,
      "step": 25665
    },
    {
      "epoch": 0.5256691171953393,
      "grad_norm": 0.24341432750225067,
      "learning_rate": 0.0002001477461373678,
      "loss": 4.6833,
      "step": 25670
    },
    {
      "epoch": 0.5257715069727439,
      "grad_norm": 0.24989259243011475,
      "learning_rate": 0.0002000894073651932,
      "loss": 4.6567,
      "step": 25675
    },
    {
      "epoch": 0.5258738967501485,
      "grad_norm": 0.2554193437099457,
      "learning_rate": 0.00020003106006421056,
      "loss": 4.7217,
      "step": 25680
    },
    {
      "epoch": 0.5259762865275531,
      "grad_norm": 0.2519819140434265,
      "learning_rate": 0.0001999727042443548,
      "loss": 4.7194,
      "step": 25685
    },
    {
      "epoch": 0.5260786763049577,
      "grad_norm": 0.2581077218055725,
      "learning_rate": 0.00019991433991556234,
      "loss": 4.7198,
      "step": 25690
    },
    {
      "epoch": 0.5261810660823624,
      "grad_norm": 0.24898824095726013,
      "learning_rate": 0.0001998559670877709,
      "loss": 4.7215,
      "step": 25695
    },
    {
      "epoch": 0.526283455859767,
      "grad_norm": 0.24466998875141144,
      "learning_rate": 0.00019979758577091976,
      "loss": 4.6968,
      "step": 25700
    },
    {
      "epoch": 0.5263858456371716,
      "grad_norm": 0.28194326162338257,
      "learning_rate": 0.00019973919597494959,
      "loss": 4.6962,
      "step": 25705
    },
    {
      "epoch": 0.5264882354145762,
      "grad_norm": 0.2566097676753998,
      "learning_rate": 0.0001996807977098025,
      "loss": 4.6931,
      "step": 25710
    },
    {
      "epoch": 0.5265906251919809,
      "grad_norm": 0.23906411230564117,
      "learning_rate": 0.00019962239098542207,
      "loss": 4.685,
      "step": 25715
    },
    {
      "epoch": 0.5266930149693855,
      "grad_norm": 0.2488403469324112,
      "learning_rate": 0.00019956397581175333,
      "loss": 4.6711,
      "step": 25720
    },
    {
      "epoch": 0.5267954047467901,
      "grad_norm": 0.2994610369205475,
      "learning_rate": 0.00019950555219874273,
      "loss": 4.64,
      "step": 25725
    },
    {
      "epoch": 0.5268977945241947,
      "grad_norm": 0.26433423161506653,
      "learning_rate": 0.00019944712015633817,
      "loss": 4.6997,
      "step": 25730
    },
    {
      "epoch": 0.5270001843015993,
      "grad_norm": 0.26021307706832886,
      "learning_rate": 0.00019938867969448892,
      "loss": 4.7127,
      "step": 25735
    },
    {
      "epoch": 0.527102574079004,
      "grad_norm": 0.2602417767047882,
      "learning_rate": 0.00019933023082314573,
      "loss": 4.7322,
      "step": 25740
    },
    {
      "epoch": 0.5272049638564086,
      "grad_norm": 0.25909242033958435,
      "learning_rate": 0.0001992717735522609,
      "loss": 4.7043,
      "step": 25745
    },
    {
      "epoch": 0.5273073536338132,
      "grad_norm": 0.2980313301086426,
      "learning_rate": 0.00019921330789178794,
      "loss": 4.7074,
      "step": 25750
    },
    {
      "epoch": 0.5274097434112178,
      "grad_norm": 0.2667423188686371,
      "learning_rate": 0.00019915483385168196,
      "loss": 4.6994,
      "step": 25755
    },
    {
      "epoch": 0.5275121331886224,
      "grad_norm": 0.25480931997299194,
      "learning_rate": 0.00019909635144189932,
      "loss": 4.7004,
      "step": 25760
    },
    {
      "epoch": 0.5276145229660271,
      "grad_norm": 0.2579711973667145,
      "learning_rate": 0.00019903786067239807,
      "loss": 4.6608,
      "step": 25765
    },
    {
      "epoch": 0.5277169127434317,
      "grad_norm": 0.26290157437324524,
      "learning_rate": 0.00019897936155313748,
      "loss": 4.7055,
      "step": 25770
    },
    {
      "epoch": 0.5278193025208363,
      "grad_norm": 0.23719623684883118,
      "learning_rate": 0.00019892085409407826,
      "loss": 4.6872,
      "step": 25775
    },
    {
      "epoch": 0.5279216922982409,
      "grad_norm": 0.2714715600013733,
      "learning_rate": 0.00019886233830518257,
      "loss": 4.7071,
      "step": 25780
    },
    {
      "epoch": 0.5280240820756456,
      "grad_norm": 0.25175049901008606,
      "learning_rate": 0.000198803814196414,
      "loss": 4.6765,
      "step": 25785
    },
    {
      "epoch": 0.5281264718530502,
      "grad_norm": 0.2570139169692993,
      "learning_rate": 0.00019874528177773763,
      "loss": 4.7269,
      "step": 25790
    },
    {
      "epoch": 0.5282288616304548,
      "grad_norm": 0.2545175552368164,
      "learning_rate": 0.00019868674105911972,
      "loss": 4.6839,
      "step": 25795
    },
    {
      "epoch": 0.5283312514078594,
      "grad_norm": 0.24411524832248688,
      "learning_rate": 0.0001986281920505283,
      "loss": 4.7008,
      "step": 25800
    },
    {
      "epoch": 0.528433641185264,
      "grad_norm": 0.2793164849281311,
      "learning_rate": 0.00019856963476193236,
      "loss": 4.6892,
      "step": 25805
    },
    {
      "epoch": 0.5285360309626687,
      "grad_norm": 0.2509508728981018,
      "learning_rate": 0.00019851106920330277,
      "loss": 4.7038,
      "step": 25810
    },
    {
      "epoch": 0.5286384207400733,
      "grad_norm": 0.2581671476364136,
      "learning_rate": 0.00019845249538461145,
      "loss": 4.6889,
      "step": 25815
    },
    {
      "epoch": 0.5287408105174779,
      "grad_norm": 0.2587572932243347,
      "learning_rate": 0.00019839391331583192,
      "loss": 4.711,
      "step": 25820
    },
    {
      "epoch": 0.5288432002948825,
      "grad_norm": 0.24888691306114197,
      "learning_rate": 0.00019833532300693907,
      "loss": 4.6986,
      "step": 25825
    },
    {
      "epoch": 0.5289455900722871,
      "grad_norm": 0.2639336585998535,
      "learning_rate": 0.00019827672446790914,
      "loss": 4.7135,
      "step": 25830
    },
    {
      "epoch": 0.5290479798496918,
      "grad_norm": 0.2600058913230896,
      "learning_rate": 0.00019821811770871976,
      "loss": 4.6807,
      "step": 25835
    },
    {
      "epoch": 0.5291503696270964,
      "grad_norm": 0.2666216492652893,
      "learning_rate": 0.0001981595027393501,
      "loss": 4.6908,
      "step": 25840
    },
    {
      "epoch": 0.529252759404501,
      "grad_norm": 0.2627016305923462,
      "learning_rate": 0.00019810087956978055,
      "loss": 4.712,
      "step": 25845
    },
    {
      "epoch": 0.5293551491819056,
      "grad_norm": 0.27691221237182617,
      "learning_rate": 0.00019804224820999295,
      "loss": 4.7209,
      "step": 25850
    },
    {
      "epoch": 0.5294575389593102,
      "grad_norm": 0.24884915351867676,
      "learning_rate": 0.00019798360866997065,
      "loss": 4.6783,
      "step": 25855
    },
    {
      "epoch": 0.529559928736715,
      "grad_norm": 0.25666943192481995,
      "learning_rate": 0.00019792496095969827,
      "loss": 4.6969,
      "step": 25860
    },
    {
      "epoch": 0.5296623185141196,
      "grad_norm": 0.23978044092655182,
      "learning_rate": 0.00019786630508916183,
      "loss": 4.7119,
      "step": 25865
    },
    {
      "epoch": 0.5297647082915242,
      "grad_norm": 0.2427990585565567,
      "learning_rate": 0.0001978076410683488,
      "loss": 4.7305,
      "step": 25870
    },
    {
      "epoch": 0.5298670980689288,
      "grad_norm": 0.2595253884792328,
      "learning_rate": 0.00019774896890724798,
      "loss": 4.6835,
      "step": 25875
    },
    {
      "epoch": 0.5299694878463335,
      "grad_norm": 0.2647393047809601,
      "learning_rate": 0.00019769028861584955,
      "loss": 4.7471,
      "step": 25880
    },
    {
      "epoch": 0.5300718776237381,
      "grad_norm": 0.2709347903728485,
      "learning_rate": 0.00019763160020414518,
      "loss": 4.709,
      "step": 25885
    },
    {
      "epoch": 0.5301742674011427,
      "grad_norm": 0.2865538001060486,
      "learning_rate": 0.00019757290368212777,
      "loss": 4.7528,
      "step": 25890
    },
    {
      "epoch": 0.5302766571785473,
      "grad_norm": 0.2557232081890106,
      "learning_rate": 0.00019751419905979177,
      "loss": 4.7577,
      "step": 25895
    },
    {
      "epoch": 0.530379046955952,
      "grad_norm": 0.26313456892967224,
      "learning_rate": 0.00019745548634713276,
      "loss": 4.7332,
      "step": 25900
    },
    {
      "epoch": 0.5304814367333566,
      "grad_norm": 0.2618166506290436,
      "learning_rate": 0.00019739676555414795,
      "loss": 4.7336,
      "step": 25905
    },
    {
      "epoch": 0.5305838265107612,
      "grad_norm": 0.26226091384887695,
      "learning_rate": 0.0001973380366908359,
      "loss": 4.7233,
      "step": 25910
    },
    {
      "epoch": 0.5306862162881658,
      "grad_norm": 0.24330765008926392,
      "learning_rate": 0.00019727929976719633,
      "loss": 4.7354,
      "step": 25915
    },
    {
      "epoch": 0.5307886060655704,
      "grad_norm": 0.2726133465766907,
      "learning_rate": 0.00019722055479323057,
      "loss": 4.7441,
      "step": 25920
    },
    {
      "epoch": 0.530890995842975,
      "grad_norm": 0.2631099224090576,
      "learning_rate": 0.00019716180177894118,
      "loss": 4.7255,
      "step": 25925
    },
    {
      "epoch": 0.5309933856203797,
      "grad_norm": 0.28416886925697327,
      "learning_rate": 0.00019710304073433215,
      "loss": 4.6626,
      "step": 25930
    },
    {
      "epoch": 0.5310957753977843,
      "grad_norm": 0.25636976957321167,
      "learning_rate": 0.0001970442716694088,
      "loss": 4.6742,
      "step": 25935
    },
    {
      "epoch": 0.5311981651751889,
      "grad_norm": 0.25970378518104553,
      "learning_rate": 0.00019698549459417792,
      "loss": 4.6955,
      "step": 25940
    },
    {
      "epoch": 0.5313005549525935,
      "grad_norm": 0.26659950613975525,
      "learning_rate": 0.0001969267095186475,
      "loss": 4.7669,
      "step": 25945
    },
    {
      "epoch": 0.5314029447299982,
      "grad_norm": 0.3476724624633789,
      "learning_rate": 0.00019686791645282697,
      "loss": 4.6512,
      "step": 25950
    },
    {
      "epoch": 0.5315053345074028,
      "grad_norm": 0.2607884705066681,
      "learning_rate": 0.00019680911540672716,
      "loss": 4.6727,
      "step": 25955
    },
    {
      "epoch": 0.5316077242848074,
      "grad_norm": 0.3115827143192291,
      "learning_rate": 0.00019675030639036022,
      "loss": 4.7441,
      "step": 25960
    },
    {
      "epoch": 0.531710114062212,
      "grad_norm": 0.26234233379364014,
      "learning_rate": 0.00019669148941373966,
      "loss": 4.7156,
      "step": 25965
    },
    {
      "epoch": 0.5318125038396166,
      "grad_norm": 0.25020137429237366,
      "learning_rate": 0.0001966326644868803,
      "loss": 4.7013,
      "step": 25970
    },
    {
      "epoch": 0.5319148936170213,
      "grad_norm": 0.26293113827705383,
      "learning_rate": 0.00019657383161979846,
      "loss": 4.7357,
      "step": 25975
    },
    {
      "epoch": 0.5320172833944259,
      "grad_norm": 0.2511340081691742,
      "learning_rate": 0.00019651499082251167,
      "loss": 4.686,
      "step": 25980
    },
    {
      "epoch": 0.5321196731718305,
      "grad_norm": 0.26991066336631775,
      "learning_rate": 0.0001964561421050388,
      "loss": 4.6865,
      "step": 25985
    },
    {
      "epoch": 0.5322220629492351,
      "grad_norm": 0.26026806235313416,
      "learning_rate": 0.00019639728547740014,
      "loss": 4.7252,
      "step": 25990
    },
    {
      "epoch": 0.5323244527266398,
      "grad_norm": 0.2697601914405823,
      "learning_rate": 0.00019633842094961732,
      "loss": 4.6772,
      "step": 25995
    },
    {
      "epoch": 0.5324268425040444,
      "grad_norm": 0.27490803599357605,
      "learning_rate": 0.00019627954853171332,
      "loss": 4.6808,
      "step": 26000
    },
    {
      "epoch": 0.5324268425040444,
      "eval_loss": 4.715832233428955,
      "eval_runtime": 17.1217,
      "eval_samples_per_second": 33.992,
      "eval_steps_per_second": 4.264,
      "step": 26000
    },
    {
      "epoch": 0.532529232281449,
      "grad_norm": 0.2695653438568115,
      "learning_rate": 0.00019622066823371247,
      "loss": 4.7279,
      "step": 26005
    },
    {
      "epoch": 0.5326316220588536,
      "grad_norm": 0.2591971457004547,
      "learning_rate": 0.00019616178006564032,
      "loss": 4.7927,
      "step": 26010
    },
    {
      "epoch": 0.5327340118362582,
      "grad_norm": 0.26702797412872314,
      "learning_rate": 0.00019610288403752393,
      "loss": 4.6334,
      "step": 26015
    },
    {
      "epoch": 0.5328364016136629,
      "grad_norm": 0.26594123244285583,
      "learning_rate": 0.0001960439801593916,
      "loss": 4.7251,
      "step": 26020
    },
    {
      "epoch": 0.5329387913910675,
      "grad_norm": 0.25022926926612854,
      "learning_rate": 0.00019598506844127305,
      "loss": 4.6757,
      "step": 26025
    },
    {
      "epoch": 0.5330411811684721,
      "grad_norm": 0.26219940185546875,
      "learning_rate": 0.0001959261488931992,
      "loss": 4.7274,
      "step": 26030
    },
    {
      "epoch": 0.5331435709458767,
      "grad_norm": 0.26517900824546814,
      "learning_rate": 0.0001958672215252024,
      "loss": 4.6714,
      "step": 26035
    },
    {
      "epoch": 0.5332459607232813,
      "grad_norm": 0.24350258708000183,
      "learning_rate": 0.0001958082863473164,
      "loss": 4.7222,
      "step": 26040
    },
    {
      "epoch": 0.533348350500686,
      "grad_norm": 0.2614833414554596,
      "learning_rate": 0.00019574934336957604,
      "loss": 4.717,
      "step": 26045
    },
    {
      "epoch": 0.5334507402780906,
      "grad_norm": 0.2601943910121918,
      "learning_rate": 0.00019569039260201775,
      "loss": 4.74,
      "step": 26050
    },
    {
      "epoch": 0.5335531300554952,
      "grad_norm": 0.28414246439933777,
      "learning_rate": 0.00019563143405467915,
      "loss": 4.7131,
      "step": 26055
    },
    {
      "epoch": 0.5336555198328999,
      "grad_norm": 0.2536947727203369,
      "learning_rate": 0.00019557246773759923,
      "loss": 4.7276,
      "step": 26060
    },
    {
      "epoch": 0.5337579096103046,
      "grad_norm": 0.2547471523284912,
      "learning_rate": 0.00019551349366081822,
      "loss": 4.6709,
      "step": 26065
    },
    {
      "epoch": 0.5338602993877092,
      "grad_norm": 0.27495211362838745,
      "learning_rate": 0.00019545451183437777,
      "loss": 4.7397,
      "step": 26070
    },
    {
      "epoch": 0.5339626891651138,
      "grad_norm": 0.2624116837978363,
      "learning_rate": 0.00019539552226832086,
      "loss": 4.6874,
      "step": 26075
    },
    {
      "epoch": 0.5340650789425184,
      "grad_norm": 0.25751790404319763,
      "learning_rate": 0.0001953365249726917,
      "loss": 4.6688,
      "step": 26080
    },
    {
      "epoch": 0.534167468719923,
      "grad_norm": 0.2581915557384491,
      "learning_rate": 0.00019527751995753587,
      "loss": 4.7396,
      "step": 26085
    },
    {
      "epoch": 0.5342698584973277,
      "grad_norm": 0.26121985912323,
      "learning_rate": 0.0001952185072329002,
      "loss": 4.7678,
      "step": 26090
    },
    {
      "epoch": 0.5343722482747323,
      "grad_norm": 0.2548855245113373,
      "learning_rate": 0.00019515948680883297,
      "loss": 4.7034,
      "step": 26095
    },
    {
      "epoch": 0.5344746380521369,
      "grad_norm": 0.24417440593242645,
      "learning_rate": 0.00019510045869538365,
      "loss": 4.7148,
      "step": 26100
    },
    {
      "epoch": 0.5345770278295415,
      "grad_norm": 0.2615807056427002,
      "learning_rate": 0.0001950414229026031,
      "loss": 4.6572,
      "step": 26105
    },
    {
      "epoch": 0.5346794176069462,
      "grad_norm": 0.27709394693374634,
      "learning_rate": 0.00019498237944054337,
      "loss": 4.7087,
      "step": 26110
    },
    {
      "epoch": 0.5347818073843508,
      "grad_norm": 0.44070112705230713,
      "learning_rate": 0.00019492332831925785,
      "loss": 4.7618,
      "step": 26115
    },
    {
      "epoch": 0.5348841971617554,
      "grad_norm": 0.2976377308368683,
      "learning_rate": 0.00019486426954880143,
      "loss": 4.6753,
      "step": 26120
    },
    {
      "epoch": 0.53498658693916,
      "grad_norm": 0.241999089717865,
      "learning_rate": 0.0001948052031392301,
      "loss": 4.6477,
      "step": 26125
    },
    {
      "epoch": 0.5350889767165646,
      "grad_norm": 0.2862304449081421,
      "learning_rate": 0.0001947461291006011,
      "loss": 4.7064,
      "step": 26130
    },
    {
      "epoch": 0.5351913664939693,
      "grad_norm": 0.2612556219100952,
      "learning_rate": 0.0001946870474429731,
      "loss": 4.6663,
      "step": 26135
    },
    {
      "epoch": 0.5352937562713739,
      "grad_norm": 0.2638161778450012,
      "learning_rate": 0.00019462795817640612,
      "loss": 4.6824,
      "step": 26140
    },
    {
      "epoch": 0.5353961460487785,
      "grad_norm": 0.2660267949104309,
      "learning_rate": 0.0001945688613109613,
      "loss": 4.6729,
      "step": 26145
    },
    {
      "epoch": 0.5354985358261831,
      "grad_norm": 0.26595360040664673,
      "learning_rate": 0.00019450975685670122,
      "loss": 4.7145,
      "step": 26150
    },
    {
      "epoch": 0.5356009256035877,
      "grad_norm": 0.24708572030067444,
      "learning_rate": 0.00019445064482368965,
      "loss": 4.749,
      "step": 26155
    },
    {
      "epoch": 0.5357033153809924,
      "grad_norm": 0.26008662581443787,
      "learning_rate": 0.0001943915252219917,
      "loss": 4.7328,
      "step": 26160
    },
    {
      "epoch": 0.535805705158397,
      "grad_norm": 0.2674421966075897,
      "learning_rate": 0.0001943323980616738,
      "loss": 4.7129,
      "step": 26165
    },
    {
      "epoch": 0.5359080949358016,
      "grad_norm": 0.2631687521934509,
      "learning_rate": 0.0001942732633528036,
      "loss": 4.6784,
      "step": 26170
    },
    {
      "epoch": 0.5360104847132062,
      "grad_norm": 0.26723283529281616,
      "learning_rate": 0.00019421412110545008,
      "loss": 4.7509,
      "step": 26175
    },
    {
      "epoch": 0.5361128744906108,
      "grad_norm": 0.2441859096288681,
      "learning_rate": 0.00019415497132968353,
      "loss": 4.6971,
      "step": 26180
    },
    {
      "epoch": 0.5362152642680155,
      "grad_norm": 0.2712872624397278,
      "learning_rate": 0.00019409581403557533,
      "loss": 4.6796,
      "step": 26185
    },
    {
      "epoch": 0.5363176540454201,
      "grad_norm": 0.6035029292106628,
      "learning_rate": 0.0001940366492331985,
      "loss": 4.6581,
      "step": 26190
    },
    {
      "epoch": 0.5364200438228247,
      "grad_norm": 0.27305108308792114,
      "learning_rate": 0.000193977476932627,
      "loss": 4.6978,
      "step": 26195
    },
    {
      "epoch": 0.5365224336002293,
      "grad_norm": 0.2726241648197174,
      "learning_rate": 0.00019391829714393626,
      "loss": 4.7578,
      "step": 26200
    },
    {
      "epoch": 0.536624823377634,
      "grad_norm": 0.25249484181404114,
      "learning_rate": 0.00019385910987720282,
      "loss": 4.7464,
      "step": 26205
    },
    {
      "epoch": 0.5367272131550386,
      "grad_norm": 0.2539021074771881,
      "learning_rate": 0.00019379991514250472,
      "loss": 4.6595,
      "step": 26210
    },
    {
      "epoch": 0.5368296029324432,
      "grad_norm": 0.2574290633201599,
      "learning_rate": 0.0001937407129499211,
      "loss": 4.6557,
      "step": 26215
    },
    {
      "epoch": 0.5369319927098478,
      "grad_norm": 0.3053162097930908,
      "learning_rate": 0.00019368150330953242,
      "loss": 4.6835,
      "step": 26220
    },
    {
      "epoch": 0.5370343824872524,
      "grad_norm": 0.24399952590465546,
      "learning_rate": 0.00019362228623142039,
      "loss": 4.69,
      "step": 26225
    },
    {
      "epoch": 0.5371367722646571,
      "grad_norm": 0.2516840994358063,
      "learning_rate": 0.00019356306172566796,
      "loss": 4.7066,
      "step": 26230
    },
    {
      "epoch": 0.5372391620420617,
      "grad_norm": 0.26467376947402954,
      "learning_rate": 0.0001935038298023595,
      "loss": 4.7172,
      "step": 26235
    },
    {
      "epoch": 0.5373415518194663,
      "grad_norm": 0.25264087319374084,
      "learning_rate": 0.00019344459047158046,
      "loss": 4.7694,
      "step": 26240
    },
    {
      "epoch": 0.5374439415968709,
      "grad_norm": 0.2514682710170746,
      "learning_rate": 0.0001933853437434176,
      "loss": 4.7084,
      "step": 26245
    },
    {
      "epoch": 0.5375463313742755,
      "grad_norm": 0.2545846402645111,
      "learning_rate": 0.00019332608962795904,
      "loss": 4.7126,
      "step": 26250
    },
    {
      "epoch": 0.5376487211516802,
      "grad_norm": 0.254913866519928,
      "learning_rate": 0.00019326682813529392,
      "loss": 4.6826,
      "step": 26255
    },
    {
      "epoch": 0.5377511109290848,
      "grad_norm": 0.27251002192497253,
      "learning_rate": 0.000193207559275513,
      "loss": 4.669,
      "step": 26260
    },
    {
      "epoch": 0.5378535007064895,
      "grad_norm": 0.26807644963264465,
      "learning_rate": 0.00019314828305870796,
      "loss": 4.7196,
      "step": 26265
    },
    {
      "epoch": 0.5379558904838941,
      "grad_norm": 0.2559725046157837,
      "learning_rate": 0.00019308899949497187,
      "loss": 4.6049,
      "step": 26270
    },
    {
      "epoch": 0.5380582802612988,
      "grad_norm": 0.2622315585613251,
      "learning_rate": 0.0001930297085943991,
      "loss": 4.7031,
      "step": 26275
    },
    {
      "epoch": 0.5381606700387034,
      "grad_norm": 0.2548885941505432,
      "learning_rate": 0.00019297041036708517,
      "loss": 4.7116,
      "step": 26280
    },
    {
      "epoch": 0.538263059816108,
      "grad_norm": 0.2712334096431732,
      "learning_rate": 0.00019291110482312686,
      "loss": 4.6726,
      "step": 26285
    },
    {
      "epoch": 0.5383654495935126,
      "grad_norm": 0.34245362877845764,
      "learning_rate": 0.00019285179197262232,
      "loss": 4.7,
      "step": 26290
    },
    {
      "epoch": 0.5384678393709172,
      "grad_norm": 0.2666618824005127,
      "learning_rate": 0.00019279247182567077,
      "loss": 4.6762,
      "step": 26295
    },
    {
      "epoch": 0.5385702291483219,
      "grad_norm": 0.24612635374069214,
      "learning_rate": 0.0001927331443923727,
      "loss": 4.746,
      "step": 26300
    },
    {
      "epoch": 0.5386726189257265,
      "grad_norm": 0.2689967453479767,
      "learning_rate": 0.00019267380968283004,
      "loss": 4.7046,
      "step": 26305
    },
    {
      "epoch": 0.5387750087031311,
      "grad_norm": 0.25115466117858887,
      "learning_rate": 0.00019261446770714571,
      "loss": 4.6499,
      "step": 26310
    },
    {
      "epoch": 0.5388773984805357,
      "grad_norm": 0.2397141456604004,
      "learning_rate": 0.00019255511847542404,
      "loss": 4.6902,
      "step": 26315
    },
    {
      "epoch": 0.5389797882579404,
      "grad_norm": 0.2797822654247284,
      "learning_rate": 0.0001924957619977704,
      "loss": 4.7158,
      "step": 26320
    },
    {
      "epoch": 0.539082178035345,
      "grad_norm": 0.27893519401550293,
      "learning_rate": 0.00019243639828429163,
      "loss": 4.6728,
      "step": 26325
    },
    {
      "epoch": 0.5391845678127496,
      "grad_norm": 0.2753532826900482,
      "learning_rate": 0.00019237702734509568,
      "loss": 4.7019,
      "step": 26330
    },
    {
      "epoch": 0.5392869575901542,
      "grad_norm": 0.2707187533378601,
      "learning_rate": 0.0001923176491902917,
      "loss": 4.729,
      "step": 26335
    },
    {
      "epoch": 0.5393893473675588,
      "grad_norm": 0.26808810234069824,
      "learning_rate": 0.0001922582638299901,
      "loss": 4.6933,
      "step": 26340
    },
    {
      "epoch": 0.5394917371449635,
      "grad_norm": 0.2533009946346283,
      "learning_rate": 0.00019219887127430255,
      "loss": 4.6881,
      "step": 26345
    },
    {
      "epoch": 0.5395941269223681,
      "grad_norm": 0.27040114998817444,
      "learning_rate": 0.00019213947153334194,
      "loss": 4.6711,
      "step": 26350
    },
    {
      "epoch": 0.5396965166997727,
      "grad_norm": 0.26400721073150635,
      "learning_rate": 0.00019208006461722232,
      "loss": 4.6366,
      "step": 26355
    },
    {
      "epoch": 0.5397989064771773,
      "grad_norm": 0.2703004777431488,
      "learning_rate": 0.00019202065053605904,
      "loss": 4.6477,
      "step": 26360
    },
    {
      "epoch": 0.5399012962545819,
      "grad_norm": 0.2598544657230377,
      "learning_rate": 0.00019196122929996865,
      "loss": 4.6492,
      "step": 26365
    },
    {
      "epoch": 0.5400036860319866,
      "grad_norm": 0.2817550599575043,
      "learning_rate": 0.00019190180091906885,
      "loss": 4.7068,
      "step": 26370
    },
    {
      "epoch": 0.5401060758093912,
      "grad_norm": 0.2642737925052643,
      "learning_rate": 0.00019184236540347862,
      "loss": 4.7371,
      "step": 26375
    },
    {
      "epoch": 0.5402084655867958,
      "grad_norm": 0.24548190832138062,
      "learning_rate": 0.00019178292276331822,
      "loss": 4.7364,
      "step": 26380
    },
    {
      "epoch": 0.5403108553642004,
      "grad_norm": 0.6843188405036926,
      "learning_rate": 0.00019172347300870898,
      "loss": 4.717,
      "step": 26385
    },
    {
      "epoch": 0.540413245141605,
      "grad_norm": 0.269150048494339,
      "learning_rate": 0.00019166401614977355,
      "loss": 4.6323,
      "step": 26390
    },
    {
      "epoch": 0.5405156349190097,
      "grad_norm": 0.26181933283805847,
      "learning_rate": 0.00019160455219663562,
      "loss": 4.7094,
      "step": 26395
    },
    {
      "epoch": 0.5406180246964143,
      "grad_norm": 0.2523230016231537,
      "learning_rate": 0.00019154508115942044,
      "loss": 4.6977,
      "step": 26400
    },
    {
      "epoch": 0.5407204144738189,
      "grad_norm": 0.25122886896133423,
      "learning_rate": 0.00019148560304825405,
      "loss": 4.6931,
      "step": 26405
    },
    {
      "epoch": 0.5408228042512235,
      "grad_norm": 0.25125694274902344,
      "learning_rate": 0.000191426117873264,
      "loss": 4.7033,
      "step": 26410
    },
    {
      "epoch": 0.5409251940286282,
      "grad_norm": 0.25516924262046814,
      "learning_rate": 0.00019136662564457892,
      "loss": 4.7436,
      "step": 26415
    },
    {
      "epoch": 0.5410275838060328,
      "grad_norm": 0.2656814455986023,
      "learning_rate": 0.00019130712637232855,
      "loss": 4.7096,
      "step": 26420
    },
    {
      "epoch": 0.5411299735834374,
      "grad_norm": 0.27654868364334106,
      "learning_rate": 0.00019124762006664405,
      "loss": 4.6982,
      "step": 26425
    },
    {
      "epoch": 0.541232363360842,
      "grad_norm": 0.26052772998809814,
      "learning_rate": 0.00019118810673765765,
      "loss": 4.7112,
      "step": 26430
    },
    {
      "epoch": 0.5413347531382466,
      "grad_norm": 0.24654258787631989,
      "learning_rate": 0.00019112858639550276,
      "loss": 4.6829,
      "step": 26435
    },
    {
      "epoch": 0.5414371429156513,
      "grad_norm": 0.2553759217262268,
      "learning_rate": 0.00019106905905031396,
      "loss": 4.71,
      "step": 26440
    },
    {
      "epoch": 0.5415395326930559,
      "grad_norm": 0.2568046450614929,
      "learning_rate": 0.00019100952471222713,
      "loss": 4.7044,
      "step": 26445
    },
    {
      "epoch": 0.5416419224704605,
      "grad_norm": 0.3430255949497223,
      "learning_rate": 0.00019094998339137925,
      "loss": 4.6998,
      "step": 26450
    },
    {
      "epoch": 0.5417443122478651,
      "grad_norm": 0.25559189915657043,
      "learning_rate": 0.00019089043509790859,
      "loss": 4.6574,
      "step": 26455
    },
    {
      "epoch": 0.5418467020252697,
      "grad_norm": 0.283968985080719,
      "learning_rate": 0.00019083087984195443,
      "loss": 4.7017,
      "step": 26460
    },
    {
      "epoch": 0.5419490918026745,
      "grad_norm": 0.2678341269493103,
      "learning_rate": 0.0001907713176336574,
      "loss": 4.6707,
      "step": 26465
    },
    {
      "epoch": 0.5420514815800791,
      "grad_norm": 0.2516711950302124,
      "learning_rate": 0.00019071174848315928,
      "loss": 4.7012,
      "step": 26470
    },
    {
      "epoch": 0.5421538713574837,
      "grad_norm": 0.25756508111953735,
      "learning_rate": 0.00019065217240060302,
      "loss": 4.7477,
      "step": 26475
    },
    {
      "epoch": 0.5422562611348883,
      "grad_norm": 0.2792092263698578,
      "learning_rate": 0.00019059258939613266,
      "loss": 4.7309,
      "step": 26480
    },
    {
      "epoch": 0.542358650912293,
      "grad_norm": 0.2740444242954254,
      "learning_rate": 0.00019053299947989358,
      "loss": 4.7099,
      "step": 26485
    },
    {
      "epoch": 0.5424610406896976,
      "grad_norm": 0.2703920602798462,
      "learning_rate": 0.0001904734026620322,
      "loss": 4.7371,
      "step": 26490
    },
    {
      "epoch": 0.5425634304671022,
      "grad_norm": 0.2684652805328369,
      "learning_rate": 0.00019041379895269622,
      "loss": 4.7206,
      "step": 26495
    },
    {
      "epoch": 0.5426658202445068,
      "grad_norm": 0.27988335490226746,
      "learning_rate": 0.00019035418836203444,
      "loss": 4.7138,
      "step": 26500
    },
    {
      "epoch": 0.5427682100219114,
      "grad_norm": 0.27285775542259216,
      "learning_rate": 0.00019029457090019686,
      "loss": 4.7015,
      "step": 26505
    },
    {
      "epoch": 0.5428705997993161,
      "grad_norm": 0.28652337193489075,
      "learning_rate": 0.0001902349465773346,
      "loss": 4.665,
      "step": 26510
    },
    {
      "epoch": 0.5429729895767207,
      "grad_norm": 0.2635771632194519,
      "learning_rate": 0.00019017531540360002,
      "loss": 4.7252,
      "step": 26515
    },
    {
      "epoch": 0.5430753793541253,
      "grad_norm": 0.2735646069049835,
      "learning_rate": 0.0001901156773891467,
      "loss": 4.6917,
      "step": 26520
    },
    {
      "epoch": 0.5431777691315299,
      "grad_norm": 0.25290530920028687,
      "learning_rate": 0.0001900560325441292,
      "loss": 4.6999,
      "step": 26525
    },
    {
      "epoch": 0.5432801589089346,
      "grad_norm": 0.26772749423980713,
      "learning_rate": 0.0001899963808787034,
      "loss": 4.7049,
      "step": 26530
    },
    {
      "epoch": 0.5433825486863392,
      "grad_norm": 0.2506352663040161,
      "learning_rate": 0.00018993672240302627,
      "loss": 4.6646,
      "step": 26535
    },
    {
      "epoch": 0.5434849384637438,
      "grad_norm": 0.2651379108428955,
      "learning_rate": 0.00018987705712725597,
      "loss": 4.7177,
      "step": 26540
    },
    {
      "epoch": 0.5435873282411484,
      "grad_norm": 0.25979286432266235,
      "learning_rate": 0.00018981738506155182,
      "loss": 4.7103,
      "step": 26545
    },
    {
      "epoch": 0.543689718018553,
      "grad_norm": 0.2624056935310364,
      "learning_rate": 0.00018975770621607423,
      "loss": 4.7129,
      "step": 26550
    },
    {
      "epoch": 0.5437921077959577,
      "grad_norm": 0.26970383524894714,
      "learning_rate": 0.00018969802060098492,
      "loss": 4.7168,
      "step": 26555
    },
    {
      "epoch": 0.5438944975733623,
      "grad_norm": 0.2662735879421234,
      "learning_rate": 0.00018963832822644657,
      "loss": 4.6698,
      "step": 26560
    },
    {
      "epoch": 0.5439968873507669,
      "grad_norm": 0.2513354420661926,
      "learning_rate": 0.00018957862910262306,
      "loss": 4.6928,
      "step": 26565
    },
    {
      "epoch": 0.5440992771281715,
      "grad_norm": 0.2497677505016327,
      "learning_rate": 0.00018951892323967962,
      "loss": 4.644,
      "step": 26570
    },
    {
      "epoch": 0.5442016669055761,
      "grad_norm": 0.24549132585525513,
      "learning_rate": 0.00018945921064778237,
      "loss": 4.6796,
      "step": 26575
    },
    {
      "epoch": 0.5443040566829808,
      "grad_norm": 0.2421257197856903,
      "learning_rate": 0.00018939949133709865,
      "loss": 4.6875,
      "step": 26580
    },
    {
      "epoch": 0.5444064464603854,
      "grad_norm": 0.2439357191324234,
      "learning_rate": 0.00018933976531779702,
      "loss": 4.704,
      "step": 26585
    },
    {
      "epoch": 0.54450883623779,
      "grad_norm": 0.29698094725608826,
      "learning_rate": 0.00018928003260004712,
      "loss": 4.5639,
      "step": 26590
    },
    {
      "epoch": 0.5446112260151946,
      "grad_norm": 0.27436554431915283,
      "learning_rate": 0.00018922029319401966,
      "loss": 4.6947,
      "step": 26595
    },
    {
      "epoch": 0.5447136157925992,
      "grad_norm": 0.24455125629901886,
      "learning_rate": 0.00018916054710988673,
      "loss": 4.6753,
      "step": 26600
    },
    {
      "epoch": 0.5448160055700039,
      "grad_norm": 0.2663513720035553,
      "learning_rate": 0.00018910079435782124,
      "loss": 4.7759,
      "step": 26605
    },
    {
      "epoch": 0.5449183953474085,
      "grad_norm": 0.2627456486225128,
      "learning_rate": 0.00018904103494799744,
      "loss": 4.6826,
      "step": 26610
    },
    {
      "epoch": 0.5450207851248131,
      "grad_norm": 0.2518863379955292,
      "learning_rate": 0.00018898126889059074,
      "loss": 4.6251,
      "step": 26615
    },
    {
      "epoch": 0.5451231749022177,
      "grad_norm": 0.2908294200897217,
      "learning_rate": 0.00018892149619577757,
      "loss": 4.6404,
      "step": 26620
    },
    {
      "epoch": 0.5452255646796224,
      "grad_norm": 0.2405155599117279,
      "learning_rate": 0.00018886171687373547,
      "loss": 4.6741,
      "step": 26625
    },
    {
      "epoch": 0.545327954457027,
      "grad_norm": 0.24741297960281372,
      "learning_rate": 0.00018880193093464323,
      "loss": 4.7229,
      "step": 26630
    },
    {
      "epoch": 0.5454303442344316,
      "grad_norm": 0.26555123925209045,
      "learning_rate": 0.00018874213838868063,
      "loss": 4.7053,
      "step": 26635
    },
    {
      "epoch": 0.5455327340118362,
      "grad_norm": 0.26296570897102356,
      "learning_rate": 0.00018868233924602872,
      "loss": 4.6912,
      "step": 26640
    },
    {
      "epoch": 0.5456351237892408,
      "grad_norm": 0.2616727650165558,
      "learning_rate": 0.00018862253351686962,
      "loss": 4.6611,
      "step": 26645
    },
    {
      "epoch": 0.5457375135666455,
      "grad_norm": 0.2673822343349457,
      "learning_rate": 0.00018856272121138646,
      "loss": 4.671,
      "step": 26650
    },
    {
      "epoch": 0.5458399033440501,
      "grad_norm": 0.25794851779937744,
      "learning_rate": 0.00018850290233976362,
      "loss": 4.6675,
      "step": 26655
    },
    {
      "epoch": 0.5459422931214547,
      "grad_norm": 0.2626766860485077,
      "learning_rate": 0.0001884430769121866,
      "loss": 4.6574,
      "step": 26660
    },
    {
      "epoch": 0.5460446828988593,
      "grad_norm": 0.25485607981681824,
      "learning_rate": 0.00018838324493884197,
      "loss": 4.6978,
      "step": 26665
    },
    {
      "epoch": 0.5461470726762641,
      "grad_norm": 0.24853378534317017,
      "learning_rate": 0.00018832340642991737,
      "loss": 4.6955,
      "step": 26670
    },
    {
      "epoch": 0.5462494624536687,
      "grad_norm": 0.25438764691352844,
      "learning_rate": 0.00018826356139560165,
      "loss": 4.6484,
      "step": 26675
    },
    {
      "epoch": 0.5463518522310733,
      "grad_norm": 0.28246045112609863,
      "learning_rate": 0.00018820370984608468,
      "loss": 4.6657,
      "step": 26680
    },
    {
      "epoch": 0.5464542420084779,
      "grad_norm": 0.261288583278656,
      "learning_rate": 0.00018814385179155755,
      "loss": 4.6753,
      "step": 26685
    },
    {
      "epoch": 0.5465566317858825,
      "grad_norm": 0.26489171385765076,
      "learning_rate": 0.00018808398724221232,
      "loss": 4.7225,
      "step": 26690
    },
    {
      "epoch": 0.5466590215632872,
      "grad_norm": 0.2542685568332672,
      "learning_rate": 0.0001880241162082423,
      "loss": 4.6645,
      "step": 26695
    },
    {
      "epoch": 0.5467614113406918,
      "grad_norm": 0.31563809514045715,
      "learning_rate": 0.00018796423869984176,
      "loss": 4.6155,
      "step": 26700
    },
    {
      "epoch": 0.5468638011180964,
      "grad_norm": 0.2422453612089157,
      "learning_rate": 0.00018790435472720618,
      "loss": 4.7036,
      "step": 26705
    },
    {
      "epoch": 0.546966190895501,
      "grad_norm": 0.2594524025917053,
      "learning_rate": 0.00018784446430053213,
      "loss": 4.6983,
      "step": 26710
    },
    {
      "epoch": 0.5470685806729056,
      "grad_norm": 0.26316750049591064,
      "learning_rate": 0.00018778456743001723,
      "loss": 4.6909,
      "step": 26715
    },
    {
      "epoch": 0.5471709704503103,
      "grad_norm": 0.270597904920578,
      "learning_rate": 0.0001877246641258602,
      "loss": 4.6347,
      "step": 26720
    },
    {
      "epoch": 0.5472733602277149,
      "grad_norm": 0.26570531725883484,
      "learning_rate": 0.00018766475439826089,
      "loss": 4.7231,
      "step": 26725
    },
    {
      "epoch": 0.5473757500051195,
      "grad_norm": 0.26886647939682007,
      "learning_rate": 0.00018760483825742028,
      "loss": 4.6587,
      "step": 26730
    },
    {
      "epoch": 0.5474781397825241,
      "grad_norm": 0.24525582790374756,
      "learning_rate": 0.0001875449157135403,
      "loss": 4.6629,
      "step": 26735
    },
    {
      "epoch": 0.5475805295599288,
      "grad_norm": 0.2805182635784149,
      "learning_rate": 0.00018748498677682418,
      "loss": 4.6969,
      "step": 26740
    },
    {
      "epoch": 0.5476829193373334,
      "grad_norm": 0.24619364738464355,
      "learning_rate": 0.00018742505145747596,
      "loss": 4.6606,
      "step": 26745
    },
    {
      "epoch": 0.547785309114738,
      "grad_norm": 0.2541234493255615,
      "learning_rate": 0.00018736510976570105,
      "loss": 4.644,
      "step": 26750
    },
    {
      "epoch": 0.5478876988921426,
      "grad_norm": 0.32706886529922485,
      "learning_rate": 0.00018730516171170587,
      "loss": 4.6714,
      "step": 26755
    },
    {
      "epoch": 0.5479900886695472,
      "grad_norm": 0.25250479578971863,
      "learning_rate": 0.00018724520730569773,
      "loss": 4.7627,
      "step": 26760
    },
    {
      "epoch": 0.5480924784469519,
      "grad_norm": 0.24195893108844757,
      "learning_rate": 0.00018718524655788525,
      "loss": 4.6815,
      "step": 26765
    },
    {
      "epoch": 0.5481948682243565,
      "grad_norm": 0.24525094032287598,
      "learning_rate": 0.00018712527947847805,
      "loss": 4.7774,
      "step": 26770
    },
    {
      "epoch": 0.5482972580017611,
      "grad_norm": 0.25464239716529846,
      "learning_rate": 0.00018706530607768678,
      "loss": 4.6538,
      "step": 26775
    },
    {
      "epoch": 0.5483996477791657,
      "grad_norm": 0.2611331045627594,
      "learning_rate": 0.0001870053263657233,
      "loss": 4.6961,
      "step": 26780
    },
    {
      "epoch": 0.5485020375565703,
      "grad_norm": 0.24742406606674194,
      "learning_rate": 0.00018694534035280038,
      "loss": 4.6935,
      "step": 26785
    },
    {
      "epoch": 0.548604427333975,
      "grad_norm": 0.26102006435394287,
      "learning_rate": 0.00018688534804913197,
      "loss": 4.7044,
      "step": 26790
    },
    {
      "epoch": 0.5487068171113796,
      "grad_norm": 0.26690933108329773,
      "learning_rate": 0.000186825349464933,
      "loss": 4.7174,
      "step": 26795
    },
    {
      "epoch": 0.5488092068887842,
      "grad_norm": 0.2613051235675812,
      "learning_rate": 0.00018676534461041966,
      "loss": 4.7217,
      "step": 26800
    },
    {
      "epoch": 0.5489115966661888,
      "grad_norm": 0.2524051070213318,
      "learning_rate": 0.00018670533349580894,
      "loss": 4.6833,
      "step": 26805
    },
    {
      "epoch": 0.5490139864435934,
      "grad_norm": 0.2654576003551483,
      "learning_rate": 0.00018664531613131913,
      "loss": 4.6963,
      "step": 26810
    },
    {
      "epoch": 0.5491163762209981,
      "grad_norm": 0.27269455790519714,
      "learning_rate": 0.00018658529252716944,
      "loss": 4.5475,
      "step": 26815
    },
    {
      "epoch": 0.5492187659984027,
      "grad_norm": 0.27000007033348083,
      "learning_rate": 0.0001865252626935802,
      "loss": 4.6638,
      "step": 26820
    },
    {
      "epoch": 0.5493211557758073,
      "grad_norm": 0.26834815740585327,
      "learning_rate": 0.00018646522664077282,
      "loss": 4.6984,
      "step": 26825
    },
    {
      "epoch": 0.5494235455532119,
      "grad_norm": 0.2604410946369171,
      "learning_rate": 0.0001864051843789697,
      "loss": 4.6812,
      "step": 26830
    },
    {
      "epoch": 0.5495259353306166,
      "grad_norm": 0.2676681876182556,
      "learning_rate": 0.00018634513591839436,
      "loss": 4.6832,
      "step": 26835
    },
    {
      "epoch": 0.5496283251080212,
      "grad_norm": 0.26295915246009827,
      "learning_rate": 0.0001862850812692713,
      "loss": 4.6604,
      "step": 26840
    },
    {
      "epoch": 0.5497307148854258,
      "grad_norm": 0.25580817461013794,
      "learning_rate": 0.0001862250204418262,
      "loss": 4.638,
      "step": 26845
    },
    {
      "epoch": 0.5498331046628304,
      "grad_norm": 0.24330177903175354,
      "learning_rate": 0.0001861649534462857,
      "loss": 4.7265,
      "step": 26850
    },
    {
      "epoch": 0.549935494440235,
      "grad_norm": 0.25567755103111267,
      "learning_rate": 0.00018610488029287752,
      "loss": 4.6356,
      "step": 26855
    },
    {
      "epoch": 0.5500378842176397,
      "grad_norm": 0.280666321516037,
      "learning_rate": 0.00018604480099183033,
      "loss": 4.6588,
      "step": 26860
    },
    {
      "epoch": 0.5501402739950443,
      "grad_norm": 0.23846778273582458,
      "learning_rate": 0.00018598471555337403,
      "loss": 4.7061,
      "step": 26865
    },
    {
      "epoch": 0.5502426637724489,
      "grad_norm": 0.27296262979507446,
      "learning_rate": 0.00018592462398773945,
      "loss": 4.6588,
      "step": 26870
    },
    {
      "epoch": 0.5503450535498536,
      "grad_norm": 0.2844870686531067,
      "learning_rate": 0.00018586452630515847,
      "loss": 4.6414,
      "step": 26875
    },
    {
      "epoch": 0.5504474433272583,
      "grad_norm": 0.26245835423469543,
      "learning_rate": 0.00018580442251586406,
      "loss": 4.6617,
      "step": 26880
    },
    {
      "epoch": 0.5505498331046629,
      "grad_norm": 0.2403590828180313,
      "learning_rate": 0.00018574431263009009,
      "loss": 4.6952,
      "step": 26885
    },
    {
      "epoch": 0.5506522228820675,
      "grad_norm": 0.28531143069267273,
      "learning_rate": 0.00018568419665807167,
      "loss": 4.6971,
      "step": 26890
    },
    {
      "epoch": 0.5507546126594721,
      "grad_norm": 0.2594854533672333,
      "learning_rate": 0.00018562407461004482,
      "loss": 4.6822,
      "step": 26895
    },
    {
      "epoch": 0.5508570024368767,
      "grad_norm": 0.2994579076766968,
      "learning_rate": 0.00018556394649624664,
      "loss": 4.6721,
      "step": 26900
    },
    {
      "epoch": 0.5509593922142814,
      "grad_norm": 0.25801777839660645,
      "learning_rate": 0.00018550381232691524,
      "loss": 4.6886,
      "step": 26905
    },
    {
      "epoch": 0.551061781991686,
      "grad_norm": 0.28385981917381287,
      "learning_rate": 0.00018544367211228969,
      "loss": 4.6774,
      "step": 26910
    },
    {
      "epoch": 0.5511641717690906,
      "grad_norm": 0.2582709789276123,
      "learning_rate": 0.00018538352586261025,
      "loss": 4.6501,
      "step": 26915
    },
    {
      "epoch": 0.5512665615464952,
      "grad_norm": 0.2655521631240845,
      "learning_rate": 0.00018532337358811816,
      "loss": 4.6338,
      "step": 26920
    },
    {
      "epoch": 0.5513689513238998,
      "grad_norm": 0.2572382986545563,
      "learning_rate": 0.00018526321529905554,
      "loss": 4.6787,
      "step": 26925
    },
    {
      "epoch": 0.5514713411013045,
      "grad_norm": 0.25258669257164,
      "learning_rate": 0.00018520305100566567,
      "loss": 4.6881,
      "step": 26930
    },
    {
      "epoch": 0.5515737308787091,
      "grad_norm": 0.7811693549156189,
      "learning_rate": 0.0001851428807181929,
      "loss": 4.6534,
      "step": 26935
    },
    {
      "epoch": 0.5516761206561137,
      "grad_norm": 0.2604113519191742,
      "learning_rate": 0.0001850827044468825,
      "loss": 4.7078,
      "step": 26940
    },
    {
      "epoch": 0.5517785104335183,
      "grad_norm": 0.2545107305049896,
      "learning_rate": 0.0001850225222019807,
      "loss": 4.7246,
      "step": 26945
    },
    {
      "epoch": 0.551880900210923,
      "grad_norm": 0.24491417407989502,
      "learning_rate": 0.00018496233399373495,
      "loss": 4.6983,
      "step": 26950
    },
    {
      "epoch": 0.5519832899883276,
      "grad_norm": 0.4022284150123596,
      "learning_rate": 0.0001849021398323935,
      "loss": 4.753,
      "step": 26955
    },
    {
      "epoch": 0.5520856797657322,
      "grad_norm": 0.2897582948207855,
      "learning_rate": 0.00018484193972820575,
      "loss": 4.7317,
      "step": 26960
    },
    {
      "epoch": 0.5521880695431368,
      "grad_norm": 0.24569040536880493,
      "learning_rate": 0.00018478173369142212,
      "loss": 4.6981,
      "step": 26965
    },
    {
      "epoch": 0.5522904593205414,
      "grad_norm": 0.25419145822525024,
      "learning_rate": 0.00018472152173229392,
      "loss": 4.7112,
      "step": 26970
    },
    {
      "epoch": 0.5523928490979461,
      "grad_norm": 0.2585573196411133,
      "learning_rate": 0.00018466130386107362,
      "loss": 4.6837,
      "step": 26975
    },
    {
      "epoch": 0.5524952388753507,
      "grad_norm": 0.24350248277187347,
      "learning_rate": 0.00018460108008801454,
      "loss": 4.7195,
      "step": 26980
    },
    {
      "epoch": 0.5525976286527553,
      "grad_norm": 0.26366153359413147,
      "learning_rate": 0.00018454085042337107,
      "loss": 4.6877,
      "step": 26985
    },
    {
      "epoch": 0.5527000184301599,
      "grad_norm": 3.328211545944214,
      "learning_rate": 0.00018448061487739875,
      "loss": 4.6177,
      "step": 26990
    },
    {
      "epoch": 0.5528024082075645,
      "grad_norm": 0.2649926245212555,
      "learning_rate": 0.00018442037346035387,
      "loss": 4.6967,
      "step": 26995
    },
    {
      "epoch": 0.5529047979849692,
      "grad_norm": 0.26315954327583313,
      "learning_rate": 0.00018436012618249385,
      "loss": 4.6918,
      "step": 27000
    },
    {
      "epoch": 0.5529047979849692,
      "eval_loss": 4.707661151885986,
      "eval_runtime": 17.1438,
      "eval_samples_per_second": 33.948,
      "eval_steps_per_second": 4.258,
      "step": 27000
    },
    {
      "epoch": 0.5530071877623738,
      "grad_norm": 0.2600516974925995,
      "learning_rate": 0.0001842998730540771,
      "loss": 4.6866,
      "step": 27005
    },
    {
      "epoch": 0.5531095775397784,
      "grad_norm": 0.24640217423439026,
      "learning_rate": 0.0001842396140853631,
      "loss": 4.7314,
      "step": 27010
    },
    {
      "epoch": 0.553211967317183,
      "grad_norm": 0.2561544179916382,
      "learning_rate": 0.0001841793492866121,
      "loss": 4.6893,
      "step": 27015
    },
    {
      "epoch": 0.5533143570945876,
      "grad_norm": 0.2652214467525482,
      "learning_rate": 0.00018411907866808567,
      "loss": 4.6634,
      "step": 27020
    },
    {
      "epoch": 0.5534167468719923,
      "grad_norm": 0.2577473521232605,
      "learning_rate": 0.000184058802240046,
      "loss": 4.7232,
      "step": 27025
    },
    {
      "epoch": 0.5535191366493969,
      "grad_norm": 0.24600835144519806,
      "learning_rate": 0.00018399852001275662,
      "loss": 4.6832,
      "step": 27030
    },
    {
      "epoch": 0.5536215264268015,
      "grad_norm": 0.24829618632793427,
      "learning_rate": 0.00018393823199648179,
      "loss": 4.6178,
      "step": 27035
    },
    {
      "epoch": 0.5537239162042061,
      "grad_norm": 0.2764621078968048,
      "learning_rate": 0.00018387793820148687,
      "loss": 4.723,
      "step": 27040
    },
    {
      "epoch": 0.5538263059816108,
      "grad_norm": 0.25059133768081665,
      "learning_rate": 0.00018381763863803822,
      "loss": 4.6999,
      "step": 27045
    },
    {
      "epoch": 0.5539286957590154,
      "grad_norm": 0.2450833022594452,
      "learning_rate": 0.0001837573333164031,
      "loss": 4.6377,
      "step": 27050
    },
    {
      "epoch": 0.55403108553642,
      "grad_norm": 0.25288718938827515,
      "learning_rate": 0.00018369702224684982,
      "loss": 4.6913,
      "step": 27055
    },
    {
      "epoch": 0.5541334753138246,
      "grad_norm": 0.24497826397418976,
      "learning_rate": 0.00018363670543964768,
      "loss": 4.6703,
      "step": 27060
    },
    {
      "epoch": 0.5542358650912292,
      "grad_norm": 0.2574631869792938,
      "learning_rate": 0.0001835763829050669,
      "loss": 4.6749,
      "step": 27065
    },
    {
      "epoch": 0.5543382548686339,
      "grad_norm": 0.28040191531181335,
      "learning_rate": 0.0001835160546533787,
      "loss": 4.7238,
      "step": 27070
    },
    {
      "epoch": 0.5544406446460386,
      "grad_norm": 0.2564941644668579,
      "learning_rate": 0.00018345572069485523,
      "loss": 4.7229,
      "step": 27075
    },
    {
      "epoch": 0.5545430344234432,
      "grad_norm": 0.2596915662288666,
      "learning_rate": 0.00018339538103976974,
      "loss": 4.6761,
      "step": 27080
    },
    {
      "epoch": 0.5546454242008478,
      "grad_norm": 0.2704494297504425,
      "learning_rate": 0.00018333503569839627,
      "loss": 4.7243,
      "step": 27085
    },
    {
      "epoch": 0.5547478139782525,
      "grad_norm": 0.25303056836128235,
      "learning_rate": 0.00018327468468101003,
      "loss": 4.7087,
      "step": 27090
    },
    {
      "epoch": 0.5548502037556571,
      "grad_norm": 0.2724940776824951,
      "learning_rate": 0.00018321432799788697,
      "loss": 4.7116,
      "step": 27095
    },
    {
      "epoch": 0.5549525935330617,
      "grad_norm": 0.26068201661109924,
      "learning_rate": 0.0001831539656593042,
      "loss": 4.657,
      "step": 27100
    },
    {
      "epoch": 0.5550549833104663,
      "grad_norm": 0.2497115582227707,
      "learning_rate": 0.00018309359767553976,
      "loss": 4.6816,
      "step": 27105
    },
    {
      "epoch": 0.5551573730878709,
      "grad_norm": 0.2723061740398407,
      "learning_rate": 0.00018303322405687248,
      "loss": 4.6453,
      "step": 27110
    },
    {
      "epoch": 0.5552597628652756,
      "grad_norm": 0.3353561758995056,
      "learning_rate": 0.00018297284481358242,
      "loss": 4.5887,
      "step": 27115
    },
    {
      "epoch": 0.5553621526426802,
      "grad_norm": 0.2825780212879181,
      "learning_rate": 0.00018291245995595033,
      "loss": 4.6267,
      "step": 27120
    },
    {
      "epoch": 0.5554645424200848,
      "grad_norm": 0.2716694474220276,
      "learning_rate": 0.00018285206949425812,
      "loss": 4.7231,
      "step": 27125
    },
    {
      "epoch": 0.5555669321974894,
      "grad_norm": 0.2531558871269226,
      "learning_rate": 0.00018279167343878859,
      "loss": 4.6276,
      "step": 27130
    },
    {
      "epoch": 0.555669321974894,
      "grad_norm": 0.2798287868499756,
      "learning_rate": 0.0001827312717998254,
      "loss": 4.7516,
      "step": 27135
    },
    {
      "epoch": 0.5557717117522987,
      "grad_norm": 0.2748793661594391,
      "learning_rate": 0.0001826708645876533,
      "loss": 4.6299,
      "step": 27140
    },
    {
      "epoch": 0.5558741015297033,
      "grad_norm": 0.24465809762477875,
      "learning_rate": 0.0001826104518125579,
      "loss": 4.6925,
      "step": 27145
    },
    {
      "epoch": 0.5559764913071079,
      "grad_norm": 0.29484617710113525,
      "learning_rate": 0.00018255003348482577,
      "loss": 4.6911,
      "step": 27150
    },
    {
      "epoch": 0.5560788810845125,
      "grad_norm": 0.28605300188064575,
      "learning_rate": 0.0001824896096147445,
      "loss": 4.7485,
      "step": 27155
    },
    {
      "epoch": 0.5561812708619172,
      "grad_norm": 0.2805459797382355,
      "learning_rate": 0.00018242918021260256,
      "loss": 4.7061,
      "step": 27160
    },
    {
      "epoch": 0.5562836606393218,
      "grad_norm": 0.3476477861404419,
      "learning_rate": 0.00018236874528868927,
      "loss": 4.6931,
      "step": 27165
    },
    {
      "epoch": 0.5563860504167264,
      "grad_norm": 0.26884692907333374,
      "learning_rate": 0.00018230830485329506,
      "loss": 4.7284,
      "step": 27170
    },
    {
      "epoch": 0.556488440194131,
      "grad_norm": 0.25742143392562866,
      "learning_rate": 0.00018224785891671124,
      "loss": 4.6757,
      "step": 27175
    },
    {
      "epoch": 0.5565908299715356,
      "grad_norm": 0.27536290884017944,
      "learning_rate": 0.00018218740748923002,
      "loss": 4.6719,
      "step": 27180
    },
    {
      "epoch": 0.5566932197489403,
      "grad_norm": 0.24896275997161865,
      "learning_rate": 0.00018212695058114457,
      "loss": 4.7062,
      "step": 27185
    },
    {
      "epoch": 0.5567956095263449,
      "grad_norm": 0.2572430968284607,
      "learning_rate": 0.00018206648820274894,
      "loss": 4.6957,
      "step": 27190
    },
    {
      "epoch": 0.5568979993037495,
      "grad_norm": 0.28539586067199707,
      "learning_rate": 0.00018200602036433825,
      "loss": 4.7002,
      "step": 27195
    },
    {
      "epoch": 0.5570003890811541,
      "grad_norm": 0.2676612138748169,
      "learning_rate": 0.00018194554707620842,
      "loss": 4.6843,
      "step": 27200
    },
    {
      "epoch": 0.5571027788585587,
      "grad_norm": 0.2554614543914795,
      "learning_rate": 0.00018188506834865636,
      "loss": 4.6578,
      "step": 27205
    },
    {
      "epoch": 0.5572051686359634,
      "grad_norm": 0.27748146653175354,
      "learning_rate": 0.00018182458419197984,
      "loss": 4.6778,
      "step": 27210
    },
    {
      "epoch": 0.557307558413368,
      "grad_norm": 0.2704891860485077,
      "learning_rate": 0.0001817640946164776,
      "loss": 4.697,
      "step": 27215
    },
    {
      "epoch": 0.5574099481907726,
      "grad_norm": 0.30365341901779175,
      "learning_rate": 0.00018170359963244934,
      "loss": 4.6481,
      "step": 27220
    },
    {
      "epoch": 0.5575123379681772,
      "grad_norm": 0.2857930362224579,
      "learning_rate": 0.00018164309925019565,
      "loss": 4.7163,
      "step": 27225
    },
    {
      "epoch": 0.5576147277455819,
      "grad_norm": 0.3207056522369385,
      "learning_rate": 0.000181582593480018,
      "loss": 4.7017,
      "step": 27230
    },
    {
      "epoch": 0.5577171175229865,
      "grad_norm": 0.462871789932251,
      "learning_rate": 0.00018152208233221878,
      "loss": 4.701,
      "step": 27235
    },
    {
      "epoch": 0.5578195073003911,
      "grad_norm": 0.26674333214759827,
      "learning_rate": 0.00018146156581710137,
      "loss": 4.7067,
      "step": 27240
    },
    {
      "epoch": 0.5579218970777957,
      "grad_norm": 0.2612088918685913,
      "learning_rate": 0.00018140104394497007,
      "loss": 4.5949,
      "step": 27245
    },
    {
      "epoch": 0.5580242868552003,
      "grad_norm": 0.2627280056476593,
      "learning_rate": 0.00018134051672612992,
      "loss": 4.6408,
      "step": 27250
    },
    {
      "epoch": 0.558126676632605,
      "grad_norm": 0.24926099181175232,
      "learning_rate": 0.0001812799841708871,
      "loss": 4.6388,
      "step": 27255
    },
    {
      "epoch": 0.5582290664100096,
      "grad_norm": 0.27249667048454285,
      "learning_rate": 0.00018121944628954852,
      "loss": 4.6831,
      "step": 27260
    },
    {
      "epoch": 0.5583314561874142,
      "grad_norm": 0.29811960458755493,
      "learning_rate": 0.0001811589030924221,
      "loss": 4.6652,
      "step": 27265
    },
    {
      "epoch": 0.5584338459648188,
      "grad_norm": 0.28687939047813416,
      "learning_rate": 0.00018109835458981666,
      "loss": 4.6839,
      "step": 27270
    },
    {
      "epoch": 0.5585362357422234,
      "grad_norm": 0.27148640155792236,
      "learning_rate": 0.00018103780079204186,
      "loss": 4.6828,
      "step": 27275
    },
    {
      "epoch": 0.5586386255196282,
      "grad_norm": 0.25214913487434387,
      "learning_rate": 0.00018097724170940824,
      "loss": 4.6627,
      "step": 27280
    },
    {
      "epoch": 0.5587410152970328,
      "grad_norm": 0.26098278164863586,
      "learning_rate": 0.00018091667735222738,
      "loss": 4.6741,
      "step": 27285
    },
    {
      "epoch": 0.5588434050744374,
      "grad_norm": 0.27879834175109863,
      "learning_rate": 0.0001808561077308117,
      "loss": 4.6534,
      "step": 27290
    },
    {
      "epoch": 0.558945794851842,
      "grad_norm": 0.2642155587673187,
      "learning_rate": 0.0001807955328554744,
      "loss": 4.7127,
      "step": 27295
    },
    {
      "epoch": 0.5590481846292467,
      "grad_norm": 0.26316457986831665,
      "learning_rate": 0.00018073495273652972,
      "loss": 4.6811,
      "step": 27300
    },
    {
      "epoch": 0.5591505744066513,
      "grad_norm": 0.25082215666770935,
      "learning_rate": 0.00018067436738429275,
      "loss": 4.699,
      "step": 27305
    },
    {
      "epoch": 0.5592529641840559,
      "grad_norm": 0.278416246175766,
      "learning_rate": 0.0001806137768090794,
      "loss": 4.6951,
      "step": 27310
    },
    {
      "epoch": 0.5593553539614605,
      "grad_norm": 0.3086649477481842,
      "learning_rate": 0.0001805531810212066,
      "loss": 4.7411,
      "step": 27315
    },
    {
      "epoch": 0.5594577437388651,
      "grad_norm": 0.2515885531902313,
      "learning_rate": 0.00018049258003099207,
      "loss": 4.7039,
      "step": 27320
    },
    {
      "epoch": 0.5595601335162698,
      "grad_norm": 0.2530098855495453,
      "learning_rate": 0.00018043197384875445,
      "loss": 4.7325,
      "step": 27325
    },
    {
      "epoch": 0.5596625232936744,
      "grad_norm": 1.3420644998550415,
      "learning_rate": 0.00018037136248481321,
      "loss": 4.6194,
      "step": 27330
    },
    {
      "epoch": 0.559764913071079,
      "grad_norm": 0.26888757944107056,
      "learning_rate": 0.0001803107459494888,
      "loss": 4.694,
      "step": 27335
    },
    {
      "epoch": 0.5598673028484836,
      "grad_norm": 0.26449164748191833,
      "learning_rate": 0.00018025012425310248,
      "loss": 4.656,
      "step": 27340
    },
    {
      "epoch": 0.5599696926258882,
      "grad_norm": 0.2640646696090698,
      "learning_rate": 0.00018018949740597647,
      "loss": 4.6354,
      "step": 27345
    },
    {
      "epoch": 0.5600720824032929,
      "grad_norm": 0.26847779750823975,
      "learning_rate": 0.00018012886541843366,
      "loss": 4.6841,
      "step": 27350
    },
    {
      "epoch": 0.5601744721806975,
      "grad_norm": 0.26787444949150085,
      "learning_rate": 0.00018006822830079808,
      "loss": 4.6685,
      "step": 27355
    },
    {
      "epoch": 0.5602768619581021,
      "grad_norm": 0.26082074642181396,
      "learning_rate": 0.00018000758606339455,
      "loss": 4.725,
      "step": 27360
    },
    {
      "epoch": 0.5603792517355067,
      "grad_norm": 0.26512712240219116,
      "learning_rate": 0.00017994693871654859,
      "loss": 4.6762,
      "step": 27365
    },
    {
      "epoch": 0.5604816415129114,
      "grad_norm": 0.2519755959510803,
      "learning_rate": 0.00017988628627058683,
      "loss": 4.6567,
      "step": 27370
    },
    {
      "epoch": 0.560584031290316,
      "grad_norm": 0.25382062792778015,
      "learning_rate": 0.00017982562873583664,
      "loss": 4.6997,
      "step": 27375
    },
    {
      "epoch": 0.5606864210677206,
      "grad_norm": 0.24935616552829742,
      "learning_rate": 0.00017976496612262622,
      "loss": 4.7334,
      "step": 27380
    },
    {
      "epoch": 0.5607888108451252,
      "grad_norm": 1.4657069444656372,
      "learning_rate": 0.0001797042984412848,
      "loss": 4.7416,
      "step": 27385
    },
    {
      "epoch": 0.5608912006225298,
      "grad_norm": 0.27165108919143677,
      "learning_rate": 0.0001796436257021423,
      "loss": 4.7026,
      "step": 27390
    },
    {
      "epoch": 0.5609935903999345,
      "grad_norm": 0.2536027431488037,
      "learning_rate": 0.00017958294791552961,
      "loss": 4.7337,
      "step": 27395
    },
    {
      "epoch": 0.5610959801773391,
      "grad_norm": 0.28293851017951965,
      "learning_rate": 0.0001795222650917784,
      "loss": 4.6477,
      "step": 27400
    },
    {
      "epoch": 0.5611983699547437,
      "grad_norm": 0.2583474814891815,
      "learning_rate": 0.0001794615772412212,
      "loss": 4.7176,
      "step": 27405
    },
    {
      "epoch": 0.5613007597321483,
      "grad_norm": 0.25620806217193604,
      "learning_rate": 0.00017940088437419155,
      "loss": 4.7288,
      "step": 27410
    },
    {
      "epoch": 0.561403149509553,
      "grad_norm": 0.29453903436660767,
      "learning_rate": 0.00017934018650102366,
      "loss": 4.6706,
      "step": 27415
    },
    {
      "epoch": 0.5615055392869576,
      "grad_norm": 0.2408190816640854,
      "learning_rate": 0.00017927948363205266,
      "loss": 4.6994,
      "step": 27420
    },
    {
      "epoch": 0.5616079290643622,
      "grad_norm": 0.2770805060863495,
      "learning_rate": 0.0001792187757776145,
      "loss": 4.6142,
      "step": 27425
    },
    {
      "epoch": 0.5617103188417668,
      "grad_norm": 0.2456747442483902,
      "learning_rate": 0.0001791580629480461,
      "loss": 4.7398,
      "step": 27430
    },
    {
      "epoch": 0.5618127086191714,
      "grad_norm": 0.2638939619064331,
      "learning_rate": 0.00017909734515368503,
      "loss": 4.7796,
      "step": 27435
    },
    {
      "epoch": 0.561915098396576,
      "grad_norm": 0.25840067863464355,
      "learning_rate": 0.0001790366224048699,
      "loss": 4.6846,
      "step": 27440
    },
    {
      "epoch": 0.5620174881739807,
      "grad_norm": 0.25280097126960754,
      "learning_rate": 0.00017897589471193998,
      "loss": 4.6698,
      "step": 27445
    },
    {
      "epoch": 0.5621198779513853,
      "grad_norm": 0.27671337127685547,
      "learning_rate": 0.00017891516208523556,
      "loss": 4.6482,
      "step": 27450
    },
    {
      "epoch": 0.5622222677287899,
      "grad_norm": 0.25754889845848083,
      "learning_rate": 0.00017885442453509766,
      "loss": 4.6554,
      "step": 27455
    },
    {
      "epoch": 0.5623246575061945,
      "grad_norm": 0.25561729073524475,
      "learning_rate": 0.00017879368207186815,
      "loss": 4.6459,
      "step": 27460
    },
    {
      "epoch": 0.5624270472835992,
      "grad_norm": 0.27110201120376587,
      "learning_rate": 0.00017873293470588978,
      "loss": 4.7197,
      "step": 27465
    },
    {
      "epoch": 0.5625294370610038,
      "grad_norm": 0.2641063332557678,
      "learning_rate": 0.0001786721824475061,
      "loss": 4.6728,
      "step": 27470
    },
    {
      "epoch": 0.5626318268384084,
      "grad_norm": 0.2695915400981903,
      "learning_rate": 0.00017861142530706145,
      "loss": 4.6747,
      "step": 27475
    },
    {
      "epoch": 0.5627342166158131,
      "grad_norm": 0.26590949296951294,
      "learning_rate": 0.00017855066329490113,
      "loss": 4.6944,
      "step": 27480
    },
    {
      "epoch": 0.5628366063932178,
      "grad_norm": 0.2543744742870331,
      "learning_rate": 0.00017848989642137118,
      "loss": 4.7019,
      "step": 27485
    },
    {
      "epoch": 0.5629389961706224,
      "grad_norm": 0.3397654592990875,
      "learning_rate": 0.00017842912469681838,
      "loss": 4.6556,
      "step": 27490
    },
    {
      "epoch": 0.563041385948027,
      "grad_norm": 0.27330344915390015,
      "learning_rate": 0.00017836834813159054,
      "loss": 4.6798,
      "step": 27495
    },
    {
      "epoch": 0.5631437757254316,
      "grad_norm": 0.27935123443603516,
      "learning_rate": 0.00017830756673603614,
      "loss": 4.6958,
      "step": 27500
    },
    {
      "epoch": 0.5632461655028362,
      "grad_norm": 0.3052602708339691,
      "learning_rate": 0.00017824678052050455,
      "loss": 4.6644,
      "step": 27505
    },
    {
      "epoch": 0.5633485552802409,
      "grad_norm": 0.27773189544677734,
      "learning_rate": 0.00017818598949534593,
      "loss": 4.7347,
      "step": 27510
    },
    {
      "epoch": 0.5634509450576455,
      "grad_norm": 0.2945863902568817,
      "learning_rate": 0.0001781251936709113,
      "loss": 4.6685,
      "step": 27515
    },
    {
      "epoch": 0.5635533348350501,
      "grad_norm": 0.26481300592422485,
      "learning_rate": 0.0001780643930575524,
      "loss": 4.7081,
      "step": 27520
    },
    {
      "epoch": 0.5636557246124547,
      "grad_norm": 0.26331692934036255,
      "learning_rate": 0.00017800358766562196,
      "loss": 4.6974,
      "step": 27525
    },
    {
      "epoch": 0.5637581143898593,
      "grad_norm": 0.25365984439849854,
      "learning_rate": 0.00017794277750547328,
      "loss": 4.7501,
      "step": 27530
    },
    {
      "epoch": 0.563860504167264,
      "grad_norm": 0.25934815406799316,
      "learning_rate": 0.00017788196258746074,
      "loss": 4.7036,
      "step": 27535
    },
    {
      "epoch": 0.5639628939446686,
      "grad_norm": 0.24910928308963776,
      "learning_rate": 0.00017782114292193933,
      "loss": 4.6434,
      "step": 27540
    },
    {
      "epoch": 0.5640652837220732,
      "grad_norm": 0.26640039682388306,
      "learning_rate": 0.0001777603185192649,
      "loss": 4.687,
      "step": 27545
    },
    {
      "epoch": 0.5641676734994778,
      "grad_norm": 0.2597857713699341,
      "learning_rate": 0.00017769948938979417,
      "loss": 4.6525,
      "step": 27550
    },
    {
      "epoch": 0.5642700632768824,
      "grad_norm": 0.2706809937953949,
      "learning_rate": 0.00017763865554388465,
      "loss": 4.6735,
      "step": 27555
    },
    {
      "epoch": 0.5643724530542871,
      "grad_norm": 0.2589665651321411,
      "learning_rate": 0.00017757781699189446,
      "loss": 4.7307,
      "step": 27560
    },
    {
      "epoch": 0.5644748428316917,
      "grad_norm": 0.26661667227745056,
      "learning_rate": 0.00017751697374418287,
      "loss": 4.6581,
      "step": 27565
    },
    {
      "epoch": 0.5645772326090963,
      "grad_norm": 0.27051541209220886,
      "learning_rate": 0.0001774561258111097,
      "loss": 4.6218,
      "step": 27570
    },
    {
      "epoch": 0.5646796223865009,
      "grad_norm": 0.26295870542526245,
      "learning_rate": 0.00017739527320303556,
      "loss": 4.7004,
      "step": 27575
    },
    {
      "epoch": 0.5647820121639056,
      "grad_norm": 0.25100234150886536,
      "learning_rate": 0.00017733441593032205,
      "loss": 4.7355,
      "step": 27580
    },
    {
      "epoch": 0.5648844019413102,
      "grad_norm": 0.2552400827407837,
      "learning_rate": 0.00017727355400333134,
      "loss": 4.6514,
      "step": 27585
    },
    {
      "epoch": 0.5649867917187148,
      "grad_norm": 0.2714831531047821,
      "learning_rate": 0.00017721268743242654,
      "loss": 4.6485,
      "step": 27590
    },
    {
      "epoch": 0.5650891814961194,
      "grad_norm": 0.24230805039405823,
      "learning_rate": 0.00017715181622797148,
      "loss": 4.6641,
      "step": 27595
    },
    {
      "epoch": 0.565191571273524,
      "grad_norm": 0.28542134165763855,
      "learning_rate": 0.0001770909404003308,
      "loss": 4.7616,
      "step": 27600
    },
    {
      "epoch": 0.5652939610509287,
      "grad_norm": 0.25475913286209106,
      "learning_rate": 0.00017703005995986997,
      "loss": 4.6521,
      "step": 27605
    },
    {
      "epoch": 0.5653963508283333,
      "grad_norm": 0.300935298204422,
      "learning_rate": 0.00017696917491695518,
      "loss": 4.6605,
      "step": 27610
    },
    {
      "epoch": 0.5654987406057379,
      "grad_norm": 0.2584412395954132,
      "learning_rate": 0.0001769082852819534,
      "loss": 4.6719,
      "step": 27615
    },
    {
      "epoch": 0.5656011303831425,
      "grad_norm": 0.25238272547721863,
      "learning_rate": 0.00017684739106523247,
      "loss": 4.7001,
      "step": 27620
    },
    {
      "epoch": 0.5657035201605471,
      "grad_norm": 0.26562386751174927,
      "learning_rate": 0.00017678649227716094,
      "loss": 4.6864,
      "step": 27625
    },
    {
      "epoch": 0.5658059099379518,
      "grad_norm": 0.2750887870788574,
      "learning_rate": 0.00017672558892810808,
      "loss": 4.6785,
      "step": 27630
    },
    {
      "epoch": 0.5659082997153564,
      "grad_norm": 0.25745710730552673,
      "learning_rate": 0.0001766646810284441,
      "loss": 4.626,
      "step": 27635
    },
    {
      "epoch": 0.566010689492761,
      "grad_norm": 0.29567331075668335,
      "learning_rate": 0.00017660376858853982,
      "loss": 4.6796,
      "step": 27640
    },
    {
      "epoch": 0.5661130792701656,
      "grad_norm": 0.26070085167884827,
      "learning_rate": 0.00017654285161876695,
      "loss": 4.6649,
      "step": 27645
    },
    {
      "epoch": 0.5662154690475703,
      "grad_norm": 0.266722172498703,
      "learning_rate": 0.0001764819301294979,
      "loss": 4.6549,
      "step": 27650
    },
    {
      "epoch": 0.5663178588249749,
      "grad_norm": 0.2570223808288574,
      "learning_rate": 0.00017642100413110587,
      "loss": 4.6951,
      "step": 27655
    },
    {
      "epoch": 0.5664202486023795,
      "grad_norm": 0.2593010365962982,
      "learning_rate": 0.00017636007363396486,
      "loss": 4.6673,
      "step": 27660
    },
    {
      "epoch": 0.5665226383797841,
      "grad_norm": 0.2559462785720825,
      "learning_rate": 0.0001762991386484496,
      "loss": 4.6072,
      "step": 27665
    },
    {
      "epoch": 0.5666250281571887,
      "grad_norm": 0.28588035702705383,
      "learning_rate": 0.00017623819918493556,
      "loss": 4.7359,
      "step": 27670
    },
    {
      "epoch": 0.5667274179345934,
      "grad_norm": 0.25574782490730286,
      "learning_rate": 0.00017617725525379907,
      "loss": 4.5986,
      "step": 27675
    },
    {
      "epoch": 0.566829807711998,
      "grad_norm": 0.2513871490955353,
      "learning_rate": 0.00017611630686541712,
      "loss": 4.6416,
      "step": 27680
    },
    {
      "epoch": 0.5669321974894027,
      "grad_norm": 0.2426123172044754,
      "learning_rate": 0.00017605535403016745,
      "loss": 4.6405,
      "step": 27685
    },
    {
      "epoch": 0.5670345872668073,
      "grad_norm": 0.25610238313674927,
      "learning_rate": 0.00017599439675842867,
      "loss": 4.707,
      "step": 27690
    },
    {
      "epoch": 0.567136977044212,
      "grad_norm": 0.24854488670825958,
      "learning_rate": 0.00017593343506058003,
      "loss": 4.7291,
      "step": 27695
    },
    {
      "epoch": 0.5672393668216166,
      "grad_norm": 0.2657903730869293,
      "learning_rate": 0.0001758724689470016,
      "loss": 4.6798,
      "step": 27700
    },
    {
      "epoch": 0.5673417565990212,
      "grad_norm": 0.27302059531211853,
      "learning_rate": 0.00017581149842807426,
      "loss": 4.5984,
      "step": 27705
    },
    {
      "epoch": 0.5674441463764258,
      "grad_norm": 0.25416168570518494,
      "learning_rate": 0.00017575052351417942,
      "loss": 4.6871,
      "step": 27710
    },
    {
      "epoch": 0.5675465361538304,
      "grad_norm": 0.26361799240112305,
      "learning_rate": 0.00017568954421569945,
      "loss": 4.7508,
      "step": 27715
    },
    {
      "epoch": 0.5676489259312351,
      "grad_norm": 0.25274792313575745,
      "learning_rate": 0.00017562856054301746,
      "loss": 4.6261,
      "step": 27720
    },
    {
      "epoch": 0.5677513157086397,
      "grad_norm": 0.2605419158935547,
      "learning_rate": 0.0001755675725065171,
      "loss": 4.7088,
      "step": 27725
    },
    {
      "epoch": 0.5678537054860443,
      "grad_norm": 0.26386305689811707,
      "learning_rate": 0.00017550658011658305,
      "loss": 4.7259,
      "step": 27730
    },
    {
      "epoch": 0.5679560952634489,
      "grad_norm": 0.26385360956192017,
      "learning_rate": 0.0001754455833836005,
      "loss": 4.6939,
      "step": 27735
    },
    {
      "epoch": 0.5680584850408535,
      "grad_norm": 0.2676510214805603,
      "learning_rate": 0.0001753845823179555,
      "loss": 4.6886,
      "step": 27740
    },
    {
      "epoch": 0.5681608748182582,
      "grad_norm": 0.2556968331336975,
      "learning_rate": 0.0001753235769300348,
      "loss": 4.7153,
      "step": 27745
    },
    {
      "epoch": 0.5682632645956628,
      "grad_norm": 0.25364434719085693,
      "learning_rate": 0.0001752625672302259,
      "loss": 4.6915,
      "step": 27750
    },
    {
      "epoch": 0.5683656543730674,
      "grad_norm": 0.3793516755104065,
      "learning_rate": 0.000175201553228917,
      "loss": 4.6868,
      "step": 27755
    },
    {
      "epoch": 0.568468044150472,
      "grad_norm": 0.35917139053344727,
      "learning_rate": 0.00017514053493649704,
      "loss": 4.6942,
      "step": 27760
    },
    {
      "epoch": 0.5685704339278767,
      "grad_norm": 0.35769927501678467,
      "learning_rate": 0.00017507951236335578,
      "loss": 4.6506,
      "step": 27765
    },
    {
      "epoch": 0.5686728237052813,
      "grad_norm": 0.2890172600746155,
      "learning_rate": 0.00017501848551988359,
      "loss": 4.6713,
      "step": 27770
    },
    {
      "epoch": 0.5687752134826859,
      "grad_norm": 0.2619645595550537,
      "learning_rate": 0.0001749574544164716,
      "loss": 4.6714,
      "step": 27775
    },
    {
      "epoch": 0.5688776032600905,
      "grad_norm": 0.2589600384235382,
      "learning_rate": 0.00017489641906351172,
      "loss": 4.5968,
      "step": 27780
    },
    {
      "epoch": 0.5689799930374951,
      "grad_norm": 0.259791761636734,
      "learning_rate": 0.00017483537947139645,
      "loss": 4.6749,
      "step": 27785
    },
    {
      "epoch": 0.5690823828148998,
      "grad_norm": 0.28064924478530884,
      "learning_rate": 0.00017477433565051925,
      "loss": 4.7284,
      "step": 27790
    },
    {
      "epoch": 0.5691847725923044,
      "grad_norm": 0.2724134027957916,
      "learning_rate": 0.00017471328761127405,
      "loss": 4.6741,
      "step": 27795
    },
    {
      "epoch": 0.569287162369709,
      "grad_norm": 0.26181870698928833,
      "learning_rate": 0.00017465223536405566,
      "loss": 4.6555,
      "step": 27800
    },
    {
      "epoch": 0.5693895521471136,
      "grad_norm": 0.26660504937171936,
      "learning_rate": 0.00017459117891925946,
      "loss": 4.6492,
      "step": 27805
    },
    {
      "epoch": 0.5694919419245182,
      "grad_norm": 0.26408490538597107,
      "learning_rate": 0.00017453011828728171,
      "loss": 4.6067,
      "step": 27810
    },
    {
      "epoch": 0.5695943317019229,
      "grad_norm": 0.25006911158561707,
      "learning_rate": 0.00017446905347851931,
      "loss": 4.6796,
      "step": 27815
    },
    {
      "epoch": 0.5696967214793275,
      "grad_norm": 0.2554749548435211,
      "learning_rate": 0.0001744079845033699,
      "loss": 4.6869,
      "step": 27820
    },
    {
      "epoch": 0.5697991112567321,
      "grad_norm": 0.2597162127494812,
      "learning_rate": 0.0001743469113722317,
      "loss": 4.6845,
      "step": 27825
    },
    {
      "epoch": 0.5699015010341367,
      "grad_norm": 0.5917623043060303,
      "learning_rate": 0.00017428583409550373,
      "loss": 4.5729,
      "step": 27830
    },
    {
      "epoch": 0.5700038908115413,
      "grad_norm": 0.25182023644447327,
      "learning_rate": 0.00017422475268358585,
      "loss": 4.693,
      "step": 27835
    },
    {
      "epoch": 0.570106280588946,
      "grad_norm": 0.2560945153236389,
      "learning_rate": 0.00017416366714687834,
      "loss": 4.7037,
      "step": 27840
    },
    {
      "epoch": 0.5702086703663506,
      "grad_norm": 0.2556745409965515,
      "learning_rate": 0.0001741025774957825,
      "loss": 4.6833,
      "step": 27845
    },
    {
      "epoch": 0.5703110601437552,
      "grad_norm": 0.2526347041130066,
      "learning_rate": 0.00017404148374070003,
      "loss": 4.6932,
      "step": 27850
    },
    {
      "epoch": 0.5704134499211598,
      "grad_norm": 0.2605088949203491,
      "learning_rate": 0.00017398038589203354,
      "loss": 4.7266,
      "step": 27855
    },
    {
      "epoch": 0.5705158396985645,
      "grad_norm": 0.25798657536506653,
      "learning_rate": 0.00017391928396018623,
      "loss": 4.6746,
      "step": 27860
    },
    {
      "epoch": 0.5706182294759691,
      "grad_norm": 0.27800747752189636,
      "learning_rate": 0.00017385817795556206,
      "loss": 4.6993,
      "step": 27865
    },
    {
      "epoch": 0.5707206192533737,
      "grad_norm": 0.2565612494945526,
      "learning_rate": 0.00017379706788856565,
      "loss": 4.6535,
      "step": 27870
    },
    {
      "epoch": 0.5708230090307783,
      "grad_norm": 0.25342145562171936,
      "learning_rate": 0.00017373595376960225,
      "loss": 4.5707,
      "step": 27875
    },
    {
      "epoch": 0.5709253988081829,
      "grad_norm": 0.27122899889945984,
      "learning_rate": 0.00017367483560907793,
      "loss": 4.7282,
      "step": 27880
    },
    {
      "epoch": 0.5710277885855877,
      "grad_norm": 0.24127766489982605,
      "learning_rate": 0.00017361371341739944,
      "loss": 4.6964,
      "step": 27885
    },
    {
      "epoch": 0.5711301783629923,
      "grad_norm": 0.2535822093486786,
      "learning_rate": 0.000173552587204974,
      "loss": 4.6916,
      "step": 27890
    },
    {
      "epoch": 0.5712325681403969,
      "grad_norm": 0.28374695777893066,
      "learning_rate": 0.00017349145698220981,
      "loss": 4.6998,
      "step": 27895
    },
    {
      "epoch": 0.5713349579178015,
      "grad_norm": 0.2652948498725891,
      "learning_rate": 0.0001734303227595155,
      "loss": 4.6927,
      "step": 27900
    },
    {
      "epoch": 0.5714373476952062,
      "grad_norm": 0.28421682119369507,
      "learning_rate": 0.00017336918454730063,
      "loss": 4.6617,
      "step": 27905
    },
    {
      "epoch": 0.5715397374726108,
      "grad_norm": 0.27349361777305603,
      "learning_rate": 0.0001733080423559752,
      "loss": 4.6623,
      "step": 27910
    },
    {
      "epoch": 0.5716421272500154,
      "grad_norm": 0.269765168428421,
      "learning_rate": 0.00017324689619595008,
      "loss": 4.6512,
      "step": 27915
    },
    {
      "epoch": 0.57174451702742,
      "grad_norm": 0.2743627727031708,
      "learning_rate": 0.00017318574607763664,
      "loss": 4.6922,
      "step": 27920
    },
    {
      "epoch": 0.5718469068048246,
      "grad_norm": 0.2587744891643524,
      "learning_rate": 0.00017312459201144706,
      "loss": 4.6939,
      "step": 27925
    },
    {
      "epoch": 0.5719492965822293,
      "grad_norm": 0.24722817540168762,
      "learning_rate": 0.00017306343400779416,
      "loss": 4.6897,
      "step": 27930
    },
    {
      "epoch": 0.5720516863596339,
      "grad_norm": 0.2554013133049011,
      "learning_rate": 0.0001730022720770914,
      "loss": 4.6531,
      "step": 27935
    },
    {
      "epoch": 0.5721540761370385,
      "grad_norm": 0.2546253800392151,
      "learning_rate": 0.00017294110622975294,
      "loss": 4.7333,
      "step": 27940
    },
    {
      "epoch": 0.5722564659144431,
      "grad_norm": 0.2455330789089203,
      "learning_rate": 0.00017287993647619352,
      "loss": 4.5944,
      "step": 27945
    },
    {
      "epoch": 0.5723588556918477,
      "grad_norm": 0.26898321509361267,
      "learning_rate": 0.0001728187628268287,
      "loss": 4.6866,
      "step": 27950
    },
    {
      "epoch": 0.5724612454692524,
      "grad_norm": 0.26398465037345886,
      "learning_rate": 0.0001727575852920746,
      "loss": 4.7012,
      "step": 27955
    },
    {
      "epoch": 0.572563635246657,
      "grad_norm": 0.27200570702552795,
      "learning_rate": 0.00017269640388234804,
      "loss": 4.6234,
      "step": 27960
    },
    {
      "epoch": 0.5726660250240616,
      "grad_norm": 0.2645496129989624,
      "learning_rate": 0.00017263521860806644,
      "loss": 4.6344,
      "step": 27965
    },
    {
      "epoch": 0.5727684148014662,
      "grad_norm": 0.2591147720813751,
      "learning_rate": 0.00017257402947964793,
      "loss": 4.7042,
      "step": 27970
    },
    {
      "epoch": 0.5728708045788709,
      "grad_norm": 0.254060298204422,
      "learning_rate": 0.00017251283650751133,
      "loss": 4.7294,
      "step": 27975
    },
    {
      "epoch": 0.5729731943562755,
      "grad_norm": 0.26018354296684265,
      "learning_rate": 0.00017245163970207598,
      "loss": 4.6871,
      "step": 27980
    },
    {
      "epoch": 0.5730755841336801,
      "grad_norm": 0.2611655294895172,
      "learning_rate": 0.0001723904390737621,
      "loss": 4.6886,
      "step": 27985
    },
    {
      "epoch": 0.5731779739110847,
      "grad_norm": 0.2722248136997223,
      "learning_rate": 0.0001723292346329903,
      "loss": 4.6262,
      "step": 27990
    },
    {
      "epoch": 0.5732803636884893,
      "grad_norm": 0.2603926658630371,
      "learning_rate": 0.00017226802639018195,
      "loss": 4.6824,
      "step": 27995
    },
    {
      "epoch": 0.573382753465894,
      "grad_norm": 0.25823864340782166,
      "learning_rate": 0.00017220681435575923,
      "loss": 4.6908,
      "step": 28000
    },
    {
      "epoch": 0.573382753465894,
      "eval_loss": 4.6973042488098145,
      "eval_runtime": 17.1349,
      "eval_samples_per_second": 33.966,
      "eval_steps_per_second": 4.26,
      "step": 28000
    },
    {
      "epoch": 0.5734851432432986,
      "grad_norm": 0.2666439712047577,
      "learning_rate": 0.00017214559854014468,
      "loss": 4.6527,
      "step": 28005
    },
    {
      "epoch": 0.5735875330207032,
      "grad_norm": 0.26047781109809875,
      "learning_rate": 0.00017208437895376168,
      "loss": 4.7092,
      "step": 28010
    },
    {
      "epoch": 0.5736899227981078,
      "grad_norm": 0.25635117292404175,
      "learning_rate": 0.00017202315560703416,
      "loss": 4.7167,
      "step": 28015
    },
    {
      "epoch": 0.5737923125755124,
      "grad_norm": 0.26363733410835266,
      "learning_rate": 0.00017196192851038676,
      "loss": 4.6531,
      "step": 28020
    },
    {
      "epoch": 0.5738947023529171,
      "grad_norm": 0.2589953541755676,
      "learning_rate": 0.00017190069767424472,
      "loss": 4.6882,
      "step": 28025
    },
    {
      "epoch": 0.5739970921303217,
      "grad_norm": 0.28121355175971985,
      "learning_rate": 0.0001718394631090339,
      "loss": 4.7118,
      "step": 28030
    },
    {
      "epoch": 0.5740994819077263,
      "grad_norm": 0.2536048889160156,
      "learning_rate": 0.00017177822482518078,
      "loss": 4.6569,
      "step": 28035
    },
    {
      "epoch": 0.5742018716851309,
      "grad_norm": 0.24868465960025787,
      "learning_rate": 0.00017171698283311254,
      "loss": 4.6585,
      "step": 28040
    },
    {
      "epoch": 0.5743042614625355,
      "grad_norm": 0.2640113830566406,
      "learning_rate": 0.00017165573714325697,
      "loss": 4.656,
      "step": 28045
    },
    {
      "epoch": 0.5744066512399402,
      "grad_norm": 0.24657516181468964,
      "learning_rate": 0.0001715944877660424,
      "loss": 4.6864,
      "step": 28050
    },
    {
      "epoch": 0.5745090410173448,
      "grad_norm": 0.2633567452430725,
      "learning_rate": 0.00017153323471189803,
      "loss": 4.7001,
      "step": 28055
    },
    {
      "epoch": 0.5746114307947494,
      "grad_norm": 0.29227760434150696,
      "learning_rate": 0.00017147197799125333,
      "loss": 4.645,
      "step": 28060
    },
    {
      "epoch": 0.574713820572154,
      "grad_norm": 0.26054689288139343,
      "learning_rate": 0.00017141071761453872,
      "loss": 4.6636,
      "step": 28065
    },
    {
      "epoch": 0.5748162103495587,
      "grad_norm": 0.2804853320121765,
      "learning_rate": 0.00017134945359218508,
      "loss": 4.7059,
      "step": 28070
    },
    {
      "epoch": 0.5749186001269633,
      "grad_norm": 0.26142027974128723,
      "learning_rate": 0.0001712881859346239,
      "loss": 4.6987,
      "step": 28075
    },
    {
      "epoch": 0.5750209899043679,
      "grad_norm": 0.2628867030143738,
      "learning_rate": 0.00017122691465228733,
      "loss": 4.6591,
      "step": 28080
    },
    {
      "epoch": 0.5751233796817725,
      "grad_norm": 0.26298439502716064,
      "learning_rate": 0.00017116563975560817,
      "loss": 4.7099,
      "step": 28085
    },
    {
      "epoch": 0.5752257694591772,
      "grad_norm": 0.2536146640777588,
      "learning_rate": 0.00017110436125501976,
      "loss": 4.6849,
      "step": 28090
    },
    {
      "epoch": 0.5753281592365819,
      "grad_norm": 0.2704731225967407,
      "learning_rate": 0.00017104307916095615,
      "loss": 4.6969,
      "step": 28095
    },
    {
      "epoch": 0.5754305490139865,
      "grad_norm": 0.24779964983463287,
      "learning_rate": 0.0001709817934838519,
      "loss": 4.6344,
      "step": 28100
    },
    {
      "epoch": 0.5755329387913911,
      "grad_norm": 0.2553950250148773,
      "learning_rate": 0.00017092050423414226,
      "loss": 4.718,
      "step": 28105
    },
    {
      "epoch": 0.5756353285687957,
      "grad_norm": 0.27989503741264343,
      "learning_rate": 0.00017085921142226298,
      "loss": 4.6681,
      "step": 28110
    },
    {
      "epoch": 0.5757377183462004,
      "grad_norm": 0.26345035433769226,
      "learning_rate": 0.00017079791505865058,
      "loss": 4.6979,
      "step": 28115
    },
    {
      "epoch": 0.575840108123605,
      "grad_norm": 0.2537969648838043,
      "learning_rate": 0.00017073661515374205,
      "loss": 4.7218,
      "step": 28120
    },
    {
      "epoch": 0.5759424979010096,
      "grad_norm": 0.27907902002334595,
      "learning_rate": 0.00017067531171797504,
      "loss": 4.6075,
      "step": 28125
    },
    {
      "epoch": 0.5760448876784142,
      "grad_norm": 0.2688951790332794,
      "learning_rate": 0.00017061400476178778,
      "loss": 4.6963,
      "step": 28130
    },
    {
      "epoch": 0.5761472774558188,
      "grad_norm": 0.25790733098983765,
      "learning_rate": 0.0001705526942956191,
      "loss": 4.6751,
      "step": 28135
    },
    {
      "epoch": 0.5762496672332235,
      "grad_norm": 0.24855847656726837,
      "learning_rate": 0.00017049138032990845,
      "loss": 4.6854,
      "step": 28140
    },
    {
      "epoch": 0.5763520570106281,
      "grad_norm": 0.2617311179637909,
      "learning_rate": 0.0001704300628750959,
      "loss": 4.6834,
      "step": 28145
    },
    {
      "epoch": 0.5764544467880327,
      "grad_norm": 0.26312047243118286,
      "learning_rate": 0.00017036874194162202,
      "loss": 4.6839,
      "step": 28150
    },
    {
      "epoch": 0.5765568365654373,
      "grad_norm": 0.26814526319503784,
      "learning_rate": 0.00017030741753992806,
      "loss": 4.6945,
      "step": 28155
    },
    {
      "epoch": 0.576659226342842,
      "grad_norm": 0.399715393781662,
      "learning_rate": 0.00017024608968045578,
      "loss": 4.7197,
      "step": 28160
    },
    {
      "epoch": 0.5767616161202466,
      "grad_norm": 0.25246548652648926,
      "learning_rate": 0.00017018475837364766,
      "loss": 4.6981,
      "step": 28165
    },
    {
      "epoch": 0.5768640058976512,
      "grad_norm": 0.2590174674987793,
      "learning_rate": 0.00017012342362994663,
      "loss": 4.6296,
      "step": 28170
    },
    {
      "epoch": 0.5769663956750558,
      "grad_norm": 0.2699030041694641,
      "learning_rate": 0.0001700620854597963,
      "loss": 4.6516,
      "step": 28175
    },
    {
      "epoch": 0.5770687854524604,
      "grad_norm": 0.2671308219432831,
      "learning_rate": 0.00017000074387364073,
      "loss": 4.7291,
      "step": 28180
    },
    {
      "epoch": 0.577171175229865,
      "grad_norm": 0.2662724256515503,
      "learning_rate": 0.00016993939888192478,
      "loss": 4.6775,
      "step": 28185
    },
    {
      "epoch": 0.5772735650072697,
      "grad_norm": 0.283224493265152,
      "learning_rate": 0.0001698780504950937,
      "loss": 4.7108,
      "step": 28190
    },
    {
      "epoch": 0.5773759547846743,
      "grad_norm": 0.2565319836139679,
      "learning_rate": 0.00016981669872359335,
      "loss": 4.6752,
      "step": 28195
    },
    {
      "epoch": 0.5774783445620789,
      "grad_norm": 0.26022616028785706,
      "learning_rate": 0.00016975534357787026,
      "loss": 4.6665,
      "step": 28200
    },
    {
      "epoch": 0.5775807343394835,
      "grad_norm": 0.2849796712398529,
      "learning_rate": 0.00016969398506837144,
      "loss": 4.7026,
      "step": 28205
    },
    {
      "epoch": 0.5776831241168882,
      "grad_norm": 0.2585475444793701,
      "learning_rate": 0.00016963262320554456,
      "loss": 4.7047,
      "step": 28210
    },
    {
      "epoch": 0.5777855138942928,
      "grad_norm": 0.2459850162267685,
      "learning_rate": 0.00016957125799983775,
      "loss": 4.6862,
      "step": 28215
    },
    {
      "epoch": 0.5778879036716974,
      "grad_norm": 0.26823312044143677,
      "learning_rate": 0.00016950988946169983,
      "loss": 4.7114,
      "step": 28220
    },
    {
      "epoch": 0.577990293449102,
      "grad_norm": 0.28757771849632263,
      "learning_rate": 0.00016944851760158003,
      "loss": 4.6745,
      "step": 28225
    },
    {
      "epoch": 0.5780926832265066,
      "grad_norm": 0.2935873866081238,
      "learning_rate": 0.00016938714242992827,
      "loss": 4.6881,
      "step": 28230
    },
    {
      "epoch": 0.5781950730039113,
      "grad_norm": 0.25757095217704773,
      "learning_rate": 0.00016932576395719507,
      "loss": 4.6828,
      "step": 28235
    },
    {
      "epoch": 0.5782974627813159,
      "grad_norm": 0.2548655867576599,
      "learning_rate": 0.0001692643821938314,
      "loss": 4.6973,
      "step": 28240
    },
    {
      "epoch": 0.5783998525587205,
      "grad_norm": 0.2801399528980255,
      "learning_rate": 0.00016920299715028882,
      "loss": 4.6526,
      "step": 28245
    },
    {
      "epoch": 0.5785022423361251,
      "grad_norm": 0.2520582675933838,
      "learning_rate": 0.0001691416088370195,
      "loss": 4.6791,
      "step": 28250
    },
    {
      "epoch": 0.5786046321135297,
      "grad_norm": 0.25987929105758667,
      "learning_rate": 0.00016908021726447613,
      "loss": 4.6524,
      "step": 28255
    },
    {
      "epoch": 0.5787070218909344,
      "grad_norm": 0.2543611526489258,
      "learning_rate": 0.00016901882244311188,
      "loss": 4.668,
      "step": 28260
    },
    {
      "epoch": 0.578809411668339,
      "grad_norm": 0.2713223993778229,
      "learning_rate": 0.00016895742438338067,
      "loss": 4.6145,
      "step": 28265
    },
    {
      "epoch": 0.5789118014457436,
      "grad_norm": 0.2809205651283264,
      "learning_rate": 0.0001688960230957368,
      "loss": 4.6519,
      "step": 28270
    },
    {
      "epoch": 0.5790141912231482,
      "grad_norm": 0.27665913105010986,
      "learning_rate": 0.00016883461859063513,
      "loss": 4.7191,
      "step": 28275
    },
    {
      "epoch": 0.5791165810005529,
      "grad_norm": 0.2994121313095093,
      "learning_rate": 0.00016877321087853118,
      "loss": 4.6631,
      "step": 28280
    },
    {
      "epoch": 0.5792189707779575,
      "grad_norm": 0.2573554515838623,
      "learning_rate": 0.00016871179996988083,
      "loss": 4.6867,
      "step": 28285
    },
    {
      "epoch": 0.5793213605553622,
      "grad_norm": 0.2663307189941406,
      "learning_rate": 0.00016865038587514077,
      "loss": 4.6789,
      "step": 28290
    },
    {
      "epoch": 0.5794237503327668,
      "grad_norm": 0.2640919089317322,
      "learning_rate": 0.00016858896860476799,
      "loss": 4.6987,
      "step": 28295
    },
    {
      "epoch": 0.5795261401101715,
      "grad_norm": 0.2629927098751068,
      "learning_rate": 0.0001685275481692201,
      "loss": 4.6202,
      "step": 28300
    },
    {
      "epoch": 0.5796285298875761,
      "grad_norm": 0.5783554315567017,
      "learning_rate": 0.00016846612457895534,
      "loss": 4.6934,
      "step": 28305
    },
    {
      "epoch": 0.5797309196649807,
      "grad_norm": 0.29603996872901917,
      "learning_rate": 0.0001684046978444324,
      "loss": 4.6555,
      "step": 28310
    },
    {
      "epoch": 0.5798333094423853,
      "grad_norm": 0.2576543986797333,
      "learning_rate": 0.00016834326797611037,
      "loss": 4.6764,
      "step": 28315
    },
    {
      "epoch": 0.5799356992197899,
      "grad_norm": 0.2601107656955719,
      "learning_rate": 0.00016828183498444915,
      "loss": 4.6808,
      "step": 28320
    },
    {
      "epoch": 0.5800380889971946,
      "grad_norm": 0.25905126333236694,
      "learning_rate": 0.00016822039887990907,
      "loss": 4.6454,
      "step": 28325
    },
    {
      "epoch": 0.5801404787745992,
      "grad_norm": 0.27695906162261963,
      "learning_rate": 0.00016815895967295085,
      "loss": 4.6744,
      "step": 28330
    },
    {
      "epoch": 0.5802428685520038,
      "grad_norm": 0.2757084369659424,
      "learning_rate": 0.00016809751737403592,
      "loss": 4.7141,
      "step": 28335
    },
    {
      "epoch": 0.5803452583294084,
      "grad_norm": 0.27058765292167664,
      "learning_rate": 0.00016803607199362614,
      "loss": 4.6815,
      "step": 28340
    },
    {
      "epoch": 0.580447648106813,
      "grad_norm": 0.26202329993247986,
      "learning_rate": 0.00016797462354218388,
      "loss": 4.6967,
      "step": 28345
    },
    {
      "epoch": 0.5805500378842177,
      "grad_norm": 0.2569383382797241,
      "learning_rate": 0.00016791317203017217,
      "loss": 4.6194,
      "step": 28350
    },
    {
      "epoch": 0.5806524276616223,
      "grad_norm": 0.2605290710926056,
      "learning_rate": 0.0001678517174680544,
      "loss": 4.707,
      "step": 28355
    },
    {
      "epoch": 0.5807548174390269,
      "grad_norm": 0.31713876128196716,
      "learning_rate": 0.00016779025986629452,
      "loss": 4.6597,
      "step": 28360
    },
    {
      "epoch": 0.5808572072164315,
      "grad_norm": 0.25696828961372375,
      "learning_rate": 0.00016772879923535706,
      "loss": 4.69,
      "step": 28365
    },
    {
      "epoch": 0.5809595969938361,
      "grad_norm": 0.31100475788116455,
      "learning_rate": 0.00016766733558570696,
      "loss": 4.6482,
      "step": 28370
    },
    {
      "epoch": 0.5810619867712408,
      "grad_norm": 0.26407870650291443,
      "learning_rate": 0.0001676058689278099,
      "loss": 4.7087,
      "step": 28375
    },
    {
      "epoch": 0.5811643765486454,
      "grad_norm": 0.26597467064857483,
      "learning_rate": 0.00016754439927213174,
      "loss": 4.6945,
      "step": 28380
    },
    {
      "epoch": 0.58126676632605,
      "grad_norm": 0.25429606437683105,
      "learning_rate": 0.00016748292662913908,
      "loss": 4.6643,
      "step": 28385
    },
    {
      "epoch": 0.5813691561034546,
      "grad_norm": 0.2809353172779083,
      "learning_rate": 0.00016742145100929898,
      "loss": 4.7229,
      "step": 28390
    },
    {
      "epoch": 0.5814715458808593,
      "grad_norm": 0.2482432723045349,
      "learning_rate": 0.00016735997242307903,
      "loss": 4.6838,
      "step": 28395
    },
    {
      "epoch": 0.5815739356582639,
      "grad_norm": 0.265804648399353,
      "learning_rate": 0.0001672984908809472,
      "loss": 4.6401,
      "step": 28400
    },
    {
      "epoch": 0.5816763254356685,
      "grad_norm": 0.2800675928592682,
      "learning_rate": 0.00016723700639337217,
      "loss": 4.6847,
      "step": 28405
    },
    {
      "epoch": 0.5817787152130731,
      "grad_norm": 0.38202622532844543,
      "learning_rate": 0.00016717551897082287,
      "loss": 4.7052,
      "step": 28410
    },
    {
      "epoch": 0.5818811049904777,
      "grad_norm": 0.2673313319683075,
      "learning_rate": 0.00016711402862376902,
      "loss": 4.6676,
      "step": 28415
    },
    {
      "epoch": 0.5819834947678824,
      "grad_norm": 0.27206218242645264,
      "learning_rate": 0.00016705253536268062,
      "loss": 4.6282,
      "step": 28420
    },
    {
      "epoch": 0.582085884545287,
      "grad_norm": 0.26041173934936523,
      "learning_rate": 0.00016699103919802817,
      "loss": 4.6551,
      "step": 28425
    },
    {
      "epoch": 0.5821882743226916,
      "grad_norm": 0.2748391926288605,
      "learning_rate": 0.0001669295401402828,
      "loss": 4.6714,
      "step": 28430
    },
    {
      "epoch": 0.5822906641000962,
      "grad_norm": 0.2595105469226837,
      "learning_rate": 0.00016686803819991605,
      "loss": 4.6219,
      "step": 28435
    },
    {
      "epoch": 0.5823930538775008,
      "grad_norm": 0.2541680335998535,
      "learning_rate": 0.00016680653338739992,
      "loss": 4.6717,
      "step": 28440
    },
    {
      "epoch": 0.5824954436549055,
      "grad_norm": 0.2788894474506378,
      "learning_rate": 0.000166745025713207,
      "loss": 4.6475,
      "step": 28445
    },
    {
      "epoch": 0.5825978334323101,
      "grad_norm": 0.2586911916732788,
      "learning_rate": 0.0001666835151878103,
      "loss": 4.6912,
      "step": 28450
    },
    {
      "epoch": 0.5827002232097147,
      "grad_norm": 0.4290415942668915,
      "learning_rate": 0.00016662200182168322,
      "loss": 4.7443,
      "step": 28455
    },
    {
      "epoch": 0.5828026129871193,
      "grad_norm": 0.26729732751846313,
      "learning_rate": 0.00016656048562529982,
      "loss": 4.6921,
      "step": 28460
    },
    {
      "epoch": 0.582905002764524,
      "grad_norm": 0.27923545241355896,
      "learning_rate": 0.00016649896660913462,
      "loss": 4.6826,
      "step": 28465
    },
    {
      "epoch": 0.5830073925419286,
      "grad_norm": 0.25673234462738037,
      "learning_rate": 0.00016643744478366245,
      "loss": 4.667,
      "step": 28470
    },
    {
      "epoch": 0.5831097823193332,
      "grad_norm": 0.2606159448623657,
      "learning_rate": 0.00016637592015935886,
      "loss": 4.6494,
      "step": 28475
    },
    {
      "epoch": 0.5832121720967378,
      "grad_norm": 0.25604692101478577,
      "learning_rate": 0.00016631439274669964,
      "loss": 4.6766,
      "step": 28480
    },
    {
      "epoch": 0.5833145618741424,
      "grad_norm": 0.25201162695884705,
      "learning_rate": 0.00016625286255616124,
      "loss": 4.7532,
      "step": 28485
    },
    {
      "epoch": 0.583416951651547,
      "grad_norm": 0.24749386310577393,
      "learning_rate": 0.00016619132959822046,
      "loss": 4.7172,
      "step": 28490
    },
    {
      "epoch": 0.5835193414289518,
      "grad_norm": 0.25692281126976013,
      "learning_rate": 0.0001661297938833547,
      "loss": 4.6548,
      "step": 28495
    },
    {
      "epoch": 0.5836217312063564,
      "grad_norm": 0.24823105335235596,
      "learning_rate": 0.00016606825542204163,
      "loss": 4.6655,
      "step": 28500
    },
    {
      "epoch": 0.583724120983761,
      "grad_norm": 0.25307193398475647,
      "learning_rate": 0.00016600671422475959,
      "loss": 4.7401,
      "step": 28505
    },
    {
      "epoch": 0.5838265107611657,
      "grad_norm": 0.25224077701568604,
      "learning_rate": 0.00016594517030198727,
      "loss": 4.7756,
      "step": 28510
    },
    {
      "epoch": 0.5839289005385703,
      "grad_norm": 0.2719027101993561,
      "learning_rate": 0.0001658836236642039,
      "loss": 4.6432,
      "step": 28515
    },
    {
      "epoch": 0.5840312903159749,
      "grad_norm": 0.2612490653991699,
      "learning_rate": 0.00016582207432188912,
      "loss": 4.759,
      "step": 28520
    },
    {
      "epoch": 0.5841336800933795,
      "grad_norm": 0.25117358565330505,
      "learning_rate": 0.00016576052228552298,
      "loss": 4.7374,
      "step": 28525
    },
    {
      "epoch": 0.5842360698707841,
      "grad_norm": 0.24579162895679474,
      "learning_rate": 0.0001656989675655861,
      "loss": 4.6565,
      "step": 28530
    },
    {
      "epoch": 0.5843384596481888,
      "grad_norm": 0.3122329115867615,
      "learning_rate": 0.00016563741017255952,
      "loss": 4.669,
      "step": 28535
    },
    {
      "epoch": 0.5844408494255934,
      "grad_norm": 0.2683679461479187,
      "learning_rate": 0.0001655758501169247,
      "loss": 4.664,
      "step": 28540
    },
    {
      "epoch": 0.584543239202998,
      "grad_norm": 0.275326669216156,
      "learning_rate": 0.00016551428740916357,
      "loss": 4.7266,
      "step": 28545
    },
    {
      "epoch": 0.5846456289804026,
      "grad_norm": 0.30032962560653687,
      "learning_rate": 0.00016545272205975852,
      "loss": 4.661,
      "step": 28550
    },
    {
      "epoch": 0.5847480187578072,
      "grad_norm": 0.36011090874671936,
      "learning_rate": 0.00016539115407919236,
      "loss": 4.6642,
      "step": 28555
    },
    {
      "epoch": 0.5848504085352119,
      "grad_norm": 0.30276724696159363,
      "learning_rate": 0.00016532958347794844,
      "loss": 4.6804,
      "step": 28560
    },
    {
      "epoch": 0.5849527983126165,
      "grad_norm": 0.2697668969631195,
      "learning_rate": 0.00016526801026651045,
      "loss": 4.6549,
      "step": 28565
    },
    {
      "epoch": 0.5850551880900211,
      "grad_norm": 0.28318217396736145,
      "learning_rate": 0.00016520643445536257,
      "loss": 4.6579,
      "step": 28570
    },
    {
      "epoch": 0.5851575778674257,
      "grad_norm": 0.26577475666999817,
      "learning_rate": 0.00016514485605498936,
      "loss": 4.6787,
      "step": 28575
    },
    {
      "epoch": 0.5852599676448303,
      "grad_norm": 0.27212128043174744,
      "learning_rate": 0.00016508327507587595,
      "loss": 4.6718,
      "step": 28580
    },
    {
      "epoch": 0.585362357422235,
      "grad_norm": 0.27527156472206116,
      "learning_rate": 0.00016502169152850786,
      "loss": 4.6573,
      "step": 28585
    },
    {
      "epoch": 0.5854647471996396,
      "grad_norm": 0.2678073048591614,
      "learning_rate": 0.000164960105423371,
      "loss": 4.6498,
      "step": 28590
    },
    {
      "epoch": 0.5855671369770442,
      "grad_norm": 0.2557547688484192,
      "learning_rate": 0.00016489851677095165,
      "loss": 4.6869,
      "step": 28595
    },
    {
      "epoch": 0.5856695267544488,
      "grad_norm": 0.3396357595920563,
      "learning_rate": 0.0001648369255817367,
      "loss": 4.6218,
      "step": 28600
    },
    {
      "epoch": 0.5857719165318535,
      "grad_norm": 0.2767936885356903,
      "learning_rate": 0.00016477533186621345,
      "loss": 4.7138,
      "step": 28605
    },
    {
      "epoch": 0.5858743063092581,
      "grad_norm": 0.28629255294799805,
      "learning_rate": 0.0001647137356348694,
      "loss": 4.7251,
      "step": 28610
    },
    {
      "epoch": 0.5859766960866627,
      "grad_norm": 0.30186545848846436,
      "learning_rate": 0.0001646521368981928,
      "loss": 4.6946,
      "step": 28615
    },
    {
      "epoch": 0.5860790858640673,
      "grad_norm": 0.3243028223514557,
      "learning_rate": 0.0001645905356666721,
      "loss": 4.6657,
      "step": 28620
    },
    {
      "epoch": 0.5861814756414719,
      "grad_norm": 0.3595491349697113,
      "learning_rate": 0.00016452893195079618,
      "loss": 4.5852,
      "step": 28625
    },
    {
      "epoch": 0.5862838654188766,
      "grad_norm": 0.25961947441101074,
      "learning_rate": 0.00016446732576105453,
      "loss": 4.6496,
      "step": 28630
    },
    {
      "epoch": 0.5863862551962812,
      "grad_norm": 0.4880898892879486,
      "learning_rate": 0.00016440571710793692,
      "loss": 4.6133,
      "step": 28635
    },
    {
      "epoch": 0.5864886449736858,
      "grad_norm": 0.2825993001461029,
      "learning_rate": 0.00016434410600193347,
      "loss": 4.7074,
      "step": 28640
    },
    {
      "epoch": 0.5865910347510904,
      "grad_norm": 0.2694116532802582,
      "learning_rate": 0.00016428249245353487,
      "loss": 4.6548,
      "step": 28645
    },
    {
      "epoch": 0.586693424528495,
      "grad_norm": 0.27538102865219116,
      "learning_rate": 0.00016422087647323218,
      "loss": 4.6847,
      "step": 28650
    },
    {
      "epoch": 0.5867958143058997,
      "grad_norm": 0.25950124859809875,
      "learning_rate": 0.00016415925807151683,
      "loss": 4.6639,
      "step": 28655
    },
    {
      "epoch": 0.5868982040833043,
      "grad_norm": 0.25048917531967163,
      "learning_rate": 0.00016409763725888074,
      "loss": 4.6673,
      "step": 28660
    },
    {
      "epoch": 0.5870005938607089,
      "grad_norm": 0.2542320787906647,
      "learning_rate": 0.00016403601404581608,
      "loss": 4.6901,
      "step": 28665
    },
    {
      "epoch": 0.5871029836381135,
      "grad_norm": 0.25401172041893005,
      "learning_rate": 0.00016397438844281562,
      "loss": 4.6929,
      "step": 28670
    },
    {
      "epoch": 0.5872053734155182,
      "grad_norm": 0.2795599400997162,
      "learning_rate": 0.00016391276046037243,
      "loss": 4.6901,
      "step": 28675
    },
    {
      "epoch": 0.5873077631929228,
      "grad_norm": 0.26690590381622314,
      "learning_rate": 0.00016385113010898003,
      "loss": 4.6625,
      "step": 28680
    },
    {
      "epoch": 0.5874101529703274,
      "grad_norm": 0.2658803164958954,
      "learning_rate": 0.00016378949739913234,
      "loss": 4.6648,
      "step": 28685
    },
    {
      "epoch": 0.587512542747732,
      "grad_norm": 0.2667374908924103,
      "learning_rate": 0.00016372786234132363,
      "loss": 4.6663,
      "step": 28690
    },
    {
      "epoch": 0.5876149325251366,
      "grad_norm": 0.2675791084766388,
      "learning_rate": 0.00016366622494604858,
      "loss": 4.6718,
      "step": 28695
    },
    {
      "epoch": 0.5877173223025414,
      "grad_norm": 0.2973770499229431,
      "learning_rate": 0.00016360458522380234,
      "loss": 4.6598,
      "step": 28700
    },
    {
      "epoch": 0.587819712079946,
      "grad_norm": 0.2763424217700958,
      "learning_rate": 0.00016354294318508044,
      "loss": 4.5858,
      "step": 28705
    },
    {
      "epoch": 0.5879221018573506,
      "grad_norm": 0.2831302285194397,
      "learning_rate": 0.00016348129884037867,
      "loss": 4.6891,
      "step": 28710
    },
    {
      "epoch": 0.5880244916347552,
      "grad_norm": 0.2575315237045288,
      "learning_rate": 0.00016341965220019342,
      "loss": 4.6859,
      "step": 28715
    },
    {
      "epoch": 0.5881268814121599,
      "grad_norm": 0.28888994455337524,
      "learning_rate": 0.0001633580032750213,
      "loss": 4.6963,
      "step": 28720
    },
    {
      "epoch": 0.5882292711895645,
      "grad_norm": 0.2642802894115448,
      "learning_rate": 0.0001632963520753594,
      "loss": 4.6827,
      "step": 28725
    },
    {
      "epoch": 0.5883316609669691,
      "grad_norm": 0.2706809341907501,
      "learning_rate": 0.00016323469861170522,
      "loss": 4.6905,
      "step": 28730
    },
    {
      "epoch": 0.5884340507443737,
      "grad_norm": 0.3463543653488159,
      "learning_rate": 0.0001631730428945565,
      "loss": 4.6254,
      "step": 28735
    },
    {
      "epoch": 0.5885364405217783,
      "grad_norm": 0.26060375571250916,
      "learning_rate": 0.00016311138493441154,
      "loss": 4.6735,
      "step": 28740
    },
    {
      "epoch": 0.588638830299183,
      "grad_norm": 0.2775377333164215,
      "learning_rate": 0.00016304972474176895,
      "loss": 4.6673,
      "step": 28745
    },
    {
      "epoch": 0.5887412200765876,
      "grad_norm": 0.29451581835746765,
      "learning_rate": 0.00016298806232712762,
      "loss": 4.647,
      "step": 28750
    },
    {
      "epoch": 0.5888436098539922,
      "grad_norm": 0.28171879053115845,
      "learning_rate": 0.00016292639770098708,
      "loss": 4.6576,
      "step": 28755
    },
    {
      "epoch": 0.5889459996313968,
      "grad_norm": 0.2790437638759613,
      "learning_rate": 0.00016286473087384694,
      "loss": 4.7157,
      "step": 28760
    },
    {
      "epoch": 0.5890483894088014,
      "grad_norm": 0.27042701840400696,
      "learning_rate": 0.0001628030618562073,
      "loss": 4.7097,
      "step": 28765
    },
    {
      "epoch": 0.5891507791862061,
      "grad_norm": 0.35401538014411926,
      "learning_rate": 0.00016274139065856872,
      "loss": 4.725,
      "step": 28770
    },
    {
      "epoch": 0.5892531689636107,
      "grad_norm": 0.2562393546104431,
      "learning_rate": 0.00016267971729143205,
      "loss": 4.6841,
      "step": 28775
    },
    {
      "epoch": 0.5893555587410153,
      "grad_norm": 0.2634783387184143,
      "learning_rate": 0.00016261804176529848,
      "loss": 4.6525,
      "step": 28780
    },
    {
      "epoch": 0.5894579485184199,
      "grad_norm": 0.37353622913360596,
      "learning_rate": 0.00016255636409066965,
      "loss": 4.6772,
      "step": 28785
    },
    {
      "epoch": 0.5895603382958245,
      "grad_norm": 0.31675294041633606,
      "learning_rate": 0.00016249468427804745,
      "loss": 4.6367,
      "step": 28790
    },
    {
      "epoch": 0.5896627280732292,
      "grad_norm": 0.2585216760635376,
      "learning_rate": 0.0001624330023379343,
      "loss": 4.6942,
      "step": 28795
    },
    {
      "epoch": 0.5897651178506338,
      "grad_norm": 0.28349608182907104,
      "learning_rate": 0.0001623713182808329,
      "loss": 4.7013,
      "step": 28800
    },
    {
      "epoch": 0.5898675076280384,
      "grad_norm": 0.25947660207748413,
      "learning_rate": 0.00016230963211724613,
      "loss": 4.6381,
      "step": 28805
    },
    {
      "epoch": 0.589969897405443,
      "grad_norm": 0.2591255009174347,
      "learning_rate": 0.0001622479438576776,
      "loss": 4.6237,
      "step": 28810
    },
    {
      "epoch": 0.5900722871828477,
      "grad_norm": 0.2591240406036377,
      "learning_rate": 0.00016218625351263093,
      "loss": 4.6259,
      "step": 28815
    },
    {
      "epoch": 0.5901746769602523,
      "grad_norm": 0.275452196598053,
      "learning_rate": 0.0001621245610926103,
      "loss": 4.6352,
      "step": 28820
    },
    {
      "epoch": 0.5902770667376569,
      "grad_norm": 0.2508513331413269,
      "learning_rate": 0.0001620628666081202,
      "loss": 4.6222,
      "step": 28825
    },
    {
      "epoch": 0.5903794565150615,
      "grad_norm": 0.25792360305786133,
      "learning_rate": 0.00016200117006966543,
      "loss": 4.6654,
      "step": 28830
    },
    {
      "epoch": 0.5904818462924661,
      "grad_norm": 0.263846218585968,
      "learning_rate": 0.0001619394714877511,
      "loss": 4.6983,
      "step": 28835
    },
    {
      "epoch": 0.5905842360698708,
      "grad_norm": 0.26363155245780945,
      "learning_rate": 0.00016187777087288286,
      "loss": 4.7044,
      "step": 28840
    },
    {
      "epoch": 0.5906866258472754,
      "grad_norm": 0.29336559772491455,
      "learning_rate": 0.0001618160682355665,
      "loss": 4.7013,
      "step": 28845
    },
    {
      "epoch": 0.59078901562468,
      "grad_norm": 0.24546614289283752,
      "learning_rate": 0.00016175436358630823,
      "loss": 4.669,
      "step": 28850
    },
    {
      "epoch": 0.5908914054020846,
      "grad_norm": 0.2472277730703354,
      "learning_rate": 0.00016169265693561464,
      "loss": 4.589,
      "step": 28855
    },
    {
      "epoch": 0.5909937951794892,
      "grad_norm": 0.37090569734573364,
      "learning_rate": 0.00016163094829399258,
      "loss": 4.6549,
      "step": 28860
    },
    {
      "epoch": 0.5910961849568939,
      "grad_norm": 0.25700247287750244,
      "learning_rate": 0.00016156923767194937,
      "loss": 4.6596,
      "step": 28865
    },
    {
      "epoch": 0.5911985747342985,
      "grad_norm": 0.27771201729774475,
      "learning_rate": 0.00016150752507999248,
      "loss": 4.7326,
      "step": 28870
    },
    {
      "epoch": 0.5913009645117031,
      "grad_norm": 0.26109370589256287,
      "learning_rate": 0.00016144581052862983,
      "loss": 4.6741,
      "step": 28875
    },
    {
      "epoch": 0.5914033542891077,
      "grad_norm": 0.2527990937232971,
      "learning_rate": 0.00016138409402836974,
      "loss": 4.7112,
      "step": 28880
    },
    {
      "epoch": 0.5915057440665124,
      "grad_norm": 0.2642720639705658,
      "learning_rate": 0.0001613223755897207,
      "loss": 4.6787,
      "step": 28885
    },
    {
      "epoch": 0.591608133843917,
      "grad_norm": 0.2582034170627594,
      "learning_rate": 0.0001612606552231917,
      "loss": 4.6967,
      "step": 28890
    },
    {
      "epoch": 0.5917105236213216,
      "grad_norm": 0.26001131534576416,
      "learning_rate": 0.00016119893293929186,
      "loss": 4.6871,
      "step": 28895
    },
    {
      "epoch": 0.5918129133987263,
      "grad_norm": 0.2612674832344055,
      "learning_rate": 0.00016113720874853083,
      "loss": 4.6856,
      "step": 28900
    },
    {
      "epoch": 0.591915303176131,
      "grad_norm": 0.269471675157547,
      "learning_rate": 0.00016107548266141843,
      "loss": 4.7022,
      "step": 28905
    },
    {
      "epoch": 0.5920176929535356,
      "grad_norm": 0.26346471905708313,
      "learning_rate": 0.0001610137546884649,
      "loss": 4.7134,
      "step": 28910
    },
    {
      "epoch": 0.5921200827309402,
      "grad_norm": 0.2597687840461731,
      "learning_rate": 0.00016095202484018076,
      "loss": 4.6904,
      "step": 28915
    },
    {
      "epoch": 0.5922224725083448,
      "grad_norm": 0.27877506613731384,
      "learning_rate": 0.00016089029312707685,
      "loss": 4.5869,
      "step": 28920
    },
    {
      "epoch": 0.5923248622857494,
      "grad_norm": 0.27964040637016296,
      "learning_rate": 0.00016082855955966438,
      "loss": 4.7057,
      "step": 28925
    },
    {
      "epoch": 0.592427252063154,
      "grad_norm": 0.7393932342529297,
      "learning_rate": 0.00016076682414845468,
      "loss": 4.6607,
      "step": 28930
    },
    {
      "epoch": 0.5925296418405587,
      "grad_norm": 0.27063611149787903,
      "learning_rate": 0.00016070508690395978,
      "loss": 4.6405,
      "step": 28935
    },
    {
      "epoch": 0.5926320316179633,
      "grad_norm": 0.28568169474601746,
      "learning_rate": 0.00016064334783669158,
      "loss": 4.7347,
      "step": 28940
    },
    {
      "epoch": 0.5927344213953679,
      "grad_norm": 0.27155593037605286,
      "learning_rate": 0.00016058160695716256,
      "loss": 4.6198,
      "step": 28945
    },
    {
      "epoch": 0.5928368111727725,
      "grad_norm": 0.2821517884731293,
      "learning_rate": 0.0001605198642758855,
      "loss": 4.7014,
      "step": 28950
    },
    {
      "epoch": 0.5929392009501772,
      "grad_norm": 0.27491268515586853,
      "learning_rate": 0.00016045811980337335,
      "loss": 4.691,
      "step": 28955
    },
    {
      "epoch": 0.5930415907275818,
      "grad_norm": 0.27691519260406494,
      "learning_rate": 0.00016039637355013946,
      "loss": 4.6568,
      "step": 28960
    },
    {
      "epoch": 0.5931439805049864,
      "grad_norm": 0.25760793685913086,
      "learning_rate": 0.0001603346255266975,
      "loss": 4.6677,
      "step": 28965
    },
    {
      "epoch": 0.593246370282391,
      "grad_norm": 0.297439306974411,
      "learning_rate": 0.00016027287574356148,
      "loss": 4.6604,
      "step": 28970
    },
    {
      "epoch": 0.5933487600597956,
      "grad_norm": 0.27333223819732666,
      "learning_rate": 0.00016021112421124546,
      "loss": 4.6476,
      "step": 28975
    },
    {
      "epoch": 0.5934511498372003,
      "grad_norm": 0.28570982813835144,
      "learning_rate": 0.00016014937094026407,
      "loss": 4.6237,
      "step": 28980
    },
    {
      "epoch": 0.5935535396146049,
      "grad_norm": 0.25136297941207886,
      "learning_rate": 0.00016008761594113222,
      "loss": 4.6771,
      "step": 28985
    },
    {
      "epoch": 0.5936559293920095,
      "grad_norm": 0.26670604944229126,
      "learning_rate": 0.00016002585922436492,
      "loss": 4.7002,
      "step": 28990
    },
    {
      "epoch": 0.5937583191694141,
      "grad_norm": 0.26492464542388916,
      "learning_rate": 0.00015996410080047765,
      "loss": 4.685,
      "step": 28995
    },
    {
      "epoch": 0.5938607089468187,
      "grad_norm": 0.26494765281677246,
      "learning_rate": 0.0001599023406799861,
      "loss": 4.6326,
      "step": 29000
    },
    {
      "epoch": 0.5938607089468187,
      "eval_loss": 4.6867289543151855,
      "eval_runtime": 17.1353,
      "eval_samples_per_second": 33.965,
      "eval_steps_per_second": 4.26,
      "step": 29000
    },
    {
      "epoch": 0.5939630987242234,
      "grad_norm": 0.28002622723579407,
      "learning_rate": 0.00015984057887340626,
      "loss": 4.6923,
      "step": 29005
    },
    {
      "epoch": 0.594065488501628,
      "grad_norm": 0.2913799285888672,
      "learning_rate": 0.00015977881539125448,
      "loss": 4.6041,
      "step": 29010
    },
    {
      "epoch": 0.5941678782790326,
      "grad_norm": 0.27965831756591797,
      "learning_rate": 0.00015971705024404726,
      "loss": 4.6498,
      "step": 29015
    },
    {
      "epoch": 0.5942702680564372,
      "grad_norm": 0.27786508202552795,
      "learning_rate": 0.00015965528344230148,
      "loss": 4.6857,
      "step": 29020
    },
    {
      "epoch": 0.5943726578338419,
      "grad_norm": 0.26868125796318054,
      "learning_rate": 0.0001595935149965343,
      "loss": 4.7213,
      "step": 29025
    },
    {
      "epoch": 0.5944750476112465,
      "grad_norm": 0.28863799571990967,
      "learning_rate": 0.00015953174491726307,
      "loss": 4.7019,
      "step": 29030
    },
    {
      "epoch": 0.5945774373886511,
      "grad_norm": 0.3015957176685333,
      "learning_rate": 0.00015946997321500563,
      "loss": 4.6739,
      "step": 29035
    },
    {
      "epoch": 0.5946798271660557,
      "grad_norm": 0.2558874189853668,
      "learning_rate": 0.0001594081999002798,
      "loss": 4.6626,
      "step": 29040
    },
    {
      "epoch": 0.5947822169434603,
      "grad_norm": 0.26330509781837463,
      "learning_rate": 0.00015934642498360388,
      "loss": 4.7061,
      "step": 29045
    },
    {
      "epoch": 0.594884606720865,
      "grad_norm": 0.2671198546886444,
      "learning_rate": 0.00015928464847549635,
      "loss": 4.627,
      "step": 29050
    },
    {
      "epoch": 0.5949869964982696,
      "grad_norm": 0.27151599526405334,
      "learning_rate": 0.00015922287038647615,
      "loss": 4.6833,
      "step": 29055
    },
    {
      "epoch": 0.5950893862756742,
      "grad_norm": 0.24983324110507965,
      "learning_rate": 0.00015916109072706215,
      "loss": 4.6917,
      "step": 29060
    },
    {
      "epoch": 0.5951917760530788,
      "grad_norm": 0.26332369446754456,
      "learning_rate": 0.0001590993095077738,
      "loss": 4.6526,
      "step": 29065
    },
    {
      "epoch": 0.5952941658304834,
      "grad_norm": 0.26414334774017334,
      "learning_rate": 0.00015903752673913061,
      "loss": 4.7179,
      "step": 29070
    },
    {
      "epoch": 0.5953965556078881,
      "grad_norm": 0.2591826021671295,
      "learning_rate": 0.0001589757424316525,
      "loss": 4.6453,
      "step": 29075
    },
    {
      "epoch": 0.5954989453852927,
      "grad_norm": 0.26921385526657104,
      "learning_rate": 0.00015891395659585963,
      "loss": 4.6801,
      "step": 29080
    },
    {
      "epoch": 0.5956013351626973,
      "grad_norm": 0.2525412142276764,
      "learning_rate": 0.00015885216924227223,
      "loss": 4.7421,
      "step": 29085
    },
    {
      "epoch": 0.5957037249401019,
      "grad_norm": 0.2510041892528534,
      "learning_rate": 0.00015879038038141106,
      "loss": 4.7037,
      "step": 29090
    },
    {
      "epoch": 0.5958061147175066,
      "grad_norm": 0.27487054467201233,
      "learning_rate": 0.00015872859002379695,
      "loss": 4.7657,
      "step": 29095
    },
    {
      "epoch": 0.5959085044949112,
      "grad_norm": 0.25881633162498474,
      "learning_rate": 0.00015866679817995108,
      "loss": 4.6379,
      "step": 29100
    },
    {
      "epoch": 0.5960108942723159,
      "grad_norm": 0.27736732363700867,
      "learning_rate": 0.00015860500486039489,
      "loss": 4.6497,
      "step": 29105
    },
    {
      "epoch": 0.5961132840497205,
      "grad_norm": 0.2656888961791992,
      "learning_rate": 0.00015854321007564997,
      "loss": 4.6758,
      "step": 29110
    },
    {
      "epoch": 0.5962156738271251,
      "grad_norm": 0.25772184133529663,
      "learning_rate": 0.00015848141383623823,
      "loss": 4.6871,
      "step": 29115
    },
    {
      "epoch": 0.5963180636045298,
      "grad_norm": 1.7596418857574463,
      "learning_rate": 0.00015841961615268183,
      "loss": 4.6907,
      "step": 29120
    },
    {
      "epoch": 0.5964204533819344,
      "grad_norm": 0.26713013648986816,
      "learning_rate": 0.0001583578170355032,
      "loss": 4.6423,
      "step": 29125
    },
    {
      "epoch": 0.596522843159339,
      "grad_norm": 0.2562316358089447,
      "learning_rate": 0.0001582960164952249,
      "loss": 4.6231,
      "step": 29130
    },
    {
      "epoch": 0.5966252329367436,
      "grad_norm": 0.2903379201889038,
      "learning_rate": 0.00015823421454236992,
      "loss": 4.6779,
      "step": 29135
    },
    {
      "epoch": 0.5967276227141483,
      "grad_norm": 0.2811809480190277,
      "learning_rate": 0.00015817241118746124,
      "loss": 4.6525,
      "step": 29140
    },
    {
      "epoch": 0.5968300124915529,
      "grad_norm": 0.29069632291793823,
      "learning_rate": 0.00015811060644102233,
      "loss": 4.6847,
      "step": 29145
    },
    {
      "epoch": 0.5969324022689575,
      "grad_norm": 0.25295203924179077,
      "learning_rate": 0.0001580488003135768,
      "loss": 4.6821,
      "step": 29150
    },
    {
      "epoch": 0.5970347920463621,
      "grad_norm": 0.2610161602497101,
      "learning_rate": 0.00015798699281564845,
      "loss": 4.6196,
      "step": 29155
    },
    {
      "epoch": 0.5971371818237667,
      "grad_norm": 0.26808518171310425,
      "learning_rate": 0.00015792518395776133,
      "loss": 4.6069,
      "step": 29160
    },
    {
      "epoch": 0.5972395716011714,
      "grad_norm": 0.260791152715683,
      "learning_rate": 0.00015786337375043973,
      "loss": 4.7065,
      "step": 29165
    },
    {
      "epoch": 0.597341961378576,
      "grad_norm": 0.26363709568977356,
      "learning_rate": 0.00015780156220420818,
      "loss": 4.6931,
      "step": 29170
    },
    {
      "epoch": 0.5974443511559806,
      "grad_norm": 0.26608097553253174,
      "learning_rate": 0.00015773974932959154,
      "loss": 4.6345,
      "step": 29175
    },
    {
      "epoch": 0.5975467409333852,
      "grad_norm": 0.2536289095878601,
      "learning_rate": 0.0001576779351371147,
      "loss": 4.693,
      "step": 29180
    },
    {
      "epoch": 0.5976491307107898,
      "grad_norm": 0.24805046617984772,
      "learning_rate": 0.00015761611963730282,
      "loss": 4.6448,
      "step": 29185
    },
    {
      "epoch": 0.5977515204881945,
      "grad_norm": 0.26312559843063354,
      "learning_rate": 0.00015755430284068137,
      "loss": 4.6967,
      "step": 29190
    },
    {
      "epoch": 0.5978539102655991,
      "grad_norm": 0.2719932794570923,
      "learning_rate": 0.0001574924847577761,
      "loss": 4.7166,
      "step": 29195
    },
    {
      "epoch": 0.5979563000430037,
      "grad_norm": 0.39098620414733887,
      "learning_rate": 0.00015743066539911275,
      "loss": 4.6323,
      "step": 29200
    },
    {
      "epoch": 0.5980586898204083,
      "grad_norm": 0.2763932943344116,
      "learning_rate": 0.00015736884477521749,
      "loss": 4.6671,
      "step": 29205
    },
    {
      "epoch": 0.598161079597813,
      "grad_norm": 0.27419716119766235,
      "learning_rate": 0.0001573070228966165,
      "loss": 4.6826,
      "step": 29210
    },
    {
      "epoch": 0.5982634693752176,
      "grad_norm": 0.26688000559806824,
      "learning_rate": 0.00015724519977383642,
      "loss": 4.6959,
      "step": 29215
    },
    {
      "epoch": 0.5983658591526222,
      "grad_norm": 0.2649906575679779,
      "learning_rate": 0.00015718337541740398,
      "loss": 4.6427,
      "step": 29220
    },
    {
      "epoch": 0.5984682489300268,
      "grad_norm": 0.25840264558792114,
      "learning_rate": 0.00015712154983784602,
      "loss": 4.6831,
      "step": 29225
    },
    {
      "epoch": 0.5985706387074314,
      "grad_norm": 0.2627979815006256,
      "learning_rate": 0.00015705972304568978,
      "loss": 4.6888,
      "step": 29230
    },
    {
      "epoch": 0.5986730284848361,
      "grad_norm": 0.2715907692909241,
      "learning_rate": 0.00015699789505146258,
      "loss": 4.6857,
      "step": 29235
    },
    {
      "epoch": 0.5987754182622407,
      "grad_norm": 0.28489550948143005,
      "learning_rate": 0.00015693606586569193,
      "loss": 4.6919,
      "step": 29240
    },
    {
      "epoch": 0.5988778080396453,
      "grad_norm": 0.2671533524990082,
      "learning_rate": 0.00015687423549890566,
      "loss": 4.6655,
      "step": 29245
    },
    {
      "epoch": 0.5989801978170499,
      "grad_norm": 0.26588207483291626,
      "learning_rate": 0.00015681240396163174,
      "loss": 4.6856,
      "step": 29250
    },
    {
      "epoch": 0.5990825875944545,
      "grad_norm": 0.3119635581970215,
      "learning_rate": 0.00015675057126439824,
      "loss": 4.6705,
      "step": 29255
    },
    {
      "epoch": 0.5991849773718592,
      "grad_norm": 0.25209978222846985,
      "learning_rate": 0.00015668873741773355,
      "loss": 4.6643,
      "step": 29260
    },
    {
      "epoch": 0.5992873671492638,
      "grad_norm": 0.25916942954063416,
      "learning_rate": 0.00015662690243216628,
      "loss": 4.6447,
      "step": 29265
    },
    {
      "epoch": 0.5993897569266684,
      "grad_norm": 0.26691484451293945,
      "learning_rate": 0.00015656506631822514,
      "loss": 4.6826,
      "step": 29270
    },
    {
      "epoch": 0.599492146704073,
      "grad_norm": 0.35600513219833374,
      "learning_rate": 0.00015650322908643908,
      "loss": 4.6915,
      "step": 29275
    },
    {
      "epoch": 0.5995945364814776,
      "grad_norm": 0.2651926875114441,
      "learning_rate": 0.00015644139074733716,
      "loss": 4.6604,
      "step": 29280
    },
    {
      "epoch": 0.5996969262588823,
      "grad_norm": 0.2603910565376282,
      "learning_rate": 0.0001563795513114488,
      "loss": 4.6681,
      "step": 29285
    },
    {
      "epoch": 0.5997993160362869,
      "grad_norm": 0.26804107427597046,
      "learning_rate": 0.00015631771078930347,
      "loss": 4.675,
      "step": 29290
    },
    {
      "epoch": 0.5999017058136915,
      "grad_norm": 0.2554345428943634,
      "learning_rate": 0.00015625586919143085,
      "loss": 4.6594,
      "step": 29295
    },
    {
      "epoch": 0.6000040955910961,
      "grad_norm": 0.2568386197090149,
      "learning_rate": 0.00015619402652836085,
      "loss": 4.6619,
      "step": 29300
    },
    {
      "epoch": 0.6001064853685009,
      "grad_norm": 0.28163617849349976,
      "learning_rate": 0.00015613218281062343,
      "loss": 4.637,
      "step": 29305
    },
    {
      "epoch": 0.6002088751459055,
      "grad_norm": 0.2598473131656647,
      "learning_rate": 0.00015607033804874892,
      "loss": 4.7283,
      "step": 29310
    },
    {
      "epoch": 0.6003112649233101,
      "grad_norm": 0.2979266345500946,
      "learning_rate": 0.0001560084922532677,
      "loss": 4.6401,
      "step": 29315
    },
    {
      "epoch": 0.6004136547007147,
      "grad_norm": 0.251546174287796,
      "learning_rate": 0.00015594664543471039,
      "loss": 4.6493,
      "step": 29320
    },
    {
      "epoch": 0.6005160444781193,
      "grad_norm": 0.26767000555992126,
      "learning_rate": 0.00015588479760360766,
      "loss": 4.6874,
      "step": 29325
    },
    {
      "epoch": 0.600618434255524,
      "grad_norm": 0.26352837681770325,
      "learning_rate": 0.0001558229487704905,
      "loss": 4.6588,
      "step": 29330
    },
    {
      "epoch": 0.6007208240329286,
      "grad_norm": 0.319023996591568,
      "learning_rate": 0.00015576109894589008,
      "loss": 4.7051,
      "step": 29335
    },
    {
      "epoch": 0.6008232138103332,
      "grad_norm": 0.24961285293102264,
      "learning_rate": 0.00015569924814033752,
      "loss": 4.6927,
      "step": 29340
    },
    {
      "epoch": 0.6009256035877378,
      "grad_norm": 0.26609739661216736,
      "learning_rate": 0.00015563739636436445,
      "loss": 4.6607,
      "step": 29345
    },
    {
      "epoch": 0.6010279933651425,
      "grad_norm": 0.2591767907142639,
      "learning_rate": 0.00015557554362850232,
      "loss": 4.6741,
      "step": 29350
    },
    {
      "epoch": 0.6011303831425471,
      "grad_norm": 0.2783168852329254,
      "learning_rate": 0.00015551368994328298,
      "loss": 4.6915,
      "step": 29355
    },
    {
      "epoch": 0.6012327729199517,
      "grad_norm": 0.30141308903694153,
      "learning_rate": 0.00015545183531923834,
      "loss": 4.6651,
      "step": 29360
    },
    {
      "epoch": 0.6013351626973563,
      "grad_norm": 0.2937859296798706,
      "learning_rate": 0.0001553899797669005,
      "loss": 4.6587,
      "step": 29365
    },
    {
      "epoch": 0.6014375524747609,
      "grad_norm": 0.2556871175765991,
      "learning_rate": 0.00015532812329680168,
      "loss": 4.6832,
      "step": 29370
    },
    {
      "epoch": 0.6015399422521656,
      "grad_norm": 0.2573263943195343,
      "learning_rate": 0.0001552662659194743,
      "loss": 4.6928,
      "step": 29375
    },
    {
      "epoch": 0.6016423320295702,
      "grad_norm": 0.249510258436203,
      "learning_rate": 0.00015520440764545096,
      "loss": 4.604,
      "step": 29380
    },
    {
      "epoch": 0.6017447218069748,
      "grad_norm": 0.27031224966049194,
      "learning_rate": 0.00015514254848526433,
      "loss": 4.667,
      "step": 29385
    },
    {
      "epoch": 0.6018471115843794,
      "grad_norm": 0.2602192461490631,
      "learning_rate": 0.00015508068844944734,
      "loss": 4.6744,
      "step": 29390
    },
    {
      "epoch": 0.601949501361784,
      "grad_norm": 0.2823655605316162,
      "learning_rate": 0.00015501882754853286,
      "loss": 4.6708,
      "step": 29395
    },
    {
      "epoch": 0.6020518911391887,
      "grad_norm": 0.2946068048477173,
      "learning_rate": 0.00015495696579305413,
      "loss": 4.6657,
      "step": 29400
    },
    {
      "epoch": 0.6021542809165933,
      "grad_norm": 0.2830444574356079,
      "learning_rate": 0.00015489510319354456,
      "loss": 4.659,
      "step": 29405
    },
    {
      "epoch": 0.6022566706939979,
      "grad_norm": 0.27121251821517944,
      "learning_rate": 0.00015483323976053743,
      "loss": 4.6996,
      "step": 29410
    },
    {
      "epoch": 0.6023590604714025,
      "grad_norm": 0.2876390218734741,
      "learning_rate": 0.00015477137550456643,
      "loss": 4.6174,
      "step": 29415
    },
    {
      "epoch": 0.6024614502488072,
      "grad_norm": 0.2655010223388672,
      "learning_rate": 0.00015470951043616523,
      "loss": 4.6378,
      "step": 29420
    },
    {
      "epoch": 0.6025638400262118,
      "grad_norm": 0.2648998200893402,
      "learning_rate": 0.00015464764456586775,
      "loss": 4.6921,
      "step": 29425
    },
    {
      "epoch": 0.6026662298036164,
      "grad_norm": 0.26296836137771606,
      "learning_rate": 0.000154585777904208,
      "loss": 4.682,
      "step": 29430
    },
    {
      "epoch": 0.602768619581021,
      "grad_norm": 0.2766362428665161,
      "learning_rate": 0.0001545239104617201,
      "loss": 4.6228,
      "step": 29435
    },
    {
      "epoch": 0.6028710093584256,
      "grad_norm": 0.2620740532875061,
      "learning_rate": 0.00015446204224893833,
      "loss": 4.6544,
      "step": 29440
    },
    {
      "epoch": 0.6029733991358303,
      "grad_norm": 0.264326810836792,
      "learning_rate": 0.00015440017327639704,
      "loss": 4.7159,
      "step": 29445
    },
    {
      "epoch": 0.6030757889132349,
      "grad_norm": 0.2602043151855469,
      "learning_rate": 0.00015433830355463083,
      "loss": 4.672,
      "step": 29450
    },
    {
      "epoch": 0.6031781786906395,
      "grad_norm": 0.26989299058914185,
      "learning_rate": 0.00015427643309417439,
      "loss": 4.679,
      "step": 29455
    },
    {
      "epoch": 0.6032805684680441,
      "grad_norm": 0.2631540298461914,
      "learning_rate": 0.00015421456190556246,
      "loss": 4.6572,
      "step": 29460
    },
    {
      "epoch": 0.6033829582454487,
      "grad_norm": 0.2787879407405853,
      "learning_rate": 0.00015415268999932993,
      "loss": 4.6719,
      "step": 29465
    },
    {
      "epoch": 0.6034853480228534,
      "grad_norm": 0.261043906211853,
      "learning_rate": 0.00015409081738601183,
      "loss": 4.6791,
      "step": 29470
    },
    {
      "epoch": 0.603587737800258,
      "grad_norm": 0.2684559226036072,
      "learning_rate": 0.00015402894407614342,
      "loss": 4.6567,
      "step": 29475
    },
    {
      "epoch": 0.6036901275776626,
      "grad_norm": 0.28989318013191223,
      "learning_rate": 0.00015396707008025982,
      "loss": 4.7154,
      "step": 29480
    },
    {
      "epoch": 0.6037925173550672,
      "grad_norm": 0.26652899384498596,
      "learning_rate": 0.00015390519540889658,
      "loss": 4.6232,
      "step": 29485
    },
    {
      "epoch": 0.6038949071324718,
      "grad_norm": 0.269520103931427,
      "learning_rate": 0.00015384332007258904,
      "loss": 4.6293,
      "step": 29490
    },
    {
      "epoch": 0.6039972969098765,
      "grad_norm": 0.29781582951545715,
      "learning_rate": 0.00015378144408187294,
      "loss": 4.6637,
      "step": 29495
    },
    {
      "epoch": 0.6040996866872811,
      "grad_norm": 0.2606029212474823,
      "learning_rate": 0.000153719567447284,
      "loss": 4.7466,
      "step": 29500
    },
    {
      "epoch": 0.6042020764646857,
      "grad_norm": 0.27624937891960144,
      "learning_rate": 0.000153657690179358,
      "loss": 4.5986,
      "step": 29505
    },
    {
      "epoch": 0.6043044662420904,
      "grad_norm": 0.28331923484802246,
      "learning_rate": 0.0001535958122886309,
      "loss": 4.6918,
      "step": 29510
    },
    {
      "epoch": 0.6044068560194951,
      "grad_norm": 0.26128119230270386,
      "learning_rate": 0.0001535339337856388,
      "loss": 4.7142,
      "step": 29515
    },
    {
      "epoch": 0.6045092457968997,
      "grad_norm": 0.3133602440357208,
      "learning_rate": 0.0001534720546809178,
      "loss": 4.6729,
      "step": 29520
    },
    {
      "epoch": 0.6046116355743043,
      "grad_norm": 0.265758752822876,
      "learning_rate": 0.0001534101749850042,
      "loss": 4.708,
      "step": 29525
    },
    {
      "epoch": 0.6047140253517089,
      "grad_norm": 0.25712907314300537,
      "learning_rate": 0.00015334829470843437,
      "loss": 4.674,
      "step": 29530
    },
    {
      "epoch": 0.6048164151291135,
      "grad_norm": 0.27626919746398926,
      "learning_rate": 0.00015328641386174466,
      "loss": 4.6511,
      "step": 29535
    },
    {
      "epoch": 0.6049188049065182,
      "grad_norm": 0.30588817596435547,
      "learning_rate": 0.00015322453245547176,
      "loss": 4.6458,
      "step": 29540
    },
    {
      "epoch": 0.6050211946839228,
      "grad_norm": 0.25631025433540344,
      "learning_rate": 0.00015316265050015234,
      "loss": 4.6855,
      "step": 29545
    },
    {
      "epoch": 0.6051235844613274,
      "grad_norm": 0.26386743783950806,
      "learning_rate": 0.000153100768006323,
      "loss": 4.6843,
      "step": 29550
    },
    {
      "epoch": 0.605225974238732,
      "grad_norm": 0.310792475938797,
      "learning_rate": 0.00015303888498452068,
      "loss": 4.6306,
      "step": 29555
    },
    {
      "epoch": 0.6053283640161367,
      "grad_norm": 0.2991289496421814,
      "learning_rate": 0.00015297700144528226,
      "loss": 4.6873,
      "step": 29560
    },
    {
      "epoch": 0.6054307537935413,
      "grad_norm": 0.2646171450614929,
      "learning_rate": 0.00015291511739914476,
      "loss": 4.6461,
      "step": 29565
    },
    {
      "epoch": 0.6055331435709459,
      "grad_norm": 0.2500417232513428,
      "learning_rate": 0.00015285323285664536,
      "loss": 4.6622,
      "step": 29570
    },
    {
      "epoch": 0.6056355333483505,
      "grad_norm": 0.26394060254096985,
      "learning_rate": 0.00015279134782832113,
      "loss": 4.6331,
      "step": 29575
    },
    {
      "epoch": 0.6057379231257551,
      "grad_norm": 0.25219351053237915,
      "learning_rate": 0.00015272946232470943,
      "loss": 4.6639,
      "step": 29580
    },
    {
      "epoch": 0.6058403129031598,
      "grad_norm": 0.2596396803855896,
      "learning_rate": 0.00015266757635634752,
      "loss": 4.6249,
      "step": 29585
    },
    {
      "epoch": 0.6059427026805644,
      "grad_norm": 0.33561503887176514,
      "learning_rate": 0.00015260568993377292,
      "loss": 4.6586,
      "step": 29590
    },
    {
      "epoch": 0.606045092457969,
      "grad_norm": 0.2690029740333557,
      "learning_rate": 0.00015254380306752312,
      "loss": 4.6338,
      "step": 29595
    },
    {
      "epoch": 0.6061474822353736,
      "grad_norm": 0.27321794629096985,
      "learning_rate": 0.00015248191576813563,
      "loss": 4.6541,
      "step": 29600
    },
    {
      "epoch": 0.6062498720127782,
      "grad_norm": 0.26653915643692017,
      "learning_rate": 0.00015242002804614815,
      "loss": 4.6559,
      "step": 29605
    },
    {
      "epoch": 0.6063522617901829,
      "grad_norm": 0.44555196166038513,
      "learning_rate": 0.0001523581399120984,
      "loss": 4.5864,
      "step": 29610
    },
    {
      "epoch": 0.6064546515675875,
      "grad_norm": 0.26160576939582825,
      "learning_rate": 0.00015229625137652423,
      "loss": 4.688,
      "step": 29615
    },
    {
      "epoch": 0.6065570413449921,
      "grad_norm": 0.26186686754226685,
      "learning_rate": 0.00015223436244996341,
      "loss": 4.6661,
      "step": 29620
    },
    {
      "epoch": 0.6066594311223967,
      "grad_norm": 0.2802843749523163,
      "learning_rate": 0.00015217247314295397,
      "loss": 4.6789,
      "step": 29625
    },
    {
      "epoch": 0.6067618208998014,
      "grad_norm": 0.6387742161750793,
      "learning_rate": 0.00015211058346603382,
      "loss": 4.784,
      "step": 29630
    },
    {
      "epoch": 0.606864210677206,
      "grad_norm": 0.2868768274784088,
      "learning_rate": 0.00015204869342974104,
      "loss": 4.6126,
      "step": 29635
    },
    {
      "epoch": 0.6069666004546106,
      "grad_norm": 0.2657756805419922,
      "learning_rate": 0.00015198680304461386,
      "loss": 4.6796,
      "step": 29640
    },
    {
      "epoch": 0.6070689902320152,
      "grad_norm": 0.25940439105033875,
      "learning_rate": 0.00015192491232119025,
      "loss": 4.6374,
      "step": 29645
    },
    {
      "epoch": 0.6071713800094198,
      "grad_norm": 0.2724342346191406,
      "learning_rate": 0.0001518630212700087,
      "loss": 4.6236,
      "step": 29650
    },
    {
      "epoch": 0.6072737697868245,
      "grad_norm": 0.26338621973991394,
      "learning_rate": 0.00015180112990160726,
      "loss": 4.6229,
      "step": 29655
    },
    {
      "epoch": 0.6073761595642291,
      "grad_norm": 0.2676234543323517,
      "learning_rate": 0.00015173923822652444,
      "loss": 4.6505,
      "step": 29660
    },
    {
      "epoch": 0.6074785493416337,
      "grad_norm": 0.2560276389122009,
      "learning_rate": 0.0001516773462552986,
      "loss": 4.6728,
      "step": 29665
    },
    {
      "epoch": 0.6075809391190383,
      "grad_norm": 0.25123026967048645,
      "learning_rate": 0.00015161545399846815,
      "loss": 4.6781,
      "step": 29670
    },
    {
      "epoch": 0.6076833288964429,
      "grad_norm": 0.2650813162326813,
      "learning_rate": 0.00015155356146657163,
      "loss": 4.6428,
      "step": 29675
    },
    {
      "epoch": 0.6077857186738476,
      "grad_norm": 0.2570786476135254,
      "learning_rate": 0.00015149166867014754,
      "loss": 4.654,
      "step": 29680
    },
    {
      "epoch": 0.6078881084512522,
      "grad_norm": 0.26854678988456726,
      "learning_rate": 0.0001514297756197346,
      "loss": 4.6582,
      "step": 29685
    },
    {
      "epoch": 0.6079904982286568,
      "grad_norm": 0.25913646817207336,
      "learning_rate": 0.0001513678823258713,
      "loss": 4.6373,
      "step": 29690
    },
    {
      "epoch": 0.6080928880060614,
      "grad_norm": 0.2833520174026489,
      "learning_rate": 0.00015130598879909636,
      "loss": 4.6458,
      "step": 29695
    },
    {
      "epoch": 0.608195277783466,
      "grad_norm": 0.27404841780662537,
      "learning_rate": 0.00015124409504994847,
      "loss": 4.696,
      "step": 29700
    },
    {
      "epoch": 0.6082976675608707,
      "grad_norm": 0.27438125014305115,
      "learning_rate": 0.00015118220108896645,
      "loss": 4.698,
      "step": 29705
    },
    {
      "epoch": 0.6084000573382754,
      "grad_norm": 0.2868211269378662,
      "learning_rate": 0.00015112030692668906,
      "loss": 4.6449,
      "step": 29710
    },
    {
      "epoch": 0.60850244711568,
      "grad_norm": 0.2659040093421936,
      "learning_rate": 0.0001510584125736551,
      "loss": 4.674,
      "step": 29715
    },
    {
      "epoch": 0.6086048368930846,
      "grad_norm": 0.27274179458618164,
      "learning_rate": 0.00015099651804040346,
      "loss": 4.7177,
      "step": 29720
    },
    {
      "epoch": 0.6087072266704893,
      "grad_norm": 0.24936515092849731,
      "learning_rate": 0.00015093462333747298,
      "loss": 4.6933,
      "step": 29725
    },
    {
      "epoch": 0.6088096164478939,
      "grad_norm": 0.2781531810760498,
      "learning_rate": 0.00015087272847540261,
      "loss": 4.6862,
      "step": 29730
    },
    {
      "epoch": 0.6089120062252985,
      "grad_norm": 0.2787362039089203,
      "learning_rate": 0.00015081083346473134,
      "loss": 4.6542,
      "step": 29735
    },
    {
      "epoch": 0.6090143960027031,
      "grad_norm": 0.2624753713607788,
      "learning_rate": 0.00015074893831599806,
      "loss": 4.6621,
      "step": 29740
    },
    {
      "epoch": 0.6091167857801078,
      "grad_norm": 0.26399970054626465,
      "learning_rate": 0.00015068704303974176,
      "loss": 4.6525,
      "step": 29745
    },
    {
      "epoch": 0.6092191755575124,
      "grad_norm": 0.2662067413330078,
      "learning_rate": 0.0001506251476465015,
      "loss": 4.6695,
      "step": 29750
    },
    {
      "epoch": 0.609321565334917,
      "grad_norm": 0.27489542961120605,
      "learning_rate": 0.0001505632521468163,
      "loss": 4.6658,
      "step": 29755
    },
    {
      "epoch": 0.6094239551123216,
      "grad_norm": 0.2619228661060333,
      "learning_rate": 0.0001505013565512252,
      "loss": 4.6896,
      "step": 29760
    },
    {
      "epoch": 0.6095263448897262,
      "grad_norm": 0.3220798671245575,
      "learning_rate": 0.00015043946087026728,
      "loss": 4.6468,
      "step": 29765
    },
    {
      "epoch": 0.6096287346671309,
      "grad_norm": 0.26916709542274475,
      "learning_rate": 0.0001503775651144816,
      "loss": 4.7052,
      "step": 29770
    },
    {
      "epoch": 0.6097311244445355,
      "grad_norm": 0.2660595178604126,
      "learning_rate": 0.0001503156692944073,
      "loss": 4.6184,
      "step": 29775
    },
    {
      "epoch": 0.6098335142219401,
      "grad_norm": 0.26139718294143677,
      "learning_rate": 0.00015025377342058342,
      "loss": 4.6833,
      "step": 29780
    },
    {
      "epoch": 0.6099359039993447,
      "grad_norm": 0.25445857644081116,
      "learning_rate": 0.00015019187750354914,
      "loss": 4.6651,
      "step": 29785
    },
    {
      "epoch": 0.6100382937767493,
      "grad_norm": 0.24757958948612213,
      "learning_rate": 0.00015012998155384353,
      "loss": 4.6792,
      "step": 29790
    },
    {
      "epoch": 0.610140683554154,
      "grad_norm": 0.24935737252235413,
      "learning_rate": 0.00015006808558200576,
      "loss": 4.6747,
      "step": 29795
    },
    {
      "epoch": 0.6102430733315586,
      "grad_norm": 0.2751321792602539,
      "learning_rate": 0.00015000618959857493,
      "loss": 4.6873,
      "step": 29800
    },
    {
      "epoch": 0.6103454631089632,
      "grad_norm": 0.2626931667327881,
      "learning_rate": 0.0001499442936140902,
      "loss": 4.7037,
      "step": 29805
    },
    {
      "epoch": 0.6104478528863678,
      "grad_norm": 0.26975181698799133,
      "learning_rate": 0.00014988239763909068,
      "loss": 4.621,
      "step": 29810
    },
    {
      "epoch": 0.6105502426637724,
      "grad_norm": 0.25670450925827026,
      "learning_rate": 0.00014982050168411555,
      "loss": 4.5853,
      "step": 29815
    },
    {
      "epoch": 0.6106526324411771,
      "grad_norm": 0.2627800703048706,
      "learning_rate": 0.00014975860575970384,
      "loss": 4.6774,
      "step": 29820
    },
    {
      "epoch": 0.6107550222185817,
      "grad_norm": 0.2657567858695984,
      "learning_rate": 0.00014969670987639473,
      "loss": 4.6183,
      "step": 29825
    },
    {
      "epoch": 0.6108574119959863,
      "grad_norm": 0.25893691182136536,
      "learning_rate": 0.00014963481404472743,
      "loss": 4.6916,
      "step": 29830
    },
    {
      "epoch": 0.6109598017733909,
      "grad_norm": 0.26803627610206604,
      "learning_rate": 0.0001495729182752409,
      "loss": 4.6461,
      "step": 29835
    },
    {
      "epoch": 0.6110621915507956,
      "grad_norm": 0.26486125588417053,
      "learning_rate": 0.00014951102257847426,
      "loss": 4.6816,
      "step": 29840
    },
    {
      "epoch": 0.6111645813282002,
      "grad_norm": 0.2670956552028656,
      "learning_rate": 0.0001494491269649667,
      "loss": 4.5994,
      "step": 29845
    },
    {
      "epoch": 0.6112669711056048,
      "grad_norm": 0.2605704665184021,
      "learning_rate": 0.00014938723144525718,
      "loss": 4.6656,
      "step": 29850
    },
    {
      "epoch": 0.6113693608830094,
      "grad_norm": 0.2715194821357727,
      "learning_rate": 0.00014932533602988478,
      "loss": 4.6504,
      "step": 29855
    },
    {
      "epoch": 0.611471750660414,
      "grad_norm": 0.245295450091362,
      "learning_rate": 0.00014926344072938865,
      "loss": 4.6806,
      "step": 29860
    },
    {
      "epoch": 0.6115741404378187,
      "grad_norm": 0.6133278012275696,
      "learning_rate": 0.0001492015455543076,
      "loss": 4.6174,
      "step": 29865
    },
    {
      "epoch": 0.6116765302152233,
      "grad_norm": 0.2752784192562103,
      "learning_rate": 0.00014913965051518077,
      "loss": 4.6169,
      "step": 29870
    },
    {
      "epoch": 0.6117789199926279,
      "grad_norm": 0.2706621587276459,
      "learning_rate": 0.00014907775562254707,
      "loss": 4.6717,
      "step": 29875
    },
    {
      "epoch": 0.6118813097700325,
      "grad_norm": 0.3045152425765991,
      "learning_rate": 0.0001490158608869456,
      "loss": 4.6855,
      "step": 29880
    },
    {
      "epoch": 0.6119836995474371,
      "grad_norm": 0.7316267490386963,
      "learning_rate": 0.00014895396631891506,
      "loss": 4.6194,
      "step": 29885
    },
    {
      "epoch": 0.6120860893248418,
      "grad_norm": 0.29746928811073303,
      "learning_rate": 0.00014889207192899444,
      "loss": 4.6297,
      "step": 29890
    },
    {
      "epoch": 0.6121884791022464,
      "grad_norm": 0.26595771312713623,
      "learning_rate": 0.00014883017772772264,
      "loss": 4.6982,
      "step": 29895
    },
    {
      "epoch": 0.612290868879651,
      "grad_norm": 0.27452102303504944,
      "learning_rate": 0.00014876828372563844,
      "loss": 4.6291,
      "step": 29900
    },
    {
      "epoch": 0.6123932586570556,
      "grad_norm": 0.28861257433891296,
      "learning_rate": 0.00014870638993328062,
      "loss": 4.6327,
      "step": 29905
    },
    {
      "epoch": 0.6124956484344602,
      "grad_norm": 0.27628979086875916,
      "learning_rate": 0.00014864449636118804,
      "loss": 4.6293,
      "step": 29910
    },
    {
      "epoch": 0.612598038211865,
      "grad_norm": 0.28083112835884094,
      "learning_rate": 0.00014858260301989928,
      "loss": 4.5978,
      "step": 29915
    },
    {
      "epoch": 0.6127004279892696,
      "grad_norm": 0.2867166996002197,
      "learning_rate": 0.0001485207099199531,
      "loss": 4.6514,
      "step": 29920
    },
    {
      "epoch": 0.6128028177666742,
      "grad_norm": 0.27758774161338806,
      "learning_rate": 0.00014845881707188822,
      "loss": 4.6544,
      "step": 29925
    },
    {
      "epoch": 0.6129052075440788,
      "grad_norm": 0.26441365480422974,
      "learning_rate": 0.00014839692448624309,
      "loss": 4.645,
      "step": 29930
    },
    {
      "epoch": 0.6130075973214835,
      "grad_norm": 0.26363876461982727,
      "learning_rate": 0.00014833503217355636,
      "loss": 4.6959,
      "step": 29935
    },
    {
      "epoch": 0.6131099870988881,
      "grad_norm": 0.2617667019367218,
      "learning_rate": 0.00014827314014436652,
      "loss": 4.6219,
      "step": 29940
    },
    {
      "epoch": 0.6132123768762927,
      "grad_norm": 0.2581588625907898,
      "learning_rate": 0.000148211248409212,
      "loss": 4.6411,
      "step": 29945
    },
    {
      "epoch": 0.6133147666536973,
      "grad_norm": 0.25130411982536316,
      "learning_rate": 0.00014814935697863123,
      "loss": 4.6319,
      "step": 29950
    },
    {
      "epoch": 0.613417156431102,
      "grad_norm": 0.26718178391456604,
      "learning_rate": 0.00014808746586316264,
      "loss": 4.6441,
      "step": 29955
    },
    {
      "epoch": 0.6135195462085066,
      "grad_norm": 0.2654898166656494,
      "learning_rate": 0.00014802557507334437,
      "loss": 4.7215,
      "step": 29960
    },
    {
      "epoch": 0.6136219359859112,
      "grad_norm": 0.2652888000011444,
      "learning_rate": 0.0001479636846197148,
      "loss": 4.6066,
      "step": 29965
    },
    {
      "epoch": 0.6137243257633158,
      "grad_norm": 0.2645012140274048,
      "learning_rate": 0.00014790179451281215,
      "loss": 4.7078,
      "step": 29970
    },
    {
      "epoch": 0.6138267155407204,
      "grad_norm": 0.29193446040153503,
      "learning_rate": 0.00014783990476317443,
      "loss": 4.5795,
      "step": 29975
    },
    {
      "epoch": 0.6139291053181251,
      "grad_norm": 0.26642873883247375,
      "learning_rate": 0.00014777801538133978,
      "loss": 4.6555,
      "step": 29980
    },
    {
      "epoch": 0.6140314950955297,
      "grad_norm": 0.26247337460517883,
      "learning_rate": 0.00014771612637784622,
      "loss": 4.6602,
      "step": 29985
    },
    {
      "epoch": 0.6141338848729343,
      "grad_norm": 0.28664398193359375,
      "learning_rate": 0.00014765423776323168,
      "loss": 4.6469,
      "step": 29990
    },
    {
      "epoch": 0.6142362746503389,
      "grad_norm": 0.2641092538833618,
      "learning_rate": 0.000147592349548034,
      "loss": 4.6224,
      "step": 29995
    },
    {
      "epoch": 0.6143386644277435,
      "grad_norm": 0.2592602074146271,
      "learning_rate": 0.0001475304617427911,
      "loss": 4.6271,
      "step": 30000
    },
    {
      "epoch": 0.6143386644277435,
      "eval_loss": 4.678186416625977,
      "eval_runtime": 17.1354,
      "eval_samples_per_second": 33.965,
      "eval_steps_per_second": 4.26,
      "step": 30000
    },
    {
      "epoch": 0.6144410542051482,
      "grad_norm": 0.27276498079299927,
      "learning_rate": 0.0001474685743580406,
      "loss": 4.6502,
      "step": 30005
    },
    {
      "epoch": 0.6145434439825528,
      "grad_norm": 0.30016595125198364,
      "learning_rate": 0.00014740668740432017,
      "loss": 4.6169,
      "step": 30010
    },
    {
      "epoch": 0.6146458337599574,
      "grad_norm": 0.27015364170074463,
      "learning_rate": 0.00014734480089216752,
      "loss": 4.6303,
      "step": 30015
    },
    {
      "epoch": 0.614748223537362,
      "grad_norm": 0.25933071970939636,
      "learning_rate": 0.00014728291483212015,
      "loss": 4.655,
      "step": 30020
    },
    {
      "epoch": 0.6148506133147666,
      "grad_norm": 0.2642268240451813,
      "learning_rate": 0.00014722102923471538,
      "loss": 4.6966,
      "step": 30025
    },
    {
      "epoch": 0.6149530030921713,
      "grad_norm": 0.26965948939323425,
      "learning_rate": 0.0001471591441104907,
      "loss": 4.6566,
      "step": 30030
    },
    {
      "epoch": 0.6150553928695759,
      "grad_norm": 0.26817694306373596,
      "learning_rate": 0.00014709725946998335,
      "loss": 4.6389,
      "step": 30035
    },
    {
      "epoch": 0.6151577826469805,
      "grad_norm": 0.27362197637557983,
      "learning_rate": 0.0001470353753237305,
      "loss": 4.633,
      "step": 30040
    },
    {
      "epoch": 0.6152601724243851,
      "grad_norm": 0.25554922223091125,
      "learning_rate": 0.00014697349168226928,
      "loss": 4.6694,
      "step": 30045
    },
    {
      "epoch": 0.6153625622017898,
      "grad_norm": 0.25772884488105774,
      "learning_rate": 0.00014691160855613683,
      "loss": 4.6689,
      "step": 30050
    },
    {
      "epoch": 0.6154649519791944,
      "grad_norm": 0.27397212386131287,
      "learning_rate": 0.00014684972595586994,
      "loss": 4.6297,
      "step": 30055
    },
    {
      "epoch": 0.615567341756599,
      "grad_norm": 0.2920845150947571,
      "learning_rate": 0.0001467878438920055,
      "loss": 4.6813,
      "step": 30060
    },
    {
      "epoch": 0.6156697315340036,
      "grad_norm": 0.27796587347984314,
      "learning_rate": 0.00014672596237508043,
      "loss": 4.6721,
      "step": 30065
    },
    {
      "epoch": 0.6157721213114082,
      "grad_norm": 0.27973082661628723,
      "learning_rate": 0.00014666408141563117,
      "loss": 4.6396,
      "step": 30070
    },
    {
      "epoch": 0.6158745110888129,
      "grad_norm": 0.2719156742095947,
      "learning_rate": 0.00014660220102419443,
      "loss": 4.6355,
      "step": 30075
    },
    {
      "epoch": 0.6159769008662175,
      "grad_norm": 0.26830217242240906,
      "learning_rate": 0.00014654032121130667,
      "loss": 4.55,
      "step": 30080
    },
    {
      "epoch": 0.6160792906436221,
      "grad_norm": 0.28855401277542114,
      "learning_rate": 0.00014647844198750423,
      "loss": 4.692,
      "step": 30085
    },
    {
      "epoch": 0.6161816804210267,
      "grad_norm": 0.27030399441719055,
      "learning_rate": 0.00014641656336332346,
      "loss": 4.6713,
      "step": 30090
    },
    {
      "epoch": 0.6162840701984313,
      "grad_norm": 0.25906726717948914,
      "learning_rate": 0.00014635468534930052,
      "loss": 4.7163,
      "step": 30095
    },
    {
      "epoch": 0.616386459975836,
      "grad_norm": 0.2601703405380249,
      "learning_rate": 0.0001462928079559714,
      "loss": 4.6709,
      "step": 30100
    },
    {
      "epoch": 0.6164888497532406,
      "grad_norm": 0.2649032473564148,
      "learning_rate": 0.0001462309311938721,
      "loss": 4.6223,
      "step": 30105
    },
    {
      "epoch": 0.6165912395306452,
      "grad_norm": 0.26456424593925476,
      "learning_rate": 0.00014616905507353864,
      "loss": 4.6261,
      "step": 30110
    },
    {
      "epoch": 0.6166936293080498,
      "grad_norm": 0.2632828950881958,
      "learning_rate": 0.00014610717960550655,
      "loss": 4.662,
      "step": 30115
    },
    {
      "epoch": 0.6167960190854546,
      "grad_norm": 0.28897950053215027,
      "learning_rate": 0.00014604530480031154,
      "loss": 4.6395,
      "step": 30120
    },
    {
      "epoch": 0.6168984088628592,
      "grad_norm": 0.2776760756969452,
      "learning_rate": 0.00014598343066848923,
      "loss": 4.6749,
      "step": 30125
    },
    {
      "epoch": 0.6170007986402638,
      "grad_norm": 0.3099904954433441,
      "learning_rate": 0.0001459215572205749,
      "loss": 4.6695,
      "step": 30130
    },
    {
      "epoch": 0.6171031884176684,
      "grad_norm": 0.28448745608329773,
      "learning_rate": 0.00014585968446710395,
      "loss": 4.6232,
      "step": 30135
    },
    {
      "epoch": 0.617205578195073,
      "grad_norm": 0.2703884541988373,
      "learning_rate": 0.00014579781241861158,
      "loss": 4.6903,
      "step": 30140
    },
    {
      "epoch": 0.6173079679724777,
      "grad_norm": 0.2632048428058624,
      "learning_rate": 0.0001457359410856327,
      "loss": 4.6358,
      "step": 30145
    },
    {
      "epoch": 0.6174103577498823,
      "grad_norm": 0.2635573148727417,
      "learning_rate": 0.0001456740704787023,
      "loss": 4.6238,
      "step": 30150
    },
    {
      "epoch": 0.6175127475272869,
      "grad_norm": 0.27382394671440125,
      "learning_rate": 0.00014561220060835527,
      "loss": 4.6477,
      "step": 30155
    },
    {
      "epoch": 0.6176151373046915,
      "grad_norm": 0.2560978829860687,
      "learning_rate": 0.0001455503314851263,
      "loss": 4.653,
      "step": 30160
    },
    {
      "epoch": 0.6177175270820962,
      "grad_norm": 0.26267510652542114,
      "learning_rate": 0.00014548846311954986,
      "loss": 4.6569,
      "step": 30165
    },
    {
      "epoch": 0.6178199168595008,
      "grad_norm": 0.2728870213031769,
      "learning_rate": 0.0001454265955221604,
      "loss": 4.6395,
      "step": 30170
    },
    {
      "epoch": 0.6179223066369054,
      "grad_norm": 0.2571091651916504,
      "learning_rate": 0.00014536472870349228,
      "loss": 4.6942,
      "step": 30175
    },
    {
      "epoch": 0.61802469641431,
      "grad_norm": 0.27143728733062744,
      "learning_rate": 0.00014530286267407962,
      "loss": 4.5975,
      "step": 30180
    },
    {
      "epoch": 0.6181270861917146,
      "grad_norm": 0.2803422808647156,
      "learning_rate": 0.00014524099744445647,
      "loss": 4.6274,
      "step": 30185
    },
    {
      "epoch": 0.6182294759691193,
      "grad_norm": 0.25826266407966614,
      "learning_rate": 0.00014517913302515678,
      "loss": 4.6414,
      "step": 30190
    },
    {
      "epoch": 0.6183318657465239,
      "grad_norm": 0.2646903693675995,
      "learning_rate": 0.00014511726942671417,
      "loss": 4.7091,
      "step": 30195
    },
    {
      "epoch": 0.6184342555239285,
      "grad_norm": 0.27455615997314453,
      "learning_rate": 0.0001450554066596623,
      "loss": 4.6893,
      "step": 30200
    },
    {
      "epoch": 0.6185366453013331,
      "grad_norm": 0.2511138617992401,
      "learning_rate": 0.0001449935447345348,
      "loss": 4.682,
      "step": 30205
    },
    {
      "epoch": 0.6186390350787377,
      "grad_norm": 0.27153676748275757,
      "learning_rate": 0.00014493168366186483,
      "loss": 4.6948,
      "step": 30210
    },
    {
      "epoch": 0.6187414248561424,
      "grad_norm": 0.2816159725189209,
      "learning_rate": 0.00014486982345218564,
      "loss": 4.6846,
      "step": 30215
    },
    {
      "epoch": 0.618843814633547,
      "grad_norm": 0.2648754119873047,
      "learning_rate": 0.00014480796411603034,
      "loss": 4.6257,
      "step": 30220
    },
    {
      "epoch": 0.6189462044109516,
      "grad_norm": 0.2981020212173462,
      "learning_rate": 0.0001447461056639317,
      "loss": 4.6035,
      "step": 30225
    },
    {
      "epoch": 0.6190485941883562,
      "grad_norm": 0.270654559135437,
      "learning_rate": 0.00014468424810642252,
      "loss": 4.6823,
      "step": 30230
    },
    {
      "epoch": 0.6191509839657608,
      "grad_norm": 0.30308738350868225,
      "learning_rate": 0.00014462239145403546,
      "loss": 4.6935,
      "step": 30235
    },
    {
      "epoch": 0.6192533737431655,
      "grad_norm": 0.25525858998298645,
      "learning_rate": 0.0001445605357173028,
      "loss": 4.6432,
      "step": 30240
    },
    {
      "epoch": 0.6193557635205701,
      "grad_norm": 0.2812856435775757,
      "learning_rate": 0.0001444986809067569,
      "loss": 4.6421,
      "step": 30245
    },
    {
      "epoch": 0.6194581532979747,
      "grad_norm": 0.2733539044857025,
      "learning_rate": 0.00014443682703293,
      "loss": 4.6383,
      "step": 30250
    },
    {
      "epoch": 0.6195605430753793,
      "grad_norm": 0.2560039758682251,
      "learning_rate": 0.00014437497410635389,
      "loss": 4.6515,
      "step": 30255
    },
    {
      "epoch": 0.619662932852784,
      "grad_norm": 0.2572624981403351,
      "learning_rate": 0.00014431312213756043,
      "loss": 4.6698,
      "step": 30260
    },
    {
      "epoch": 0.6197653226301886,
      "grad_norm": 0.25209107995033264,
      "learning_rate": 0.0001442512711370813,
      "loss": 4.6784,
      "step": 30265
    },
    {
      "epoch": 0.6198677124075932,
      "grad_norm": 0.3089885115623474,
      "learning_rate": 0.0001441894211154479,
      "loss": 4.6849,
      "step": 30270
    },
    {
      "epoch": 0.6199701021849978,
      "grad_norm": 0.27978983521461487,
      "learning_rate": 0.0001441275720831916,
      "loss": 4.6624,
      "step": 30275
    },
    {
      "epoch": 0.6200724919624024,
      "grad_norm": 0.29287540912628174,
      "learning_rate": 0.00014406572405084358,
      "loss": 4.6392,
      "step": 30280
    },
    {
      "epoch": 0.6201748817398071,
      "grad_norm": 0.25806277990341187,
      "learning_rate": 0.0001440038770289347,
      "loss": 4.6299,
      "step": 30285
    },
    {
      "epoch": 0.6202772715172117,
      "grad_norm": 0.2711588740348816,
      "learning_rate": 0.00014394203102799575,
      "loss": 4.6432,
      "step": 30290
    },
    {
      "epoch": 0.6203796612946163,
      "grad_norm": 0.261165052652359,
      "learning_rate": 0.0001438801860585574,
      "loss": 4.712,
      "step": 30295
    },
    {
      "epoch": 0.6204820510720209,
      "grad_norm": 0.2696019411087036,
      "learning_rate": 0.00014381834213115022,
      "loss": 4.6437,
      "step": 30300
    },
    {
      "epoch": 0.6205844408494255,
      "grad_norm": 0.2878361642360687,
      "learning_rate": 0.0001437564992563043,
      "loss": 4.6677,
      "step": 30305
    },
    {
      "epoch": 0.6206868306268302,
      "grad_norm": 0.285234272480011,
      "learning_rate": 0.00014369465744454977,
      "loss": 4.6823,
      "step": 30310
    },
    {
      "epoch": 0.6207892204042348,
      "grad_norm": 0.27212634682655334,
      "learning_rate": 0.00014363281670641664,
      "loss": 4.6405,
      "step": 30315
    },
    {
      "epoch": 0.6208916101816395,
      "grad_norm": 0.2737378776073456,
      "learning_rate": 0.0001435709770524345,
      "loss": 4.6519,
      "step": 30320
    },
    {
      "epoch": 0.6209939999590441,
      "grad_norm": 0.2617710828781128,
      "learning_rate": 0.00014350913849313295,
      "loss": 4.6914,
      "step": 30325
    },
    {
      "epoch": 0.6210963897364488,
      "grad_norm": 0.30267632007598877,
      "learning_rate": 0.00014344730103904143,
      "loss": 4.6453,
      "step": 30330
    },
    {
      "epoch": 0.6211987795138534,
      "grad_norm": 0.2610156834125519,
      "learning_rate": 0.00014338546470068895,
      "loss": 4.6167,
      "step": 30335
    },
    {
      "epoch": 0.621301169291258,
      "grad_norm": 0.2700619101524353,
      "learning_rate": 0.00014332362948860458,
      "loss": 4.6627,
      "step": 30340
    },
    {
      "epoch": 0.6214035590686626,
      "grad_norm": 0.27404138445854187,
      "learning_rate": 0.00014326179541331716,
      "loss": 4.6283,
      "step": 30345
    },
    {
      "epoch": 0.6215059488460672,
      "grad_norm": 0.2615413963794708,
      "learning_rate": 0.00014319996248535514,
      "loss": 4.6532,
      "step": 30350
    },
    {
      "epoch": 0.6216083386234719,
      "grad_norm": 0.27365297079086304,
      "learning_rate": 0.000143138130715247,
      "loss": 4.6558,
      "step": 30355
    },
    {
      "epoch": 0.6217107284008765,
      "grad_norm": 0.25237101316452026,
      "learning_rate": 0.00014307630011352093,
      "loss": 4.6647,
      "step": 30360
    },
    {
      "epoch": 0.6218131181782811,
      "grad_norm": 0.2741037607192993,
      "learning_rate": 0.00014301447069070496,
      "loss": 4.668,
      "step": 30365
    },
    {
      "epoch": 0.6219155079556857,
      "grad_norm": 0.2604462802410126,
      "learning_rate": 0.00014295264245732682,
      "loss": 4.6666,
      "step": 30370
    },
    {
      "epoch": 0.6220178977330904,
      "grad_norm": 0.26137575507164,
      "learning_rate": 0.00014289081542391425,
      "loss": 4.6485,
      "step": 30375
    },
    {
      "epoch": 0.622120287510495,
      "grad_norm": 0.2824261486530304,
      "learning_rate": 0.00014282898960099445,
      "loss": 4.5958,
      "step": 30380
    },
    {
      "epoch": 0.6222226772878996,
      "grad_norm": 0.2548956871032715,
      "learning_rate": 0.0001427671649990947,
      "loss": 4.6923,
      "step": 30385
    },
    {
      "epoch": 0.6223250670653042,
      "grad_norm": 0.2591424882411957,
      "learning_rate": 0.00014270534162874203,
      "loss": 4.6416,
      "step": 30390
    },
    {
      "epoch": 0.6224274568427088,
      "grad_norm": 0.2611396014690399,
      "learning_rate": 0.00014264351950046312,
      "loss": 4.6187,
      "step": 30395
    },
    {
      "epoch": 0.6225298466201135,
      "grad_norm": 0.27166223526000977,
      "learning_rate": 0.0001425816986247846,
      "loss": 4.6821,
      "step": 30400
    },
    {
      "epoch": 0.6226322363975181,
      "grad_norm": 0.2582108676433563,
      "learning_rate": 0.00014251987901223276,
      "loss": 4.6494,
      "step": 30405
    },
    {
      "epoch": 0.6227346261749227,
      "grad_norm": 0.2771899700164795,
      "learning_rate": 0.00014245806067333377,
      "loss": 4.6585,
      "step": 30410
    },
    {
      "epoch": 0.6228370159523273,
      "grad_norm": 0.29010680317878723,
      "learning_rate": 0.0001423962436186135,
      "loss": 4.6465,
      "step": 30415
    },
    {
      "epoch": 0.6229394057297319,
      "grad_norm": 0.2566431164741516,
      "learning_rate": 0.00014233442785859774,
      "loss": 4.6721,
      "step": 30420
    },
    {
      "epoch": 0.6230417955071366,
      "grad_norm": 0.27273184061050415,
      "learning_rate": 0.00014227261340381186,
      "loss": 4.6872,
      "step": 30425
    },
    {
      "epoch": 0.6231441852845412,
      "grad_norm": 0.26049673557281494,
      "learning_rate": 0.0001422108002647811,
      "loss": 4.6581,
      "step": 30430
    },
    {
      "epoch": 0.6232465750619458,
      "grad_norm": 0.2618211507797241,
      "learning_rate": 0.00014214898845203052,
      "loss": 4.6511,
      "step": 30435
    },
    {
      "epoch": 0.6233489648393504,
      "grad_norm": 0.2608950734138489,
      "learning_rate": 0.00014208717797608504,
      "loss": 4.6518,
      "step": 30440
    },
    {
      "epoch": 0.623451354616755,
      "grad_norm": 0.2676299214363098,
      "learning_rate": 0.00014202536884746906,
      "loss": 4.6916,
      "step": 30445
    },
    {
      "epoch": 0.6235537443941597,
      "grad_norm": 0.2924249768257141,
      "learning_rate": 0.00014196356107670698,
      "loss": 4.6485,
      "step": 30450
    },
    {
      "epoch": 0.6236561341715643,
      "grad_norm": 0.27814215421676636,
      "learning_rate": 0.00014190175467432297,
      "loss": 4.6598,
      "step": 30455
    },
    {
      "epoch": 0.6237585239489689,
      "grad_norm": 0.2750813663005829,
      "learning_rate": 0.00014183994965084084,
      "loss": 4.6518,
      "step": 30460
    },
    {
      "epoch": 0.6238609137263735,
      "grad_norm": 0.2613241672515869,
      "learning_rate": 0.00014177814601678429,
      "loss": 4.6314,
      "step": 30465
    },
    {
      "epoch": 0.6239633035037782,
      "grad_norm": 0.25416532158851624,
      "learning_rate": 0.00014171634378267673,
      "loss": 4.6913,
      "step": 30470
    },
    {
      "epoch": 0.6240656932811828,
      "grad_norm": 0.2595835030078888,
      "learning_rate": 0.00014165454295904124,
      "loss": 4.6661,
      "step": 30475
    },
    {
      "epoch": 0.6241680830585874,
      "grad_norm": 0.27630704641342163,
      "learning_rate": 0.00014159274355640084,
      "loss": 4.7037,
      "step": 30480
    },
    {
      "epoch": 0.624270472835992,
      "grad_norm": 0.3200695514678955,
      "learning_rate": 0.0001415309455852782,
      "loss": 4.6411,
      "step": 30485
    },
    {
      "epoch": 0.6243728626133966,
      "grad_norm": 0.26501190662384033,
      "learning_rate": 0.00014146914905619574,
      "loss": 4.6098,
      "step": 30490
    },
    {
      "epoch": 0.6244752523908013,
      "grad_norm": 0.26884472370147705,
      "learning_rate": 0.00014140735397967567,
      "loss": 4.627,
      "step": 30495
    },
    {
      "epoch": 0.6245776421682059,
      "grad_norm": 0.2895987033843994,
      "learning_rate": 0.00014134556036624,
      "loss": 4.6695,
      "step": 30500
    },
    {
      "epoch": 0.6246800319456105,
      "grad_norm": 0.30213648080825806,
      "learning_rate": 0.00014128376822641032,
      "loss": 4.6489,
      "step": 30505
    },
    {
      "epoch": 0.6247824217230151,
      "grad_norm": 0.25780001282691956,
      "learning_rate": 0.0001412219775707082,
      "loss": 4.6279,
      "step": 30510
    },
    {
      "epoch": 0.6248848115004197,
      "grad_norm": 0.2536289393901825,
      "learning_rate": 0.00014116018840965478,
      "loss": 4.6697,
      "step": 30515
    },
    {
      "epoch": 0.6249872012778244,
      "grad_norm": 0.2722889482975006,
      "learning_rate": 0.000141098400753771,
      "loss": 4.679,
      "step": 30520
    },
    {
      "epoch": 0.6250895910552291,
      "grad_norm": 0.26849550008773804,
      "learning_rate": 0.00014103661461357752,
      "loss": 4.6523,
      "step": 30525
    },
    {
      "epoch": 0.6251919808326337,
      "grad_norm": 0.28604739904403687,
      "learning_rate": 0.00014097482999959483,
      "loss": 4.6449,
      "step": 30530
    },
    {
      "epoch": 0.6252943706100383,
      "grad_norm": 0.2754095196723938,
      "learning_rate": 0.0001409130469223431,
      "loss": 4.6632,
      "step": 30535
    },
    {
      "epoch": 0.625396760387443,
      "grad_norm": 0.26504576206207275,
      "learning_rate": 0.00014085126539234216,
      "loss": 4.742,
      "step": 30540
    },
    {
      "epoch": 0.6254991501648476,
      "grad_norm": 0.26037588715553284,
      "learning_rate": 0.0001407894854201118,
      "loss": 4.6251,
      "step": 30545
    },
    {
      "epoch": 0.6256015399422522,
      "grad_norm": 0.25871458649635315,
      "learning_rate": 0.00014072770701617124,
      "loss": 4.685,
      "step": 30550
    },
    {
      "epoch": 0.6257039297196568,
      "grad_norm": 0.27008435130119324,
      "learning_rate": 0.00014066593019103967,
      "loss": 4.6656,
      "step": 30555
    },
    {
      "epoch": 0.6258063194970614,
      "grad_norm": 0.29850319027900696,
      "learning_rate": 0.00014060415495523602,
      "loss": 4.713,
      "step": 30560
    },
    {
      "epoch": 0.6259087092744661,
      "grad_norm": 0.3362254798412323,
      "learning_rate": 0.0001405423813192787,
      "loss": 4.6315,
      "step": 30565
    },
    {
      "epoch": 0.6260110990518707,
      "grad_norm": 0.26641160249710083,
      "learning_rate": 0.00014048060929368607,
      "loss": 4.643,
      "step": 30570
    },
    {
      "epoch": 0.6261134888292753,
      "grad_norm": 0.2786027491092682,
      "learning_rate": 0.00014041883888897623,
      "loss": 4.6421,
      "step": 30575
    },
    {
      "epoch": 0.6262158786066799,
      "grad_norm": 0.25587013363838196,
      "learning_rate": 0.0001403570701156668,
      "loss": 4.6483,
      "step": 30580
    },
    {
      "epoch": 0.6263182683840846,
      "grad_norm": 0.2661448121070862,
      "learning_rate": 0.00014029530298427536,
      "loss": 4.6201,
      "step": 30585
    },
    {
      "epoch": 0.6264206581614892,
      "grad_norm": 0.26119163632392883,
      "learning_rate": 0.00014023353750531906,
      "loss": 4.6333,
      "step": 30590
    },
    {
      "epoch": 0.6265230479388938,
      "grad_norm": 0.26306217908859253,
      "learning_rate": 0.00014017177368931487,
      "loss": 4.6623,
      "step": 30595
    },
    {
      "epoch": 0.6266254377162984,
      "grad_norm": 0.2618086338043213,
      "learning_rate": 0.0001401100115467793,
      "loss": 4.6762,
      "step": 30600
    },
    {
      "epoch": 0.626727827493703,
      "grad_norm": 0.2748504877090454,
      "learning_rate": 0.0001400482510882288,
      "loss": 4.7071,
      "step": 30605
    },
    {
      "epoch": 0.6268302172711077,
      "grad_norm": 0.2736903727054596,
      "learning_rate": 0.00013998649232417948,
      "loss": 4.6199,
      "step": 30610
    },
    {
      "epoch": 0.6269326070485123,
      "grad_norm": 0.27250659465789795,
      "learning_rate": 0.00013992473526514693,
      "loss": 4.643,
      "step": 30615
    },
    {
      "epoch": 0.6270349968259169,
      "grad_norm": 0.2631660997867584,
      "learning_rate": 0.0001398629799216467,
      "loss": 4.6248,
      "step": 30620
    },
    {
      "epoch": 0.6271373866033215,
      "grad_norm": 0.262580007314682,
      "learning_rate": 0.00013980122630419408,
      "loss": 4.6678,
      "step": 30625
    },
    {
      "epoch": 0.6272397763807261,
      "grad_norm": 0.26323434710502625,
      "learning_rate": 0.0001397394744233038,
      "loss": 4.6388,
      "step": 30630
    },
    {
      "epoch": 0.6273421661581308,
      "grad_norm": 0.26835930347442627,
      "learning_rate": 0.00013967772428949058,
      "loss": 4.6852,
      "step": 30635
    },
    {
      "epoch": 0.6274445559355354,
      "grad_norm": 0.2563098073005676,
      "learning_rate": 0.0001396159759132687,
      "loss": 4.6606,
      "step": 30640
    },
    {
      "epoch": 0.62754694571294,
      "grad_norm": 0.2582773268222809,
      "learning_rate": 0.0001395542293051521,
      "loss": 4.6492,
      "step": 30645
    },
    {
      "epoch": 0.6276493354903446,
      "grad_norm": 0.26118898391723633,
      "learning_rate": 0.0001394924844756545,
      "loss": 4.6497,
      "step": 30650
    },
    {
      "epoch": 0.6277517252677492,
      "grad_norm": 0.2593550980091095,
      "learning_rate": 0.00013943074143528942,
      "loss": 4.6605,
      "step": 30655
    },
    {
      "epoch": 0.6278541150451539,
      "grad_norm": 0.274387389421463,
      "learning_rate": 0.00013936900019456976,
      "loss": 4.6924,
      "step": 30660
    },
    {
      "epoch": 0.6279565048225585,
      "grad_norm": 0.28171506524086,
      "learning_rate": 0.0001393072607640084,
      "loss": 4.6535,
      "step": 30665
    },
    {
      "epoch": 0.6280588945999631,
      "grad_norm": 0.27562254667282104,
      "learning_rate": 0.00013924552315411784,
      "loss": 4.6853,
      "step": 30670
    },
    {
      "epoch": 0.6281612843773677,
      "grad_norm": 0.2885560095310211,
      "learning_rate": 0.00013918378737541018,
      "loss": 4.6677,
      "step": 30675
    },
    {
      "epoch": 0.6282636741547724,
      "grad_norm": 0.28729793429374695,
      "learning_rate": 0.0001391220534383973,
      "loss": 4.6241,
      "step": 30680
    },
    {
      "epoch": 0.628366063932177,
      "grad_norm": 0.2732503414154053,
      "learning_rate": 0.00013906032135359078,
      "loss": 4.6551,
      "step": 30685
    },
    {
      "epoch": 0.6284684537095816,
      "grad_norm": 0.2720591127872467,
      "learning_rate": 0.0001389985911315018,
      "loss": 4.6453,
      "step": 30690
    },
    {
      "epoch": 0.6285708434869862,
      "grad_norm": 0.2801700234413147,
      "learning_rate": 0.00013893686278264127,
      "loss": 4.6355,
      "step": 30695
    },
    {
      "epoch": 0.6286732332643908,
      "grad_norm": 0.25980231165885925,
      "learning_rate": 0.0001388751363175199,
      "loss": 4.6317,
      "step": 30700
    },
    {
      "epoch": 0.6287756230417955,
      "grad_norm": 0.2581658363342285,
      "learning_rate": 0.00013881341174664776,
      "loss": 4.6678,
      "step": 30705
    },
    {
      "epoch": 0.6288780128192001,
      "grad_norm": 0.27167370915412903,
      "learning_rate": 0.00013875168908053493,
      "loss": 4.6547,
      "step": 30710
    },
    {
      "epoch": 0.6289804025966047,
      "grad_norm": 0.2577280104160309,
      "learning_rate": 0.00013868996832969098,
      "loss": 4.6499,
      "step": 30715
    },
    {
      "epoch": 0.6290827923740093,
      "grad_norm": 0.2653399109840393,
      "learning_rate": 0.00013862824950462523,
      "loss": 4.626,
      "step": 30720
    },
    {
      "epoch": 0.6291851821514141,
      "grad_norm": 0.2757514417171478,
      "learning_rate": 0.00013856653261584662,
      "loss": 4.5893,
      "step": 30725
    },
    {
      "epoch": 0.6292875719288187,
      "grad_norm": 0.27251172065734863,
      "learning_rate": 0.00013850481767386383,
      "loss": 4.7225,
      "step": 30730
    },
    {
      "epoch": 0.6293899617062233,
      "grad_norm": 0.2618646025657654,
      "learning_rate": 0.00013844310468918517,
      "loss": 4.6658,
      "step": 30735
    },
    {
      "epoch": 0.6294923514836279,
      "grad_norm": 0.256133109331131,
      "learning_rate": 0.00013838139367231857,
      "loss": 4.7011,
      "step": 30740
    },
    {
      "epoch": 0.6295947412610325,
      "grad_norm": 0.2629673480987549,
      "learning_rate": 0.0001383196846337717,
      "loss": 4.649,
      "step": 30745
    },
    {
      "epoch": 0.6296971310384372,
      "grad_norm": 0.2708331346511841,
      "learning_rate": 0.00013825797758405188,
      "loss": 4.6626,
      "step": 30750
    },
    {
      "epoch": 0.6297995208158418,
      "grad_norm": 0.26925358176231384,
      "learning_rate": 0.000138196272533666,
      "loss": 4.6328,
      "step": 30755
    },
    {
      "epoch": 0.6299019105932464,
      "grad_norm": 0.25351065397262573,
      "learning_rate": 0.00013813456949312075,
      "loss": 4.6229,
      "step": 30760
    },
    {
      "epoch": 0.630004300370651,
      "grad_norm": 0.257271945476532,
      "learning_rate": 0.00013807286847292242,
      "loss": 4.6823,
      "step": 30765
    },
    {
      "epoch": 0.6301066901480556,
      "grad_norm": 0.2567111551761627,
      "learning_rate": 0.00013801116948357686,
      "loss": 4.6937,
      "step": 30770
    },
    {
      "epoch": 0.6302090799254603,
      "grad_norm": 0.2745841443538666,
      "learning_rate": 0.00013794947253558974,
      "loss": 4.6696,
      "step": 30775
    },
    {
      "epoch": 0.6303114697028649,
      "grad_norm": 0.26817578077316284,
      "learning_rate": 0.0001378877776394663,
      "loss": 4.7009,
      "step": 30780
    },
    {
      "epoch": 0.6304138594802695,
      "grad_norm": 0.25737127661705017,
      "learning_rate": 0.0001378260848057114,
      "loss": 4.6116,
      "step": 30785
    },
    {
      "epoch": 0.6305162492576741,
      "grad_norm": 0.2685026228427887,
      "learning_rate": 0.0001377643940448296,
      "loss": 4.6796,
      "step": 30790
    },
    {
      "epoch": 0.6306186390350788,
      "grad_norm": 0.2860910892486572,
      "learning_rate": 0.00013770270536732517,
      "loss": 4.697,
      "step": 30795
    },
    {
      "epoch": 0.6307210288124834,
      "grad_norm": 0.2825358808040619,
      "learning_rate": 0.00013764101878370176,
      "loss": 4.6613,
      "step": 30800
    },
    {
      "epoch": 0.630823418589888,
      "grad_norm": 0.2780124247074127,
      "learning_rate": 0.00013757933430446297,
      "loss": 4.6359,
      "step": 30805
    },
    {
      "epoch": 0.6309258083672926,
      "grad_norm": 0.26248306035995483,
      "learning_rate": 0.00013751765194011195,
      "loss": 4.6537,
      "step": 30810
    },
    {
      "epoch": 0.6310281981446972,
      "grad_norm": 0.27073562145233154,
      "learning_rate": 0.0001374559717011514,
      "loss": 4.6388,
      "step": 30815
    },
    {
      "epoch": 0.6311305879221019,
      "grad_norm": 0.25821781158447266,
      "learning_rate": 0.0001373942935980837,
      "loss": 4.6805,
      "step": 30820
    },
    {
      "epoch": 0.6312329776995065,
      "grad_norm": 0.30033940076828003,
      "learning_rate": 0.00013733261764141096,
      "loss": 4.6187,
      "step": 30825
    },
    {
      "epoch": 0.6313353674769111,
      "grad_norm": 0.26460811495780945,
      "learning_rate": 0.00013727094384163478,
      "loss": 4.7039,
      "step": 30830
    },
    {
      "epoch": 0.6314377572543157,
      "grad_norm": 0.2702391445636749,
      "learning_rate": 0.0001372092722092565,
      "loss": 4.6402,
      "step": 30835
    },
    {
      "epoch": 0.6315401470317203,
      "grad_norm": 0.2818358838558197,
      "learning_rate": 0.0001371476027547771,
      "loss": 4.643,
      "step": 30840
    },
    {
      "epoch": 0.631642536809125,
      "grad_norm": 0.26885858178138733,
      "learning_rate": 0.000137085935488697,
      "loss": 4.6741,
      "step": 30845
    },
    {
      "epoch": 0.6317449265865296,
      "grad_norm": 0.2621515989303589,
      "learning_rate": 0.0001370242704215165,
      "loss": 4.6976,
      "step": 30850
    },
    {
      "epoch": 0.6318473163639342,
      "grad_norm": 0.2500268816947937,
      "learning_rate": 0.0001369626075637354,
      "loss": 4.7203,
      "step": 30855
    },
    {
      "epoch": 0.6319497061413388,
      "grad_norm": 0.26177504658699036,
      "learning_rate": 0.0001369009469258531,
      "loss": 4.6571,
      "step": 30860
    },
    {
      "epoch": 0.6320520959187435,
      "grad_norm": 0.27457505464553833,
      "learning_rate": 0.0001368392885183687,
      "loss": 4.6319,
      "step": 30865
    },
    {
      "epoch": 0.6321544856961481,
      "grad_norm": 0.26758092641830444,
      "learning_rate": 0.0001367776323517809,
      "loss": 4.6879,
      "step": 30870
    },
    {
      "epoch": 0.6322568754735527,
      "grad_norm": 0.2595585882663727,
      "learning_rate": 0.00013671597843658794,
      "loss": 4.7071,
      "step": 30875
    },
    {
      "epoch": 0.6323592652509573,
      "grad_norm": 0.25526174902915955,
      "learning_rate": 0.0001366543267832878,
      "loss": 4.6196,
      "step": 30880
    },
    {
      "epoch": 0.6324616550283619,
      "grad_norm": 0.26654160022735596,
      "learning_rate": 0.00013659267740237792,
      "loss": 4.7092,
      "step": 30885
    },
    {
      "epoch": 0.6325640448057666,
      "grad_norm": 0.25316905975341797,
      "learning_rate": 0.0001365310303043556,
      "loss": 4.6671,
      "step": 30890
    },
    {
      "epoch": 0.6326664345831712,
      "grad_norm": 0.27131474018096924,
      "learning_rate": 0.00013646938549971742,
      "loss": 4.5842,
      "step": 30895
    },
    {
      "epoch": 0.6327688243605758,
      "grad_norm": 0.2656306028366089,
      "learning_rate": 0.0001364077429989598,
      "loss": 4.6664,
      "step": 30900
    },
    {
      "epoch": 0.6328712141379804,
      "grad_norm": 0.2851974368095398,
      "learning_rate": 0.00013634610281257882,
      "loss": 4.6328,
      "step": 30905
    },
    {
      "epoch": 0.632973603915385,
      "grad_norm": 0.28098243474960327,
      "learning_rate": 0.0001362844649510699,
      "loss": 4.6705,
      "step": 30910
    },
    {
      "epoch": 0.6330759936927897,
      "grad_norm": 0.2572101056575775,
      "learning_rate": 0.00013622282942492831,
      "loss": 4.6928,
      "step": 30915
    },
    {
      "epoch": 0.6331783834701943,
      "grad_norm": 0.27140459418296814,
      "learning_rate": 0.00013616119624464888,
      "loss": 4.6777,
      "step": 30920
    },
    {
      "epoch": 0.6332807732475989,
      "grad_norm": 0.2552447021007538,
      "learning_rate": 0.00013609956542072588,
      "loss": 4.6599,
      "step": 30925
    },
    {
      "epoch": 0.6333831630250036,
      "grad_norm": 0.2926422655582428,
      "learning_rate": 0.00013603793696365335,
      "loss": 4.6575,
      "step": 30930
    },
    {
      "epoch": 0.6334855528024083,
      "grad_norm": 0.3174777925014496,
      "learning_rate": 0.00013597631088392494,
      "loss": 4.6245,
      "step": 30935
    },
    {
      "epoch": 0.6335879425798129,
      "grad_norm": 0.2757773697376251,
      "learning_rate": 0.0001359146871920337,
      "loss": 4.7168,
      "step": 30940
    },
    {
      "epoch": 0.6336903323572175,
      "grad_norm": 0.2805185616016388,
      "learning_rate": 0.00013585306589847246,
      "loss": 4.6643,
      "step": 30945
    },
    {
      "epoch": 0.6337927221346221,
      "grad_norm": 0.2652919888496399,
      "learning_rate": 0.0001357914470137336,
      "loss": 4.6943,
      "step": 30950
    },
    {
      "epoch": 0.6338951119120267,
      "grad_norm": 0.271441787481308,
      "learning_rate": 0.00013572983054830904,
      "loss": 4.6667,
      "step": 30955
    },
    {
      "epoch": 0.6339975016894314,
      "grad_norm": 0.25050222873687744,
      "learning_rate": 0.00013566821651269032,
      "loss": 4.6832,
      "step": 30960
    },
    {
      "epoch": 0.634099891466836,
      "grad_norm": 0.25222840905189514,
      "learning_rate": 0.00013560660491736864,
      "loss": 4.6669,
      "step": 30965
    },
    {
      "epoch": 0.6342022812442406,
      "grad_norm": 0.27370142936706543,
      "learning_rate": 0.00013554499577283455,
      "loss": 4.6698,
      "step": 30970
    },
    {
      "epoch": 0.6343046710216452,
      "grad_norm": 0.2555899918079376,
      "learning_rate": 0.0001354833890895785,
      "loss": 4.6676,
      "step": 30975
    },
    {
      "epoch": 0.6344070607990498,
      "grad_norm": 0.2767821252346039,
      "learning_rate": 0.00013542178487809035,
      "loss": 4.6232,
      "step": 30980
    },
    {
      "epoch": 0.6345094505764545,
      "grad_norm": 0.26810166239738464,
      "learning_rate": 0.00013536018314885947,
      "loss": 4.6715,
      "step": 30985
    },
    {
      "epoch": 0.6346118403538591,
      "grad_norm": 0.26725059747695923,
      "learning_rate": 0.00013529858391237492,
      "loss": 4.6095,
      "step": 30990
    },
    {
      "epoch": 0.6347142301312637,
      "grad_norm": 0.26766064763069153,
      "learning_rate": 0.00013523698717912534,
      "loss": 4.6691,
      "step": 30995
    },
    {
      "epoch": 0.6348166199086683,
      "grad_norm": 0.27194923162460327,
      "learning_rate": 0.00013517539295959885,
      "loss": 4.6947,
      "step": 31000
    },
    {
      "epoch": 0.6348166199086683,
      "eval_loss": 4.668798923492432,
      "eval_runtime": 17.1246,
      "eval_samples_per_second": 33.986,
      "eval_steps_per_second": 4.263,
      "step": 31000
    },
    {
      "epoch": 0.634919009686073,
      "grad_norm": 0.27915051579475403,
      "learning_rate": 0.00013511380126428324,
      "loss": 4.6686,
      "step": 31005
    },
    {
      "epoch": 0.6350213994634776,
      "grad_norm": 0.27530568838119507,
      "learning_rate": 0.00013505221210366583,
      "loss": 4.647,
      "step": 31010
    },
    {
      "epoch": 0.6351237892408822,
      "grad_norm": 0.271456778049469,
      "learning_rate": 0.00013499062548823353,
      "loss": 4.6884,
      "step": 31015
    },
    {
      "epoch": 0.6352261790182868,
      "grad_norm": 0.26434803009033203,
      "learning_rate": 0.00013492904142847276,
      "loss": 4.6459,
      "step": 31020
    },
    {
      "epoch": 0.6353285687956914,
      "grad_norm": 0.2877468168735504,
      "learning_rate": 0.00013486745993486956,
      "loss": 4.6195,
      "step": 31025
    },
    {
      "epoch": 0.6354309585730961,
      "grad_norm": 0.2787456214427948,
      "learning_rate": 0.00013480588101790958,
      "loss": 4.586,
      "step": 31030
    },
    {
      "epoch": 0.6355333483505007,
      "grad_norm": 0.2634255588054657,
      "learning_rate": 0.00013474430468807784,
      "loss": 4.6536,
      "step": 31035
    },
    {
      "epoch": 0.6356357381279053,
      "grad_norm": 0.2897188365459442,
      "learning_rate": 0.0001346827309558591,
      "loss": 4.6444,
      "step": 31040
    },
    {
      "epoch": 0.6357381279053099,
      "grad_norm": 0.28307461738586426,
      "learning_rate": 0.00013462115983173766,
      "loss": 4.6021,
      "step": 31045
    },
    {
      "epoch": 0.6358405176827145,
      "grad_norm": 0.2803022563457489,
      "learning_rate": 0.00013455959132619726,
      "loss": 4.6446,
      "step": 31050
    },
    {
      "epoch": 0.6359429074601192,
      "grad_norm": 0.2702583074569702,
      "learning_rate": 0.00013449802544972134,
      "loss": 4.6254,
      "step": 31055
    },
    {
      "epoch": 0.6360452972375238,
      "grad_norm": 0.2827287018299103,
      "learning_rate": 0.00013443646221279287,
      "loss": 4.6596,
      "step": 31060
    },
    {
      "epoch": 0.6361476870149284,
      "grad_norm": 0.2646181285381317,
      "learning_rate": 0.00013437490162589416,
      "loss": 4.6375,
      "step": 31065
    },
    {
      "epoch": 0.636250076792333,
      "grad_norm": 0.26009178161621094,
      "learning_rate": 0.00013431334369950739,
      "loss": 4.64,
      "step": 31070
    },
    {
      "epoch": 0.6363524665697377,
      "grad_norm": 0.2934807240962982,
      "learning_rate": 0.00013425178844411414,
      "loss": 4.6307,
      "step": 31075
    },
    {
      "epoch": 0.6364548563471423,
      "grad_norm": 0.2529505789279938,
      "learning_rate": 0.00013419023587019538,
      "loss": 4.6737,
      "step": 31080
    },
    {
      "epoch": 0.6365572461245469,
      "grad_norm": 0.2632039189338684,
      "learning_rate": 0.00013412868598823186,
      "loss": 4.6603,
      "step": 31085
    },
    {
      "epoch": 0.6366596359019515,
      "grad_norm": 0.2729152739048004,
      "learning_rate": 0.00013406713880870385,
      "loss": 4.6461,
      "step": 31090
    },
    {
      "epoch": 0.6367620256793561,
      "grad_norm": 0.29835352301597595,
      "learning_rate": 0.00013400559434209095,
      "loss": 4.7061,
      "step": 31095
    },
    {
      "epoch": 0.6368644154567608,
      "grad_norm": 0.28775349259376526,
      "learning_rate": 0.00013394405259887257,
      "loss": 4.5874,
      "step": 31100
    },
    {
      "epoch": 0.6369668052341654,
      "grad_norm": 0.2684108018875122,
      "learning_rate": 0.00013388251358952748,
      "loss": 4.6305,
      "step": 31105
    },
    {
      "epoch": 0.63706919501157,
      "grad_norm": 0.26177436113357544,
      "learning_rate": 0.00013382097732453393,
      "loss": 4.6794,
      "step": 31110
    },
    {
      "epoch": 0.6371715847889746,
      "grad_norm": 0.2835049331188202,
      "learning_rate": 0.00013375944381436997,
      "loss": 4.6515,
      "step": 31115
    },
    {
      "epoch": 0.6372739745663792,
      "grad_norm": 0.26999524235725403,
      "learning_rate": 0.00013369791306951296,
      "loss": 4.6284,
      "step": 31120
    },
    {
      "epoch": 0.6373763643437839,
      "grad_norm": 0.26571357250213623,
      "learning_rate": 0.00013363638510043978,
      "loss": 4.6305,
      "step": 31125
    },
    {
      "epoch": 0.6374787541211886,
      "grad_norm": 0.2659294605255127,
      "learning_rate": 0.00013357485991762695,
      "loss": 4.6506,
      "step": 31130
    },
    {
      "epoch": 0.6375811438985932,
      "grad_norm": 0.26096558570861816,
      "learning_rate": 0.0001335133375315505,
      "loss": 4.6697,
      "step": 31135
    },
    {
      "epoch": 0.6376835336759978,
      "grad_norm": 0.26418620347976685,
      "learning_rate": 0.00013345181795268584,
      "loss": 4.6154,
      "step": 31140
    },
    {
      "epoch": 0.6377859234534025,
      "grad_norm": 0.2718810439109802,
      "learning_rate": 0.00013339030119150812,
      "loss": 4.5965,
      "step": 31145
    },
    {
      "epoch": 0.6378883132308071,
      "grad_norm": 0.27645042538642883,
      "learning_rate": 0.00013332878725849187,
      "loss": 4.6343,
      "step": 31150
    },
    {
      "epoch": 0.6379907030082117,
      "grad_norm": 0.2620978355407715,
      "learning_rate": 0.0001332672761641112,
      "loss": 4.6612,
      "step": 31155
    },
    {
      "epoch": 0.6380930927856163,
      "grad_norm": 0.27411043643951416,
      "learning_rate": 0.0001332057679188396,
      "loss": 4.6513,
      "step": 31160
    },
    {
      "epoch": 0.6381954825630209,
      "grad_norm": 0.25224819779396057,
      "learning_rate": 0.0001331442625331503,
      "loss": 4.5905,
      "step": 31165
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 0.29067060351371765,
      "learning_rate": 0.00013308276001751596,
      "loss": 4.6709,
      "step": 31170
    },
    {
      "epoch": 0.6384002621178302,
      "grad_norm": 0.2866936922073364,
      "learning_rate": 0.00013302126038240855,
      "loss": 4.612,
      "step": 31175
    },
    {
      "epoch": 0.6385026518952348,
      "grad_norm": 0.2631337642669678,
      "learning_rate": 0.00013295976363829982,
      "loss": 4.6359,
      "step": 31180
    },
    {
      "epoch": 0.6386050416726394,
      "grad_norm": 0.2646958827972412,
      "learning_rate": 0.00013289826979566096,
      "loss": 4.6162,
      "step": 31185
    },
    {
      "epoch": 0.638707431450044,
      "grad_norm": 0.2835996448993683,
      "learning_rate": 0.00013283677886496252,
      "loss": 4.6625,
      "step": 31190
    },
    {
      "epoch": 0.6388098212274487,
      "grad_norm": 0.26449674367904663,
      "learning_rate": 0.00013277529085667475,
      "loss": 4.6393,
      "step": 31195
    },
    {
      "epoch": 0.6389122110048533,
      "grad_norm": 0.27873530983924866,
      "learning_rate": 0.00013271380578126738,
      "loss": 4.6502,
      "step": 31200
    },
    {
      "epoch": 0.6390146007822579,
      "grad_norm": 0.2593189775943756,
      "learning_rate": 0.00013265232364920938,
      "loss": 4.6423,
      "step": 31205
    },
    {
      "epoch": 0.6391169905596625,
      "grad_norm": 0.2723395824432373,
      "learning_rate": 0.00013259084447096955,
      "loss": 4.6721,
      "step": 31210
    },
    {
      "epoch": 0.6392193803370672,
      "grad_norm": 0.2658664882183075,
      "learning_rate": 0.0001325293682570161,
      "loss": 4.6312,
      "step": 31215
    },
    {
      "epoch": 0.6393217701144718,
      "grad_norm": 0.2618985176086426,
      "learning_rate": 0.00013246789501781658,
      "loss": 4.6702,
      "step": 31220
    },
    {
      "epoch": 0.6394241598918764,
      "grad_norm": 0.27947092056274414,
      "learning_rate": 0.00013240642476383818,
      "loss": 4.6387,
      "step": 31225
    },
    {
      "epoch": 0.639526549669281,
      "grad_norm": 0.27498918771743774,
      "learning_rate": 0.00013234495750554757,
      "loss": 4.5731,
      "step": 31230
    },
    {
      "epoch": 0.6396289394466856,
      "grad_norm": 0.27096331119537354,
      "learning_rate": 0.00013228349325341082,
      "loss": 4.6752,
      "step": 31235
    },
    {
      "epoch": 0.6397313292240903,
      "grad_norm": 0.26269224286079407,
      "learning_rate": 0.0001322220320178936,
      "loss": 4.687,
      "step": 31240
    },
    {
      "epoch": 0.6398337190014949,
      "grad_norm": 0.26830875873565674,
      "learning_rate": 0.0001321605738094611,
      "loss": 4.6535,
      "step": 31245
    },
    {
      "epoch": 0.6399361087788995,
      "grad_norm": 0.26473885774612427,
      "learning_rate": 0.00013209911863857775,
      "loss": 4.6585,
      "step": 31250
    },
    {
      "epoch": 0.6400384985563041,
      "grad_norm": 0.27840396761894226,
      "learning_rate": 0.00013203766651570767,
      "loss": 4.6332,
      "step": 31255
    },
    {
      "epoch": 0.6401408883337087,
      "grad_norm": 0.26939570903778076,
      "learning_rate": 0.00013197621745131453,
      "loss": 4.6491,
      "step": 31260
    },
    {
      "epoch": 0.6402432781111134,
      "grad_norm": 0.2661779820919037,
      "learning_rate": 0.00013191477145586123,
      "loss": 4.6901,
      "step": 31265
    },
    {
      "epoch": 0.640345667888518,
      "grad_norm": 0.2732379734516144,
      "learning_rate": 0.00013185332853981035,
      "loss": 4.6541,
      "step": 31270
    },
    {
      "epoch": 0.6404480576659226,
      "grad_norm": 0.2726680040359497,
      "learning_rate": 0.0001317918887136239,
      "loss": 4.6936,
      "step": 31275
    },
    {
      "epoch": 0.6405504474433272,
      "grad_norm": 0.2657804787158966,
      "learning_rate": 0.00013173045198776326,
      "loss": 4.6607,
      "step": 31280
    },
    {
      "epoch": 0.6406528372207319,
      "grad_norm": 0.26409026980400085,
      "learning_rate": 0.0001316690183726894,
      "loss": 4.6207,
      "step": 31285
    },
    {
      "epoch": 0.6407552269981365,
      "grad_norm": 0.26010459661483765,
      "learning_rate": 0.00013160758787886278,
      "loss": 4.5902,
      "step": 31290
    },
    {
      "epoch": 0.6408576167755411,
      "grad_norm": 0.26215532422065735,
      "learning_rate": 0.00013154616051674328,
      "loss": 4.6221,
      "step": 31295
    },
    {
      "epoch": 0.6409600065529457,
      "grad_norm": 0.24972034990787506,
      "learning_rate": 0.0001314847362967901,
      "loss": 4.6785,
      "step": 31300
    },
    {
      "epoch": 0.6410623963303503,
      "grad_norm": 0.2610864043235779,
      "learning_rate": 0.00013142331522946217,
      "loss": 4.6598,
      "step": 31305
    },
    {
      "epoch": 0.641164786107755,
      "grad_norm": 0.2775006592273712,
      "learning_rate": 0.00013136189732521782,
      "loss": 4.6107,
      "step": 31310
    },
    {
      "epoch": 0.6412671758851596,
      "grad_norm": 0.27934861183166504,
      "learning_rate": 0.00013130048259451458,
      "loss": 4.6204,
      "step": 31315
    },
    {
      "epoch": 0.6413695656625642,
      "grad_norm": 0.2705919146537781,
      "learning_rate": 0.0001312390710478098,
      "loss": 4.6791,
      "step": 31320
    },
    {
      "epoch": 0.6414719554399688,
      "grad_norm": 0.26356828212738037,
      "learning_rate": 0.00013117766269556008,
      "loss": 4.6094,
      "step": 31325
    },
    {
      "epoch": 0.6415743452173734,
      "grad_norm": 0.281583309173584,
      "learning_rate": 0.0001311162575482215,
      "loss": 4.6482,
      "step": 31330
    },
    {
      "epoch": 0.6416767349947782,
      "grad_norm": 0.2770513594150543,
      "learning_rate": 0.0001310548556162496,
      "loss": 4.6195,
      "step": 31335
    },
    {
      "epoch": 0.6417791247721828,
      "grad_norm": 0.27134597301483154,
      "learning_rate": 0.00013099345691009953,
      "loss": 4.6189,
      "step": 31340
    },
    {
      "epoch": 0.6418815145495874,
      "grad_norm": 0.2557509243488312,
      "learning_rate": 0.00013093206144022557,
      "loss": 4.6669,
      "step": 31345
    },
    {
      "epoch": 0.641983904326992,
      "grad_norm": 0.2638404965400696,
      "learning_rate": 0.0001308706692170817,
      "loss": 4.6664,
      "step": 31350
    },
    {
      "epoch": 0.6420862941043967,
      "grad_norm": 0.26745322346687317,
      "learning_rate": 0.00013080928025112137,
      "loss": 4.6851,
      "step": 31355
    },
    {
      "epoch": 0.6421886838818013,
      "grad_norm": 0.26370060443878174,
      "learning_rate": 0.00013074789455279725,
      "loss": 4.6362,
      "step": 31360
    },
    {
      "epoch": 0.6422910736592059,
      "grad_norm": 0.2882365882396698,
      "learning_rate": 0.00013068651213256165,
      "loss": 4.6646,
      "step": 31365
    },
    {
      "epoch": 0.6423934634366105,
      "grad_norm": 0.2712881863117218,
      "learning_rate": 0.00013062513300086625,
      "loss": 4.7021,
      "step": 31370
    },
    {
      "epoch": 0.6424958532140151,
      "grad_norm": 0.2631036043167114,
      "learning_rate": 0.00013056375716816217,
      "loss": 4.611,
      "step": 31375
    },
    {
      "epoch": 0.6425982429914198,
      "grad_norm": 0.28645893931388855,
      "learning_rate": 0.00013050238464489997,
      "loss": 4.6319,
      "step": 31380
    },
    {
      "epoch": 0.6427006327688244,
      "grad_norm": 0.2656135559082031,
      "learning_rate": 0.00013044101544152975,
      "loss": 4.6208,
      "step": 31385
    },
    {
      "epoch": 0.642803022546229,
      "grad_norm": 0.2631455361843109,
      "learning_rate": 0.00013037964956850076,
      "loss": 4.6291,
      "step": 31390
    },
    {
      "epoch": 0.6429054123236336,
      "grad_norm": 0.26318174600601196,
      "learning_rate": 0.00013031828703626198,
      "loss": 4.6321,
      "step": 31395
    },
    {
      "epoch": 0.6430078021010383,
      "grad_norm": 0.26336249709129333,
      "learning_rate": 0.00013025692785526176,
      "loss": 4.6368,
      "step": 31400
    },
    {
      "epoch": 0.6431101918784429,
      "grad_norm": 0.2661740183830261,
      "learning_rate": 0.00013019557203594774,
      "loss": 4.6368,
      "step": 31405
    },
    {
      "epoch": 0.6432125816558475,
      "grad_norm": 0.2788700461387634,
      "learning_rate": 0.0001301342195887671,
      "loss": 4.6237,
      "step": 31410
    },
    {
      "epoch": 0.6433149714332521,
      "grad_norm": 0.2701438069343567,
      "learning_rate": 0.00013007287052416652,
      "loss": 4.6335,
      "step": 31415
    },
    {
      "epoch": 0.6434173612106567,
      "grad_norm": 0.27556711435317993,
      "learning_rate": 0.00013001152485259186,
      "loss": 4.6613,
      "step": 31420
    },
    {
      "epoch": 0.6435197509880614,
      "grad_norm": 0.2829678952693939,
      "learning_rate": 0.00012995018258448864,
      "loss": 4.6782,
      "step": 31425
    },
    {
      "epoch": 0.643622140765466,
      "grad_norm": 0.27381062507629395,
      "learning_rate": 0.00012988884373030165,
      "loss": 4.6606,
      "step": 31430
    },
    {
      "epoch": 0.6437245305428706,
      "grad_norm": 0.2785673141479492,
      "learning_rate": 0.0001298275083004753,
      "loss": 4.6144,
      "step": 31435
    },
    {
      "epoch": 0.6438269203202752,
      "grad_norm": 0.2944762110710144,
      "learning_rate": 0.00012976617630545314,
      "loss": 4.6269,
      "step": 31440
    },
    {
      "epoch": 0.6439293100976798,
      "grad_norm": 0.2634918689727783,
      "learning_rate": 0.00012970484775567826,
      "loss": 4.6674,
      "step": 31445
    },
    {
      "epoch": 0.6440316998750845,
      "grad_norm": 0.2722005248069763,
      "learning_rate": 0.0001296435226615933,
      "loss": 4.5888,
      "step": 31450
    },
    {
      "epoch": 0.6441340896524891,
      "grad_norm": 0.2757498025894165,
      "learning_rate": 0.00012958220103364006,
      "loss": 4.6541,
      "step": 31455
    },
    {
      "epoch": 0.6442364794298937,
      "grad_norm": 0.27535200119018555,
      "learning_rate": 0.00012952088288225997,
      "loss": 4.6466,
      "step": 31460
    },
    {
      "epoch": 0.6443388692072983,
      "grad_norm": 0.26694583892822266,
      "learning_rate": 0.00012945956821789373,
      "loss": 4.6304,
      "step": 31465
    },
    {
      "epoch": 0.644441258984703,
      "grad_norm": 0.2681289315223694,
      "learning_rate": 0.0001293982570509815,
      "loss": 4.6285,
      "step": 31470
    },
    {
      "epoch": 0.6445436487621076,
      "grad_norm": 0.2661706805229187,
      "learning_rate": 0.0001293369493919628,
      "loss": 4.6666,
      "step": 31475
    },
    {
      "epoch": 0.6446460385395122,
      "grad_norm": 0.2730381190776825,
      "learning_rate": 0.0001292756452512767,
      "loss": 4.6437,
      "step": 31480
    },
    {
      "epoch": 0.6447484283169168,
      "grad_norm": 0.27071261405944824,
      "learning_rate": 0.00012921434463936143,
      "loss": 4.6378,
      "step": 31485
    },
    {
      "epoch": 0.6448508180943214,
      "grad_norm": 0.25690731406211853,
      "learning_rate": 0.00012915304756665479,
      "loss": 4.5901,
      "step": 31490
    },
    {
      "epoch": 0.644953207871726,
      "grad_norm": 0.26359280943870544,
      "learning_rate": 0.00012909175404359402,
      "loss": 4.6265,
      "step": 31495
    },
    {
      "epoch": 0.6450555976491307,
      "grad_norm": 0.2577619254589081,
      "learning_rate": 0.00012903046408061555,
      "loss": 4.6133,
      "step": 31500
    },
    {
      "epoch": 0.6451579874265353,
      "grad_norm": 0.27889639139175415,
      "learning_rate": 0.00012896917768815536,
      "loss": 4.6308,
      "step": 31505
    },
    {
      "epoch": 0.6452603772039399,
      "grad_norm": 0.2662416398525238,
      "learning_rate": 0.00012890789487664882,
      "loss": 4.6571,
      "step": 31510
    },
    {
      "epoch": 0.6453627669813445,
      "grad_norm": 0.2627926468849182,
      "learning_rate": 0.00012884661565653064,
      "loss": 4.6088,
      "step": 31515
    },
    {
      "epoch": 0.6454651567587492,
      "grad_norm": 0.26997891068458557,
      "learning_rate": 0.00012878534003823492,
      "loss": 4.6122,
      "step": 31520
    },
    {
      "epoch": 0.6455675465361538,
      "grad_norm": 0.2604016363620758,
      "learning_rate": 0.00012872406803219526,
      "loss": 4.6726,
      "step": 31525
    },
    {
      "epoch": 0.6456699363135584,
      "grad_norm": 0.2653314471244812,
      "learning_rate": 0.00012866279964884438,
      "loss": 4.6571,
      "step": 31530
    },
    {
      "epoch": 0.6457723260909631,
      "grad_norm": 0.25445958971977234,
      "learning_rate": 0.00012860153489861468,
      "loss": 4.6389,
      "step": 31535
    },
    {
      "epoch": 0.6458747158683678,
      "grad_norm": 0.26626482605934143,
      "learning_rate": 0.00012854027379193778,
      "loss": 4.6169,
      "step": 31540
    },
    {
      "epoch": 0.6459771056457724,
      "grad_norm": 0.2648249864578247,
      "learning_rate": 0.00012847901633924466,
      "loss": 4.6308,
      "step": 31545
    },
    {
      "epoch": 0.646079495423177,
      "grad_norm": 0.27532705664634705,
      "learning_rate": 0.00012841776255096579,
      "loss": 4.6544,
      "step": 31550
    },
    {
      "epoch": 0.6461818852005816,
      "grad_norm": 0.26371702551841736,
      "learning_rate": 0.00012835651243753098,
      "loss": 4.683,
      "step": 31555
    },
    {
      "epoch": 0.6462842749779862,
      "grad_norm": 0.26055431365966797,
      "learning_rate": 0.0001282952660093693,
      "loss": 4.6081,
      "step": 31560
    },
    {
      "epoch": 0.6463866647553909,
      "grad_norm": 0.31222742795944214,
      "learning_rate": 0.00012823402327690934,
      "loss": 4.677,
      "step": 31565
    },
    {
      "epoch": 0.6464890545327955,
      "grad_norm": 0.2880314886569977,
      "learning_rate": 0.000128172784250579,
      "loss": 4.6435,
      "step": 31570
    },
    {
      "epoch": 0.6465914443102001,
      "grad_norm": 0.2676008641719818,
      "learning_rate": 0.0001281115489408056,
      "loss": 4.6519,
      "step": 31575
    },
    {
      "epoch": 0.6466938340876047,
      "grad_norm": 0.25955820083618164,
      "learning_rate": 0.00012805031735801568,
      "loss": 4.7021,
      "step": 31580
    },
    {
      "epoch": 0.6467962238650093,
      "grad_norm": 0.26532554626464844,
      "learning_rate": 0.00012798908951263527,
      "loss": 4.6321,
      "step": 31585
    },
    {
      "epoch": 0.646898613642414,
      "grad_norm": 0.27449071407318115,
      "learning_rate": 0.00012792786541508983,
      "loss": 4.6442,
      "step": 31590
    },
    {
      "epoch": 0.6470010034198186,
      "grad_norm": 0.2585086226463318,
      "learning_rate": 0.00012786664507580397,
      "loss": 4.6519,
      "step": 31595
    },
    {
      "epoch": 0.6471033931972232,
      "grad_norm": 0.2651681900024414,
      "learning_rate": 0.00012780542850520185,
      "loss": 4.6639,
      "step": 31600
    },
    {
      "epoch": 0.6472057829746278,
      "grad_norm": 0.2532927989959717,
      "learning_rate": 0.0001277442157137069,
      "loss": 4.6369,
      "step": 31605
    },
    {
      "epoch": 0.6473081727520325,
      "grad_norm": 0.27088263630867004,
      "learning_rate": 0.00012768300671174194,
      "loss": 4.5084,
      "step": 31610
    },
    {
      "epoch": 0.6474105625294371,
      "grad_norm": 0.2733481824398041,
      "learning_rate": 0.00012762180150972907,
      "loss": 4.6762,
      "step": 31615
    },
    {
      "epoch": 0.6475129523068417,
      "grad_norm": 0.2534061372280121,
      "learning_rate": 0.00012756060011808997,
      "loss": 4.6468,
      "step": 31620
    },
    {
      "epoch": 0.6476153420842463,
      "grad_norm": 0.32298529148101807,
      "learning_rate": 0.0001274994025472453,
      "loss": 4.6615,
      "step": 31625
    },
    {
      "epoch": 0.6477177318616509,
      "grad_norm": 0.270778626203537,
      "learning_rate": 0.00012743820880761533,
      "loss": 4.6509,
      "step": 31630
    },
    {
      "epoch": 0.6478201216390556,
      "grad_norm": 0.277631551027298,
      "learning_rate": 0.0001273770189096197,
      "loss": 4.6263,
      "step": 31635
    },
    {
      "epoch": 0.6479225114164602,
      "grad_norm": 0.2573779821395874,
      "learning_rate": 0.00012731583286367725,
      "loss": 4.6196,
      "step": 31640
    },
    {
      "epoch": 0.6480249011938648,
      "grad_norm": 0.2725055515766144,
      "learning_rate": 0.00012725465068020623,
      "loss": 4.6395,
      "step": 31645
    },
    {
      "epoch": 0.6481272909712694,
      "grad_norm": 0.27602913975715637,
      "learning_rate": 0.00012719347236962432,
      "loss": 4.6314,
      "step": 31650
    },
    {
      "epoch": 0.648229680748674,
      "grad_norm": 0.2726741433143616,
      "learning_rate": 0.0001271322979423483,
      "loss": 4.6858,
      "step": 31655
    },
    {
      "epoch": 0.6483320705260787,
      "grad_norm": 0.27213045954704285,
      "learning_rate": 0.00012707112740879456,
      "loss": 4.6656,
      "step": 31660
    },
    {
      "epoch": 0.6484344603034833,
      "grad_norm": 0.27257275581359863,
      "learning_rate": 0.00012700996077937875,
      "loss": 4.6502,
      "step": 31665
    },
    {
      "epoch": 0.6485368500808879,
      "grad_norm": 0.2729254961013794,
      "learning_rate": 0.00012694879806451567,
      "loss": 4.6305,
      "step": 31670
    },
    {
      "epoch": 0.6486392398582925,
      "grad_norm": 0.2568114399909973,
      "learning_rate": 0.00012688763927461968,
      "loss": 4.6921,
      "step": 31675
    },
    {
      "epoch": 0.6487416296356971,
      "grad_norm": 0.2689872086048126,
      "learning_rate": 0.00012682648442010442,
      "loss": 4.6759,
      "step": 31680
    },
    {
      "epoch": 0.6488440194131018,
      "grad_norm": 0.260543555021286,
      "learning_rate": 0.00012676533351138275,
      "loss": 4.6728,
      "step": 31685
    },
    {
      "epoch": 0.6489464091905064,
      "grad_norm": 0.2777576744556427,
      "learning_rate": 0.00012670418655886699,
      "loss": 4.6378,
      "step": 31690
    },
    {
      "epoch": 0.649048798967911,
      "grad_norm": 0.2914818525314331,
      "learning_rate": 0.00012664304357296876,
      "loss": 4.6359,
      "step": 31695
    },
    {
      "epoch": 0.6491511887453156,
      "grad_norm": 0.2843390107154846,
      "learning_rate": 0.00012658190456409893,
      "loss": 4.6131,
      "step": 31700
    },
    {
      "epoch": 0.6492535785227203,
      "grad_norm": 0.279347687959671,
      "learning_rate": 0.00012652076954266774,
      "loss": 4.583,
      "step": 31705
    },
    {
      "epoch": 0.6493559683001249,
      "grad_norm": 0.2818780541419983,
      "learning_rate": 0.00012645963851908476,
      "loss": 4.6393,
      "step": 31710
    },
    {
      "epoch": 0.6494583580775295,
      "grad_norm": 0.2638985514640808,
      "learning_rate": 0.00012639851150375898,
      "loss": 4.6603,
      "step": 31715
    },
    {
      "epoch": 0.6495607478549341,
      "grad_norm": 0.2956162691116333,
      "learning_rate": 0.00012633738850709845,
      "loss": 4.635,
      "step": 31720
    },
    {
      "epoch": 0.6496631376323387,
      "grad_norm": 0.2939032316207886,
      "learning_rate": 0.0001262762695395107,
      "loss": 4.7239,
      "step": 31725
    },
    {
      "epoch": 0.6497655274097434,
      "grad_norm": 0.28255006670951843,
      "learning_rate": 0.00012621515461140267,
      "loss": 4.598,
      "step": 31730
    },
    {
      "epoch": 0.649867917187148,
      "grad_norm": 0.274516224861145,
      "learning_rate": 0.0001261540437331804,
      "loss": 4.6531,
      "step": 31735
    },
    {
      "epoch": 0.6499703069645527,
      "grad_norm": 0.2611735761165619,
      "learning_rate": 0.00012609293691524936,
      "loss": 4.5758,
      "step": 31740
    },
    {
      "epoch": 0.6500726967419573,
      "grad_norm": 0.28692975640296936,
      "learning_rate": 0.00012603183416801437,
      "loss": 4.64,
      "step": 31745
    },
    {
      "epoch": 0.650175086519362,
      "grad_norm": 0.2735917568206787,
      "learning_rate": 0.00012597073550187941,
      "loss": 4.6833,
      "step": 31750
    },
    {
      "epoch": 0.6502774762967666,
      "grad_norm": 0.26063647866249084,
      "learning_rate": 0.0001259096409272479,
      "loss": 4.6374,
      "step": 31755
    },
    {
      "epoch": 0.6503798660741712,
      "grad_norm": 0.3492627441883087,
      "learning_rate": 0.0001258485504545226,
      "loss": 4.5286,
      "step": 31760
    },
    {
      "epoch": 0.6504822558515758,
      "grad_norm": 0.30432385206222534,
      "learning_rate": 0.00012578746409410533,
      "loss": 4.6764,
      "step": 31765
    },
    {
      "epoch": 0.6505846456289804,
      "grad_norm": 0.2966804802417755,
      "learning_rate": 0.00012572638185639744,
      "loss": 4.6625,
      "step": 31770
    },
    {
      "epoch": 0.6506870354063851,
      "grad_norm": 0.26955991983413696,
      "learning_rate": 0.00012566530375179952,
      "loss": 4.6665,
      "step": 31775
    },
    {
      "epoch": 0.6507894251837897,
      "grad_norm": 0.27826356887817383,
      "learning_rate": 0.00012560422979071141,
      "loss": 4.6067,
      "step": 31780
    },
    {
      "epoch": 0.6508918149611943,
      "grad_norm": 0.26850879192352295,
      "learning_rate": 0.00012554315998353232,
      "loss": 4.6274,
      "step": 31785
    },
    {
      "epoch": 0.6509942047385989,
      "grad_norm": 0.34333401918411255,
      "learning_rate": 0.0001254820943406607,
      "loss": 4.6176,
      "step": 31790
    },
    {
      "epoch": 0.6510965945160035,
      "grad_norm": 0.26371055841445923,
      "learning_rate": 0.00012542103287249424,
      "loss": 4.6695,
      "step": 31795
    },
    {
      "epoch": 0.6511989842934082,
      "grad_norm": 0.2705726623535156,
      "learning_rate": 0.00012535997558943,
      "loss": 4.6457,
      "step": 31800
    },
    {
      "epoch": 0.6513013740708128,
      "grad_norm": 0.26995766162872314,
      "learning_rate": 0.00012529892250186443,
      "loss": 4.6459,
      "step": 31805
    },
    {
      "epoch": 0.6514037638482174,
      "grad_norm": 0.26318830251693726,
      "learning_rate": 0.00012523787362019297,
      "loss": 4.6526,
      "step": 31810
    },
    {
      "epoch": 0.651506153625622,
      "grad_norm": 0.2708052098751068,
      "learning_rate": 0.0001251768289548106,
      "loss": 4.6319,
      "step": 31815
    },
    {
      "epoch": 0.6516085434030267,
      "grad_norm": 0.25832897424697876,
      "learning_rate": 0.0001251157885161115,
      "loss": 4.6209,
      "step": 31820
    },
    {
      "epoch": 0.6517109331804313,
      "grad_norm": 0.2967028319835663,
      "learning_rate": 0.0001250547523144891,
      "loss": 4.6093,
      "step": 31825
    },
    {
      "epoch": 0.6518133229578359,
      "grad_norm": 0.27143245935440063,
      "learning_rate": 0.00012499372036033613,
      "loss": 4.6182,
      "step": 31830
    },
    {
      "epoch": 0.6519157127352405,
      "grad_norm": 0.28250429034233093,
      "learning_rate": 0.00012493269266404468,
      "loss": 4.6537,
      "step": 31835
    },
    {
      "epoch": 0.6520181025126451,
      "grad_norm": 0.26262298226356506,
      "learning_rate": 0.00012487166923600597,
      "loss": 4.7006,
      "step": 31840
    },
    {
      "epoch": 0.6521204922900498,
      "grad_norm": 0.27226799726486206,
      "learning_rate": 0.00012481065008661054,
      "loss": 4.7019,
      "step": 31845
    },
    {
      "epoch": 0.6522228820674544,
      "grad_norm": 0.2654237151145935,
      "learning_rate": 0.0001247496352262483,
      "loss": 4.647,
      "step": 31850
    },
    {
      "epoch": 0.652325271844859,
      "grad_norm": 0.2802788019180298,
      "learning_rate": 0.00012468862466530835,
      "loss": 4.6323,
      "step": 31855
    },
    {
      "epoch": 0.6524276616222636,
      "grad_norm": 0.26382607221603394,
      "learning_rate": 0.00012462761841417895,
      "loss": 4.5944,
      "step": 31860
    },
    {
      "epoch": 0.6525300513996682,
      "grad_norm": 0.278029203414917,
      "learning_rate": 0.00012456661648324786,
      "loss": 4.6604,
      "step": 31865
    },
    {
      "epoch": 0.6526324411770729,
      "grad_norm": 0.2683279514312744,
      "learning_rate": 0.00012450561888290192,
      "loss": 4.6052,
      "step": 31870
    },
    {
      "epoch": 0.6527348309544775,
      "grad_norm": 0.2615223824977875,
      "learning_rate": 0.0001244446256235273,
      "loss": 4.6708,
      "step": 31875
    },
    {
      "epoch": 0.6528372207318821,
      "grad_norm": 0.25342893600463867,
      "learning_rate": 0.00012438363671550944,
      "loss": 4.6668,
      "step": 31880
    },
    {
      "epoch": 0.6529396105092867,
      "grad_norm": 0.2844807803630829,
      "learning_rate": 0.00012432265216923304,
      "loss": 4.6415,
      "step": 31885
    },
    {
      "epoch": 0.6530420002866913,
      "grad_norm": 0.27546849846839905,
      "learning_rate": 0.000124261671995082,
      "loss": 4.5932,
      "step": 31890
    },
    {
      "epoch": 0.653144390064096,
      "grad_norm": 0.26301658153533936,
      "learning_rate": 0.00012420069620343953,
      "loss": 4.5966,
      "step": 31895
    },
    {
      "epoch": 0.6532467798415006,
      "grad_norm": 0.27280715107917786,
      "learning_rate": 0.00012413972480468816,
      "loss": 4.6455,
      "step": 31900
    },
    {
      "epoch": 0.6533491696189052,
      "grad_norm": 0.29577556252479553,
      "learning_rate": 0.00012407875780920944,
      "loss": 4.5637,
      "step": 31905
    },
    {
      "epoch": 0.6534515593963098,
      "grad_norm": 0.2669891119003296,
      "learning_rate": 0.0001240177952273844,
      "loss": 4.6018,
      "step": 31910
    },
    {
      "epoch": 0.6535539491737145,
      "grad_norm": 0.27085086703300476,
      "learning_rate": 0.00012395683706959332,
      "loss": 4.6722,
      "step": 31915
    },
    {
      "epoch": 0.6536563389511191,
      "grad_norm": 0.3017023503780365,
      "learning_rate": 0.00012389588334621552,
      "loss": 4.6263,
      "step": 31920
    },
    {
      "epoch": 0.6537587287285237,
      "grad_norm": 0.2614699900150299,
      "learning_rate": 0.00012383493406762974,
      "loss": 4.6629,
      "step": 31925
    },
    {
      "epoch": 0.6538611185059283,
      "grad_norm": 0.2590034306049347,
      "learning_rate": 0.00012377398924421398,
      "loss": 4.6495,
      "step": 31930
    },
    {
      "epoch": 0.6539635082833329,
      "grad_norm": 0.27455946803092957,
      "learning_rate": 0.00012371304888634532,
      "loss": 4.6404,
      "step": 31935
    },
    {
      "epoch": 0.6540658980607376,
      "grad_norm": 0.2596089541912079,
      "learning_rate": 0.00012365211300440022,
      "loss": 4.6562,
      "step": 31940
    },
    {
      "epoch": 0.6541682878381423,
      "grad_norm": 0.2656812071800232,
      "learning_rate": 0.00012359118160875442,
      "loss": 4.6422,
      "step": 31945
    },
    {
      "epoch": 0.6542706776155469,
      "grad_norm": 0.5823690295219421,
      "learning_rate": 0.00012353025470978265,
      "loss": 4.6762,
      "step": 31950
    },
    {
      "epoch": 0.6543730673929515,
      "grad_norm": 0.3039045035839081,
      "learning_rate": 0.0001234693323178591,
      "loss": 4.6236,
      "step": 31955
    },
    {
      "epoch": 0.6544754571703562,
      "grad_norm": 0.288870245218277,
      "learning_rate": 0.00012340841444335716,
      "loss": 4.6372,
      "step": 31960
    },
    {
      "epoch": 0.6545778469477608,
      "grad_norm": 0.2619956433773041,
      "learning_rate": 0.00012334750109664941,
      "loss": 4.6564,
      "step": 31965
    },
    {
      "epoch": 0.6546802367251654,
      "grad_norm": 0.28278493881225586,
      "learning_rate": 0.00012328659228810765,
      "loss": 4.6393,
      "step": 31970
    },
    {
      "epoch": 0.65478262650257,
      "grad_norm": 0.2665233016014099,
      "learning_rate": 0.00012322568802810297,
      "loss": 4.6199,
      "step": 31975
    },
    {
      "epoch": 0.6548850162799746,
      "grad_norm": 0.2708400785923004,
      "learning_rate": 0.00012316478832700558,
      "loss": 4.6483,
      "step": 31980
    },
    {
      "epoch": 0.6549874060573793,
      "grad_norm": 0.28332042694091797,
      "learning_rate": 0.000123103893195185,
      "loss": 4.6458,
      "step": 31985
    },
    {
      "epoch": 0.6550897958347839,
      "grad_norm": 0.26691529154777527,
      "learning_rate": 0.00012304300264300995,
      "loss": 4.6489,
      "step": 31990
    },
    {
      "epoch": 0.6551921856121885,
      "grad_norm": 0.2727983593940735,
      "learning_rate": 0.0001229821166808484,
      "loss": 4.6138,
      "step": 31995
    },
    {
      "epoch": 0.6552945753895931,
      "grad_norm": 0.268181711435318,
      "learning_rate": 0.00012292123531906744,
      "loss": 4.6083,
      "step": 32000
    },
    {
      "epoch": 0.6552945753895931,
      "eval_loss": 4.658926963806152,
      "eval_runtime": 17.137,
      "eval_samples_per_second": 33.962,
      "eval_steps_per_second": 4.26,
      "step": 32000
    },
    {
      "epoch": 0.6553969651669977,
      "grad_norm": 0.2773132920265198,
      "learning_rate": 0.00012286035856803347,
      "loss": 4.6312,
      "step": 32005
    },
    {
      "epoch": 0.6554993549444024,
      "grad_norm": 0.27755749225616455,
      "learning_rate": 0.00012279948643811214,
      "loss": 4.6845,
      "step": 32010
    },
    {
      "epoch": 0.655601744721807,
      "grad_norm": 0.2796289920806885,
      "learning_rate": 0.00012273861893966813,
      "loss": 4.6156,
      "step": 32015
    },
    {
      "epoch": 0.6557041344992116,
      "grad_norm": 0.2649158239364624,
      "learning_rate": 0.0001226777560830655,
      "loss": 4.6517,
      "step": 32020
    },
    {
      "epoch": 0.6558065242766162,
      "grad_norm": 0.26124560832977295,
      "learning_rate": 0.00012261689787866755,
      "loss": 4.6151,
      "step": 32025
    },
    {
      "epoch": 0.6559089140540209,
      "grad_norm": 0.27674198150634766,
      "learning_rate": 0.0001225560443368366,
      "loss": 4.6402,
      "step": 32030
    },
    {
      "epoch": 0.6560113038314255,
      "grad_norm": 0.26916542649269104,
      "learning_rate": 0.00012249519546793432,
      "loss": 4.6554,
      "step": 32035
    },
    {
      "epoch": 0.6561136936088301,
      "grad_norm": 0.3102434277534485,
      "learning_rate": 0.00012243435128232166,
      "loss": 4.6373,
      "step": 32040
    },
    {
      "epoch": 0.6562160833862347,
      "grad_norm": 0.26134636998176575,
      "learning_rate": 0.00012237351179035847,
      "loss": 4.6307,
      "step": 32045
    },
    {
      "epoch": 0.6563184731636393,
      "grad_norm": 0.2624574899673462,
      "learning_rate": 0.00012231267700240408,
      "loss": 4.6192,
      "step": 32050
    },
    {
      "epoch": 0.656420862941044,
      "grad_norm": 0.2781785726547241,
      "learning_rate": 0.00012225184692881696,
      "loss": 4.6371,
      "step": 32055
    },
    {
      "epoch": 0.6565232527184486,
      "grad_norm": 0.2835623025894165,
      "learning_rate": 0.0001221910215799547,
      "loss": 4.6352,
      "step": 32060
    },
    {
      "epoch": 0.6566256424958532,
      "grad_norm": 0.2875203490257263,
      "learning_rate": 0.0001221302009661742,
      "loss": 4.6118,
      "step": 32065
    },
    {
      "epoch": 0.6567280322732578,
      "grad_norm": 0.27794086933135986,
      "learning_rate": 0.00012206938509783146,
      "loss": 4.6267,
      "step": 32070
    },
    {
      "epoch": 0.6568304220506624,
      "grad_norm": 0.26146331429481506,
      "learning_rate": 0.00012200857398528166,
      "loss": 4.6389,
      "step": 32075
    },
    {
      "epoch": 0.6569328118280671,
      "grad_norm": 0.2839362621307373,
      "learning_rate": 0.00012194776763887922,
      "loss": 4.6356,
      "step": 32080
    },
    {
      "epoch": 0.6570352016054717,
      "grad_norm": 0.279504656791687,
      "learning_rate": 0.00012188696606897784,
      "loss": 4.633,
      "step": 32085
    },
    {
      "epoch": 0.6571375913828763,
      "grad_norm": 0.2745186686515808,
      "learning_rate": 0.00012182616928593016,
      "loss": 4.6342,
      "step": 32090
    },
    {
      "epoch": 0.6572399811602809,
      "grad_norm": 0.2650163769721985,
      "learning_rate": 0.00012176537730008822,
      "loss": 4.6658,
      "step": 32095
    },
    {
      "epoch": 0.6573423709376855,
      "grad_norm": 0.2790212333202362,
      "learning_rate": 0.0001217045901218032,
      "loss": 4.5968,
      "step": 32100
    },
    {
      "epoch": 0.6574447607150902,
      "grad_norm": 0.2662540674209595,
      "learning_rate": 0.00012164380776142539,
      "loss": 4.5886,
      "step": 32105
    },
    {
      "epoch": 0.6575471504924948,
      "grad_norm": 0.30496945977211,
      "learning_rate": 0.00012158303022930432,
      "loss": 4.6244,
      "step": 32110
    },
    {
      "epoch": 0.6576495402698994,
      "grad_norm": 0.2908116579055786,
      "learning_rate": 0.00012152225753578871,
      "loss": 4.6374,
      "step": 32115
    },
    {
      "epoch": 0.657751930047304,
      "grad_norm": 0.27005699276924133,
      "learning_rate": 0.00012146148969122637,
      "loss": 4.5542,
      "step": 32120
    },
    {
      "epoch": 0.6578543198247087,
      "grad_norm": 0.26197630167007446,
      "learning_rate": 0.00012140072670596438,
      "loss": 4.6129,
      "step": 32125
    },
    {
      "epoch": 0.6579567096021133,
      "grad_norm": 0.2653680145740509,
      "learning_rate": 0.00012133996859034896,
      "loss": 4.6552,
      "step": 32130
    },
    {
      "epoch": 0.6580590993795179,
      "grad_norm": 0.2568564713001251,
      "learning_rate": 0.00012127921535472554,
      "loss": 4.6844,
      "step": 32135
    },
    {
      "epoch": 0.6581614891569225,
      "grad_norm": 0.29861655831336975,
      "learning_rate": 0.00012121846700943857,
      "loss": 4.6407,
      "step": 32140
    },
    {
      "epoch": 0.6582638789343273,
      "grad_norm": 0.2732132077217102,
      "learning_rate": 0.00012115772356483185,
      "loss": 4.6709,
      "step": 32145
    },
    {
      "epoch": 0.6583662687117319,
      "grad_norm": 0.2853640019893646,
      "learning_rate": 0.00012109698503124825,
      "loss": 4.5996,
      "step": 32150
    },
    {
      "epoch": 0.6584686584891365,
      "grad_norm": 0.27273640036582947,
      "learning_rate": 0.00012103625141902981,
      "loss": 4.6275,
      "step": 32155
    },
    {
      "epoch": 0.6585710482665411,
      "grad_norm": 0.274423748254776,
      "learning_rate": 0.00012097552273851775,
      "loss": 4.6608,
      "step": 32160
    },
    {
      "epoch": 0.6586734380439457,
      "grad_norm": 0.2623249292373657,
      "learning_rate": 0.00012091479900005251,
      "loss": 4.6841,
      "step": 32165
    },
    {
      "epoch": 0.6587758278213504,
      "grad_norm": 0.25459566712379456,
      "learning_rate": 0.00012085408021397349,
      "loss": 4.6704,
      "step": 32170
    },
    {
      "epoch": 0.658878217598755,
      "grad_norm": 0.2660377323627472,
      "learning_rate": 0.00012079336639061948,
      "loss": 4.6507,
      "step": 32175
    },
    {
      "epoch": 0.6589806073761596,
      "grad_norm": 0.2769128084182739,
      "learning_rate": 0.00012073265754032839,
      "loss": 4.6678,
      "step": 32180
    },
    {
      "epoch": 0.6590829971535642,
      "grad_norm": 0.27020207047462463,
      "learning_rate": 0.00012067195367343701,
      "loss": 4.6802,
      "step": 32185
    },
    {
      "epoch": 0.6591853869309688,
      "grad_norm": 0.29727283120155334,
      "learning_rate": 0.00012061125480028164,
      "loss": 4.6547,
      "step": 32190
    },
    {
      "epoch": 0.6592877767083735,
      "grad_norm": 0.28506189584732056,
      "learning_rate": 0.0001205505609311976,
      "loss": 4.643,
      "step": 32195
    },
    {
      "epoch": 0.6593901664857781,
      "grad_norm": 0.27035385370254517,
      "learning_rate": 0.00012048987207651921,
      "loss": 4.6403,
      "step": 32200
    },
    {
      "epoch": 0.6594925562631827,
      "grad_norm": 0.2715008556842804,
      "learning_rate": 0.00012042918824658017,
      "loss": 4.641,
      "step": 32205
    },
    {
      "epoch": 0.6595949460405873,
      "grad_norm": 0.35396960377693176,
      "learning_rate": 0.00012036850945171322,
      "loss": 4.6404,
      "step": 32210
    },
    {
      "epoch": 0.659697335817992,
      "grad_norm": 0.2653484046459198,
      "learning_rate": 0.00012030783570225017,
      "loss": 4.6788,
      "step": 32215
    },
    {
      "epoch": 0.6597997255953966,
      "grad_norm": 0.25991418957710266,
      "learning_rate": 0.0001202471670085221,
      "loss": 4.6357,
      "step": 32220
    },
    {
      "epoch": 0.6599021153728012,
      "grad_norm": 0.25958484411239624,
      "learning_rate": 0.00012018650338085918,
      "loss": 4.6591,
      "step": 32225
    },
    {
      "epoch": 0.6600045051502058,
      "grad_norm": 0.25597479939460754,
      "learning_rate": 0.00012012584482959064,
      "loss": 4.5933,
      "step": 32230
    },
    {
      "epoch": 0.6601068949276104,
      "grad_norm": 0.28442850708961487,
      "learning_rate": 0.00012006519136504495,
      "loss": 4.5905,
      "step": 32235
    },
    {
      "epoch": 0.660209284705015,
      "grad_norm": 0.2714941203594208,
      "learning_rate": 0.00012000454299754971,
      "loss": 4.6265,
      "step": 32240
    },
    {
      "epoch": 0.6603116744824197,
      "grad_norm": 0.2652340531349182,
      "learning_rate": 0.00011994389973743159,
      "loss": 4.5818,
      "step": 32245
    },
    {
      "epoch": 0.6604140642598243,
      "grad_norm": 0.28526750206947327,
      "learning_rate": 0.00011988326159501638,
      "loss": 4.6545,
      "step": 32250
    },
    {
      "epoch": 0.6605164540372289,
      "grad_norm": 0.27121856808662415,
      "learning_rate": 0.00011982262858062918,
      "loss": 4.6917,
      "step": 32255
    },
    {
      "epoch": 0.6606188438146335,
      "grad_norm": 0.27800634503364563,
      "learning_rate": 0.00011976200070459385,
      "loss": 4.6577,
      "step": 32260
    },
    {
      "epoch": 0.6607212335920382,
      "grad_norm": 0.277902752161026,
      "learning_rate": 0.00011970137797723378,
      "loss": 4.5837,
      "step": 32265
    },
    {
      "epoch": 0.6608236233694428,
      "grad_norm": 0.2635326087474823,
      "learning_rate": 0.00011964076040887125,
      "loss": 4.6866,
      "step": 32270
    },
    {
      "epoch": 0.6609260131468474,
      "grad_norm": 0.27133452892303467,
      "learning_rate": 0.00011958014800982777,
      "loss": 4.5943,
      "step": 32275
    },
    {
      "epoch": 0.661028402924252,
      "grad_norm": 0.27321186661720276,
      "learning_rate": 0.00011951954079042381,
      "loss": 4.6676,
      "step": 32280
    },
    {
      "epoch": 0.6611307927016566,
      "grad_norm": 0.26956549286842346,
      "learning_rate": 0.00011945893876097913,
      "loss": 4.6727,
      "step": 32285
    },
    {
      "epoch": 0.6612331824790613,
      "grad_norm": 0.29847729206085205,
      "learning_rate": 0.00011939834193181255,
      "loss": 4.6154,
      "step": 32290
    },
    {
      "epoch": 0.6613355722564659,
      "grad_norm": 0.31802478432655334,
      "learning_rate": 0.00011933775031324192,
      "loss": 4.6107,
      "step": 32295
    },
    {
      "epoch": 0.6614379620338705,
      "grad_norm": 0.2739986777305603,
      "learning_rate": 0.00011927716391558436,
      "loss": 4.6971,
      "step": 32300
    },
    {
      "epoch": 0.6615403518112751,
      "grad_norm": 0.2841625213623047,
      "learning_rate": 0.000119216582749156,
      "loss": 4.6881,
      "step": 32305
    },
    {
      "epoch": 0.6616427415886798,
      "grad_norm": 0.26190415024757385,
      "learning_rate": 0.00011915600682427206,
      "loss": 4.7229,
      "step": 32310
    },
    {
      "epoch": 0.6617451313660844,
      "grad_norm": 0.2650793492794037,
      "learning_rate": 0.00011909543615124691,
      "loss": 4.6629,
      "step": 32315
    },
    {
      "epoch": 0.661847521143489,
      "grad_norm": 0.2658609449863434,
      "learning_rate": 0.00011903487074039413,
      "loss": 4.6527,
      "step": 32320
    },
    {
      "epoch": 0.6619499109208936,
      "grad_norm": 0.2587004601955414,
      "learning_rate": 0.00011897431060202613,
      "loss": 4.6361,
      "step": 32325
    },
    {
      "epoch": 0.6620523006982982,
      "grad_norm": 0.25448375940322876,
      "learning_rate": 0.00011891375574645467,
      "loss": 4.6343,
      "step": 32330
    },
    {
      "epoch": 0.6621546904757029,
      "grad_norm": 0.25848206877708435,
      "learning_rate": 0.00011885320618399056,
      "loss": 4.6221,
      "step": 32335
    },
    {
      "epoch": 0.6622570802531075,
      "grad_norm": 0.28528285026550293,
      "learning_rate": 0.00011879266192494358,
      "loss": 4.631,
      "step": 32340
    },
    {
      "epoch": 0.6623594700305121,
      "grad_norm": 0.2573157548904419,
      "learning_rate": 0.00011873212297962277,
      "loss": 4.6226,
      "step": 32345
    },
    {
      "epoch": 0.6624618598079168,
      "grad_norm": 0.2667756676673889,
      "learning_rate": 0.00011867158935833625,
      "loss": 4.6758,
      "step": 32350
    },
    {
      "epoch": 0.6625642495853215,
      "grad_norm": 0.26147395372390747,
      "learning_rate": 0.00011861106107139103,
      "loss": 4.5983,
      "step": 32355
    },
    {
      "epoch": 0.6626666393627261,
      "grad_norm": 0.2658372223377228,
      "learning_rate": 0.00011855053812909347,
      "loss": 4.6626,
      "step": 32360
    },
    {
      "epoch": 0.6627690291401307,
      "grad_norm": 0.26490965485572815,
      "learning_rate": 0.00011849002054174898,
      "loss": 4.6241,
      "step": 32365
    },
    {
      "epoch": 0.6628714189175353,
      "grad_norm": 0.26870647072792053,
      "learning_rate": 0.00011842950831966185,
      "loss": 4.6698,
      "step": 32370
    },
    {
      "epoch": 0.6629738086949399,
      "grad_norm": 0.2598232924938202,
      "learning_rate": 0.00011836900147313564,
      "loss": 4.6495,
      "step": 32375
    },
    {
      "epoch": 0.6630761984723446,
      "grad_norm": 0.26024144887924194,
      "learning_rate": 0.000118308500012473,
      "loss": 4.63,
      "step": 32380
    },
    {
      "epoch": 0.6631785882497492,
      "grad_norm": 0.30041414499282837,
      "learning_rate": 0.00011824800394797556,
      "loss": 4.6471,
      "step": 32385
    },
    {
      "epoch": 0.6632809780271538,
      "grad_norm": 0.2771242558956146,
      "learning_rate": 0.0001181875132899441,
      "loss": 4.6601,
      "step": 32390
    },
    {
      "epoch": 0.6633833678045584,
      "grad_norm": 0.26687443256378174,
      "learning_rate": 0.00011812702804867856,
      "loss": 4.6366,
      "step": 32395
    },
    {
      "epoch": 0.663485757581963,
      "grad_norm": 0.2735798954963684,
      "learning_rate": 0.0001180665482344777,
      "loss": 4.6597,
      "step": 32400
    },
    {
      "epoch": 0.6635881473593677,
      "grad_norm": 0.29402485489845276,
      "learning_rate": 0.00011800607385763962,
      "loss": 4.6251,
      "step": 32405
    },
    {
      "epoch": 0.6636905371367723,
      "grad_norm": 0.28436583280563354,
      "learning_rate": 0.00011794560492846138,
      "loss": 4.6099,
      "step": 32410
    },
    {
      "epoch": 0.6637929269141769,
      "grad_norm": 0.26935702562332153,
      "learning_rate": 0.00011788514145723918,
      "loss": 4.6544,
      "step": 32415
    },
    {
      "epoch": 0.6638953166915815,
      "grad_norm": 0.26539263129234314,
      "learning_rate": 0.00011782468345426813,
      "loss": 4.6665,
      "step": 32420
    },
    {
      "epoch": 0.6639977064689861,
      "grad_norm": 0.2789909541606903,
      "learning_rate": 0.00011776423092984256,
      "loss": 4.6838,
      "step": 32425
    },
    {
      "epoch": 0.6641000962463908,
      "grad_norm": 0.8912553191184998,
      "learning_rate": 0.00011770378389425586,
      "loss": 4.6717,
      "step": 32430
    },
    {
      "epoch": 0.6642024860237954,
      "grad_norm": 0.27865198254585266,
      "learning_rate": 0.0001176433423578004,
      "loss": 4.6275,
      "step": 32435
    },
    {
      "epoch": 0.6643048758012,
      "grad_norm": 0.33352676033973694,
      "learning_rate": 0.00011758290633076768,
      "loss": 4.6398,
      "step": 32440
    },
    {
      "epoch": 0.6644072655786046,
      "grad_norm": 0.2718777060508728,
      "learning_rate": 0.00011752247582344831,
      "loss": 4.6421,
      "step": 32445
    },
    {
      "epoch": 0.6645096553560093,
      "grad_norm": 0.2849903702735901,
      "learning_rate": 0.00011746205084613173,
      "loss": 4.6293,
      "step": 32450
    },
    {
      "epoch": 0.6646120451334139,
      "grad_norm": 0.27411556243896484,
      "learning_rate": 0.00011740163140910676,
      "loss": 4.6501,
      "step": 32455
    },
    {
      "epoch": 0.6647144349108185,
      "grad_norm": 0.27114495635032654,
      "learning_rate": 0.0001173412175226611,
      "loss": 4.6542,
      "step": 32460
    },
    {
      "epoch": 0.6648168246882231,
      "grad_norm": 0.2645331919193268,
      "learning_rate": 0.00011728080919708144,
      "loss": 4.6172,
      "step": 32465
    },
    {
      "epoch": 0.6649192144656277,
      "grad_norm": 0.2525201737880707,
      "learning_rate": 0.00011722040644265365,
      "loss": 4.6177,
      "step": 32470
    },
    {
      "epoch": 0.6650216042430324,
      "grad_norm": 0.3081844747066498,
      "learning_rate": 0.00011716000926966264,
      "loss": 4.693,
      "step": 32475
    },
    {
      "epoch": 0.665123994020437,
      "grad_norm": 0.37288954854011536,
      "learning_rate": 0.00011709961768839231,
      "loss": 4.6399,
      "step": 32480
    },
    {
      "epoch": 0.6652263837978416,
      "grad_norm": 0.3004577159881592,
      "learning_rate": 0.00011703923170912562,
      "loss": 4.6216,
      "step": 32485
    },
    {
      "epoch": 0.6653287735752462,
      "grad_norm": 0.27851006388664246,
      "learning_rate": 0.00011697885134214468,
      "loss": 4.6834,
      "step": 32490
    },
    {
      "epoch": 0.6654311633526508,
      "grad_norm": 0.2763761579990387,
      "learning_rate": 0.0001169184765977304,
      "loss": 4.4737,
      "step": 32495
    },
    {
      "epoch": 0.6655335531300555,
      "grad_norm": 0.2804718315601349,
      "learning_rate": 0.000116858107486163,
      "loss": 4.6298,
      "step": 32500
    },
    {
      "epoch": 0.6656359429074601,
      "grad_norm": 0.25102970004081726,
      "learning_rate": 0.0001167977440177217,
      "loss": 4.6373,
      "step": 32505
    },
    {
      "epoch": 0.6657383326848647,
      "grad_norm": 0.2638339698314667,
      "learning_rate": 0.0001167373862026845,
      "loss": 4.6466,
      "step": 32510
    },
    {
      "epoch": 0.6658407224622693,
      "grad_norm": 0.276727557182312,
      "learning_rate": 0.00011667703405132877,
      "loss": 4.6038,
      "step": 32515
    },
    {
      "epoch": 0.665943112239674,
      "grad_norm": 0.26491910219192505,
      "learning_rate": 0.00011661668757393073,
      "loss": 4.6227,
      "step": 32520
    },
    {
      "epoch": 0.6660455020170786,
      "grad_norm": 0.26587238907814026,
      "learning_rate": 0.00011655634678076566,
      "loss": 4.6304,
      "step": 32525
    },
    {
      "epoch": 0.6661478917944832,
      "grad_norm": 0.2602909803390503,
      "learning_rate": 0.00011649601168210791,
      "loss": 4.6159,
      "step": 32530
    },
    {
      "epoch": 0.6662502815718878,
      "grad_norm": 0.2717461585998535,
      "learning_rate": 0.00011643568228823087,
      "loss": 4.7061,
      "step": 32535
    },
    {
      "epoch": 0.6663526713492924,
      "grad_norm": 0.2562514841556549,
      "learning_rate": 0.00011637535860940687,
      "loss": 4.6409,
      "step": 32540
    },
    {
      "epoch": 0.6664550611266971,
      "grad_norm": 0.28094133734703064,
      "learning_rate": 0.00011631504065590727,
      "loss": 4.615,
      "step": 32545
    },
    {
      "epoch": 0.6665574509041018,
      "grad_norm": 0.27160724997520447,
      "learning_rate": 0.00011625472843800265,
      "loss": 4.646,
      "step": 32550
    },
    {
      "epoch": 0.6666598406815064,
      "grad_norm": 0.2829086482524872,
      "learning_rate": 0.00011619442196596248,
      "loss": 4.6315,
      "step": 32555
    },
    {
      "epoch": 0.666762230458911,
      "grad_norm": 0.6439187526702881,
      "learning_rate": 0.00011613412125005509,
      "loss": 4.6353,
      "step": 32560
    },
    {
      "epoch": 0.6668646202363157,
      "grad_norm": 0.31718677282333374,
      "learning_rate": 0.00011607382630054808,
      "loss": 4.6793,
      "step": 32565
    },
    {
      "epoch": 0.6669670100137203,
      "grad_norm": 0.2708340585231781,
      "learning_rate": 0.000116013537127708,
      "loss": 4.6164,
      "step": 32570
    },
    {
      "epoch": 0.6670693997911249,
      "grad_norm": 0.2751806378364563,
      "learning_rate": 0.0001159532537418003,
      "loss": 4.6222,
      "step": 32575
    },
    {
      "epoch": 0.6671717895685295,
      "grad_norm": 0.273276150226593,
      "learning_rate": 0.0001158929761530896,
      "loss": 4.6342,
      "step": 32580
    },
    {
      "epoch": 0.6672741793459341,
      "grad_norm": 0.2833123505115509,
      "learning_rate": 0.00011583270437183953,
      "loss": 4.6526,
      "step": 32585
    },
    {
      "epoch": 0.6673765691233388,
      "grad_norm": 0.2649253010749817,
      "learning_rate": 0.00011577243840831248,
      "loss": 4.6608,
      "step": 32590
    },
    {
      "epoch": 0.6674789589007434,
      "grad_norm": 0.2850389778614044,
      "learning_rate": 0.00011571217827277018,
      "loss": 4.6305,
      "step": 32595
    },
    {
      "epoch": 0.667581348678148,
      "grad_norm": 0.2866162061691284,
      "learning_rate": 0.00011565192397547326,
      "loss": 4.7226,
      "step": 32600
    },
    {
      "epoch": 0.6676837384555526,
      "grad_norm": 0.2693106532096863,
      "learning_rate": 0.0001155916755266812,
      "loss": 4.6469,
      "step": 32605
    },
    {
      "epoch": 0.6677861282329572,
      "grad_norm": 0.2595615088939667,
      "learning_rate": 0.00011553143293665264,
      "loss": 4.6919,
      "step": 32610
    },
    {
      "epoch": 0.6678885180103619,
      "grad_norm": 0.2672061324119568,
      "learning_rate": 0.00011547119621564525,
      "loss": 4.6689,
      "step": 32615
    },
    {
      "epoch": 0.6679909077877665,
      "grad_norm": 0.276113897562027,
      "learning_rate": 0.00011541096537391555,
      "loss": 4.6552,
      "step": 32620
    },
    {
      "epoch": 0.6680932975651711,
      "grad_norm": 0.2687895894050598,
      "learning_rate": 0.00011535074042171921,
      "loss": 4.6789,
      "step": 32625
    },
    {
      "epoch": 0.6681956873425757,
      "grad_norm": 0.2709292769432068,
      "learning_rate": 0.00011529052136931089,
      "loss": 4.6898,
      "step": 32630
    },
    {
      "epoch": 0.6682980771199803,
      "grad_norm": 0.28728988766670227,
      "learning_rate": 0.00011523030822694403,
      "loss": 4.668,
      "step": 32635
    },
    {
      "epoch": 0.668400466897385,
      "grad_norm": 0.2673599421977997,
      "learning_rate": 0.00011517010100487129,
      "loss": 4.604,
      "step": 32640
    },
    {
      "epoch": 0.6685028566747896,
      "grad_norm": 0.26948410272598267,
      "learning_rate": 0.00011510989971334435,
      "loss": 4.6317,
      "step": 32645
    },
    {
      "epoch": 0.6686052464521942,
      "grad_norm": 0.25881901383399963,
      "learning_rate": 0.00011504970436261367,
      "loss": 4.614,
      "step": 32650
    },
    {
      "epoch": 0.6687076362295988,
      "grad_norm": 0.29786792397499084,
      "learning_rate": 0.00011498951496292886,
      "loss": 4.5969,
      "step": 32655
    },
    {
      "epoch": 0.6688100260070035,
      "grad_norm": 0.2857189178466797,
      "learning_rate": 0.00011492933152453849,
      "loss": 4.6861,
      "step": 32660
    },
    {
      "epoch": 0.6689124157844081,
      "grad_norm": 0.2728564143180847,
      "learning_rate": 0.00011486915405769003,
      "loss": 4.6531,
      "step": 32665
    },
    {
      "epoch": 0.6690148055618127,
      "grad_norm": 0.26709920167922974,
      "learning_rate": 0.00011480898257263005,
      "loss": 4.615,
      "step": 32670
    },
    {
      "epoch": 0.6691171953392173,
      "grad_norm": 0.32166990637779236,
      "learning_rate": 0.00011474881707960411,
      "loss": 4.6531,
      "step": 32675
    },
    {
      "epoch": 0.6692195851166219,
      "grad_norm": 0.2602331340312958,
      "learning_rate": 0.00011468865758885653,
      "loss": 4.6367,
      "step": 32680
    },
    {
      "epoch": 0.6693219748940266,
      "grad_norm": 0.2882189154624939,
      "learning_rate": 0.00011462850411063084,
      "loss": 4.6099,
      "step": 32685
    },
    {
      "epoch": 0.6694243646714312,
      "grad_norm": 0.26194214820861816,
      "learning_rate": 0.0001145683566551696,
      "loss": 4.6382,
      "step": 32690
    },
    {
      "epoch": 0.6695267544488358,
      "grad_norm": 0.25957241654396057,
      "learning_rate": 0.00011450821523271404,
      "loss": 4.6092,
      "step": 32695
    },
    {
      "epoch": 0.6696291442262404,
      "grad_norm": 0.26672765612602234,
      "learning_rate": 0.00011444807985350463,
      "loss": 4.6781,
      "step": 32700
    },
    {
      "epoch": 0.669731534003645,
      "grad_norm": 0.2787538468837738,
      "learning_rate": 0.00011438795052778069,
      "loss": 4.6599,
      "step": 32705
    },
    {
      "epoch": 0.6698339237810497,
      "grad_norm": 0.25827673077583313,
      "learning_rate": 0.00011432782726578058,
      "loss": 4.6233,
      "step": 32710
    },
    {
      "epoch": 0.6699363135584543,
      "grad_norm": 0.2902181148529053,
      "learning_rate": 0.00011426771007774154,
      "loss": 4.5983,
      "step": 32715
    },
    {
      "epoch": 0.6700387033358589,
      "grad_norm": 0.27394747734069824,
      "learning_rate": 0.00011420759897389986,
      "loss": 4.6603,
      "step": 32720
    },
    {
      "epoch": 0.6701410931132635,
      "grad_norm": 0.2899582087993622,
      "learning_rate": 0.00011414749396449081,
      "loss": 4.6078,
      "step": 32725
    },
    {
      "epoch": 0.6702434828906682,
      "grad_norm": 0.27544349431991577,
      "learning_rate": 0.00011408739505974846,
      "loss": 4.6579,
      "step": 32730
    },
    {
      "epoch": 0.6703458726680728,
      "grad_norm": 0.2853153944015503,
      "learning_rate": 0.00011402730226990598,
      "loss": 4.6325,
      "step": 32735
    },
    {
      "epoch": 0.6704482624454774,
      "grad_norm": 0.27195265889167786,
      "learning_rate": 0.00011396721560519558,
      "loss": 4.6751,
      "step": 32740
    },
    {
      "epoch": 0.670550652222882,
      "grad_norm": 0.2665818929672241,
      "learning_rate": 0.00011390713507584815,
      "loss": 4.6516,
      "step": 32745
    },
    {
      "epoch": 0.6706530420002866,
      "grad_norm": 0.26640045642852783,
      "learning_rate": 0.0001138470606920938,
      "loss": 4.6279,
      "step": 32750
    },
    {
      "epoch": 0.6707554317776914,
      "grad_norm": 0.26836442947387695,
      "learning_rate": 0.00011378699246416152,
      "loss": 4.6299,
      "step": 32755
    },
    {
      "epoch": 0.670857821555096,
      "grad_norm": 0.26381194591522217,
      "learning_rate": 0.00011372693040227916,
      "loss": 4.6103,
      "step": 32760
    },
    {
      "epoch": 0.6709602113325006,
      "grad_norm": 0.27477508783340454,
      "learning_rate": 0.00011366687451667363,
      "loss": 4.6513,
      "step": 32765
    },
    {
      "epoch": 0.6710626011099052,
      "grad_norm": 0.26479724049568176,
      "learning_rate": 0.00011360682481757081,
      "loss": 4.6292,
      "step": 32770
    },
    {
      "epoch": 0.6711649908873099,
      "grad_norm": 0.2636280059814453,
      "learning_rate": 0.0001135467813151953,
      "loss": 4.6685,
      "step": 32775
    },
    {
      "epoch": 0.6712673806647145,
      "grad_norm": 0.26332464814186096,
      "learning_rate": 0.00011348674401977089,
      "loss": 4.6522,
      "step": 32780
    },
    {
      "epoch": 0.6713697704421191,
      "grad_norm": 0.25292447209358215,
      "learning_rate": 0.00011342671294152031,
      "loss": 4.6101,
      "step": 32785
    },
    {
      "epoch": 0.6714721602195237,
      "grad_norm": 0.2679852247238159,
      "learning_rate": 0.00011336668809066505,
      "loss": 4.6559,
      "step": 32790
    },
    {
      "epoch": 0.6715745499969283,
      "grad_norm": 0.2763931453227997,
      "learning_rate": 0.00011330666947742567,
      "loss": 4.6216,
      "step": 32795
    },
    {
      "epoch": 0.671676939774333,
      "grad_norm": 0.2601395547389984,
      "learning_rate": 0.00011324665711202168,
      "loss": 4.6413,
      "step": 32800
    },
    {
      "epoch": 0.6717793295517376,
      "grad_norm": 0.31791824102401733,
      "learning_rate": 0.00011318665100467142,
      "loss": 4.6098,
      "step": 32805
    },
    {
      "epoch": 0.6718817193291422,
      "grad_norm": 0.2943015396595001,
      "learning_rate": 0.00011312665116559227,
      "loss": 4.6116,
      "step": 32810
    },
    {
      "epoch": 0.6719841091065468,
      "grad_norm": 0.2961193919181824,
      "learning_rate": 0.00011306665760500056,
      "loss": 4.6739,
      "step": 32815
    },
    {
      "epoch": 0.6720864988839514,
      "grad_norm": 0.26749667525291443,
      "learning_rate": 0.00011300667033311137,
      "loss": 4.6054,
      "step": 32820
    },
    {
      "epoch": 0.6721888886613561,
      "grad_norm": 0.2678840458393097,
      "learning_rate": 0.0001129466893601389,
      "loss": 4.6751,
      "step": 32825
    },
    {
      "epoch": 0.6722912784387607,
      "grad_norm": 0.2867529094219208,
      "learning_rate": 0.00011288671469629626,
      "loss": 4.6906,
      "step": 32830
    },
    {
      "epoch": 0.6723936682161653,
      "grad_norm": 0.2722114026546478,
      "learning_rate": 0.00011282674635179534,
      "loss": 4.6143,
      "step": 32835
    },
    {
      "epoch": 0.6724960579935699,
      "grad_norm": 0.2897540032863617,
      "learning_rate": 0.00011276678433684711,
      "loss": 4.6943,
      "step": 32840
    },
    {
      "epoch": 0.6725984477709746,
      "grad_norm": 0.27606886625289917,
      "learning_rate": 0.00011270682866166139,
      "loss": 4.658,
      "step": 32845
    },
    {
      "epoch": 0.6727008375483792,
      "grad_norm": 0.2667522430419922,
      "learning_rate": 0.00011264687933644695,
      "loss": 4.6368,
      "step": 32850
    },
    {
      "epoch": 0.6728032273257838,
      "grad_norm": 0.2723812758922577,
      "learning_rate": 0.00011258693637141145,
      "loss": 4.6668,
      "step": 32855
    },
    {
      "epoch": 0.6729056171031884,
      "grad_norm": 0.27237269282341003,
      "learning_rate": 0.00011252699977676146,
      "loss": 4.6197,
      "step": 32860
    },
    {
      "epoch": 0.673008006880593,
      "grad_norm": 0.2706470191478729,
      "learning_rate": 0.00011246706956270257,
      "loss": 4.6622,
      "step": 32865
    },
    {
      "epoch": 0.6731103966579977,
      "grad_norm": 0.2632223963737488,
      "learning_rate": 0.00011240714573943907,
      "loss": 4.6628,
      "step": 32870
    },
    {
      "epoch": 0.6732127864354023,
      "grad_norm": 0.27151986956596375,
      "learning_rate": 0.00011234722831717432,
      "loss": 4.613,
      "step": 32875
    },
    {
      "epoch": 0.6733151762128069,
      "grad_norm": 0.2898189425468445,
      "learning_rate": 0.00011228731730611071,
      "loss": 4.682,
      "step": 32880
    },
    {
      "epoch": 0.6734175659902115,
      "grad_norm": 0.2605278491973877,
      "learning_rate": 0.00011222741271644919,
      "loss": 4.6921,
      "step": 32885
    },
    {
      "epoch": 0.6735199557676161,
      "grad_norm": 0.26354148983955383,
      "learning_rate": 0.0001121675145583899,
      "loss": 4.6374,
      "step": 32890
    },
    {
      "epoch": 0.6736223455450208,
      "grad_norm": 0.2641417980194092,
      "learning_rate": 0.00011210762284213185,
      "loss": 4.6463,
      "step": 32895
    },
    {
      "epoch": 0.6737247353224254,
      "grad_norm": 0.2970120906829834,
      "learning_rate": 0.00011204773757787282,
      "loss": 4.6127,
      "step": 32900
    },
    {
      "epoch": 0.67382712509983,
      "grad_norm": 0.2714480757713318,
      "learning_rate": 0.00011198785877580959,
      "loss": 4.5852,
      "step": 32905
    },
    {
      "epoch": 0.6739295148772346,
      "grad_norm": 0.2645569145679474,
      "learning_rate": 0.00011192798644613792,
      "loss": 4.5893,
      "step": 32910
    },
    {
      "epoch": 0.6740319046546392,
      "grad_norm": 0.2686305046081543,
      "learning_rate": 0.00011186812059905222,
      "loss": 4.6661,
      "step": 32915
    },
    {
      "epoch": 0.6741342944320439,
      "grad_norm": 0.27101930975914,
      "learning_rate": 0.00011180826124474606,
      "loss": 4.6242,
      "step": 32920
    },
    {
      "epoch": 0.6742366842094485,
      "grad_norm": 0.2710999846458435,
      "learning_rate": 0.00011174840839341177,
      "loss": 4.6523,
      "step": 32925
    },
    {
      "epoch": 0.6743390739868531,
      "grad_norm": 0.5384253263473511,
      "learning_rate": 0.00011168856205524054,
      "loss": 4.6621,
      "step": 32930
    },
    {
      "epoch": 0.6744414637642577,
      "grad_norm": 0.2712736427783966,
      "learning_rate": 0.00011162872224042257,
      "loss": 4.6247,
      "step": 32935
    },
    {
      "epoch": 0.6745438535416624,
      "grad_norm": 0.2693881392478943,
      "learning_rate": 0.0001115688889591469,
      "loss": 4.6727,
      "step": 32940
    },
    {
      "epoch": 0.674646243319067,
      "grad_norm": 0.28540852665901184,
      "learning_rate": 0.00011150906222160136,
      "loss": 4.625,
      "step": 32945
    },
    {
      "epoch": 0.6747486330964716,
      "grad_norm": 0.3191269636154175,
      "learning_rate": 0.00011144924203797283,
      "loss": 4.6696,
      "step": 32950
    },
    {
      "epoch": 0.6748510228738763,
      "grad_norm": 0.26214319467544556,
      "learning_rate": 0.00011138942841844703,
      "loss": 4.6142,
      "step": 32955
    },
    {
      "epoch": 0.674953412651281,
      "grad_norm": 0.28582412004470825,
      "learning_rate": 0.00011132962137320837,
      "loss": 4.6199,
      "step": 32960
    },
    {
      "epoch": 0.6750558024286856,
      "grad_norm": 0.26200976967811584,
      "learning_rate": 0.00011126982091244041,
      "loss": 4.6284,
      "step": 32965
    },
    {
      "epoch": 0.6751581922060902,
      "grad_norm": 0.2682289481163025,
      "learning_rate": 0.00011121002704632548,
      "loss": 4.6267,
      "step": 32970
    },
    {
      "epoch": 0.6752605819834948,
      "grad_norm": 0.26642361283302307,
      "learning_rate": 0.0001111502397850447,
      "loss": 4.6857,
      "step": 32975
    },
    {
      "epoch": 0.6753629717608994,
      "grad_norm": 0.27904707193374634,
      "learning_rate": 0.00011109045913877825,
      "loss": 4.6061,
      "step": 32980
    },
    {
      "epoch": 0.675465361538304,
      "grad_norm": 0.26362839341163635,
      "learning_rate": 0.00011103068511770499,
      "loss": 4.6184,
      "step": 32985
    },
    {
      "epoch": 0.6755677513157087,
      "grad_norm": 0.2662644386291504,
      "learning_rate": 0.00011097091773200285,
      "loss": 4.6761,
      "step": 32990
    },
    {
      "epoch": 0.6756701410931133,
      "grad_norm": 0.28967827558517456,
      "learning_rate": 0.00011091115699184843,
      "loss": 4.6337,
      "step": 32995
    },
    {
      "epoch": 0.6757725308705179,
      "grad_norm": 0.2763548791408539,
      "learning_rate": 0.00011085140290741731,
      "loss": 4.6565,
      "step": 33000
    },
    {
      "epoch": 0.6757725308705179,
      "eval_loss": 4.649252891540527,
      "eval_runtime": 17.1281,
      "eval_samples_per_second": 33.979,
      "eval_steps_per_second": 4.262,
      "step": 33000
    },
    {
      "epoch": 0.6758749206479225,
      "grad_norm": 0.26872551441192627,
      "learning_rate": 0.00011079165548888403,
      "loss": 4.6717,
      "step": 33005
    },
    {
      "epoch": 0.6759773104253272,
      "grad_norm": 0.3147912323474884,
      "learning_rate": 0.00011073191474642168,
      "loss": 4.6426,
      "step": 33010
    },
    {
      "epoch": 0.6760797002027318,
      "grad_norm": 0.2672083377838135,
      "learning_rate": 0.00011067218069020256,
      "loss": 4.5943,
      "step": 33015
    },
    {
      "epoch": 0.6761820899801364,
      "grad_norm": 0.304654985666275,
      "learning_rate": 0.00011061245333039767,
      "loss": 4.6745,
      "step": 33020
    },
    {
      "epoch": 0.676284479757541,
      "grad_norm": 0.2691044807434082,
      "learning_rate": 0.00011055273267717684,
      "loss": 4.6056,
      "step": 33025
    },
    {
      "epoch": 0.6763868695349456,
      "grad_norm": 0.2770857810974121,
      "learning_rate": 0.00011049301874070886,
      "loss": 4.6658,
      "step": 33030
    },
    {
      "epoch": 0.6764892593123503,
      "grad_norm": 0.2694748640060425,
      "learning_rate": 0.0001104333115311613,
      "loss": 4.6105,
      "step": 33035
    },
    {
      "epoch": 0.6765916490897549,
      "grad_norm": 0.2787577211856842,
      "learning_rate": 0.00011037361105870058,
      "loss": 4.6421,
      "step": 33040
    },
    {
      "epoch": 0.6766940388671595,
      "grad_norm": 0.27521374821662903,
      "learning_rate": 0.00011031391733349206,
      "loss": 4.6528,
      "step": 33045
    },
    {
      "epoch": 0.6767964286445641,
      "grad_norm": 0.26650211215019226,
      "learning_rate": 0.00011025423036569988,
      "loss": 4.6218,
      "step": 33050
    },
    {
      "epoch": 0.6768988184219688,
      "grad_norm": 0.26677361130714417,
      "learning_rate": 0.000110194550165487,
      "loss": 4.6694,
      "step": 33055
    },
    {
      "epoch": 0.6770012081993734,
      "grad_norm": 0.262565940618515,
      "learning_rate": 0.00011013487674301526,
      "loss": 4.6456,
      "step": 33060
    },
    {
      "epoch": 0.677103597976778,
      "grad_norm": 0.2637573778629303,
      "learning_rate": 0.00011007521010844543,
      "loss": 4.642,
      "step": 33065
    },
    {
      "epoch": 0.6772059877541826,
      "grad_norm": 0.28802940249443054,
      "learning_rate": 0.00011001555027193698,
      "loss": 4.6428,
      "step": 33070
    },
    {
      "epoch": 0.6773083775315872,
      "grad_norm": 0.26708462834358215,
      "learning_rate": 0.0001099558972436483,
      "loss": 4.648,
      "step": 33075
    },
    {
      "epoch": 0.6774107673089919,
      "grad_norm": 0.2635321021080017,
      "learning_rate": 0.00010989625103373665,
      "loss": 4.6278,
      "step": 33080
    },
    {
      "epoch": 0.6775131570863965,
      "grad_norm": 0.27121537923812866,
      "learning_rate": 0.00010983661165235806,
      "loss": 4.6934,
      "step": 33085
    },
    {
      "epoch": 0.6776155468638011,
      "grad_norm": 0.2633886933326721,
      "learning_rate": 0.00010977697910966741,
      "loss": 4.6117,
      "step": 33090
    },
    {
      "epoch": 0.6777179366412057,
      "grad_norm": 0.2765393555164337,
      "learning_rate": 0.00010971735341581854,
      "loss": 4.6651,
      "step": 33095
    },
    {
      "epoch": 0.6778203264186103,
      "grad_norm": 0.2615356147289276,
      "learning_rate": 0.00010965773458096387,
      "loss": 4.6445,
      "step": 33100
    },
    {
      "epoch": 0.677922716196015,
      "grad_norm": 0.2716805636882782,
      "learning_rate": 0.00010959812261525484,
      "loss": 4.6622,
      "step": 33105
    },
    {
      "epoch": 0.6780251059734196,
      "grad_norm": 0.2718418836593628,
      "learning_rate": 0.00010953851752884178,
      "loss": 4.5962,
      "step": 33110
    },
    {
      "epoch": 0.6781274957508242,
      "grad_norm": 0.2732146084308624,
      "learning_rate": 0.00010947891933187362,
      "loss": 4.6422,
      "step": 33115
    },
    {
      "epoch": 0.6782298855282288,
      "grad_norm": 0.25753968954086304,
      "learning_rate": 0.0001094193280344983,
      "loss": 4.556,
      "step": 33120
    },
    {
      "epoch": 0.6783322753056334,
      "grad_norm": 0.2574491798877716,
      "learning_rate": 0.00010935974364686253,
      "loss": 4.636,
      "step": 33125
    },
    {
      "epoch": 0.6784346650830381,
      "grad_norm": 0.5101560354232788,
      "learning_rate": 0.00010930016617911188,
      "loss": 4.6327,
      "step": 33130
    },
    {
      "epoch": 0.6785370548604427,
      "grad_norm": 0.2720508873462677,
      "learning_rate": 0.00010924059564139067,
      "loss": 4.6171,
      "step": 33135
    },
    {
      "epoch": 0.6786394446378473,
      "grad_norm": 0.2660973072052002,
      "learning_rate": 0.00010918103204384204,
      "loss": 4.6107,
      "step": 33140
    },
    {
      "epoch": 0.6787418344152519,
      "grad_norm": 0.26955458521842957,
      "learning_rate": 0.00010912147539660808,
      "loss": 4.622,
      "step": 33145
    },
    {
      "epoch": 0.6788442241926566,
      "grad_norm": 0.2750113606452942,
      "learning_rate": 0.0001090619257098295,
      "loss": 4.6351,
      "step": 33150
    },
    {
      "epoch": 0.6789466139700612,
      "grad_norm": 0.28483280539512634,
      "learning_rate": 0.00010900238299364595,
      "loss": 4.6206,
      "step": 33155
    },
    {
      "epoch": 0.6790490037474659,
      "grad_norm": 0.26131486892700195,
      "learning_rate": 0.00010894284725819595,
      "loss": 4.6404,
      "step": 33160
    },
    {
      "epoch": 0.6791513935248705,
      "grad_norm": 0.26914942264556885,
      "learning_rate": 0.00010888331851361663,
      "loss": 4.6868,
      "step": 33165
    },
    {
      "epoch": 0.6792537833022751,
      "grad_norm": 0.28682154417037964,
      "learning_rate": 0.00010882379677004412,
      "loss": 4.6411,
      "step": 33170
    },
    {
      "epoch": 0.6793561730796798,
      "grad_norm": 0.26749324798583984,
      "learning_rate": 0.00010876428203761329,
      "loss": 4.6421,
      "step": 33175
    },
    {
      "epoch": 0.6794585628570844,
      "grad_norm": 0.26501280069351196,
      "learning_rate": 0.00010870477432645779,
      "loss": 4.6195,
      "step": 33180
    },
    {
      "epoch": 0.679560952634489,
      "grad_norm": 0.26755979657173157,
      "learning_rate": 0.00010864527364671011,
      "loss": 4.6563,
      "step": 33185
    },
    {
      "epoch": 0.6796633424118936,
      "grad_norm": 0.2696143686771393,
      "learning_rate": 0.00010858578000850157,
      "loss": 4.5667,
      "step": 33190
    },
    {
      "epoch": 0.6797657321892983,
      "grad_norm": 0.26739928126335144,
      "learning_rate": 0.00010852629342196217,
      "loss": 4.6025,
      "step": 33195
    },
    {
      "epoch": 0.6798681219667029,
      "grad_norm": 0.27433258295059204,
      "learning_rate": 0.00010846681389722083,
      "loss": 4.669,
      "step": 33200
    },
    {
      "epoch": 0.6799705117441075,
      "grad_norm": 0.2671515941619873,
      "learning_rate": 0.00010840734144440524,
      "loss": 4.6127,
      "step": 33205
    },
    {
      "epoch": 0.6800729015215121,
      "grad_norm": 0.26458939909935,
      "learning_rate": 0.00010834787607364187,
      "loss": 4.6186,
      "step": 33210
    },
    {
      "epoch": 0.6801752912989167,
      "grad_norm": 0.274859219789505,
      "learning_rate": 0.00010828841779505596,
      "loss": 4.5913,
      "step": 33215
    },
    {
      "epoch": 0.6802776810763214,
      "grad_norm": 0.29292112588882446,
      "learning_rate": 0.00010822896661877165,
      "loss": 4.5322,
      "step": 33220
    },
    {
      "epoch": 0.680380070853726,
      "grad_norm": 0.2590983808040619,
      "learning_rate": 0.00010816952255491167,
      "loss": 4.6205,
      "step": 33225
    },
    {
      "epoch": 0.6804824606311306,
      "grad_norm": 0.27183979749679565,
      "learning_rate": 0.00010811008561359777,
      "loss": 4.5831,
      "step": 33230
    },
    {
      "epoch": 0.6805848504085352,
      "grad_norm": 0.2869648337364197,
      "learning_rate": 0.0001080506558049504,
      "loss": 4.6256,
      "step": 33235
    },
    {
      "epoch": 0.6806872401859398,
      "grad_norm": 0.2832396626472473,
      "learning_rate": 0.00010799123313908863,
      "loss": 4.6307,
      "step": 33240
    },
    {
      "epoch": 0.6807896299633445,
      "grad_norm": 0.2963630259037018,
      "learning_rate": 0.00010793181762613053,
      "loss": 4.6324,
      "step": 33245
    },
    {
      "epoch": 0.6808920197407491,
      "grad_norm": 0.2685511112213135,
      "learning_rate": 0.00010787240927619295,
      "loss": 4.5991,
      "step": 33250
    },
    {
      "epoch": 0.6809944095181537,
      "grad_norm": 0.5867858529090881,
      "learning_rate": 0.00010781300809939135,
      "loss": 4.6226,
      "step": 33255
    },
    {
      "epoch": 0.6810967992955583,
      "grad_norm": 0.268569678068161,
      "learning_rate": 0.00010775361410584011,
      "loss": 4.6311,
      "step": 33260
    },
    {
      "epoch": 0.681199189072963,
      "grad_norm": 0.26071876287460327,
      "learning_rate": 0.00010769422730565234,
      "loss": 4.6936,
      "step": 33265
    },
    {
      "epoch": 0.6813015788503676,
      "grad_norm": 0.2950969636440277,
      "learning_rate": 0.00010763484770894,
      "loss": 4.6456,
      "step": 33270
    },
    {
      "epoch": 0.6814039686277722,
      "grad_norm": 0.29240158200263977,
      "learning_rate": 0.00010757547532581362,
      "loss": 4.653,
      "step": 33275
    },
    {
      "epoch": 0.6815063584051768,
      "grad_norm": 0.26814985275268555,
      "learning_rate": 0.00010751611016638274,
      "loss": 4.6649,
      "step": 33280
    },
    {
      "epoch": 0.6816087481825814,
      "grad_norm": 0.29863035678863525,
      "learning_rate": 0.00010745675224075558,
      "loss": 4.5942,
      "step": 33285
    },
    {
      "epoch": 0.6817111379599861,
      "grad_norm": 0.3239007890224457,
      "learning_rate": 0.00010739740155903899,
      "loss": 4.6163,
      "step": 33290
    },
    {
      "epoch": 0.6818135277373907,
      "grad_norm": 0.30782201886177063,
      "learning_rate": 0.0001073380581313388,
      "loss": 4.6345,
      "step": 33295
    },
    {
      "epoch": 0.6819159175147953,
      "grad_norm": 0.28785884380340576,
      "learning_rate": 0.00010727872196775951,
      "loss": 4.6283,
      "step": 33300
    },
    {
      "epoch": 0.6820183072921999,
      "grad_norm": 0.30212146043777466,
      "learning_rate": 0.00010721939307840436,
      "loss": 4.6903,
      "step": 33305
    },
    {
      "epoch": 0.6821206970696045,
      "grad_norm": 0.27822816371917725,
      "learning_rate": 0.00010716007147337542,
      "loss": 4.6804,
      "step": 33310
    },
    {
      "epoch": 0.6822230868470092,
      "grad_norm": 0.2746981978416443,
      "learning_rate": 0.00010710075716277348,
      "loss": 4.6548,
      "step": 33315
    },
    {
      "epoch": 0.6823254766244138,
      "grad_norm": 0.35819563269615173,
      "learning_rate": 0.00010704145015669804,
      "loss": 4.6212,
      "step": 33320
    },
    {
      "epoch": 0.6824278664018184,
      "grad_norm": 0.2573152780532837,
      "learning_rate": 0.00010698215046524744,
      "loss": 4.6293,
      "step": 33325
    },
    {
      "epoch": 0.682530256179223,
      "grad_norm": 0.2639943063259125,
      "learning_rate": 0.0001069228580985188,
      "loss": 4.6469,
      "step": 33330
    },
    {
      "epoch": 0.6826326459566276,
      "grad_norm": 0.26783081889152527,
      "learning_rate": 0.00010686357306660777,
      "loss": 4.6176,
      "step": 33335
    },
    {
      "epoch": 0.6827350357340323,
      "grad_norm": 0.2767474353313446,
      "learning_rate": 0.00010680429537960904,
      "loss": 4.6259,
      "step": 33340
    },
    {
      "epoch": 0.6828374255114369,
      "grad_norm": 0.271543949842453,
      "learning_rate": 0.0001067450250476159,
      "loss": 4.6345,
      "step": 33345
    },
    {
      "epoch": 0.6829398152888415,
      "grad_norm": 0.2669256627559662,
      "learning_rate": 0.00010668576208072036,
      "loss": 4.6357,
      "step": 33350
    },
    {
      "epoch": 0.6830422050662461,
      "grad_norm": 0.27458763122558594,
      "learning_rate": 0.00010662650648901328,
      "loss": 4.5843,
      "step": 33355
    },
    {
      "epoch": 0.6831445948436509,
      "grad_norm": 0.2733932137489319,
      "learning_rate": 0.00010656725828258422,
      "loss": 4.662,
      "step": 33360
    },
    {
      "epoch": 0.6832469846210555,
      "grad_norm": 0.32873204350471497,
      "learning_rate": 0.0001065080174715214,
      "loss": 4.6297,
      "step": 33365
    },
    {
      "epoch": 0.6833493743984601,
      "grad_norm": 0.27415570616722107,
      "learning_rate": 0.00010644878406591192,
      "loss": 4.6236,
      "step": 33370
    },
    {
      "epoch": 0.6834517641758647,
      "grad_norm": 0.2847776710987091,
      "learning_rate": 0.00010638955807584158,
      "loss": 4.622,
      "step": 33375
    },
    {
      "epoch": 0.6835541539532694,
      "grad_norm": 0.2700304388999939,
      "learning_rate": 0.00010633033951139478,
      "loss": 4.6632,
      "step": 33380
    },
    {
      "epoch": 0.683656543730674,
      "grad_norm": 0.2543303072452545,
      "learning_rate": 0.00010627112838265482,
      "loss": 4.6415,
      "step": 33385
    },
    {
      "epoch": 0.6837589335080786,
      "grad_norm": 0.2689971625804901,
      "learning_rate": 0.00010621192469970373,
      "loss": 4.6474,
      "step": 33390
    },
    {
      "epoch": 0.6838613232854832,
      "grad_norm": 0.2835336923599243,
      "learning_rate": 0.00010615272847262213,
      "loss": 4.6392,
      "step": 33395
    },
    {
      "epoch": 0.6839637130628878,
      "grad_norm": 0.26349833607673645,
      "learning_rate": 0.00010609353971148948,
      "loss": 4.6257,
      "step": 33400
    },
    {
      "epoch": 0.6840661028402925,
      "grad_norm": 0.2775791883468628,
      "learning_rate": 0.000106034358426384,
      "loss": 4.7107,
      "step": 33405
    },
    {
      "epoch": 0.6841684926176971,
      "grad_norm": 0.3532986640930176,
      "learning_rate": 0.00010597518462738261,
      "loss": 4.5854,
      "step": 33410
    },
    {
      "epoch": 0.6842708823951017,
      "grad_norm": 0.2774605453014374,
      "learning_rate": 0.00010591601832456084,
      "loss": 4.6145,
      "step": 33415
    },
    {
      "epoch": 0.6843732721725063,
      "grad_norm": 0.27627623081207275,
      "learning_rate": 0.00010585685952799308,
      "loss": 4.6395,
      "step": 33420
    },
    {
      "epoch": 0.6844756619499109,
      "grad_norm": 0.7018665075302124,
      "learning_rate": 0.00010579770824775246,
      "loss": 4.5492,
      "step": 33425
    },
    {
      "epoch": 0.6845780517273156,
      "grad_norm": 0.2698715627193451,
      "learning_rate": 0.00010573856449391065,
      "loss": 4.5794,
      "step": 33430
    },
    {
      "epoch": 0.6846804415047202,
      "grad_norm": 0.26496490836143494,
      "learning_rate": 0.00010567942827653822,
      "loss": 4.6022,
      "step": 33435
    },
    {
      "epoch": 0.6847828312821248,
      "grad_norm": 0.27712804079055786,
      "learning_rate": 0.00010562029960570443,
      "loss": 4.6484,
      "step": 33440
    },
    {
      "epoch": 0.6848852210595294,
      "grad_norm": 0.284860759973526,
      "learning_rate": 0.00010556117849147713,
      "loss": 4.619,
      "step": 33445
    },
    {
      "epoch": 0.684987610836934,
      "grad_norm": 0.26327624917030334,
      "learning_rate": 0.00010550206494392301,
      "loss": 4.6726,
      "step": 33450
    },
    {
      "epoch": 0.6850900006143387,
      "grad_norm": 0.26512905955314636,
      "learning_rate": 0.0001054429589731075,
      "loss": 4.646,
      "step": 33455
    },
    {
      "epoch": 0.6851923903917433,
      "grad_norm": 0.28669658303260803,
      "learning_rate": 0.00010538386058909458,
      "loss": 4.6625,
      "step": 33460
    },
    {
      "epoch": 0.6852947801691479,
      "grad_norm": 0.27869224548339844,
      "learning_rate": 0.00010532476980194706,
      "loss": 4.645,
      "step": 33465
    },
    {
      "epoch": 0.6853971699465525,
      "grad_norm": 0.28113362193107605,
      "learning_rate": 0.00010526568662172649,
      "loss": 4.615,
      "step": 33470
    },
    {
      "epoch": 0.6854995597239572,
      "grad_norm": 0.3861667811870575,
      "learning_rate": 0.00010520661105849297,
      "loss": 4.6463,
      "step": 33475
    },
    {
      "epoch": 0.6856019495013618,
      "grad_norm": 0.28890690207481384,
      "learning_rate": 0.0001051475431223054,
      "loss": 4.5874,
      "step": 33480
    },
    {
      "epoch": 0.6857043392787664,
      "grad_norm": 0.28581398725509644,
      "learning_rate": 0.0001050884828232215,
      "loss": 4.6466,
      "step": 33485
    },
    {
      "epoch": 0.685806729056171,
      "grad_norm": 0.2695760428905487,
      "learning_rate": 0.00010502943017129741,
      "loss": 4.6051,
      "step": 33490
    },
    {
      "epoch": 0.6859091188335756,
      "grad_norm": 0.28361669182777405,
      "learning_rate": 0.00010497038517658821,
      "loss": 4.6244,
      "step": 33495
    },
    {
      "epoch": 0.6860115086109803,
      "grad_norm": 0.2638608515262604,
      "learning_rate": 0.00010491134784914762,
      "loss": 4.6105,
      "step": 33500
    },
    {
      "epoch": 0.6861138983883849,
      "grad_norm": 0.2701319754123688,
      "learning_rate": 0.00010485231819902791,
      "loss": 4.6511,
      "step": 33505
    },
    {
      "epoch": 0.6862162881657895,
      "grad_norm": 0.2725416123867035,
      "learning_rate": 0.00010479329623628023,
      "loss": 4.6427,
      "step": 33510
    },
    {
      "epoch": 0.6863186779431941,
      "grad_norm": 0.26230761408805847,
      "learning_rate": 0.00010473428197095443,
      "loss": 4.5727,
      "step": 33515
    },
    {
      "epoch": 0.6864210677205987,
      "grad_norm": 0.27070945501327515,
      "learning_rate": 0.00010467527541309882,
      "loss": 4.6456,
      "step": 33520
    },
    {
      "epoch": 0.6865234574980034,
      "grad_norm": 0.2823569178581238,
      "learning_rate": 0.00010461627657276063,
      "loss": 4.644,
      "step": 33525
    },
    {
      "epoch": 0.686625847275408,
      "grad_norm": 0.2608167827129364,
      "learning_rate": 0.0001045572854599857,
      "loss": 4.5949,
      "step": 33530
    },
    {
      "epoch": 0.6867282370528126,
      "grad_norm": 0.32188713550567627,
      "learning_rate": 0.00010449830208481846,
      "loss": 4.646,
      "step": 33535
    },
    {
      "epoch": 0.6868306268302172,
      "grad_norm": 0.25692039728164673,
      "learning_rate": 0.0001044393264573022,
      "loss": 4.6223,
      "step": 33540
    },
    {
      "epoch": 0.6869330166076218,
      "grad_norm": 0.2823508083820343,
      "learning_rate": 0.00010438035858747876,
      "loss": 4.6348,
      "step": 33545
    },
    {
      "epoch": 0.6870354063850265,
      "grad_norm": 0.285574346780777,
      "learning_rate": 0.00010432139848538875,
      "loss": 4.6248,
      "step": 33550
    },
    {
      "epoch": 0.6871377961624311,
      "grad_norm": 0.2840411067008972,
      "learning_rate": 0.00010426244616107133,
      "loss": 4.5983,
      "step": 33555
    },
    {
      "epoch": 0.6872401859398357,
      "grad_norm": 0.27879831194877625,
      "learning_rate": 0.00010420350162456444,
      "loss": 4.5808,
      "step": 33560
    },
    {
      "epoch": 0.6873425757172404,
      "grad_norm": 0.2666771113872528,
      "learning_rate": 0.00010414456488590476,
      "loss": 4.6369,
      "step": 33565
    },
    {
      "epoch": 0.6874449654946451,
      "grad_norm": 0.2672293186187744,
      "learning_rate": 0.00010408563595512738,
      "loss": 4.656,
      "step": 33570
    },
    {
      "epoch": 0.6875473552720497,
      "grad_norm": 0.2605060040950775,
      "learning_rate": 0.0001040267148422663,
      "loss": 4.627,
      "step": 33575
    },
    {
      "epoch": 0.6876497450494543,
      "grad_norm": 0.26467475295066833,
      "learning_rate": 0.00010396780155735417,
      "loss": 4.6612,
      "step": 33580
    },
    {
      "epoch": 0.6877521348268589,
      "grad_norm": 0.2730306386947632,
      "learning_rate": 0.00010390889611042218,
      "loss": 4.6743,
      "step": 33585
    },
    {
      "epoch": 0.6878545246042636,
      "grad_norm": 0.2668982744216919,
      "learning_rate": 0.00010384999851150029,
      "loss": 4.6168,
      "step": 33590
    },
    {
      "epoch": 0.6879569143816682,
      "grad_norm": 0.2712850272655487,
      "learning_rate": 0.00010379110877061716,
      "loss": 4.5973,
      "step": 33595
    },
    {
      "epoch": 0.6880593041590728,
      "grad_norm": 0.27507054805755615,
      "learning_rate": 0.00010373222689779988,
      "loss": 4.5955,
      "step": 33600
    },
    {
      "epoch": 0.6881616939364774,
      "grad_norm": 0.30687814950942993,
      "learning_rate": 0.0001036733529030745,
      "loss": 4.6138,
      "step": 33605
    },
    {
      "epoch": 0.688264083713882,
      "grad_norm": 0.2817098796367645,
      "learning_rate": 0.00010361448679646563,
      "loss": 4.5709,
      "step": 33610
    },
    {
      "epoch": 0.6883664734912867,
      "grad_norm": 0.3625824451446533,
      "learning_rate": 0.00010355562858799636,
      "loss": 4.6557,
      "step": 33615
    },
    {
      "epoch": 0.6884688632686913,
      "grad_norm": 0.26989036798477173,
      "learning_rate": 0.00010349677828768865,
      "loss": 4.6515,
      "step": 33620
    },
    {
      "epoch": 0.6885712530460959,
      "grad_norm": 0.2847817838191986,
      "learning_rate": 0.00010343793590556308,
      "loss": 4.6436,
      "step": 33625
    },
    {
      "epoch": 0.6886736428235005,
      "grad_norm": 0.28808367252349854,
      "learning_rate": 0.00010337910145163876,
      "loss": 4.6747,
      "step": 33630
    },
    {
      "epoch": 0.6887760326009051,
      "grad_norm": 0.2900857627391815,
      "learning_rate": 0.00010332027493593359,
      "loss": 4.6243,
      "step": 33635
    },
    {
      "epoch": 0.6888784223783098,
      "grad_norm": 0.42385783791542053,
      "learning_rate": 0.00010326145636846408,
      "loss": 4.648,
      "step": 33640
    },
    {
      "epoch": 0.6889808121557144,
      "grad_norm": 0.2676391303539276,
      "learning_rate": 0.00010320264575924525,
      "loss": 4.6579,
      "step": 33645
    },
    {
      "epoch": 0.689083201933119,
      "grad_norm": 0.2803729772567749,
      "learning_rate": 0.00010314384311829099,
      "loss": 4.6161,
      "step": 33650
    },
    {
      "epoch": 0.6891855917105236,
      "grad_norm": 0.2810579240322113,
      "learning_rate": 0.00010308504845561377,
      "loss": 4.6226,
      "step": 33655
    },
    {
      "epoch": 0.6892879814879282,
      "grad_norm": 0.27086740732192993,
      "learning_rate": 0.0001030262617812245,
      "loss": 4.6079,
      "step": 33660
    },
    {
      "epoch": 0.6893903712653329,
      "grad_norm": 0.27384790778160095,
      "learning_rate": 0.00010296748310513298,
      "loss": 4.6516,
      "step": 33665
    },
    {
      "epoch": 0.6894927610427375,
      "grad_norm": 0.270010769367218,
      "learning_rate": 0.00010290871243734759,
      "loss": 4.638,
      "step": 33670
    },
    {
      "epoch": 0.6895951508201421,
      "grad_norm": 0.2743312418460846,
      "learning_rate": 0.00010284994978787521,
      "loss": 4.6167,
      "step": 33675
    },
    {
      "epoch": 0.6896975405975467,
      "grad_norm": 0.27382251620292664,
      "learning_rate": 0.00010279119516672151,
      "loss": 4.6374,
      "step": 33680
    },
    {
      "epoch": 0.6897999303749514,
      "grad_norm": 0.26476651430130005,
      "learning_rate": 0.00010273244858389074,
      "loss": 4.6424,
      "step": 33685
    },
    {
      "epoch": 0.689902320152356,
      "grad_norm": 0.2602652311325073,
      "learning_rate": 0.00010267371004938586,
      "loss": 4.6489,
      "step": 33690
    },
    {
      "epoch": 0.6900047099297606,
      "grad_norm": 0.519544780254364,
      "learning_rate": 0.00010261497957320821,
      "loss": 4.5621,
      "step": 33695
    },
    {
      "epoch": 0.6901070997071652,
      "grad_norm": 0.28236058354377747,
      "learning_rate": 0.00010255625716535802,
      "loss": 4.5656,
      "step": 33700
    },
    {
      "epoch": 0.6902094894845698,
      "grad_norm": 0.28950345516204834,
      "learning_rate": 0.00010249754283583412,
      "loss": 4.6169,
      "step": 33705
    },
    {
      "epoch": 0.6903118792619745,
      "grad_norm": 0.3103809952735901,
      "learning_rate": 0.00010243883659463378,
      "loss": 4.6476,
      "step": 33710
    },
    {
      "epoch": 0.6904142690393791,
      "grad_norm": 0.2706397473812103,
      "learning_rate": 0.00010238013845175304,
      "loss": 4.5904,
      "step": 33715
    },
    {
      "epoch": 0.6905166588167837,
      "grad_norm": 0.26464343070983887,
      "learning_rate": 0.00010232144841718657,
      "loss": 4.5943,
      "step": 33720
    },
    {
      "epoch": 0.6906190485941883,
      "grad_norm": 0.26613783836364746,
      "learning_rate": 0.0001022627665009276,
      "loss": 4.6609,
      "step": 33725
    },
    {
      "epoch": 0.6907214383715929,
      "grad_norm": 0.28778529167175293,
      "learning_rate": 0.00010220409271296799,
      "loss": 4.6519,
      "step": 33730
    },
    {
      "epoch": 0.6908238281489976,
      "grad_norm": 0.2581515610218048,
      "learning_rate": 0.00010214542706329829,
      "loss": 4.5733,
      "step": 33735
    },
    {
      "epoch": 0.6909262179264022,
      "grad_norm": 0.27304625511169434,
      "learning_rate": 0.00010208676956190744,
      "loss": 4.6519,
      "step": 33740
    },
    {
      "epoch": 0.6910286077038068,
      "grad_norm": 0.26651403307914734,
      "learning_rate": 0.00010202812021878331,
      "loss": 4.6317,
      "step": 33745
    },
    {
      "epoch": 0.6911309974812114,
      "grad_norm": 0.274996817111969,
      "learning_rate": 0.00010196947904391222,
      "loss": 4.612,
      "step": 33750
    },
    {
      "epoch": 0.691233387258616,
      "grad_norm": 0.2680855393409729,
      "learning_rate": 0.00010191084604727899,
      "loss": 4.5888,
      "step": 33755
    },
    {
      "epoch": 0.6913357770360207,
      "grad_norm": 0.2705151438713074,
      "learning_rate": 0.00010185222123886722,
      "loss": 4.6379,
      "step": 33760
    },
    {
      "epoch": 0.6914381668134253,
      "grad_norm": 0.29698050022125244,
      "learning_rate": 0.0001017936046286591,
      "loss": 4.6247,
      "step": 33765
    },
    {
      "epoch": 0.69154055659083,
      "grad_norm": 0.27303194999694824,
      "learning_rate": 0.0001017349962266353,
      "loss": 4.6425,
      "step": 33770
    },
    {
      "epoch": 0.6916429463682346,
      "grad_norm": 0.2617448568344116,
      "learning_rate": 0.00010167639604277519,
      "loss": 4.6372,
      "step": 33775
    },
    {
      "epoch": 0.6917453361456393,
      "grad_norm": 0.2696899473667145,
      "learning_rate": 0.00010161780408705681,
      "loss": 4.6355,
      "step": 33780
    },
    {
      "epoch": 0.6918477259230439,
      "grad_norm": 0.28130942583084106,
      "learning_rate": 0.00010155922036945658,
      "loss": 4.5759,
      "step": 33785
    },
    {
      "epoch": 0.6919501157004485,
      "grad_norm": 0.28385117650032043,
      "learning_rate": 0.00010150064489994965,
      "loss": 4.6041,
      "step": 33790
    },
    {
      "epoch": 0.6920525054778531,
      "grad_norm": 0.2776240408420563,
      "learning_rate": 0.00010144207768850994,
      "loss": 4.6411,
      "step": 33795
    },
    {
      "epoch": 0.6921548952552578,
      "grad_norm": 0.272550493478775,
      "learning_rate": 0.0001013835187451096,
      "loss": 4.6128,
      "step": 33800
    },
    {
      "epoch": 0.6922572850326624,
      "grad_norm": 0.31850337982177734,
      "learning_rate": 0.00010132496807971964,
      "loss": 4.5977,
      "step": 33805
    },
    {
      "epoch": 0.692359674810067,
      "grad_norm": 0.28273338079452515,
      "learning_rate": 0.0001012664257023096,
      "loss": 4.6208,
      "step": 33810
    },
    {
      "epoch": 0.6924620645874716,
      "grad_norm": 0.267389178276062,
      "learning_rate": 0.00010120789162284753,
      "loss": 4.6402,
      "step": 33815
    },
    {
      "epoch": 0.6925644543648762,
      "grad_norm": 0.3180309236049652,
      "learning_rate": 0.00010114936585130016,
      "loss": 4.6548,
      "step": 33820
    },
    {
      "epoch": 0.6926668441422809,
      "grad_norm": 0.28237926959991455,
      "learning_rate": 0.00010109084839763277,
      "loss": 4.5276,
      "step": 33825
    },
    {
      "epoch": 0.6927692339196855,
      "grad_norm": 0.27617645263671875,
      "learning_rate": 0.00010103233927180929,
      "loss": 4.6769,
      "step": 33830
    },
    {
      "epoch": 0.6928716236970901,
      "grad_norm": 0.48459896445274353,
      "learning_rate": 0.00010097383848379198,
      "loss": 4.5951,
      "step": 33835
    },
    {
      "epoch": 0.6929740134744947,
      "grad_norm": 0.2679106593132019,
      "learning_rate": 0.00010091534604354204,
      "loss": 4.5875,
      "step": 33840
    },
    {
      "epoch": 0.6930764032518993,
      "grad_norm": 0.26665279269218445,
      "learning_rate": 0.00010085686196101909,
      "loss": 4.6357,
      "step": 33845
    },
    {
      "epoch": 0.693178793029304,
      "grad_norm": 0.30987781286239624,
      "learning_rate": 0.00010079838624618119,
      "loss": 4.6206,
      "step": 33850
    },
    {
      "epoch": 0.6932811828067086,
      "grad_norm": 0.26174652576446533,
      "learning_rate": 0.00010073991890898516,
      "loss": 4.6197,
      "step": 33855
    },
    {
      "epoch": 0.6933835725841132,
      "grad_norm": 0.2697709798812866,
      "learning_rate": 0.00010068145995938637,
      "loss": 4.6602,
      "step": 33860
    },
    {
      "epoch": 0.6934859623615178,
      "grad_norm": 0.2693204879760742,
      "learning_rate": 0.00010062300940733863,
      "loss": 4.6778,
      "step": 33865
    },
    {
      "epoch": 0.6935883521389224,
      "grad_norm": 0.2792659103870392,
      "learning_rate": 0.0001005645672627945,
      "loss": 4.6052,
      "step": 33870
    },
    {
      "epoch": 0.6936907419163271,
      "grad_norm": 0.27753615379333496,
      "learning_rate": 0.00010050613353570505,
      "loss": 4.6265,
      "step": 33875
    },
    {
      "epoch": 0.6937931316937317,
      "grad_norm": 0.25686487555503845,
      "learning_rate": 0.00010044770823601975,
      "loss": 4.673,
      "step": 33880
    },
    {
      "epoch": 0.6938955214711363,
      "grad_norm": 0.2649760842323303,
      "learning_rate": 0.00010038929137368683,
      "loss": 4.6462,
      "step": 33885
    },
    {
      "epoch": 0.6939979112485409,
      "grad_norm": 0.26999253034591675,
      "learning_rate": 0.00010033088295865316,
      "loss": 4.6194,
      "step": 33890
    },
    {
      "epoch": 0.6941003010259456,
      "grad_norm": 0.29455089569091797,
      "learning_rate": 0.00010027248300086385,
      "loss": 4.6102,
      "step": 33895
    },
    {
      "epoch": 0.6942026908033502,
      "grad_norm": 0.27291572093963623,
      "learning_rate": 0.00010021409151026287,
      "loss": 4.6452,
      "step": 33900
    },
    {
      "epoch": 0.6943050805807548,
      "grad_norm": 0.2754967510700226,
      "learning_rate": 0.00010015570849679263,
      "loss": 4.6245,
      "step": 33905
    },
    {
      "epoch": 0.6944074703581594,
      "grad_norm": 0.2637985944747925,
      "learning_rate": 0.00010009733397039405,
      "loss": 4.6501,
      "step": 33910
    },
    {
      "epoch": 0.694509860135564,
      "grad_norm": 0.27101102471351624,
      "learning_rate": 0.0001000389679410067,
      "loss": 4.6049,
      "step": 33915
    },
    {
      "epoch": 0.6946122499129687,
      "grad_norm": 0.2719937264919281,
      "learning_rate": 9.998061041856871e-05,
      "loss": 4.632,
      "step": 33920
    },
    {
      "epoch": 0.6947146396903733,
      "grad_norm": 0.2812808156013489,
      "learning_rate": 9.992226141301661e-05,
      "loss": 4.6184,
      "step": 33925
    },
    {
      "epoch": 0.6948170294677779,
      "grad_norm": 0.28687840700149536,
      "learning_rate": 9.986392093428561e-05,
      "loss": 4.6098,
      "step": 33930
    },
    {
      "epoch": 0.6949194192451825,
      "grad_norm": 0.28334131836891174,
      "learning_rate": 9.980558899230953e-05,
      "loss": 4.5923,
      "step": 33935
    },
    {
      "epoch": 0.6950218090225871,
      "grad_norm": 0.2913331389427185,
      "learning_rate": 9.974726559702055e-05,
      "loss": 4.6653,
      "step": 33940
    },
    {
      "epoch": 0.6951241987999918,
      "grad_norm": 0.29087406396865845,
      "learning_rate": 9.96889507583495e-05,
      "loss": 4.6465,
      "step": 33945
    },
    {
      "epoch": 0.6952265885773964,
      "grad_norm": 0.2694760859012604,
      "learning_rate": 9.963064448622583e-05,
      "loss": 4.6269,
      "step": 33950
    },
    {
      "epoch": 0.695328978354801,
      "grad_norm": 0.2800590693950653,
      "learning_rate": 9.957234679057735e-05,
      "loss": 4.611,
      "step": 33955
    },
    {
      "epoch": 0.6954313681322056,
      "grad_norm": 0.2676231265068054,
      "learning_rate": 9.951405768133056e-05,
      "loss": 4.6176,
      "step": 33960
    },
    {
      "epoch": 0.6955337579096103,
      "grad_norm": 0.2661445736885071,
      "learning_rate": 9.945577716841043e-05,
      "loss": 4.582,
      "step": 33965
    },
    {
      "epoch": 0.695636147687015,
      "grad_norm": 0.2714359164237976,
      "learning_rate": 9.939750526174052e-05,
      "loss": 4.586,
      "step": 33970
    },
    {
      "epoch": 0.6957385374644196,
      "grad_norm": 0.2764875888824463,
      "learning_rate": 9.933924197124281e-05,
      "loss": 4.6132,
      "step": 33975
    },
    {
      "epoch": 0.6958409272418242,
      "grad_norm": 0.2633768916130066,
      "learning_rate": 9.92809873068379e-05,
      "loss": 4.5958,
      "step": 33980
    },
    {
      "epoch": 0.6959433170192288,
      "grad_norm": 0.33859845995903015,
      "learning_rate": 9.922274127844503e-05,
      "loss": 4.5989,
      "step": 33985
    },
    {
      "epoch": 0.6960457067966335,
      "grad_norm": 0.2747366428375244,
      "learning_rate": 9.916450389598169e-05,
      "loss": 4.6358,
      "step": 33990
    },
    {
      "epoch": 0.6961480965740381,
      "grad_norm": 0.28387874364852905,
      "learning_rate": 9.910627516936414e-05,
      "loss": 4.6491,
      "step": 33995
    },
    {
      "epoch": 0.6962504863514427,
      "grad_norm": 0.28758618235588074,
      "learning_rate": 9.904805510850709e-05,
      "loss": 4.6424,
      "step": 34000
    },
    {
      "epoch": 0.6962504863514427,
      "eval_loss": 4.641878128051758,
      "eval_runtime": 17.1317,
      "eval_samples_per_second": 33.972,
      "eval_steps_per_second": 4.261,
      "step": 34000
    },
    {
      "epoch": 0.6963528761288473,
      "grad_norm": 0.2643364369869232,
      "learning_rate": 9.898984372332372e-05,
      "loss": 4.6064,
      "step": 34005
    },
    {
      "epoch": 0.696455265906252,
      "grad_norm": 0.26672685146331787,
      "learning_rate": 9.893164102372579e-05,
      "loss": 4.6363,
      "step": 34010
    },
    {
      "epoch": 0.6965576556836566,
      "grad_norm": 0.26538074016571045,
      "learning_rate": 9.887344701962368e-05,
      "loss": 4.6558,
      "step": 34015
    },
    {
      "epoch": 0.6966600454610612,
      "grad_norm": 0.2577105164527893,
      "learning_rate": 9.881526172092603e-05,
      "loss": 4.6318,
      "step": 34020
    },
    {
      "epoch": 0.6967624352384658,
      "grad_norm": 0.2822088599205017,
      "learning_rate": 9.875708513754017e-05,
      "loss": 4.56,
      "step": 34025
    },
    {
      "epoch": 0.6968648250158704,
      "grad_norm": 0.27342158555984497,
      "learning_rate": 9.869891727937206e-05,
      "loss": 4.5965,
      "step": 34030
    },
    {
      "epoch": 0.6969672147932751,
      "grad_norm": 0.30411356687545776,
      "learning_rate": 9.864075815632589e-05,
      "loss": 4.5894,
      "step": 34035
    },
    {
      "epoch": 0.6970696045706797,
      "grad_norm": 0.2588144540786743,
      "learning_rate": 9.858260777830456e-05,
      "loss": 4.6367,
      "step": 34040
    },
    {
      "epoch": 0.6971719943480843,
      "grad_norm": 0.2642608880996704,
      "learning_rate": 9.852446615520948e-05,
      "loss": 4.6082,
      "step": 34045
    },
    {
      "epoch": 0.6972743841254889,
      "grad_norm": 0.27088019251823425,
      "learning_rate": 9.846633329694046e-05,
      "loss": 4.6136,
      "step": 34050
    },
    {
      "epoch": 0.6973767739028935,
      "grad_norm": 0.27060768008232117,
      "learning_rate": 9.840820921339587e-05,
      "loss": 4.5861,
      "step": 34055
    },
    {
      "epoch": 0.6974791636802982,
      "grad_norm": 0.26521429419517517,
      "learning_rate": 9.835009391447272e-05,
      "loss": 4.6138,
      "step": 34060
    },
    {
      "epoch": 0.6975815534577028,
      "grad_norm": 0.28483012318611145,
      "learning_rate": 9.829198741006624e-05,
      "loss": 4.5893,
      "step": 34065
    },
    {
      "epoch": 0.6976839432351074,
      "grad_norm": 0.28838014602661133,
      "learning_rate": 9.823388971007041e-05,
      "loss": 4.6379,
      "step": 34070
    },
    {
      "epoch": 0.697786333012512,
      "grad_norm": 0.26538369059562683,
      "learning_rate": 9.817580082437761e-05,
      "loss": 4.6638,
      "step": 34075
    },
    {
      "epoch": 0.6978887227899166,
      "grad_norm": 0.2743552625179291,
      "learning_rate": 9.81177207628787e-05,
      "loss": 4.6491,
      "step": 34080
    },
    {
      "epoch": 0.6979911125673213,
      "grad_norm": 0.29531776905059814,
      "learning_rate": 9.805964953546311e-05,
      "loss": 4.6233,
      "step": 34085
    },
    {
      "epoch": 0.6980935023447259,
      "grad_norm": 0.27311116456985474,
      "learning_rate": 9.800158715201874e-05,
      "loss": 4.6743,
      "step": 34090
    },
    {
      "epoch": 0.6981958921221305,
      "grad_norm": 0.29840296506881714,
      "learning_rate": 9.794353362243192e-05,
      "loss": 4.6264,
      "step": 34095
    },
    {
      "epoch": 0.6982982818995351,
      "grad_norm": 0.28220632672309875,
      "learning_rate": 9.788548895658754e-05,
      "loss": 4.6377,
      "step": 34100
    },
    {
      "epoch": 0.6984006716769398,
      "grad_norm": 0.2777146100997925,
      "learning_rate": 9.782745316436897e-05,
      "loss": 4.6099,
      "step": 34105
    },
    {
      "epoch": 0.6985030614543444,
      "grad_norm": 0.2898336946964264,
      "learning_rate": 9.776942625565812e-05,
      "loss": 4.6604,
      "step": 34110
    },
    {
      "epoch": 0.698605451231749,
      "grad_norm": 0.28175127506256104,
      "learning_rate": 9.771140824033523e-05,
      "loss": 4.5905,
      "step": 34115
    },
    {
      "epoch": 0.6987078410091536,
      "grad_norm": 0.2771725356578827,
      "learning_rate": 9.765339912827911e-05,
      "loss": 4.6262,
      "step": 34120
    },
    {
      "epoch": 0.6988102307865582,
      "grad_norm": 0.27140840888023376,
      "learning_rate": 9.759539892936725e-05,
      "loss": 4.5966,
      "step": 34125
    },
    {
      "epoch": 0.6989126205639629,
      "grad_norm": 0.2764374911785126,
      "learning_rate": 9.753740765347522e-05,
      "loss": 4.6608,
      "step": 34130
    },
    {
      "epoch": 0.6990150103413675,
      "grad_norm": 0.26501938700675964,
      "learning_rate": 9.747942531047742e-05,
      "loss": 4.6205,
      "step": 34135
    },
    {
      "epoch": 0.6991174001187721,
      "grad_norm": 0.27190810441970825,
      "learning_rate": 9.742145191024661e-05,
      "loss": 4.5789,
      "step": 34140
    },
    {
      "epoch": 0.6992197898961767,
      "grad_norm": 0.2597341239452362,
      "learning_rate": 9.736348746265394e-05,
      "loss": 4.6178,
      "step": 34145
    },
    {
      "epoch": 0.6993221796735813,
      "grad_norm": 0.27105921506881714,
      "learning_rate": 9.730553197756915e-05,
      "loss": 4.6153,
      "step": 34150
    },
    {
      "epoch": 0.699424569450986,
      "grad_norm": 0.26925212144851685,
      "learning_rate": 9.724758546486048e-05,
      "loss": 4.5839,
      "step": 34155
    },
    {
      "epoch": 0.6995269592283906,
      "grad_norm": 0.2827601432800293,
      "learning_rate": 9.718964793439447e-05,
      "loss": 4.6578,
      "step": 34160
    },
    {
      "epoch": 0.6996293490057952,
      "grad_norm": 0.2963789999485016,
      "learning_rate": 9.713171939603631e-05,
      "loss": 4.6487,
      "step": 34165
    },
    {
      "epoch": 0.6997317387831998,
      "grad_norm": 0.2923237383365631,
      "learning_rate": 9.70737998596496e-05,
      "loss": 4.6496,
      "step": 34170
    },
    {
      "epoch": 0.6998341285606046,
      "grad_norm": 0.274325966835022,
      "learning_rate": 9.701588933509635e-05,
      "loss": 4.6433,
      "step": 34175
    },
    {
      "epoch": 0.6999365183380092,
      "grad_norm": 0.2733668386936188,
      "learning_rate": 9.69579878322371e-05,
      "loss": 4.6413,
      "step": 34180
    },
    {
      "epoch": 0.7000389081154138,
      "grad_norm": 0.270188570022583,
      "learning_rate": 9.690009536093086e-05,
      "loss": 4.6081,
      "step": 34185
    },
    {
      "epoch": 0.7001412978928184,
      "grad_norm": 0.27390149235725403,
      "learning_rate": 9.684221193103505e-05,
      "loss": 4.6254,
      "step": 34190
    },
    {
      "epoch": 0.700243687670223,
      "grad_norm": 0.27170562744140625,
      "learning_rate": 9.678433755240558e-05,
      "loss": 4.6129,
      "step": 34195
    },
    {
      "epoch": 0.7003460774476277,
      "grad_norm": 0.330304354429245,
      "learning_rate": 9.672647223489689e-05,
      "loss": 4.6597,
      "step": 34200
    },
    {
      "epoch": 0.7004484672250323,
      "grad_norm": 0.2667689323425293,
      "learning_rate": 9.66686159883617e-05,
      "loss": 4.6014,
      "step": 34205
    },
    {
      "epoch": 0.7005508570024369,
      "grad_norm": 0.26405391097068787,
      "learning_rate": 9.661076882265132e-05,
      "loss": 4.5914,
      "step": 34210
    },
    {
      "epoch": 0.7006532467798415,
      "grad_norm": 0.27725109457969666,
      "learning_rate": 9.655293074761554e-05,
      "loss": 4.6374,
      "step": 34215
    },
    {
      "epoch": 0.7007556365572462,
      "grad_norm": 0.27563926577568054,
      "learning_rate": 9.649510177310247e-05,
      "loss": 4.6258,
      "step": 34220
    },
    {
      "epoch": 0.7008580263346508,
      "grad_norm": 0.28803661465644836,
      "learning_rate": 9.643728190895878e-05,
      "loss": 4.6535,
      "step": 34225
    },
    {
      "epoch": 0.7009604161120554,
      "grad_norm": 0.2780574858188629,
      "learning_rate": 9.637947116502959e-05,
      "loss": 4.6744,
      "step": 34230
    },
    {
      "epoch": 0.70106280588946,
      "grad_norm": 0.2649870812892914,
      "learning_rate": 9.632166955115839e-05,
      "loss": 4.6309,
      "step": 34235
    },
    {
      "epoch": 0.7011651956668646,
      "grad_norm": 0.2766774892807007,
      "learning_rate": 9.626387707718715e-05,
      "loss": 4.6788,
      "step": 34240
    },
    {
      "epoch": 0.7012675854442693,
      "grad_norm": 0.286602646112442,
      "learning_rate": 9.620609375295632e-05,
      "loss": 4.6395,
      "step": 34245
    },
    {
      "epoch": 0.7013699752216739,
      "grad_norm": 0.26506146788597107,
      "learning_rate": 9.614831958830484e-05,
      "loss": 4.6513,
      "step": 34250
    },
    {
      "epoch": 0.7014723649990785,
      "grad_norm": 0.280479371547699,
      "learning_rate": 9.609055459306986e-05,
      "loss": 4.6245,
      "step": 34255
    },
    {
      "epoch": 0.7015747547764831,
      "grad_norm": 0.28955379128456116,
      "learning_rate": 9.603279877708718e-05,
      "loss": 4.6074,
      "step": 34260
    },
    {
      "epoch": 0.7016771445538877,
      "grad_norm": 0.26712292432785034,
      "learning_rate": 9.597505215019104e-05,
      "loss": 4.6086,
      "step": 34265
    },
    {
      "epoch": 0.7017795343312924,
      "grad_norm": 0.2787074148654938,
      "learning_rate": 9.591731472221397e-05,
      "loss": 4.6254,
      "step": 34270
    },
    {
      "epoch": 0.701881924108697,
      "grad_norm": 0.26819899678230286,
      "learning_rate": 9.585958650298706e-05,
      "loss": 4.5885,
      "step": 34275
    },
    {
      "epoch": 0.7019843138861016,
      "grad_norm": 0.2676922380924225,
      "learning_rate": 9.580186750233983e-05,
      "loss": 4.5781,
      "step": 34280
    },
    {
      "epoch": 0.7020867036635062,
      "grad_norm": 0.263573557138443,
      "learning_rate": 9.574415773010015e-05,
      "loss": 4.6143,
      "step": 34285
    },
    {
      "epoch": 0.7021890934409108,
      "grad_norm": 0.5991362929344177,
      "learning_rate": 9.568645719609433e-05,
      "loss": 4.6701,
      "step": 34290
    },
    {
      "epoch": 0.7022914832183155,
      "grad_norm": 0.26259157061576843,
      "learning_rate": 9.562876591014725e-05,
      "loss": 4.5848,
      "step": 34295
    },
    {
      "epoch": 0.7023938729957201,
      "grad_norm": 0.2694006562232971,
      "learning_rate": 9.557108388208195e-05,
      "loss": 4.6143,
      "step": 34300
    },
    {
      "epoch": 0.7024962627731247,
      "grad_norm": 0.2679491341114044,
      "learning_rate": 9.551341112172013e-05,
      "loss": 4.6394,
      "step": 34305
    },
    {
      "epoch": 0.7025986525505293,
      "grad_norm": 0.2616468667984009,
      "learning_rate": 9.545574763888186e-05,
      "loss": 4.6475,
      "step": 34310
    },
    {
      "epoch": 0.702701042327934,
      "grad_norm": 0.2682160437107086,
      "learning_rate": 9.539809344338551e-05,
      "loss": 4.6331,
      "step": 34315
    },
    {
      "epoch": 0.7028034321053386,
      "grad_norm": 0.2596127390861511,
      "learning_rate": 9.534044854504802e-05,
      "loss": 4.5799,
      "step": 34320
    },
    {
      "epoch": 0.7029058218827432,
      "grad_norm": 0.28490668535232544,
      "learning_rate": 9.528281295368469e-05,
      "loss": 4.6401,
      "step": 34325
    },
    {
      "epoch": 0.7030082116601478,
      "grad_norm": 0.26502755284309387,
      "learning_rate": 9.522518667910917e-05,
      "loss": 4.6217,
      "step": 34330
    },
    {
      "epoch": 0.7031106014375524,
      "grad_norm": 0.2625313997268677,
      "learning_rate": 9.516756973113364e-05,
      "loss": 4.5792,
      "step": 34335
    },
    {
      "epoch": 0.7032129912149571,
      "grad_norm": 0.3207727074623108,
      "learning_rate": 9.510996211956867e-05,
      "loss": 4.6167,
      "step": 34340
    },
    {
      "epoch": 0.7033153809923617,
      "grad_norm": 0.30994778871536255,
      "learning_rate": 9.50523638542231e-05,
      "loss": 4.6362,
      "step": 34345
    },
    {
      "epoch": 0.7034177707697663,
      "grad_norm": 0.26821982860565186,
      "learning_rate": 9.499477494490433e-05,
      "loss": 4.6061,
      "step": 34350
    },
    {
      "epoch": 0.7035201605471709,
      "grad_norm": 0.2646442949771881,
      "learning_rate": 9.493719540141815e-05,
      "loss": 4.6474,
      "step": 34355
    },
    {
      "epoch": 0.7036225503245755,
      "grad_norm": 0.25915464758872986,
      "learning_rate": 9.487962523356868e-05,
      "loss": 4.6719,
      "step": 34360
    },
    {
      "epoch": 0.7037249401019802,
      "grad_norm": 0.2941591143608093,
      "learning_rate": 9.48220644511585e-05,
      "loss": 4.6098,
      "step": 34365
    },
    {
      "epoch": 0.7038273298793848,
      "grad_norm": 0.26016825437545776,
      "learning_rate": 9.476451306398864e-05,
      "loss": 4.6346,
      "step": 34370
    },
    {
      "epoch": 0.7039297196567895,
      "grad_norm": 0.2864300608634949,
      "learning_rate": 9.470697108185838e-05,
      "loss": 4.678,
      "step": 34375
    },
    {
      "epoch": 0.7040321094341941,
      "grad_norm": 0.2741391658782959,
      "learning_rate": 9.464943851456556e-05,
      "loss": 4.6368,
      "step": 34380
    },
    {
      "epoch": 0.7041344992115988,
      "grad_norm": 0.2671283781528473,
      "learning_rate": 9.459191537190631e-05,
      "loss": 4.6042,
      "step": 34385
    },
    {
      "epoch": 0.7042368889890034,
      "grad_norm": 0.2896394729614258,
      "learning_rate": 9.453440166367529e-05,
      "loss": 4.6453,
      "step": 34390
    },
    {
      "epoch": 0.704339278766408,
      "grad_norm": 0.2839994728565216,
      "learning_rate": 9.447689739966527e-05,
      "loss": 4.6129,
      "step": 34395
    },
    {
      "epoch": 0.7044416685438126,
      "grad_norm": 0.27805691957473755,
      "learning_rate": 9.441940258966774e-05,
      "loss": 4.5951,
      "step": 34400
    },
    {
      "epoch": 0.7045440583212172,
      "grad_norm": 0.27397221326828003,
      "learning_rate": 9.43619172434724e-05,
      "loss": 4.5941,
      "step": 34405
    },
    {
      "epoch": 0.7046464480986219,
      "grad_norm": 0.2726888358592987,
      "learning_rate": 9.430444137086738e-05,
      "loss": 4.6141,
      "step": 34410
    },
    {
      "epoch": 0.7047488378760265,
      "grad_norm": 0.2829132080078125,
      "learning_rate": 9.424697498163917e-05,
      "loss": 4.6236,
      "step": 34415
    },
    {
      "epoch": 0.7048512276534311,
      "grad_norm": 0.30598700046539307,
      "learning_rate": 9.418951808557271e-05,
      "loss": 4.6069,
      "step": 34420
    },
    {
      "epoch": 0.7049536174308357,
      "grad_norm": 0.27091097831726074,
      "learning_rate": 9.413207069245125e-05,
      "loss": 4.6398,
      "step": 34425
    },
    {
      "epoch": 0.7050560072082404,
      "grad_norm": 0.26590096950531006,
      "learning_rate": 9.407463281205647e-05,
      "loss": 4.6303,
      "step": 34430
    },
    {
      "epoch": 0.705158396985645,
      "grad_norm": 0.2824147939682007,
      "learning_rate": 9.401720445416847e-05,
      "loss": 4.5699,
      "step": 34435
    },
    {
      "epoch": 0.7052607867630496,
      "grad_norm": 0.28604599833488464,
      "learning_rate": 9.395978562856553e-05,
      "loss": 4.6345,
      "step": 34440
    },
    {
      "epoch": 0.7053631765404542,
      "grad_norm": 0.2786464989185333,
      "learning_rate": 9.390237634502457e-05,
      "loss": 4.6476,
      "step": 34445
    },
    {
      "epoch": 0.7054655663178588,
      "grad_norm": 0.2748449742794037,
      "learning_rate": 9.384497661332075e-05,
      "loss": 4.6295,
      "step": 34450
    },
    {
      "epoch": 0.7055679560952635,
      "grad_norm": 0.2649158537387848,
      "learning_rate": 9.378758644322756e-05,
      "loss": 4.57,
      "step": 34455
    },
    {
      "epoch": 0.7056703458726681,
      "grad_norm": 0.29101142287254333,
      "learning_rate": 9.373020584451698e-05,
      "loss": 4.6764,
      "step": 34460
    },
    {
      "epoch": 0.7057727356500727,
      "grad_norm": 0.2751317322254181,
      "learning_rate": 9.36728348269593e-05,
      "loss": 4.6075,
      "step": 34465
    },
    {
      "epoch": 0.7058751254274773,
      "grad_norm": 0.27171221375465393,
      "learning_rate": 9.361547340032315e-05,
      "loss": 4.6013,
      "step": 34470
    },
    {
      "epoch": 0.705977515204882,
      "grad_norm": 0.2699592113494873,
      "learning_rate": 9.355812157437553e-05,
      "loss": 4.5751,
      "step": 34475
    },
    {
      "epoch": 0.7060799049822866,
      "grad_norm": 0.5079491138458252,
      "learning_rate": 9.350077935888198e-05,
      "loss": 4.6284,
      "step": 34480
    },
    {
      "epoch": 0.7061822947596912,
      "grad_norm": 0.26962020993232727,
      "learning_rate": 9.344344676360606e-05,
      "loss": 4.626,
      "step": 34485
    },
    {
      "epoch": 0.7062846845370958,
      "grad_norm": 0.2690640091896057,
      "learning_rate": 9.338612379830999e-05,
      "loss": 4.6348,
      "step": 34490
    },
    {
      "epoch": 0.7063870743145004,
      "grad_norm": 0.28714755177497864,
      "learning_rate": 9.332881047275427e-05,
      "loss": 4.6122,
      "step": 34495
    },
    {
      "epoch": 0.706489464091905,
      "grad_norm": 0.27294567227363586,
      "learning_rate": 9.327150679669766e-05,
      "loss": 4.6603,
      "step": 34500
    },
    {
      "epoch": 0.7065918538693097,
      "grad_norm": 0.27827540040016174,
      "learning_rate": 9.32142127798974e-05,
      "loss": 4.6319,
      "step": 34505
    },
    {
      "epoch": 0.7066942436467143,
      "grad_norm": 0.2705303132534027,
      "learning_rate": 9.315692843210905e-05,
      "loss": 4.628,
      "step": 34510
    },
    {
      "epoch": 0.7067966334241189,
      "grad_norm": 0.28062865138053894,
      "learning_rate": 9.309965376308648e-05,
      "loss": 4.6471,
      "step": 34515
    },
    {
      "epoch": 0.7068990232015235,
      "grad_norm": 0.2768203020095825,
      "learning_rate": 9.304238878258196e-05,
      "loss": 4.6557,
      "step": 34520
    },
    {
      "epoch": 0.7070014129789282,
      "grad_norm": 0.26975247263908386,
      "learning_rate": 9.29851335003461e-05,
      "loss": 4.64,
      "step": 34525
    },
    {
      "epoch": 0.7071038027563328,
      "grad_norm": 0.3466828763484955,
      "learning_rate": 9.292788792612791e-05,
      "loss": 4.6212,
      "step": 34530
    },
    {
      "epoch": 0.7072061925337374,
      "grad_norm": 0.2951856851577759,
      "learning_rate": 9.287065206967457e-05,
      "loss": 4.6391,
      "step": 34535
    },
    {
      "epoch": 0.707308582311142,
      "grad_norm": 0.27098989486694336,
      "learning_rate": 9.281342594073176e-05,
      "loss": 4.6481,
      "step": 34540
    },
    {
      "epoch": 0.7074109720885466,
      "grad_norm": 0.26753920316696167,
      "learning_rate": 9.275620954904355e-05,
      "loss": 4.634,
      "step": 34545
    },
    {
      "epoch": 0.7075133618659513,
      "grad_norm": 0.26743990182876587,
      "learning_rate": 9.269900290435219e-05,
      "loss": 4.6,
      "step": 34550
    },
    {
      "epoch": 0.7076157516433559,
      "grad_norm": 0.26732194423675537,
      "learning_rate": 9.264180601639838e-05,
      "loss": 4.5985,
      "step": 34555
    },
    {
      "epoch": 0.7077181414207605,
      "grad_norm": 0.25865405797958374,
      "learning_rate": 9.258461889492116e-05,
      "loss": 4.5589,
      "step": 34560
    },
    {
      "epoch": 0.7078205311981651,
      "grad_norm": 0.2651233673095703,
      "learning_rate": 9.252744154965783e-05,
      "loss": 4.629,
      "step": 34565
    },
    {
      "epoch": 0.7079229209755697,
      "grad_norm": 0.26940491795539856,
      "learning_rate": 9.247027399034409e-05,
      "loss": 4.5988,
      "step": 34570
    },
    {
      "epoch": 0.7080253107529744,
      "grad_norm": 0.2711552679538727,
      "learning_rate": 9.241311622671397e-05,
      "loss": 4.581,
      "step": 34575
    },
    {
      "epoch": 0.7081277005303791,
      "grad_norm": 0.27430421113967896,
      "learning_rate": 9.23559682684998e-05,
      "loss": 4.6122,
      "step": 34580
    },
    {
      "epoch": 0.7082300903077837,
      "grad_norm": 0.3444492220878601,
      "learning_rate": 9.229883012543225e-05,
      "loss": 4.6314,
      "step": 34585
    },
    {
      "epoch": 0.7083324800851883,
      "grad_norm": 0.2723939120769501,
      "learning_rate": 9.224170180724037e-05,
      "loss": 4.6284,
      "step": 34590
    },
    {
      "epoch": 0.708434869862593,
      "grad_norm": 0.274140864610672,
      "learning_rate": 9.218458332365144e-05,
      "loss": 4.6356,
      "step": 34595
    },
    {
      "epoch": 0.7085372596399976,
      "grad_norm": 0.2797831594944,
      "learning_rate": 9.212747468439117e-05,
      "loss": 4.6264,
      "step": 34600
    },
    {
      "epoch": 0.7086396494174022,
      "grad_norm": 0.27547451853752136,
      "learning_rate": 9.207037589918353e-05,
      "loss": 4.6597,
      "step": 34605
    },
    {
      "epoch": 0.7087420391948068,
      "grad_norm": 0.2748309373855591,
      "learning_rate": 9.20132869777508e-05,
      "loss": 4.6356,
      "step": 34610
    },
    {
      "epoch": 0.7088444289722114,
      "grad_norm": 0.2922535538673401,
      "learning_rate": 9.195620792981362e-05,
      "loss": 4.6006,
      "step": 34615
    },
    {
      "epoch": 0.7089468187496161,
      "grad_norm": 0.27265986800193787,
      "learning_rate": 9.189913876509102e-05,
      "loss": 4.6749,
      "step": 34620
    },
    {
      "epoch": 0.7090492085270207,
      "grad_norm": 0.26631078124046326,
      "learning_rate": 9.184207949330012e-05,
      "loss": 4.6095,
      "step": 34625
    },
    {
      "epoch": 0.7091515983044253,
      "grad_norm": 0.2759389579296112,
      "learning_rate": 9.178503012415658e-05,
      "loss": 4.5712,
      "step": 34630
    },
    {
      "epoch": 0.7092539880818299,
      "grad_norm": 0.2792028784751892,
      "learning_rate": 9.172799066737429e-05,
      "loss": 4.6401,
      "step": 34635
    },
    {
      "epoch": 0.7093563778592346,
      "grad_norm": 0.2706100642681122,
      "learning_rate": 9.167096113266541e-05,
      "loss": 4.6037,
      "step": 34640
    },
    {
      "epoch": 0.7094587676366392,
      "grad_norm": 0.27416983246803284,
      "learning_rate": 9.16139415297405e-05,
      "loss": 4.5819,
      "step": 34645
    },
    {
      "epoch": 0.7095611574140438,
      "grad_norm": 0.28448888659477234,
      "learning_rate": 9.155693186830839e-05,
      "loss": 4.6489,
      "step": 34650
    },
    {
      "epoch": 0.7096635471914484,
      "grad_norm": 0.27139759063720703,
      "learning_rate": 9.149993215807619e-05,
      "loss": 4.6525,
      "step": 34655
    },
    {
      "epoch": 0.709765936968853,
      "grad_norm": 0.2676253616809845,
      "learning_rate": 9.144294240874933e-05,
      "loss": 4.66,
      "step": 34660
    },
    {
      "epoch": 0.7098683267462577,
      "grad_norm": 0.2800234854221344,
      "learning_rate": 9.138596263003153e-05,
      "loss": 4.6374,
      "step": 34665
    },
    {
      "epoch": 0.7099707165236623,
      "grad_norm": 0.2750464081764221,
      "learning_rate": 9.132899283162498e-05,
      "loss": 4.643,
      "step": 34670
    },
    {
      "epoch": 0.7100731063010669,
      "grad_norm": 0.3019641041755676,
      "learning_rate": 9.127203302322981e-05,
      "loss": 4.6366,
      "step": 34675
    },
    {
      "epoch": 0.7101754960784715,
      "grad_norm": 0.39563891291618347,
      "learning_rate": 9.121508321454478e-05,
      "loss": 4.6529,
      "step": 34680
    },
    {
      "epoch": 0.7102778858558761,
      "grad_norm": 0.279168039560318,
      "learning_rate": 9.115814341526683e-05,
      "loss": 4.5976,
      "step": 34685
    },
    {
      "epoch": 0.7103802756332808,
      "grad_norm": 0.2793745994567871,
      "learning_rate": 9.110121363509114e-05,
      "loss": 4.5832,
      "step": 34690
    },
    {
      "epoch": 0.7104826654106854,
      "grad_norm": 0.264337956905365,
      "learning_rate": 9.104429388371129e-05,
      "loss": 4.6195,
      "step": 34695
    },
    {
      "epoch": 0.71058505518809,
      "grad_norm": 0.27568039298057556,
      "learning_rate": 9.098738417081911e-05,
      "loss": 4.6585,
      "step": 34700
    },
    {
      "epoch": 0.7106874449654946,
      "grad_norm": 0.31890881061553955,
      "learning_rate": 9.093048450610467e-05,
      "loss": 4.6064,
      "step": 34705
    },
    {
      "epoch": 0.7107898347428993,
      "grad_norm": 0.2892616391181946,
      "learning_rate": 9.08735948992564e-05,
      "loss": 4.5892,
      "step": 34710
    },
    {
      "epoch": 0.7108922245203039,
      "grad_norm": 0.28401005268096924,
      "learning_rate": 9.081671535996104e-05,
      "loss": 4.5841,
      "step": 34715
    },
    {
      "epoch": 0.7109946142977085,
      "grad_norm": 0.27535170316696167,
      "learning_rate": 9.075984589790344e-05,
      "loss": 4.6577,
      "step": 34720
    },
    {
      "epoch": 0.7110970040751131,
      "grad_norm": 0.272706538438797,
      "learning_rate": 9.070298652276693e-05,
      "loss": 4.5934,
      "step": 34725
    },
    {
      "epoch": 0.7111993938525177,
      "grad_norm": 0.26996663212776184,
      "learning_rate": 9.064613724423308e-05,
      "loss": 4.6322,
      "step": 34730
    },
    {
      "epoch": 0.7113017836299224,
      "grad_norm": 0.30381351709365845,
      "learning_rate": 9.058929807198166e-05,
      "loss": 4.5868,
      "step": 34735
    },
    {
      "epoch": 0.711404173407327,
      "grad_norm": 0.2938435971736908,
      "learning_rate": 9.053246901569081e-05,
      "loss": 4.6256,
      "step": 34740
    },
    {
      "epoch": 0.7115065631847316,
      "grad_norm": 0.2768428325653076,
      "learning_rate": 9.047565008503689e-05,
      "loss": 4.5957,
      "step": 34745
    },
    {
      "epoch": 0.7116089529621362,
      "grad_norm": 0.3060249388217926,
      "learning_rate": 9.041884128969455e-05,
      "loss": 4.5915,
      "step": 34750
    },
    {
      "epoch": 0.7117113427395408,
      "grad_norm": 0.273645281791687,
      "learning_rate": 9.036204263933672e-05,
      "loss": 4.6285,
      "step": 34755
    },
    {
      "epoch": 0.7118137325169455,
      "grad_norm": 0.2825133502483368,
      "learning_rate": 9.030525414363468e-05,
      "loss": 4.6175,
      "step": 34760
    },
    {
      "epoch": 0.7119161222943501,
      "grad_norm": 0.26800164580345154,
      "learning_rate": 9.024847581225777e-05,
      "loss": 4.6281,
      "step": 34765
    },
    {
      "epoch": 0.7120185120717547,
      "grad_norm": 0.2778143584728241,
      "learning_rate": 9.019170765487379e-05,
      "loss": 4.6035,
      "step": 34770
    },
    {
      "epoch": 0.7121209018491593,
      "grad_norm": 0.26916933059692383,
      "learning_rate": 9.013494968114879e-05,
      "loss": 4.6818,
      "step": 34775
    },
    {
      "epoch": 0.7122232916265641,
      "grad_norm": 0.2809462547302246,
      "learning_rate": 9.007820190074699e-05,
      "loss": 4.4804,
      "step": 34780
    },
    {
      "epoch": 0.7123256814039687,
      "grad_norm": 0.3131517469882965,
      "learning_rate": 9.002146432333095e-05,
      "loss": 4.6172,
      "step": 34785
    },
    {
      "epoch": 0.7124280711813733,
      "grad_norm": 0.29829517006874084,
      "learning_rate": 8.996473695856151e-05,
      "loss": 4.6227,
      "step": 34790
    },
    {
      "epoch": 0.7125304609587779,
      "grad_norm": 0.2847922742366791,
      "learning_rate": 8.990801981609765e-05,
      "loss": 4.5335,
      "step": 34795
    },
    {
      "epoch": 0.7126328507361825,
      "grad_norm": 0.2825530171394348,
      "learning_rate": 8.985131290559676e-05,
      "loss": 4.5997,
      "step": 34800
    },
    {
      "epoch": 0.7127352405135872,
      "grad_norm": 0.27248820662498474,
      "learning_rate": 8.979461623671445e-05,
      "loss": 4.6408,
      "step": 34805
    },
    {
      "epoch": 0.7128376302909918,
      "grad_norm": 0.3130579888820648,
      "learning_rate": 8.973792981910445e-05,
      "loss": 4.6386,
      "step": 34810
    },
    {
      "epoch": 0.7129400200683964,
      "grad_norm": 0.26889511942863464,
      "learning_rate": 8.968125366241895e-05,
      "loss": 4.6432,
      "step": 34815
    },
    {
      "epoch": 0.713042409845801,
      "grad_norm": 0.2863365113735199,
      "learning_rate": 8.962458777630824e-05,
      "loss": 4.5673,
      "step": 34820
    },
    {
      "epoch": 0.7131447996232056,
      "grad_norm": 0.2696453034877777,
      "learning_rate": 8.956793217042095e-05,
      "loss": 4.615,
      "step": 34825
    },
    {
      "epoch": 0.7132471894006103,
      "grad_norm": 0.28150179982185364,
      "learning_rate": 8.95112868544039e-05,
      "loss": 4.5932,
      "step": 34830
    },
    {
      "epoch": 0.7133495791780149,
      "grad_norm": 0.287253737449646,
      "learning_rate": 8.945465183790219e-05,
      "loss": 4.495,
      "step": 34835
    },
    {
      "epoch": 0.7134519689554195,
      "grad_norm": 0.2708670496940613,
      "learning_rate": 8.93980271305592e-05,
      "loss": 4.6332,
      "step": 34840
    },
    {
      "epoch": 0.7135543587328241,
      "grad_norm": 0.30999526381492615,
      "learning_rate": 8.934141274201645e-05,
      "loss": 4.5616,
      "step": 34845
    },
    {
      "epoch": 0.7136567485102288,
      "grad_norm": 0.2686518132686615,
      "learning_rate": 8.92848086819138e-05,
      "loss": 4.6334,
      "step": 34850
    },
    {
      "epoch": 0.7137591382876334,
      "grad_norm": 0.30306416749954224,
      "learning_rate": 8.922821495988936e-05,
      "loss": 4.5848,
      "step": 34855
    },
    {
      "epoch": 0.713861528065038,
      "grad_norm": 0.33884480595588684,
      "learning_rate": 8.917163158557935e-05,
      "loss": 4.5813,
      "step": 34860
    },
    {
      "epoch": 0.7139639178424426,
      "grad_norm": 0.2970154881477356,
      "learning_rate": 8.911505856861836e-05,
      "loss": 4.6241,
      "step": 34865
    },
    {
      "epoch": 0.7140663076198472,
      "grad_norm": 0.27281296253204346,
      "learning_rate": 8.90584959186392e-05,
      "loss": 4.6425,
      "step": 34870
    },
    {
      "epoch": 0.7141686973972519,
      "grad_norm": 0.28088513016700745,
      "learning_rate": 8.900194364527283e-05,
      "loss": 4.6148,
      "step": 34875
    },
    {
      "epoch": 0.7142710871746565,
      "grad_norm": 0.3154902458190918,
      "learning_rate": 8.894540175814854e-05,
      "loss": 4.6612,
      "step": 34880
    },
    {
      "epoch": 0.7143734769520611,
      "grad_norm": 0.27968519926071167,
      "learning_rate": 8.888887026689386e-05,
      "loss": 4.5898,
      "step": 34885
    },
    {
      "epoch": 0.7144758667294657,
      "grad_norm": 0.26756688952445984,
      "learning_rate": 8.883234918113436e-05,
      "loss": 4.5878,
      "step": 34890
    },
    {
      "epoch": 0.7145782565068703,
      "grad_norm": 0.28028222918510437,
      "learning_rate": 8.877583851049411e-05,
      "loss": 4.6771,
      "step": 34895
    },
    {
      "epoch": 0.714680646284275,
      "grad_norm": 0.2732871174812317,
      "learning_rate": 8.871933826459528e-05,
      "loss": 4.5864,
      "step": 34900
    },
    {
      "epoch": 0.7147830360616796,
      "grad_norm": 0.3130098581314087,
      "learning_rate": 8.866284845305818e-05,
      "loss": 4.6858,
      "step": 34905
    },
    {
      "epoch": 0.7148854258390842,
      "grad_norm": 0.2788325846195221,
      "learning_rate": 8.860636908550144e-05,
      "loss": 4.6588,
      "step": 34910
    },
    {
      "epoch": 0.7149878156164888,
      "grad_norm": 0.2613585889339447,
      "learning_rate": 8.854990017154197e-05,
      "loss": 4.5716,
      "step": 34915
    },
    {
      "epoch": 0.7150902053938935,
      "grad_norm": 0.2701241374015808,
      "learning_rate": 8.849344172079473e-05,
      "loss": 4.6231,
      "step": 34920
    },
    {
      "epoch": 0.7151925951712981,
      "grad_norm": 0.3269418179988861,
      "learning_rate": 8.843699374287307e-05,
      "loss": 4.621,
      "step": 34925
    },
    {
      "epoch": 0.7152949849487027,
      "grad_norm": 0.31243759393692017,
      "learning_rate": 8.83805562473885e-05,
      "loss": 4.6237,
      "step": 34930
    },
    {
      "epoch": 0.7153973747261073,
      "grad_norm": 0.29419705271720886,
      "learning_rate": 8.832412924395061e-05,
      "loss": 4.6194,
      "step": 34935
    },
    {
      "epoch": 0.7154997645035119,
      "grad_norm": 0.2717318534851074,
      "learning_rate": 8.826771274216741e-05,
      "loss": 4.5815,
      "step": 34940
    },
    {
      "epoch": 0.7156021542809166,
      "grad_norm": 0.27413737773895264,
      "learning_rate": 8.82113067516451e-05,
      "loss": 4.6183,
      "step": 34945
    },
    {
      "epoch": 0.7157045440583212,
      "grad_norm": 0.2781221270561218,
      "learning_rate": 8.815491128198787e-05,
      "loss": 4.6954,
      "step": 34950
    },
    {
      "epoch": 0.7158069338357258,
      "grad_norm": 0.28618308901786804,
      "learning_rate": 8.809852634279835e-05,
      "loss": 4.6715,
      "step": 34955
    },
    {
      "epoch": 0.7159093236131304,
      "grad_norm": 0.27211809158325195,
      "learning_rate": 8.804215194367728e-05,
      "loss": 4.5879,
      "step": 34960
    },
    {
      "epoch": 0.716011713390535,
      "grad_norm": 0.300509512424469,
      "learning_rate": 8.798578809422367e-05,
      "loss": 4.6073,
      "step": 34965
    },
    {
      "epoch": 0.7161141031679397,
      "grad_norm": 0.2684769332408905,
      "learning_rate": 8.792943480403463e-05,
      "loss": 4.5932,
      "step": 34970
    },
    {
      "epoch": 0.7162164929453443,
      "grad_norm": 0.28665891289711,
      "learning_rate": 8.787309208270556e-05,
      "loss": 4.5766,
      "step": 34975
    },
    {
      "epoch": 0.7163188827227489,
      "grad_norm": 0.29640817642211914,
      "learning_rate": 8.781675993983009e-05,
      "loss": 4.5813,
      "step": 34980
    },
    {
      "epoch": 0.7164212725001536,
      "grad_norm": 0.27339333295822144,
      "learning_rate": 8.776043838499982e-05,
      "loss": 4.6273,
      "step": 34985
    },
    {
      "epoch": 0.7165236622775583,
      "grad_norm": 0.27096307277679443,
      "learning_rate": 8.770412742780487e-05,
      "loss": 4.6263,
      "step": 34990
    },
    {
      "epoch": 0.7166260520549629,
      "grad_norm": 0.2680208086967468,
      "learning_rate": 8.76478270778334e-05,
      "loss": 4.7049,
      "step": 34995
    },
    {
      "epoch": 0.7167284418323675,
      "grad_norm": 0.28621193766593933,
      "learning_rate": 8.759153734467166e-05,
      "loss": 4.6445,
      "step": 35000
    },
    {
      "epoch": 0.7167284418323675,
      "eval_loss": 4.63367223739624,
      "eval_runtime": 17.1299,
      "eval_samples_per_second": 33.976,
      "eval_steps_per_second": 4.262,
      "step": 35000
    },
    {
      "epoch": 0.7168308316097721,
      "grad_norm": 0.3055669963359833,
      "learning_rate": 8.753525823790423e-05,
      "loss": 4.6183,
      "step": 35005
    },
    {
      "epoch": 0.7169332213871767,
      "grad_norm": 0.2594819664955139,
      "learning_rate": 8.747898976711399e-05,
      "loss": 4.6443,
      "step": 35010
    },
    {
      "epoch": 0.7170356111645814,
      "grad_norm": 0.26975128054618835,
      "learning_rate": 8.742273194188165e-05,
      "loss": 4.6332,
      "step": 35015
    },
    {
      "epoch": 0.717138000941986,
      "grad_norm": 0.2627929151058197,
      "learning_rate": 8.736648477178645e-05,
      "loss": 4.654,
      "step": 35020
    },
    {
      "epoch": 0.7172403907193906,
      "grad_norm": 0.2685806453227997,
      "learning_rate": 8.73102482664057e-05,
      "loss": 4.5743,
      "step": 35025
    },
    {
      "epoch": 0.7173427804967952,
      "grad_norm": 0.27302104234695435,
      "learning_rate": 8.725402243531477e-05,
      "loss": 4.6513,
      "step": 35030
    },
    {
      "epoch": 0.7174451702741999,
      "grad_norm": 0.2837328612804413,
      "learning_rate": 8.719780728808749e-05,
      "loss": 4.6101,
      "step": 35035
    },
    {
      "epoch": 0.7175475600516045,
      "grad_norm": 0.27872976660728455,
      "learning_rate": 8.714160283429561e-05,
      "loss": 4.6854,
      "step": 35040
    },
    {
      "epoch": 0.7176499498290091,
      "grad_norm": 0.284140944480896,
      "learning_rate": 8.708540908350913e-05,
      "loss": 4.6184,
      "step": 35045
    },
    {
      "epoch": 0.7177523396064137,
      "grad_norm": 0.27065593004226685,
      "learning_rate": 8.702922604529637e-05,
      "loss": 4.6628,
      "step": 35050
    },
    {
      "epoch": 0.7178547293838183,
      "grad_norm": 0.2855406701564789,
      "learning_rate": 8.697305372922363e-05,
      "loss": 4.5895,
      "step": 35055
    },
    {
      "epoch": 0.717957119161223,
      "grad_norm": 0.284546434879303,
      "learning_rate": 8.691689214485547e-05,
      "loss": 4.6093,
      "step": 35060
    },
    {
      "epoch": 0.7180595089386276,
      "grad_norm": 0.27335232496261597,
      "learning_rate": 8.686074130175456e-05,
      "loss": 4.6557,
      "step": 35065
    },
    {
      "epoch": 0.7181618987160322,
      "grad_norm": 0.27810123562812805,
      "learning_rate": 8.680460120948193e-05,
      "loss": 4.6233,
      "step": 35070
    },
    {
      "epoch": 0.7182642884934368,
      "grad_norm": 0.2812967002391815,
      "learning_rate": 8.674847187759658e-05,
      "loss": 4.6,
      "step": 35075
    },
    {
      "epoch": 0.7183666782708414,
      "grad_norm": 0.27613675594329834,
      "learning_rate": 8.669235331565569e-05,
      "loss": 4.568,
      "step": 35080
    },
    {
      "epoch": 0.7184690680482461,
      "grad_norm": 0.28897684812545776,
      "learning_rate": 8.663624553321477e-05,
      "loss": 4.5767,
      "step": 35085
    },
    {
      "epoch": 0.7185714578256507,
      "grad_norm": 0.2786373794078827,
      "learning_rate": 8.658014853982736e-05,
      "loss": 4.5984,
      "step": 35090
    },
    {
      "epoch": 0.7186738476030553,
      "grad_norm": 0.2863420844078064,
      "learning_rate": 8.652406234504512e-05,
      "loss": 4.6394,
      "step": 35095
    },
    {
      "epoch": 0.7187762373804599,
      "grad_norm": 0.29722583293914795,
      "learning_rate": 8.6467986958418e-05,
      "loss": 4.6108,
      "step": 35100
    },
    {
      "epoch": 0.7188786271578645,
      "grad_norm": 0.28767991065979004,
      "learning_rate": 8.641192238949406e-05,
      "loss": 4.6258,
      "step": 35105
    },
    {
      "epoch": 0.7189810169352692,
      "grad_norm": 0.27794715762138367,
      "learning_rate": 8.635586864781948e-05,
      "loss": 4.6028,
      "step": 35110
    },
    {
      "epoch": 0.7190834067126738,
      "grad_norm": 0.2857860326766968,
      "learning_rate": 8.629982574293858e-05,
      "loss": 4.6554,
      "step": 35115
    },
    {
      "epoch": 0.7191857964900784,
      "grad_norm": 0.2754746079444885,
      "learning_rate": 8.624379368439399e-05,
      "loss": 4.6117,
      "step": 35120
    },
    {
      "epoch": 0.719288186267483,
      "grad_norm": 0.29061755537986755,
      "learning_rate": 8.618777248172632e-05,
      "loss": 4.5841,
      "step": 35125
    },
    {
      "epoch": 0.7193905760448877,
      "grad_norm": 0.31932756304740906,
      "learning_rate": 8.613176214447435e-05,
      "loss": 4.6827,
      "step": 35130
    },
    {
      "epoch": 0.7194929658222923,
      "grad_norm": 0.28680405020713806,
      "learning_rate": 8.607576268217515e-05,
      "loss": 4.5784,
      "step": 35135
    },
    {
      "epoch": 0.7195953555996969,
      "grad_norm": 0.2824251055717468,
      "learning_rate": 8.601977410436381e-05,
      "loss": 4.5641,
      "step": 35140
    },
    {
      "epoch": 0.7196977453771015,
      "grad_norm": 0.2622056305408478,
      "learning_rate": 8.596379642057355e-05,
      "loss": 4.6209,
      "step": 35145
    },
    {
      "epoch": 0.7198001351545061,
      "grad_norm": 0.28085857629776,
      "learning_rate": 8.59078296403359e-05,
      "loss": 4.5532,
      "step": 35150
    },
    {
      "epoch": 0.7199025249319108,
      "grad_norm": 0.2631789743900299,
      "learning_rate": 8.585187377318025e-05,
      "loss": 4.6004,
      "step": 35155
    },
    {
      "epoch": 0.7200049147093154,
      "grad_norm": 0.2821058928966522,
      "learning_rate": 8.579592882863444e-05,
      "loss": 4.6997,
      "step": 35160
    },
    {
      "epoch": 0.72010730448672,
      "grad_norm": 0.29496118426322937,
      "learning_rate": 8.573999481622427e-05,
      "loss": 4.618,
      "step": 35165
    },
    {
      "epoch": 0.7202096942641246,
      "grad_norm": 0.30153515934944153,
      "learning_rate": 8.568407174547363e-05,
      "loss": 4.5909,
      "step": 35170
    },
    {
      "epoch": 0.7203120840415292,
      "grad_norm": 0.2813776433467865,
      "learning_rate": 8.562815962590481e-05,
      "loss": 4.5943,
      "step": 35175
    },
    {
      "epoch": 0.7204144738189339,
      "grad_norm": 0.29179972410202026,
      "learning_rate": 8.557225846703795e-05,
      "loss": 4.6066,
      "step": 35180
    },
    {
      "epoch": 0.7205168635963386,
      "grad_norm": 0.274844229221344,
      "learning_rate": 8.551636827839137e-05,
      "loss": 4.6262,
      "step": 35185
    },
    {
      "epoch": 0.7206192533737432,
      "grad_norm": 0.2772488296031952,
      "learning_rate": 8.546048906948176e-05,
      "loss": 4.6229,
      "step": 35190
    },
    {
      "epoch": 0.7207216431511478,
      "grad_norm": 0.26569491624832153,
      "learning_rate": 8.540462084982366e-05,
      "loss": 4.5865,
      "step": 35195
    },
    {
      "epoch": 0.7208240329285525,
      "grad_norm": 0.25651228427886963,
      "learning_rate": 8.534876362892985e-05,
      "loss": 4.6133,
      "step": 35200
    },
    {
      "epoch": 0.7209264227059571,
      "grad_norm": 0.2695825397968292,
      "learning_rate": 8.52929174163112e-05,
      "loss": 4.6394,
      "step": 35205
    },
    {
      "epoch": 0.7210288124833617,
      "grad_norm": 0.2850077152252197,
      "learning_rate": 8.523708222147684e-05,
      "loss": 4.6254,
      "step": 35210
    },
    {
      "epoch": 0.7211312022607663,
      "grad_norm": 0.2739093601703644,
      "learning_rate": 8.518125805393386e-05,
      "loss": 4.6471,
      "step": 35215
    },
    {
      "epoch": 0.721233592038171,
      "grad_norm": 0.43531596660614014,
      "learning_rate": 8.512544492318748e-05,
      "loss": 4.582,
      "step": 35220
    },
    {
      "epoch": 0.7213359818155756,
      "grad_norm": 0.27666881680488586,
      "learning_rate": 8.506964283874124e-05,
      "loss": 4.6576,
      "step": 35225
    },
    {
      "epoch": 0.7214383715929802,
      "grad_norm": 0.2962600886821747,
      "learning_rate": 8.501385181009657e-05,
      "loss": 4.7266,
      "step": 35230
    },
    {
      "epoch": 0.7215407613703848,
      "grad_norm": 0.2809500992298126,
      "learning_rate": 8.495807184675303e-05,
      "loss": 4.5417,
      "step": 35235
    },
    {
      "epoch": 0.7216431511477894,
      "grad_norm": 0.27637147903442383,
      "learning_rate": 8.49023029582085e-05,
      "loss": 4.6507,
      "step": 35240
    },
    {
      "epoch": 0.721745540925194,
      "grad_norm": 0.28118348121643066,
      "learning_rate": 8.48465451539588e-05,
      "loss": 4.6487,
      "step": 35245
    },
    {
      "epoch": 0.7218479307025987,
      "grad_norm": 0.27437835931777954,
      "learning_rate": 8.47907984434979e-05,
      "loss": 4.5611,
      "step": 35250
    },
    {
      "epoch": 0.7219503204800033,
      "grad_norm": 0.27034005522727966,
      "learning_rate": 8.473506283631781e-05,
      "loss": 4.5887,
      "step": 35255
    },
    {
      "epoch": 0.7220527102574079,
      "grad_norm": 0.2759099006652832,
      "learning_rate": 8.467933834190885e-05,
      "loss": 4.6269,
      "step": 35260
    },
    {
      "epoch": 0.7221551000348125,
      "grad_norm": 0.44436919689178467,
      "learning_rate": 8.462362496975926e-05,
      "loss": 4.5557,
      "step": 35265
    },
    {
      "epoch": 0.7222574898122172,
      "grad_norm": 0.27312707901000977,
      "learning_rate": 8.456792272935541e-05,
      "loss": 4.6757,
      "step": 35270
    },
    {
      "epoch": 0.7223598795896218,
      "grad_norm": 0.30309104919433594,
      "learning_rate": 8.451223163018192e-05,
      "loss": 4.5831,
      "step": 35275
    },
    {
      "epoch": 0.7224622693670264,
      "grad_norm": 0.2631854712963104,
      "learning_rate": 8.445655168172136e-05,
      "loss": 4.6387,
      "step": 35280
    },
    {
      "epoch": 0.722564659144431,
      "grad_norm": 0.30008718371391296,
      "learning_rate": 8.440088289345438e-05,
      "loss": 4.5845,
      "step": 35285
    },
    {
      "epoch": 0.7226670489218356,
      "grad_norm": 0.287648469209671,
      "learning_rate": 8.434522527486e-05,
      "loss": 4.646,
      "step": 35290
    },
    {
      "epoch": 0.7227694386992403,
      "grad_norm": 0.277083158493042,
      "learning_rate": 8.428957883541483e-05,
      "loss": 4.65,
      "step": 35295
    },
    {
      "epoch": 0.7228718284766449,
      "grad_norm": 0.26783913373947144,
      "learning_rate": 8.423394358459412e-05,
      "loss": 4.6158,
      "step": 35300
    },
    {
      "epoch": 0.7229742182540495,
      "grad_norm": 0.2731381058692932,
      "learning_rate": 8.417831953187093e-05,
      "loss": 4.6213,
      "step": 35305
    },
    {
      "epoch": 0.7230766080314541,
      "grad_norm": 0.2617172598838806,
      "learning_rate": 8.412270668671637e-05,
      "loss": 4.6631,
      "step": 35310
    },
    {
      "epoch": 0.7231789978088587,
      "grad_norm": 0.27751821279525757,
      "learning_rate": 8.406710505859985e-05,
      "loss": 4.6043,
      "step": 35315
    },
    {
      "epoch": 0.7232813875862634,
      "grad_norm": 0.2831033170223236,
      "learning_rate": 8.40115146569887e-05,
      "loss": 4.6168,
      "step": 35320
    },
    {
      "epoch": 0.723383777363668,
      "grad_norm": 0.26409485936164856,
      "learning_rate": 8.395593549134836e-05,
      "loss": 4.6152,
      "step": 35325
    },
    {
      "epoch": 0.7234861671410726,
      "grad_norm": 0.2689374089241028,
      "learning_rate": 8.390036757114246e-05,
      "loss": 4.6712,
      "step": 35330
    },
    {
      "epoch": 0.7235885569184772,
      "grad_norm": 0.30068737268447876,
      "learning_rate": 8.384481090583263e-05,
      "loss": 4.6699,
      "step": 35335
    },
    {
      "epoch": 0.7236909466958819,
      "grad_norm": 0.27237340807914734,
      "learning_rate": 8.378926550487855e-05,
      "loss": 4.6071,
      "step": 35340
    },
    {
      "epoch": 0.7237933364732865,
      "grad_norm": 0.27116408944129944,
      "learning_rate": 8.3733731377738e-05,
      "loss": 4.6052,
      "step": 35345
    },
    {
      "epoch": 0.7238957262506911,
      "grad_norm": 0.26923108100891113,
      "learning_rate": 8.3678208533867e-05,
      "loss": 4.6141,
      "step": 35350
    },
    {
      "epoch": 0.7239981160280957,
      "grad_norm": 0.27835944294929504,
      "learning_rate": 8.362269698271943e-05,
      "loss": 4.5921,
      "step": 35355
    },
    {
      "epoch": 0.7241005058055003,
      "grad_norm": 0.28105825185775757,
      "learning_rate": 8.356719673374729e-05,
      "loss": 4.6204,
      "step": 35360
    },
    {
      "epoch": 0.724202895582905,
      "grad_norm": 0.28705254197120667,
      "learning_rate": 8.351170779640087e-05,
      "loss": 4.6037,
      "step": 35365
    },
    {
      "epoch": 0.7243052853603096,
      "grad_norm": 0.2829848527908325,
      "learning_rate": 8.345623018012814e-05,
      "loss": 4.5822,
      "step": 35370
    },
    {
      "epoch": 0.7244076751377142,
      "grad_norm": 0.2970631420612335,
      "learning_rate": 8.340076389437548e-05,
      "loss": 4.6239,
      "step": 35375
    },
    {
      "epoch": 0.7245100649151188,
      "grad_norm": 0.27193063497543335,
      "learning_rate": 8.33453089485873e-05,
      "loss": 4.6157,
      "step": 35380
    },
    {
      "epoch": 0.7246124546925234,
      "grad_norm": 0.35156217217445374,
      "learning_rate": 8.328986535220593e-05,
      "loss": 4.562,
      "step": 35385
    },
    {
      "epoch": 0.7247148444699282,
      "grad_norm": 0.26313531398773193,
      "learning_rate": 8.323443311467184e-05,
      "loss": 4.6291,
      "step": 35390
    },
    {
      "epoch": 0.7248172342473328,
      "grad_norm": 0.28509581089019775,
      "learning_rate": 8.317901224542354e-05,
      "loss": 4.5954,
      "step": 35395
    },
    {
      "epoch": 0.7249196240247374,
      "grad_norm": 0.27369093894958496,
      "learning_rate": 8.312360275389775e-05,
      "loss": 4.6007,
      "step": 35400
    },
    {
      "epoch": 0.725022013802142,
      "grad_norm": 0.2805980443954468,
      "learning_rate": 8.306820464952902e-05,
      "loss": 4.5833,
      "step": 35405
    },
    {
      "epoch": 0.7251244035795467,
      "grad_norm": 0.2740568518638611,
      "learning_rate": 8.301281794175011e-05,
      "loss": 4.6279,
      "step": 35410
    },
    {
      "epoch": 0.7252267933569513,
      "grad_norm": 0.2633547782897949,
      "learning_rate": 8.295744263999185e-05,
      "loss": 4.6034,
      "step": 35415
    },
    {
      "epoch": 0.7253291831343559,
      "grad_norm": 0.3053842782974243,
      "learning_rate": 8.290207875368307e-05,
      "loss": 4.5261,
      "step": 35420
    },
    {
      "epoch": 0.7254315729117605,
      "grad_norm": 0.2739124596118927,
      "learning_rate": 8.284672629225061e-05,
      "loss": 4.5951,
      "step": 35425
    },
    {
      "epoch": 0.7255339626891651,
      "grad_norm": 0.27901989221572876,
      "learning_rate": 8.279138526511959e-05,
      "loss": 4.6457,
      "step": 35430
    },
    {
      "epoch": 0.7256363524665698,
      "grad_norm": 0.27331113815307617,
      "learning_rate": 8.273605568171278e-05,
      "loss": 4.6224,
      "step": 35435
    },
    {
      "epoch": 0.7257387422439744,
      "grad_norm": 0.312152624130249,
      "learning_rate": 8.268073755145145e-05,
      "loss": 4.6285,
      "step": 35440
    },
    {
      "epoch": 0.725841132021379,
      "grad_norm": 0.28793373703956604,
      "learning_rate": 8.26254308837546e-05,
      "loss": 4.6206,
      "step": 35445
    },
    {
      "epoch": 0.7259435217987836,
      "grad_norm": 0.30180874466896057,
      "learning_rate": 8.25701356880394e-05,
      "loss": 4.6198,
      "step": 35450
    },
    {
      "epoch": 0.7260459115761883,
      "grad_norm": 0.27114638686180115,
      "learning_rate": 8.25148519737211e-05,
      "loss": 4.6382,
      "step": 35455
    },
    {
      "epoch": 0.7261483013535929,
      "grad_norm": 0.2920796275138855,
      "learning_rate": 8.245957975021293e-05,
      "loss": 4.6466,
      "step": 35460
    },
    {
      "epoch": 0.7262506911309975,
      "grad_norm": 0.42375123500823975,
      "learning_rate": 8.240431902692615e-05,
      "loss": 4.5568,
      "step": 35465
    },
    {
      "epoch": 0.7263530809084021,
      "grad_norm": 0.26363077759742737,
      "learning_rate": 8.234906981327018e-05,
      "loss": 4.6527,
      "step": 35470
    },
    {
      "epoch": 0.7264554706858067,
      "grad_norm": 0.26325905323028564,
      "learning_rate": 8.229383211865234e-05,
      "loss": 4.5654,
      "step": 35475
    },
    {
      "epoch": 0.7265578604632114,
      "grad_norm": 0.27255550026893616,
      "learning_rate": 8.223860595247806e-05,
      "loss": 4.6126,
      "step": 35480
    },
    {
      "epoch": 0.726660250240616,
      "grad_norm": 0.2707035541534424,
      "learning_rate": 8.218339132415072e-05,
      "loss": 4.5827,
      "step": 35485
    },
    {
      "epoch": 0.7267626400180206,
      "grad_norm": 0.2907780706882477,
      "learning_rate": 8.212818824307192e-05,
      "loss": 4.5935,
      "step": 35490
    },
    {
      "epoch": 0.7268650297954252,
      "grad_norm": 0.27819907665252686,
      "learning_rate": 8.207299671864117e-05,
      "loss": 4.6256,
      "step": 35495
    },
    {
      "epoch": 0.7269674195728298,
      "grad_norm": 0.2932546138763428,
      "learning_rate": 8.201781676025589e-05,
      "loss": 4.6447,
      "step": 35500
    },
    {
      "epoch": 0.7270698093502345,
      "grad_norm": 0.2775144577026367,
      "learning_rate": 8.196264837731191e-05,
      "loss": 4.6417,
      "step": 35505
    },
    {
      "epoch": 0.7271721991276391,
      "grad_norm": 0.2768743634223938,
      "learning_rate": 8.190749157920257e-05,
      "loss": 4.6247,
      "step": 35510
    },
    {
      "epoch": 0.7272745889050437,
      "grad_norm": 0.28260377049446106,
      "learning_rate": 8.185234637531965e-05,
      "loss": 4.6204,
      "step": 35515
    },
    {
      "epoch": 0.7273769786824483,
      "grad_norm": 0.28319257497787476,
      "learning_rate": 8.179721277505283e-05,
      "loss": 4.6336,
      "step": 35520
    },
    {
      "epoch": 0.727479368459853,
      "grad_norm": 0.2869589924812317,
      "learning_rate": 8.174209078778982e-05,
      "loss": 4.6033,
      "step": 35525
    },
    {
      "epoch": 0.7275817582372576,
      "grad_norm": 0.27521735429763794,
      "learning_rate": 8.168698042291626e-05,
      "loss": 4.5958,
      "step": 35530
    },
    {
      "epoch": 0.7276841480146622,
      "grad_norm": 0.2832813560962677,
      "learning_rate": 8.163188168981587e-05,
      "loss": 4.6506,
      "step": 35535
    },
    {
      "epoch": 0.7277865377920668,
      "grad_norm": 0.28442350029945374,
      "learning_rate": 8.157679459787051e-05,
      "loss": 4.544,
      "step": 35540
    },
    {
      "epoch": 0.7278889275694714,
      "grad_norm": 0.2788909375667572,
      "learning_rate": 8.15217191564599e-05,
      "loss": 4.5901,
      "step": 35545
    },
    {
      "epoch": 0.727991317346876,
      "grad_norm": 0.2873976528644562,
      "learning_rate": 8.146665537496173e-05,
      "loss": 4.6138,
      "step": 35550
    },
    {
      "epoch": 0.7280937071242807,
      "grad_norm": 0.2732342481613159,
      "learning_rate": 8.141160326275205e-05,
      "loss": 4.591,
      "step": 35555
    },
    {
      "epoch": 0.7281960969016853,
      "grad_norm": 0.268839031457901,
      "learning_rate": 8.135656282920437e-05,
      "loss": 4.6524,
      "step": 35560
    },
    {
      "epoch": 0.7282984866790899,
      "grad_norm": 0.3491988778114319,
      "learning_rate": 8.130153408369068e-05,
      "loss": 4.5966,
      "step": 35565
    },
    {
      "epoch": 0.7284008764564945,
      "grad_norm": 0.7563338875770569,
      "learning_rate": 8.124651703558092e-05,
      "loss": 4.5912,
      "step": 35570
    },
    {
      "epoch": 0.7285032662338992,
      "grad_norm": 0.2689594626426697,
      "learning_rate": 8.119151169424271e-05,
      "loss": 4.6556,
      "step": 35575
    },
    {
      "epoch": 0.7286056560113038,
      "grad_norm": 0.27166950702667236,
      "learning_rate": 8.113651806904207e-05,
      "loss": 4.5771,
      "step": 35580
    },
    {
      "epoch": 0.7287080457887084,
      "grad_norm": 0.2878290116786957,
      "learning_rate": 8.108153616934279e-05,
      "loss": 4.6205,
      "step": 35585
    },
    {
      "epoch": 0.728810435566113,
      "grad_norm": 0.2817739248275757,
      "learning_rate": 8.10265660045067e-05,
      "loss": 4.5326,
      "step": 35590
    },
    {
      "epoch": 0.7289128253435178,
      "grad_norm": 0.26659125089645386,
      "learning_rate": 8.097160758389375e-05,
      "loss": 4.6098,
      "step": 35595
    },
    {
      "epoch": 0.7290152151209224,
      "grad_norm": 0.26286768913269043,
      "learning_rate": 8.091666091686175e-05,
      "loss": 4.6222,
      "step": 35600
    },
    {
      "epoch": 0.729117604898327,
      "grad_norm": 0.266060471534729,
      "learning_rate": 8.086172601276653e-05,
      "loss": 4.6103,
      "step": 35605
    },
    {
      "epoch": 0.7292199946757316,
      "grad_norm": 0.3146880865097046,
      "learning_rate": 8.080680288096204e-05,
      "loss": 4.6348,
      "step": 35610
    },
    {
      "epoch": 0.7293223844531362,
      "grad_norm": 0.2808399498462677,
      "learning_rate": 8.075189153080008e-05,
      "loss": 4.594,
      "step": 35615
    },
    {
      "epoch": 0.7294247742305409,
      "grad_norm": 0.287859171628952,
      "learning_rate": 8.06969919716305e-05,
      "loss": 4.6327,
      "step": 35620
    },
    {
      "epoch": 0.7295271640079455,
      "grad_norm": 0.27413034439086914,
      "learning_rate": 8.064210421280107e-05,
      "loss": 4.6109,
      "step": 35625
    },
    {
      "epoch": 0.7296295537853501,
      "grad_norm": 0.27877911925315857,
      "learning_rate": 8.058722826365776e-05,
      "loss": 4.5684,
      "step": 35630
    },
    {
      "epoch": 0.7297319435627547,
      "grad_norm": 0.26801979541778564,
      "learning_rate": 8.053236413354434e-05,
      "loss": 4.5709,
      "step": 35635
    },
    {
      "epoch": 0.7298343333401593,
      "grad_norm": 0.29206186532974243,
      "learning_rate": 8.047751183180252e-05,
      "loss": 4.5712,
      "step": 35640
    },
    {
      "epoch": 0.729936723117564,
      "grad_norm": 0.35441872477531433,
      "learning_rate": 8.042267136777231e-05,
      "loss": 4.5272,
      "step": 35645
    },
    {
      "epoch": 0.7300391128949686,
      "grad_norm": 0.2733302116394043,
      "learning_rate": 8.036784275079123e-05,
      "loss": 4.6058,
      "step": 35650
    },
    {
      "epoch": 0.7301415026723732,
      "grad_norm": 0.28338730335235596,
      "learning_rate": 8.031302599019516e-05,
      "loss": 4.6214,
      "step": 35655
    },
    {
      "epoch": 0.7302438924497778,
      "grad_norm": 0.2677144706249237,
      "learning_rate": 8.025822109531792e-05,
      "loss": 4.6192,
      "step": 35660
    },
    {
      "epoch": 0.7303462822271825,
      "grad_norm": 0.2829006016254425,
      "learning_rate": 8.020342807549113e-05,
      "loss": 4.6536,
      "step": 35665
    },
    {
      "epoch": 0.7304486720045871,
      "grad_norm": 0.2853788137435913,
      "learning_rate": 8.014864694004455e-05,
      "loss": 4.6486,
      "step": 35670
    },
    {
      "epoch": 0.7305510617819917,
      "grad_norm": 0.2703281044960022,
      "learning_rate": 8.009387769830575e-05,
      "loss": 4.6321,
      "step": 35675
    },
    {
      "epoch": 0.7306534515593963,
      "grad_norm": 0.3253069221973419,
      "learning_rate": 8.003912035960051e-05,
      "loss": 4.5723,
      "step": 35680
    },
    {
      "epoch": 0.7307558413368009,
      "grad_norm": 0.2917252779006958,
      "learning_rate": 7.998437493325241e-05,
      "loss": 4.6112,
      "step": 35685
    },
    {
      "epoch": 0.7308582311142056,
      "grad_norm": 0.2790091037750244,
      "learning_rate": 7.992964142858299e-05,
      "loss": 4.6069,
      "step": 35690
    },
    {
      "epoch": 0.7309606208916102,
      "grad_norm": 0.2863832712173462,
      "learning_rate": 7.987491985491197e-05,
      "loss": 4.6444,
      "step": 35695
    },
    {
      "epoch": 0.7310630106690148,
      "grad_norm": 0.27816665172576904,
      "learning_rate": 7.982021022155668e-05,
      "loss": 4.6117,
      "step": 35700
    },
    {
      "epoch": 0.7311654004464194,
      "grad_norm": 0.28824666142463684,
      "learning_rate": 7.97655125378327e-05,
      "loss": 4.6057,
      "step": 35705
    },
    {
      "epoch": 0.731267790223824,
      "grad_norm": 0.30592820048332214,
      "learning_rate": 7.971082681305363e-05,
      "loss": 4.5846,
      "step": 35710
    },
    {
      "epoch": 0.7313701800012287,
      "grad_norm": 0.27617231011390686,
      "learning_rate": 7.965615305653069e-05,
      "loss": 4.5765,
      "step": 35715
    },
    {
      "epoch": 0.7314725697786333,
      "grad_norm": 0.26955723762512207,
      "learning_rate": 7.960149127757342e-05,
      "loss": 4.6057,
      "step": 35720
    },
    {
      "epoch": 0.7315749595560379,
      "grad_norm": 0.27496469020843506,
      "learning_rate": 7.954684148548912e-05,
      "loss": 4.6161,
      "step": 35725
    },
    {
      "epoch": 0.7316773493334425,
      "grad_norm": 0.26673826575279236,
      "learning_rate": 7.949220368958304e-05,
      "loss": 4.5882,
      "step": 35730
    },
    {
      "epoch": 0.7317797391108471,
      "grad_norm": 0.27812322974205017,
      "learning_rate": 7.943757789915861e-05,
      "loss": 4.6123,
      "step": 35735
    },
    {
      "epoch": 0.7318821288882518,
      "grad_norm": 0.2739894688129425,
      "learning_rate": 7.938296412351692e-05,
      "loss": 4.6414,
      "step": 35740
    },
    {
      "epoch": 0.7319845186656564,
      "grad_norm": 0.28313854336738586,
      "learning_rate": 7.932836237195723e-05,
      "loss": 4.5572,
      "step": 35745
    },
    {
      "epoch": 0.732086908443061,
      "grad_norm": 0.2770625948905945,
      "learning_rate": 7.927377265377656e-05,
      "loss": 4.5784,
      "step": 35750
    },
    {
      "epoch": 0.7321892982204656,
      "grad_norm": 0.27078017592430115,
      "learning_rate": 7.921919497827012e-05,
      "loss": 4.6036,
      "step": 35755
    },
    {
      "epoch": 0.7322916879978703,
      "grad_norm": 0.28478506207466125,
      "learning_rate": 7.916462935473089e-05,
      "loss": 4.6117,
      "step": 35760
    },
    {
      "epoch": 0.7323940777752749,
      "grad_norm": 0.27009883522987366,
      "learning_rate": 7.911007579244978e-05,
      "loss": 4.5985,
      "step": 35765
    },
    {
      "epoch": 0.7324964675526795,
      "grad_norm": 0.2889600396156311,
      "learning_rate": 7.905553430071589e-05,
      "loss": 4.619,
      "step": 35770
    },
    {
      "epoch": 0.7325988573300841,
      "grad_norm": 0.27302753925323486,
      "learning_rate": 7.900100488881598e-05,
      "loss": 4.6127,
      "step": 35775
    },
    {
      "epoch": 0.7327012471074887,
      "grad_norm": 0.27285054326057434,
      "learning_rate": 7.89464875660348e-05,
      "loss": 4.6208,
      "step": 35780
    },
    {
      "epoch": 0.7328036368848934,
      "grad_norm": 0.2659536600112915,
      "learning_rate": 7.889198234165529e-05,
      "loss": 4.6039,
      "step": 35785
    },
    {
      "epoch": 0.732906026662298,
      "grad_norm": 0.2752051055431366,
      "learning_rate": 7.883748922495794e-05,
      "loss": 4.5993,
      "step": 35790
    },
    {
      "epoch": 0.7330084164397027,
      "grad_norm": 0.2768678069114685,
      "learning_rate": 7.87830082252215e-05,
      "loss": 4.5965,
      "step": 35795
    },
    {
      "epoch": 0.7331108062171073,
      "grad_norm": 0.2679316997528076,
      "learning_rate": 7.872853935172255e-05,
      "loss": 4.6243,
      "step": 35800
    },
    {
      "epoch": 0.733213195994512,
      "grad_norm": 0.2648327350616455,
      "learning_rate": 7.86740826137356e-05,
      "loss": 4.6027,
      "step": 35805
    },
    {
      "epoch": 0.7333155857719166,
      "grad_norm": 0.2949255108833313,
      "learning_rate": 7.861963802053305e-05,
      "loss": 4.6423,
      "step": 35810
    },
    {
      "epoch": 0.7334179755493212,
      "grad_norm": 0.29031598567962646,
      "learning_rate": 7.856520558138523e-05,
      "loss": 4.5904,
      "step": 35815
    },
    {
      "epoch": 0.7335203653267258,
      "grad_norm": 0.4511967599391937,
      "learning_rate": 7.851078530556055e-05,
      "loss": 4.6195,
      "step": 35820
    },
    {
      "epoch": 0.7336227551041304,
      "grad_norm": 0.3028342127799988,
      "learning_rate": 7.845637720232522e-05,
      "loss": 4.6299,
      "step": 35825
    },
    {
      "epoch": 0.7337251448815351,
      "grad_norm": 0.2744481861591339,
      "learning_rate": 7.84019812809433e-05,
      "loss": 4.5955,
      "step": 35830
    },
    {
      "epoch": 0.7338275346589397,
      "grad_norm": 0.2706994116306305,
      "learning_rate": 7.834759755067704e-05,
      "loss": 4.5803,
      "step": 35835
    },
    {
      "epoch": 0.7339299244363443,
      "grad_norm": 0.2765248119831085,
      "learning_rate": 7.829322602078627e-05,
      "loss": 4.5825,
      "step": 35840
    },
    {
      "epoch": 0.7340323142137489,
      "grad_norm": 0.2680410146713257,
      "learning_rate": 7.8238866700529e-05,
      "loss": 4.5614,
      "step": 35845
    },
    {
      "epoch": 0.7341347039911535,
      "grad_norm": 0.29486092925071716,
      "learning_rate": 7.818451959916122e-05,
      "loss": 4.6369,
      "step": 35850
    },
    {
      "epoch": 0.7342370937685582,
      "grad_norm": 0.3106747269630432,
      "learning_rate": 7.813018472593645e-05,
      "loss": 4.6024,
      "step": 35855
    },
    {
      "epoch": 0.7343394835459628,
      "grad_norm": 0.27558979392051697,
      "learning_rate": 7.807586209010655e-05,
      "loss": 4.5633,
      "step": 35860
    },
    {
      "epoch": 0.7344418733233674,
      "grad_norm": 0.27722612023353577,
      "learning_rate": 7.802155170092112e-05,
      "loss": 4.6287,
      "step": 35865
    },
    {
      "epoch": 0.734544263100772,
      "grad_norm": 0.2866869866847992,
      "learning_rate": 7.796725356762754e-05,
      "loss": 4.5713,
      "step": 35870
    },
    {
      "epoch": 0.7346466528781767,
      "grad_norm": 0.2874201238155365,
      "learning_rate": 7.791296769947144e-05,
      "loss": 4.6416,
      "step": 35875
    },
    {
      "epoch": 0.7347490426555813,
      "grad_norm": 0.27711090445518494,
      "learning_rate": 7.785869410569605e-05,
      "loss": 4.6211,
      "step": 35880
    },
    {
      "epoch": 0.7348514324329859,
      "grad_norm": 0.26898038387298584,
      "learning_rate": 7.780443279554268e-05,
      "loss": 4.6596,
      "step": 35885
    },
    {
      "epoch": 0.7349538222103905,
      "grad_norm": 0.2773260176181793,
      "learning_rate": 7.775018377825041e-05,
      "loss": 4.6504,
      "step": 35890
    },
    {
      "epoch": 0.7350562119877951,
      "grad_norm": 0.2722721993923187,
      "learning_rate": 7.76959470630564e-05,
      "loss": 4.588,
      "step": 35895
    },
    {
      "epoch": 0.7351586017651998,
      "grad_norm": 0.28141921758651733,
      "learning_rate": 7.764172265919561e-05,
      "loss": 4.6155,
      "step": 35900
    },
    {
      "epoch": 0.7352609915426044,
      "grad_norm": 0.27787846326828003,
      "learning_rate": 7.758751057590084e-05,
      "loss": 4.6119,
      "step": 35905
    },
    {
      "epoch": 0.735363381320009,
      "grad_norm": 0.26910316944122314,
      "learning_rate": 7.753331082240302e-05,
      "loss": 4.64,
      "step": 35910
    },
    {
      "epoch": 0.7354657710974136,
      "grad_norm": 0.26209381222724915,
      "learning_rate": 7.747912340793073e-05,
      "loss": 4.6252,
      "step": 35915
    },
    {
      "epoch": 0.7355681608748182,
      "grad_norm": 0.2831920087337494,
      "learning_rate": 7.742494834171055e-05,
      "loss": 4.6328,
      "step": 35920
    },
    {
      "epoch": 0.7356705506522229,
      "grad_norm": 0.28854113817214966,
      "learning_rate": 7.737078563296706e-05,
      "loss": 4.5855,
      "step": 35925
    },
    {
      "epoch": 0.7357729404296275,
      "grad_norm": 0.2855021059513092,
      "learning_rate": 7.731663529092248e-05,
      "loss": 4.6104,
      "step": 35930
    },
    {
      "epoch": 0.7358753302070321,
      "grad_norm": 0.3063238561153412,
      "learning_rate": 7.72624973247972e-05,
      "loss": 4.5534,
      "step": 35935
    },
    {
      "epoch": 0.7359777199844367,
      "grad_norm": 0.3179500997066498,
      "learning_rate": 7.72083717438093e-05,
      "loss": 4.6236,
      "step": 35940
    },
    {
      "epoch": 0.7360801097618414,
      "grad_norm": 0.27991044521331787,
      "learning_rate": 7.715425855717492e-05,
      "loss": 4.6413,
      "step": 35945
    },
    {
      "epoch": 0.736182499539246,
      "grad_norm": 0.27822068333625793,
      "learning_rate": 7.710015777410798e-05,
      "loss": 4.6782,
      "step": 35950
    },
    {
      "epoch": 0.7362848893166506,
      "grad_norm": 0.27126359939575195,
      "learning_rate": 7.704606940382024e-05,
      "loss": 4.662,
      "step": 35955
    },
    {
      "epoch": 0.7363872790940552,
      "grad_norm": 0.2927936017513275,
      "learning_rate": 7.699199345552151e-05,
      "loss": 4.5777,
      "step": 35960
    },
    {
      "epoch": 0.7364896688714598,
      "grad_norm": 0.27688297629356384,
      "learning_rate": 7.693792993841936e-05,
      "loss": 4.6065,
      "step": 35965
    },
    {
      "epoch": 0.7365920586488645,
      "grad_norm": 0.26905474066734314,
      "learning_rate": 7.688387886171924e-05,
      "loss": 4.6138,
      "step": 35970
    },
    {
      "epoch": 0.7366944484262691,
      "grad_norm": 0.26644816994667053,
      "learning_rate": 7.682984023462463e-05,
      "loss": 4.6613,
      "step": 35975
    },
    {
      "epoch": 0.7367968382036737,
      "grad_norm": 0.2702935039997101,
      "learning_rate": 7.677581406633661e-05,
      "loss": 4.622,
      "step": 35980
    },
    {
      "epoch": 0.7368992279810783,
      "grad_norm": 0.2773968279361725,
      "learning_rate": 7.672180036605437e-05,
      "loss": 4.5575,
      "step": 35985
    },
    {
      "epoch": 0.7370016177584829,
      "grad_norm": 0.2763274908065796,
      "learning_rate": 7.666779914297508e-05,
      "loss": 4.6339,
      "step": 35990
    },
    {
      "epoch": 0.7371040075358876,
      "grad_norm": 0.28777623176574707,
      "learning_rate": 7.661381040629338e-05,
      "loss": 4.6024,
      "step": 35995
    },
    {
      "epoch": 0.7372063973132923,
      "grad_norm": 0.3007139265537262,
      "learning_rate": 7.655983416520213e-05,
      "loss": 4.5967,
      "step": 36000
    },
    {
      "epoch": 0.7372063973132923,
      "eval_loss": 4.625242233276367,
      "eval_runtime": 17.131,
      "eval_samples_per_second": 33.974,
      "eval_steps_per_second": 4.261,
      "step": 36000
    },
    {
      "epoch": 0.7373087870906969,
      "grad_norm": 0.3373059332370758,
      "learning_rate": 7.6505870428892e-05,
      "loss": 4.6184,
      "step": 36005
    },
    {
      "epoch": 0.7374111768681015,
      "grad_norm": 0.28736239671707153,
      "learning_rate": 7.645191920655133e-05,
      "loss": 4.6159,
      "step": 36010
    },
    {
      "epoch": 0.7375135666455062,
      "grad_norm": 0.26335597038269043,
      "learning_rate": 7.639798050736667e-05,
      "loss": 4.6496,
      "step": 36015
    },
    {
      "epoch": 0.7376159564229108,
      "grad_norm": 0.3809347152709961,
      "learning_rate": 7.634405434052217e-05,
      "loss": 4.5739,
      "step": 36020
    },
    {
      "epoch": 0.7377183462003154,
      "grad_norm": 0.27838921546936035,
      "learning_rate": 7.629014071519994e-05,
      "loss": 4.6096,
      "step": 36025
    },
    {
      "epoch": 0.73782073597772,
      "grad_norm": 0.38840600848197937,
      "learning_rate": 7.623623964057987e-05,
      "loss": 4.5863,
      "step": 36030
    },
    {
      "epoch": 0.7379231257551246,
      "grad_norm": 0.267988920211792,
      "learning_rate": 7.618235112583991e-05,
      "loss": 4.5714,
      "step": 36035
    },
    {
      "epoch": 0.7380255155325293,
      "grad_norm": 0.26880499720573425,
      "learning_rate": 7.61284751801557e-05,
      "loss": 4.6244,
      "step": 36040
    },
    {
      "epoch": 0.7381279053099339,
      "grad_norm": 0.282042920589447,
      "learning_rate": 7.607461181270073e-05,
      "loss": 4.6137,
      "step": 36045
    },
    {
      "epoch": 0.7382302950873385,
      "grad_norm": 0.28289055824279785,
      "learning_rate": 7.60207610326465e-05,
      "loss": 4.6429,
      "step": 36050
    },
    {
      "epoch": 0.7383326848647431,
      "grad_norm": 0.2695794999599457,
      "learning_rate": 7.596692284916223e-05,
      "loss": 4.6405,
      "step": 36055
    },
    {
      "epoch": 0.7384350746421477,
      "grad_norm": 0.28726011514663696,
      "learning_rate": 7.5913097271415e-05,
      "loss": 4.6271,
      "step": 36060
    },
    {
      "epoch": 0.7385374644195524,
      "grad_norm": 0.26992538571357727,
      "learning_rate": 7.585928430856991e-05,
      "loss": 4.6412,
      "step": 36065
    },
    {
      "epoch": 0.738639854196957,
      "grad_norm": 0.2706857919692993,
      "learning_rate": 7.580548396978962e-05,
      "loss": 4.6197,
      "step": 36070
    },
    {
      "epoch": 0.7387422439743616,
      "grad_norm": 0.3034684956073761,
      "learning_rate": 7.575169626423491e-05,
      "loss": 4.6319,
      "step": 36075
    },
    {
      "epoch": 0.7388446337517662,
      "grad_norm": 0.27745088934898376,
      "learning_rate": 7.56979212010642e-05,
      "loss": 4.5837,
      "step": 36080
    },
    {
      "epoch": 0.7389470235291709,
      "grad_norm": 0.2573217451572418,
      "learning_rate": 7.564415878943402e-05,
      "loss": 4.6254,
      "step": 36085
    },
    {
      "epoch": 0.7390494133065755,
      "grad_norm": 0.28609785437583923,
      "learning_rate": 7.559040903849848e-05,
      "loss": 4.5843,
      "step": 36090
    },
    {
      "epoch": 0.7391518030839801,
      "grad_norm": 0.286507785320282,
      "learning_rate": 7.553667195740963e-05,
      "loss": 4.6006,
      "step": 36095
    },
    {
      "epoch": 0.7392541928613847,
      "grad_norm": 0.5472362041473389,
      "learning_rate": 7.548294755531743e-05,
      "loss": 4.6578,
      "step": 36100
    },
    {
      "epoch": 0.7393565826387893,
      "grad_norm": 0.2746458947658539,
      "learning_rate": 7.542923584136963e-05,
      "loss": 4.5932,
      "step": 36105
    },
    {
      "epoch": 0.739458972416194,
      "grad_norm": 0.28057464957237244,
      "learning_rate": 7.53755368247117e-05,
      "loss": 4.5992,
      "step": 36110
    },
    {
      "epoch": 0.7395613621935986,
      "grad_norm": 0.3029986321926117,
      "learning_rate": 7.532185051448728e-05,
      "loss": 4.6165,
      "step": 36115
    },
    {
      "epoch": 0.7396637519710032,
      "grad_norm": 0.2978089451789856,
      "learning_rate": 7.526817691983738e-05,
      "loss": 4.6251,
      "step": 36120
    },
    {
      "epoch": 0.7397661417484078,
      "grad_norm": 0.2777213156223297,
      "learning_rate": 7.521451604990127e-05,
      "loss": 4.581,
      "step": 36125
    },
    {
      "epoch": 0.7398685315258124,
      "grad_norm": 0.27353715896606445,
      "learning_rate": 7.516086791381579e-05,
      "loss": 4.6552,
      "step": 36130
    },
    {
      "epoch": 0.7399709213032171,
      "grad_norm": 0.28740575909614563,
      "learning_rate": 7.510723252071569e-05,
      "loss": 4.6224,
      "step": 36135
    },
    {
      "epoch": 0.7400733110806217,
      "grad_norm": 0.307476282119751,
      "learning_rate": 7.505360987973366e-05,
      "loss": 4.6363,
      "step": 36140
    },
    {
      "epoch": 0.7401757008580263,
      "grad_norm": 0.29532861709594727,
      "learning_rate": 7.500000000000002e-05,
      "loss": 4.6364,
      "step": 36145
    },
    {
      "epoch": 0.7402780906354309,
      "grad_norm": 0.3010016083717346,
      "learning_rate": 7.4946402890643e-05,
      "loss": 4.6357,
      "step": 36150
    },
    {
      "epoch": 0.7403804804128356,
      "grad_norm": 0.3083871901035309,
      "learning_rate": 7.489281856078879e-05,
      "loss": 4.6255,
      "step": 36155
    },
    {
      "epoch": 0.7404828701902402,
      "grad_norm": 0.3375720679759979,
      "learning_rate": 7.48392470195612e-05,
      "loss": 4.5509,
      "step": 36160
    },
    {
      "epoch": 0.7405852599676448,
      "grad_norm": 0.27293142676353455,
      "learning_rate": 7.478568827608194e-05,
      "loss": 4.6333,
      "step": 36165
    },
    {
      "epoch": 0.7406876497450494,
      "grad_norm": 0.27576547861099243,
      "learning_rate": 7.473214233947053e-05,
      "loss": 4.6031,
      "step": 36170
    },
    {
      "epoch": 0.740790039522454,
      "grad_norm": 0.292348176240921,
      "learning_rate": 7.46786092188444e-05,
      "loss": 4.6116,
      "step": 36175
    },
    {
      "epoch": 0.7408924292998587,
      "grad_norm": 0.28747648000717163,
      "learning_rate": 7.462508892331871e-05,
      "loss": 4.592,
      "step": 36180
    },
    {
      "epoch": 0.7409948190772633,
      "grad_norm": 0.2726137340068817,
      "learning_rate": 7.457158146200635e-05,
      "loss": 4.6099,
      "step": 36185
    },
    {
      "epoch": 0.7410972088546679,
      "grad_norm": 0.27743956446647644,
      "learning_rate": 7.451808684401827e-05,
      "loss": 4.5679,
      "step": 36190
    },
    {
      "epoch": 0.7411995986320725,
      "grad_norm": 0.273608922958374,
      "learning_rate": 7.446460507846303e-05,
      "loss": 4.6028,
      "step": 36195
    },
    {
      "epoch": 0.7413019884094773,
      "grad_norm": 0.30002596974372864,
      "learning_rate": 7.441113617444698e-05,
      "loss": 4.5459,
      "step": 36200
    },
    {
      "epoch": 0.7414043781868819,
      "grad_norm": 0.29475438594818115,
      "learning_rate": 7.435768014107457e-05,
      "loss": 4.6182,
      "step": 36205
    },
    {
      "epoch": 0.7415067679642865,
      "grad_norm": 0.26289767026901245,
      "learning_rate": 7.430423698744758e-05,
      "loss": 4.6427,
      "step": 36210
    },
    {
      "epoch": 0.7416091577416911,
      "grad_norm": 0.2837486267089844,
      "learning_rate": 7.425080672266606e-05,
      "loss": 4.5846,
      "step": 36215
    },
    {
      "epoch": 0.7417115475190957,
      "grad_norm": 0.29650506377220154,
      "learning_rate": 7.419738935582756e-05,
      "loss": 4.6243,
      "step": 36220
    },
    {
      "epoch": 0.7418139372965004,
      "grad_norm": 0.2735283374786377,
      "learning_rate": 7.414398489602765e-05,
      "loss": 4.6259,
      "step": 36225
    },
    {
      "epoch": 0.741916327073905,
      "grad_norm": 0.26285502314567566,
      "learning_rate": 7.409059335235955e-05,
      "loss": 4.5776,
      "step": 36230
    },
    {
      "epoch": 0.7420187168513096,
      "grad_norm": 0.2663193643093109,
      "learning_rate": 7.403721473391423e-05,
      "loss": 4.6158,
      "step": 36235
    },
    {
      "epoch": 0.7421211066287142,
      "grad_norm": 0.31137824058532715,
      "learning_rate": 7.398384904978075e-05,
      "loss": 4.6326,
      "step": 36240
    },
    {
      "epoch": 0.7422234964061188,
      "grad_norm": 0.2721947133541107,
      "learning_rate": 7.393049630904564e-05,
      "loss": 4.5878,
      "step": 36245
    },
    {
      "epoch": 0.7423258861835235,
      "grad_norm": 0.3060890734195709,
      "learning_rate": 7.387715652079335e-05,
      "loss": 4.6295,
      "step": 36250
    },
    {
      "epoch": 0.7424282759609281,
      "grad_norm": 0.2630886137485504,
      "learning_rate": 7.382382969410631e-05,
      "loss": 4.5788,
      "step": 36255
    },
    {
      "epoch": 0.7425306657383327,
      "grad_norm": 0.274039626121521,
      "learning_rate": 7.377051583806431e-05,
      "loss": 4.6033,
      "step": 36260
    },
    {
      "epoch": 0.7426330555157373,
      "grad_norm": 0.3136425018310547,
      "learning_rate": 7.371721496174538e-05,
      "loss": 4.6146,
      "step": 36265
    },
    {
      "epoch": 0.742735445293142,
      "grad_norm": 0.27234914898872375,
      "learning_rate": 7.36639270742251e-05,
      "loss": 4.6283,
      "step": 36270
    },
    {
      "epoch": 0.7428378350705466,
      "grad_norm": 0.27690574526786804,
      "learning_rate": 7.36106521845768e-05,
      "loss": 4.6369,
      "step": 36275
    },
    {
      "epoch": 0.7429402248479512,
      "grad_norm": 0.2764059603214264,
      "learning_rate": 7.355739030187184e-05,
      "loss": 4.6023,
      "step": 36280
    },
    {
      "epoch": 0.7430426146253558,
      "grad_norm": 0.27220767736434937,
      "learning_rate": 7.350414143517914e-05,
      "loss": 4.5813,
      "step": 36285
    },
    {
      "epoch": 0.7431450044027604,
      "grad_norm": 0.2708590626716614,
      "learning_rate": 7.34509055935654e-05,
      "loss": 4.6038,
      "step": 36290
    },
    {
      "epoch": 0.743247394180165,
      "grad_norm": 0.2739364504814148,
      "learning_rate": 7.339768278609531e-05,
      "loss": 4.6233,
      "step": 36295
    },
    {
      "epoch": 0.7433497839575697,
      "grad_norm": 0.2813643217086792,
      "learning_rate": 7.334447302183114e-05,
      "loss": 4.5869,
      "step": 36300
    },
    {
      "epoch": 0.7434521737349743,
      "grad_norm": 0.2877044081687927,
      "learning_rate": 7.329127630983302e-05,
      "loss": 4.6042,
      "step": 36305
    },
    {
      "epoch": 0.7435545635123789,
      "grad_norm": 0.3283853530883789,
      "learning_rate": 7.323809265915877e-05,
      "loss": 4.5801,
      "step": 36310
    },
    {
      "epoch": 0.7436569532897835,
      "grad_norm": 0.27724689245224,
      "learning_rate": 7.318492207886417e-05,
      "loss": 4.6033,
      "step": 36315
    },
    {
      "epoch": 0.7437593430671882,
      "grad_norm": 0.27958622574806213,
      "learning_rate": 7.313176457800264e-05,
      "loss": 4.6044,
      "step": 36320
    },
    {
      "epoch": 0.7438617328445928,
      "grad_norm": 0.27751678228378296,
      "learning_rate": 7.30786201656253e-05,
      "loss": 4.6069,
      "step": 36325
    },
    {
      "epoch": 0.7439641226219974,
      "grad_norm": 0.3063555955886841,
      "learning_rate": 7.302548885078128e-05,
      "loss": 4.5691,
      "step": 36330
    },
    {
      "epoch": 0.744066512399402,
      "grad_norm": 0.29780712723731995,
      "learning_rate": 7.297237064251728e-05,
      "loss": 4.6259,
      "step": 36335
    },
    {
      "epoch": 0.7441689021768066,
      "grad_norm": 0.2759324610233307,
      "learning_rate": 7.291926554987775e-05,
      "loss": 4.5851,
      "step": 36340
    },
    {
      "epoch": 0.7442712919542113,
      "grad_norm": 0.28593888878822327,
      "learning_rate": 7.286617358190519e-05,
      "loss": 4.5673,
      "step": 36345
    },
    {
      "epoch": 0.7443736817316159,
      "grad_norm": 0.2743288278579712,
      "learning_rate": 7.281309474763939e-05,
      "loss": 4.6147,
      "step": 36350
    },
    {
      "epoch": 0.7444760715090205,
      "grad_norm": 0.3501961827278137,
      "learning_rate": 7.276002905611837e-05,
      "loss": 4.6164,
      "step": 36355
    },
    {
      "epoch": 0.7445784612864251,
      "grad_norm": 0.3279406428337097,
      "learning_rate": 7.27069765163776e-05,
      "loss": 4.5896,
      "step": 36360
    },
    {
      "epoch": 0.7446808510638298,
      "grad_norm": 0.2953859567642212,
      "learning_rate": 7.265393713745052e-05,
      "loss": 4.6175,
      "step": 36365
    },
    {
      "epoch": 0.7447832408412344,
      "grad_norm": 0.3140943646430969,
      "learning_rate": 7.26009109283682e-05,
      "loss": 4.5764,
      "step": 36370
    },
    {
      "epoch": 0.744885630618639,
      "grad_norm": 0.2758210003376007,
      "learning_rate": 7.254789789815944e-05,
      "loss": 4.6308,
      "step": 36375
    },
    {
      "epoch": 0.7449880203960436,
      "grad_norm": 0.27534064650535583,
      "learning_rate": 7.249489805585096e-05,
      "loss": 4.628,
      "step": 36380
    },
    {
      "epoch": 0.7450904101734482,
      "grad_norm": 0.30455341935157776,
      "learning_rate": 7.24419114104671e-05,
      "loss": 4.5605,
      "step": 36385
    },
    {
      "epoch": 0.7451927999508529,
      "grad_norm": 0.27069637179374695,
      "learning_rate": 7.238893797102989e-05,
      "loss": 4.6278,
      "step": 36390
    },
    {
      "epoch": 0.7452951897282575,
      "grad_norm": 0.30609190464019775,
      "learning_rate": 7.233597774655942e-05,
      "loss": 4.6255,
      "step": 36395
    },
    {
      "epoch": 0.7453975795056621,
      "grad_norm": 0.6586396098136902,
      "learning_rate": 7.228303074607307e-05,
      "loss": 4.6865,
      "step": 36400
    },
    {
      "epoch": 0.7454999692830668,
      "grad_norm": 0.2929151654243469,
      "learning_rate": 7.223009697858638e-05,
      "loss": 4.663,
      "step": 36405
    },
    {
      "epoch": 0.7456023590604715,
      "grad_norm": 0.284393846988678,
      "learning_rate": 7.217717645311243e-05,
      "loss": 4.6215,
      "step": 36410
    },
    {
      "epoch": 0.7457047488378761,
      "grad_norm": 0.2875935435295105,
      "learning_rate": 7.2124269178662e-05,
      "loss": 4.6083,
      "step": 36415
    },
    {
      "epoch": 0.7458071386152807,
      "grad_norm": 0.2916814684867859,
      "learning_rate": 7.207137516424385e-05,
      "loss": 4.5626,
      "step": 36420
    },
    {
      "epoch": 0.7459095283926853,
      "grad_norm": 0.2665678858757019,
      "learning_rate": 7.201849441886426e-05,
      "loss": 4.6012,
      "step": 36425
    },
    {
      "epoch": 0.7460119181700899,
      "grad_norm": 0.2775857150554657,
      "learning_rate": 7.196562695152727e-05,
      "loss": 4.6326,
      "step": 36430
    },
    {
      "epoch": 0.7461143079474946,
      "grad_norm": 0.2888798415660858,
      "learning_rate": 7.191277277123483e-05,
      "loss": 4.6155,
      "step": 36435
    },
    {
      "epoch": 0.7462166977248992,
      "grad_norm": 0.2871415913105011,
      "learning_rate": 7.185993188698645e-05,
      "loss": 4.5875,
      "step": 36440
    },
    {
      "epoch": 0.7463190875023038,
      "grad_norm": 0.272702157497406,
      "learning_rate": 7.180710430777943e-05,
      "loss": 4.5732,
      "step": 36445
    },
    {
      "epoch": 0.7464214772797084,
      "grad_norm": 0.2818405032157898,
      "learning_rate": 7.175429004260876e-05,
      "loss": 4.5388,
      "step": 36450
    },
    {
      "epoch": 0.746523867057113,
      "grad_norm": 0.28034403920173645,
      "learning_rate": 7.170148910046734e-05,
      "loss": 4.6082,
      "step": 36455
    },
    {
      "epoch": 0.7466262568345177,
      "grad_norm": 0.2826538383960724,
      "learning_rate": 7.164870149034559e-05,
      "loss": 4.5642,
      "step": 36460
    },
    {
      "epoch": 0.7467286466119223,
      "grad_norm": 0.2753027379512787,
      "learning_rate": 7.159592722123173e-05,
      "loss": 4.6034,
      "step": 36465
    },
    {
      "epoch": 0.7468310363893269,
      "grad_norm": 0.27757155895233154,
      "learning_rate": 7.154316630211179e-05,
      "loss": 4.5861,
      "step": 36470
    },
    {
      "epoch": 0.7469334261667315,
      "grad_norm": 0.2987254559993744,
      "learning_rate": 7.149041874196945e-05,
      "loss": 4.6495,
      "step": 36475
    },
    {
      "epoch": 0.7470358159441362,
      "grad_norm": 0.28315460681915283,
      "learning_rate": 7.143768454978602e-05,
      "loss": 4.5983,
      "step": 36480
    },
    {
      "epoch": 0.7471382057215408,
      "grad_norm": 0.2767767310142517,
      "learning_rate": 7.138496373454085e-05,
      "loss": 4.6073,
      "step": 36485
    },
    {
      "epoch": 0.7472405954989454,
      "grad_norm": 0.28215596079826355,
      "learning_rate": 7.133225630521056e-05,
      "loss": 4.6359,
      "step": 36490
    },
    {
      "epoch": 0.74734298527635,
      "grad_norm": 0.2782031297683716,
      "learning_rate": 7.127956227076992e-05,
      "loss": 4.6096,
      "step": 36495
    },
    {
      "epoch": 0.7474453750537546,
      "grad_norm": 0.2703072428703308,
      "learning_rate": 7.122688164019111e-05,
      "loss": 4.6411,
      "step": 36500
    },
    {
      "epoch": 0.7475477648311593,
      "grad_norm": 0.2731671631336212,
      "learning_rate": 7.117421442244423e-05,
      "loss": 4.6308,
      "step": 36505
    },
    {
      "epoch": 0.7476501546085639,
      "grad_norm": 0.29174426198005676,
      "learning_rate": 7.112156062649701e-05,
      "loss": 4.6083,
      "step": 36510
    },
    {
      "epoch": 0.7477525443859685,
      "grad_norm": 0.28482604026794434,
      "learning_rate": 7.10689202613148e-05,
      "loss": 4.593,
      "step": 36515
    },
    {
      "epoch": 0.7478549341633731,
      "grad_norm": 0.2897595763206482,
      "learning_rate": 7.10162933358609e-05,
      "loss": 4.5042,
      "step": 36520
    },
    {
      "epoch": 0.7479573239407777,
      "grad_norm": 0.2659852206707001,
      "learning_rate": 7.096367985909615e-05,
      "loss": 4.5957,
      "step": 36525
    },
    {
      "epoch": 0.7480597137181824,
      "grad_norm": 0.2799854874610901,
      "learning_rate": 7.091107983997904e-05,
      "loss": 4.5992,
      "step": 36530
    },
    {
      "epoch": 0.748162103495587,
      "grad_norm": 0.2943868339061737,
      "learning_rate": 7.085849328746606e-05,
      "loss": 4.5843,
      "step": 36535
    },
    {
      "epoch": 0.7482644932729916,
      "grad_norm": 0.2765563726425171,
      "learning_rate": 7.080592021051096e-05,
      "loss": 4.5918,
      "step": 36540
    },
    {
      "epoch": 0.7483668830503962,
      "grad_norm": 0.2719048857688904,
      "learning_rate": 7.075336061806565e-05,
      "loss": 4.6418,
      "step": 36545
    },
    {
      "epoch": 0.7484692728278008,
      "grad_norm": 0.2761386036872864,
      "learning_rate": 7.070081451907945e-05,
      "loss": 4.6334,
      "step": 36550
    },
    {
      "epoch": 0.7485716626052055,
      "grad_norm": 0.3091813921928406,
      "learning_rate": 7.064828192249944e-05,
      "loss": 4.5794,
      "step": 36555
    },
    {
      "epoch": 0.7486740523826101,
      "grad_norm": 0.27586930990219116,
      "learning_rate": 7.059576283727054e-05,
      "loss": 4.6489,
      "step": 36560
    },
    {
      "epoch": 0.7487764421600147,
      "grad_norm": 0.27593696117401123,
      "learning_rate": 7.05432572723352e-05,
      "loss": 4.5955,
      "step": 36565
    },
    {
      "epoch": 0.7488788319374193,
      "grad_norm": 0.27841171622276306,
      "learning_rate": 7.049076523663357e-05,
      "loss": 4.6438,
      "step": 36570
    },
    {
      "epoch": 0.748981221714824,
      "grad_norm": 0.27219220995903015,
      "learning_rate": 7.04382867391037e-05,
      "loss": 4.6261,
      "step": 36575
    },
    {
      "epoch": 0.7490836114922286,
      "grad_norm": 0.27954012155532837,
      "learning_rate": 7.03858217886811e-05,
      "loss": 4.5801,
      "step": 36580
    },
    {
      "epoch": 0.7491860012696332,
      "grad_norm": 0.2719559073448181,
      "learning_rate": 7.03333703942991e-05,
      "loss": 4.5984,
      "step": 36585
    },
    {
      "epoch": 0.7492883910470378,
      "grad_norm": 0.345603883266449,
      "learning_rate": 7.02809325648886e-05,
      "loss": 4.5991,
      "step": 36590
    },
    {
      "epoch": 0.7493907808244424,
      "grad_norm": 0.2791408896446228,
      "learning_rate": 7.022850830937841e-05,
      "loss": 4.5987,
      "step": 36595
    },
    {
      "epoch": 0.7494931706018471,
      "grad_norm": 0.26974305510520935,
      "learning_rate": 7.017609763669484e-05,
      "loss": 4.5859,
      "step": 36600
    },
    {
      "epoch": 0.7495955603792518,
      "grad_norm": 0.3174270689487457,
      "learning_rate": 7.012370055576187e-05,
      "loss": 4.5971,
      "step": 36605
    },
    {
      "epoch": 0.7496979501566564,
      "grad_norm": 0.29159948229789734,
      "learning_rate": 7.007131707550143e-05,
      "loss": 4.6908,
      "step": 36610
    },
    {
      "epoch": 0.749800339934061,
      "grad_norm": 0.5217298865318298,
      "learning_rate": 7.00189472048327e-05,
      "loss": 4.5889,
      "step": 36615
    },
    {
      "epoch": 0.7499027297114657,
      "grad_norm": 0.2673492431640625,
      "learning_rate": 6.996659095267292e-05,
      "loss": 4.5983,
      "step": 36620
    },
    {
      "epoch": 0.7500051194888703,
      "grad_norm": 0.26948046684265137,
      "learning_rate": 6.991424832793696e-05,
      "loss": 4.6268,
      "step": 36625
    },
    {
      "epoch": 0.7501075092662749,
      "grad_norm": 0.2933151125907898,
      "learning_rate": 6.98619193395371e-05,
      "loss": 4.6163,
      "step": 36630
    },
    {
      "epoch": 0.7502098990436795,
      "grad_norm": 0.2934931814670563,
      "learning_rate": 6.980960399638361e-05,
      "loss": 4.6159,
      "step": 36635
    },
    {
      "epoch": 0.7503122888210841,
      "grad_norm": 0.2906499207019806,
      "learning_rate": 6.975730230738425e-05,
      "loss": 4.6167,
      "step": 36640
    },
    {
      "epoch": 0.7504146785984888,
      "grad_norm": 0.2920824885368347,
      "learning_rate": 6.970501428144457e-05,
      "loss": 4.5922,
      "step": 36645
    },
    {
      "epoch": 0.7505170683758934,
      "grad_norm": 0.27768781781196594,
      "learning_rate": 6.965273992746774e-05,
      "loss": 4.578,
      "step": 36650
    },
    {
      "epoch": 0.750619458153298,
      "grad_norm": 0.269756555557251,
      "learning_rate": 6.960047925435453e-05,
      "loss": 4.6133,
      "step": 36655
    },
    {
      "epoch": 0.7507218479307026,
      "grad_norm": 0.2730586528778076,
      "learning_rate": 6.954823227100353e-05,
      "loss": 4.5932,
      "step": 36660
    },
    {
      "epoch": 0.7508242377081072,
      "grad_norm": 0.28651493787765503,
      "learning_rate": 6.949599898631091e-05,
      "loss": 4.586,
      "step": 36665
    },
    {
      "epoch": 0.7509266274855119,
      "grad_norm": 0.3220368027687073,
      "learning_rate": 6.944377940917044e-05,
      "loss": 4.5902,
      "step": 36670
    },
    {
      "epoch": 0.7510290172629165,
      "grad_norm": 0.28390300273895264,
      "learning_rate": 6.939157354847384e-05,
      "loss": 4.5908,
      "step": 36675
    },
    {
      "epoch": 0.7511314070403211,
      "grad_norm": 0.29922792315483093,
      "learning_rate": 6.933938141311e-05,
      "loss": 4.6444,
      "step": 36680
    },
    {
      "epoch": 0.7512337968177257,
      "grad_norm": 0.29864510893821716,
      "learning_rate": 6.928720301196599e-05,
      "loss": 4.6025,
      "step": 36685
    },
    {
      "epoch": 0.7513361865951304,
      "grad_norm": 0.2763952314853668,
      "learning_rate": 6.923503835392623e-05,
      "loss": 4.5839,
      "step": 36690
    },
    {
      "epoch": 0.751438576372535,
      "grad_norm": 0.27281010150909424,
      "learning_rate": 6.918288744787284e-05,
      "loss": 4.6022,
      "step": 36695
    },
    {
      "epoch": 0.7515409661499396,
      "grad_norm": 0.26916012167930603,
      "learning_rate": 6.913075030268576e-05,
      "loss": 4.5863,
      "step": 36700
    },
    {
      "epoch": 0.7516433559273442,
      "grad_norm": 0.27656713128089905,
      "learning_rate": 6.907862692724238e-05,
      "loss": 4.6107,
      "step": 36705
    },
    {
      "epoch": 0.7517457457047488,
      "grad_norm": 0.3377551734447479,
      "learning_rate": 6.902651733041782e-05,
      "loss": 4.6203,
      "step": 36710
    },
    {
      "epoch": 0.7518481354821535,
      "grad_norm": 0.32195934653282166,
      "learning_rate": 6.897442152108496e-05,
      "loss": 4.5797,
      "step": 36715
    },
    {
      "epoch": 0.7519505252595581,
      "grad_norm": 0.28840330243110657,
      "learning_rate": 6.892233950811418e-05,
      "loss": 4.5722,
      "step": 36720
    },
    {
      "epoch": 0.7520529150369627,
      "grad_norm": 0.2755777835845947,
      "learning_rate": 6.88702713003736e-05,
      "loss": 4.6154,
      "step": 36725
    },
    {
      "epoch": 0.7521553048143673,
      "grad_norm": 0.2770107686519623,
      "learning_rate": 6.881821690672887e-05,
      "loss": 4.5714,
      "step": 36730
    },
    {
      "epoch": 0.7522576945917719,
      "grad_norm": 0.34963473677635193,
      "learning_rate": 6.876617633604352e-05,
      "loss": 4.6218,
      "step": 36735
    },
    {
      "epoch": 0.7523600843691766,
      "grad_norm": 0.3187248110771179,
      "learning_rate": 6.871414959717852e-05,
      "loss": 4.6303,
      "step": 36740
    },
    {
      "epoch": 0.7524624741465812,
      "grad_norm": 0.2622198462486267,
      "learning_rate": 6.86621366989925e-05,
      "loss": 4.5782,
      "step": 36745
    },
    {
      "epoch": 0.7525648639239858,
      "grad_norm": 0.27882781624794006,
      "learning_rate": 6.861013765034194e-05,
      "loss": 4.544,
      "step": 36750
    },
    {
      "epoch": 0.7526672537013904,
      "grad_norm": 0.27558428049087524,
      "learning_rate": 6.855815246008059e-05,
      "loss": 4.6296,
      "step": 36755
    },
    {
      "epoch": 0.752769643478795,
      "grad_norm": 0.27447035908699036,
      "learning_rate": 6.850618113706016e-05,
      "loss": 4.6102,
      "step": 36760
    },
    {
      "epoch": 0.7528720332561997,
      "grad_norm": 0.27361446619033813,
      "learning_rate": 6.845422369013e-05,
      "loss": 4.5727,
      "step": 36765
    },
    {
      "epoch": 0.7529744230336043,
      "grad_norm": 0.27006056904792786,
      "learning_rate": 6.84022801281368e-05,
      "loss": 4.6457,
      "step": 36770
    },
    {
      "epoch": 0.7530768128110089,
      "grad_norm": 0.30991798639297485,
      "learning_rate": 6.835035045992522e-05,
      "loss": 4.5653,
      "step": 36775
    },
    {
      "epoch": 0.7531792025884135,
      "grad_norm": 0.2698194682598114,
      "learning_rate": 6.829843469433732e-05,
      "loss": 4.6258,
      "step": 36780
    },
    {
      "epoch": 0.7532815923658182,
      "grad_norm": 0.2809673845767975,
      "learning_rate": 6.824653284021296e-05,
      "loss": 4.6027,
      "step": 36785
    },
    {
      "epoch": 0.7533839821432228,
      "grad_norm": 0.2807193398475647,
      "learning_rate": 6.819464490638954e-05,
      "loss": 4.6427,
      "step": 36790
    },
    {
      "epoch": 0.7534863719206274,
      "grad_norm": 0.29017332196235657,
      "learning_rate": 6.814277090170204e-05,
      "loss": 4.6518,
      "step": 36795
    },
    {
      "epoch": 0.753588761698032,
      "grad_norm": 0.28695717453956604,
      "learning_rate": 6.809091083498328e-05,
      "loss": 4.5811,
      "step": 36800
    },
    {
      "epoch": 0.7536911514754366,
      "grad_norm": 0.27444949746131897,
      "learning_rate": 6.803906471506336e-05,
      "loss": 4.6075,
      "step": 36805
    },
    {
      "epoch": 0.7537935412528414,
      "grad_norm": 0.2674769461154938,
      "learning_rate": 6.798723255077032e-05,
      "loss": 4.5853,
      "step": 36810
    },
    {
      "epoch": 0.753895931030246,
      "grad_norm": 0.29308798909187317,
      "learning_rate": 6.793541435092981e-05,
      "loss": 4.631,
      "step": 36815
    },
    {
      "epoch": 0.7539983208076506,
      "grad_norm": 0.2815810739994049,
      "learning_rate": 6.78836101243648e-05,
      "loss": 4.6146,
      "step": 36820
    },
    {
      "epoch": 0.7541007105850552,
      "grad_norm": 0.29573461413383484,
      "learning_rate": 6.783181987989621e-05,
      "loss": 4.6258,
      "step": 36825
    },
    {
      "epoch": 0.7542031003624599,
      "grad_norm": 0.2775028645992279,
      "learning_rate": 6.778004362634243e-05,
      "loss": 4.6229,
      "step": 36830
    },
    {
      "epoch": 0.7543054901398645,
      "grad_norm": 0.26205408573150635,
      "learning_rate": 6.772828137251943e-05,
      "loss": 4.5634,
      "step": 36835
    },
    {
      "epoch": 0.7544078799172691,
      "grad_norm": 0.26867425441741943,
      "learning_rate": 6.767653312724096e-05,
      "loss": 4.6122,
      "step": 36840
    },
    {
      "epoch": 0.7545102696946737,
      "grad_norm": 0.2727378308773041,
      "learning_rate": 6.762479889931823e-05,
      "loss": 4.5925,
      "step": 36845
    },
    {
      "epoch": 0.7546126594720783,
      "grad_norm": 0.28438353538513184,
      "learning_rate": 6.757307869756005e-05,
      "loss": 4.5865,
      "step": 36850
    },
    {
      "epoch": 0.754715049249483,
      "grad_norm": 0.2814081907272339,
      "learning_rate": 6.752137253077304e-05,
      "loss": 4.555,
      "step": 36855
    },
    {
      "epoch": 0.7548174390268876,
      "grad_norm": 0.32102739810943604,
      "learning_rate": 6.74696804077612e-05,
      "loss": 4.6096,
      "step": 36860
    },
    {
      "epoch": 0.7549198288042922,
      "grad_norm": 0.2817835211753845,
      "learning_rate": 6.741800233732628e-05,
      "loss": 4.6303,
      "step": 36865
    },
    {
      "epoch": 0.7550222185816968,
      "grad_norm": 0.2916010022163391,
      "learning_rate": 6.73663383282675e-05,
      "loss": 4.6123,
      "step": 36870
    },
    {
      "epoch": 0.7551246083591014,
      "grad_norm": 0.27132171392440796,
      "learning_rate": 6.731468838938193e-05,
      "loss": 4.5794,
      "step": 36875
    },
    {
      "epoch": 0.7552269981365061,
      "grad_norm": 0.2867998480796814,
      "learning_rate": 6.726305252946401e-05,
      "loss": 4.6391,
      "step": 36880
    },
    {
      "epoch": 0.7553293879139107,
      "grad_norm": 0.2729617953300476,
      "learning_rate": 6.72114307573058e-05,
      "loss": 4.6125,
      "step": 36885
    },
    {
      "epoch": 0.7554317776913153,
      "grad_norm": 0.2647687494754791,
      "learning_rate": 6.715982308169721e-05,
      "loss": 4.6241,
      "step": 36890
    },
    {
      "epoch": 0.7555341674687199,
      "grad_norm": 0.2753639817237854,
      "learning_rate": 6.710822951142535e-05,
      "loss": 4.5655,
      "step": 36895
    },
    {
      "epoch": 0.7556365572461246,
      "grad_norm": 0.3001912832260132,
      "learning_rate": 6.705665005527522e-05,
      "loss": 4.6188,
      "step": 36900
    },
    {
      "epoch": 0.7557389470235292,
      "grad_norm": 0.26426348090171814,
      "learning_rate": 6.70050847220295e-05,
      "loss": 4.6051,
      "step": 36905
    },
    {
      "epoch": 0.7558413368009338,
      "grad_norm": 0.2681485712528229,
      "learning_rate": 6.695353352046806e-05,
      "loss": 4.6461,
      "step": 36910
    },
    {
      "epoch": 0.7559437265783384,
      "grad_norm": 0.28377512097358704,
      "learning_rate": 6.690199645936876e-05,
      "loss": 4.6207,
      "step": 36915
    },
    {
      "epoch": 0.756046116355743,
      "grad_norm": 0.2729427218437195,
      "learning_rate": 6.685047354750688e-05,
      "loss": 4.6266,
      "step": 36920
    },
    {
      "epoch": 0.7561485061331477,
      "grad_norm": 0.2970889210700989,
      "learning_rate": 6.679896479365524e-05,
      "loss": 4.5379,
      "step": 36925
    },
    {
      "epoch": 0.7562508959105523,
      "grad_norm": 0.2982221245765686,
      "learning_rate": 6.674747020658443e-05,
      "loss": 4.6025,
      "step": 36930
    },
    {
      "epoch": 0.7563532856879569,
      "grad_norm": 0.2927866280078888,
      "learning_rate": 6.669598979506244e-05,
      "loss": 4.5632,
      "step": 36935
    },
    {
      "epoch": 0.7564556754653615,
      "grad_norm": 0.27343738079071045,
      "learning_rate": 6.664452356785504e-05,
      "loss": 4.5915,
      "step": 36940
    },
    {
      "epoch": 0.7565580652427661,
      "grad_norm": 0.283540278673172,
      "learning_rate": 6.659307153372528e-05,
      "loss": 4.6048,
      "step": 36945
    },
    {
      "epoch": 0.7566604550201708,
      "grad_norm": 0.2763478457927704,
      "learning_rate": 6.654163370143411e-05,
      "loss": 4.6093,
      "step": 36950
    },
    {
      "epoch": 0.7567628447975754,
      "grad_norm": 0.2957182228565216,
      "learning_rate": 6.649021007974002e-05,
      "loss": 4.5892,
      "step": 36955
    },
    {
      "epoch": 0.75686523457498,
      "grad_norm": 0.7029393911361694,
      "learning_rate": 6.643880067739879e-05,
      "loss": 4.5881,
      "step": 36960
    },
    {
      "epoch": 0.7569676243523846,
      "grad_norm": 0.2768975794315338,
      "learning_rate": 6.638740550316416e-05,
      "loss": 4.5847,
      "step": 36965
    },
    {
      "epoch": 0.7570700141297892,
      "grad_norm": 0.2871471643447876,
      "learning_rate": 6.63360245657872e-05,
      "loss": 4.6242,
      "step": 36970
    },
    {
      "epoch": 0.7571724039071939,
      "grad_norm": 0.30523785948753357,
      "learning_rate": 6.628465787401657e-05,
      "loss": 4.5973,
      "step": 36975
    },
    {
      "epoch": 0.7572747936845985,
      "grad_norm": 0.27099236845970154,
      "learning_rate": 6.623330543659869e-05,
      "loss": 4.5939,
      "step": 36980
    },
    {
      "epoch": 0.7573771834620031,
      "grad_norm": 0.3290117681026459,
      "learning_rate": 6.618196726227736e-05,
      "loss": 4.6173,
      "step": 36985
    },
    {
      "epoch": 0.7574795732394077,
      "grad_norm": 0.333617240190506,
      "learning_rate": 6.613064335979403e-05,
      "loss": 4.5882,
      "step": 36990
    },
    {
      "epoch": 0.7575819630168124,
      "grad_norm": 0.346046507358551,
      "learning_rate": 6.607933373788764e-05,
      "loss": 4.6062,
      "step": 36995
    },
    {
      "epoch": 0.757684352794217,
      "grad_norm": 0.30597689747810364,
      "learning_rate": 6.602803840529488e-05,
      "loss": 4.6224,
      "step": 37000
    },
    {
      "epoch": 0.757684352794217,
      "eval_loss": 4.618415832519531,
      "eval_runtime": 27.973,
      "eval_samples_per_second": 20.806,
      "eval_steps_per_second": 2.61,
      "step": 37000
    },
    {
      "epoch": 0.7577867425716216,
      "grad_norm": 0.2713710367679596,
      "learning_rate": 6.597675737074985e-05,
      "loss": 4.6253,
      "step": 37005
    },
    {
      "epoch": 0.7578891323490262,
      "grad_norm": 0.29117444157600403,
      "learning_rate": 6.592549064298417e-05,
      "loss": 4.6411,
      "step": 37010
    },
    {
      "epoch": 0.757991522126431,
      "grad_norm": 0.27179235219955444,
      "learning_rate": 6.587423823072727e-05,
      "loss": 4.5576,
      "step": 37015
    },
    {
      "epoch": 0.7580939119038356,
      "grad_norm": 0.2707149386405945,
      "learning_rate": 6.582300014270588e-05,
      "loss": 4.5971,
      "step": 37020
    },
    {
      "epoch": 0.7581963016812402,
      "grad_norm": 0.29081299901008606,
      "learning_rate": 6.577177638764439e-05,
      "loss": 4.5642,
      "step": 37025
    },
    {
      "epoch": 0.7582986914586448,
      "grad_norm": 0.29647204279899597,
      "learning_rate": 6.572056697426489e-05,
      "loss": 4.6166,
      "step": 37030
    },
    {
      "epoch": 0.7584010812360494,
      "grad_norm": 0.27781417965888977,
      "learning_rate": 6.566937191128666e-05,
      "loss": 4.5966,
      "step": 37035
    },
    {
      "epoch": 0.7585034710134541,
      "grad_norm": 0.3100695312023163,
      "learning_rate": 6.561819120742691e-05,
      "loss": 4.6062,
      "step": 37040
    },
    {
      "epoch": 0.7586058607908587,
      "grad_norm": 0.28153157234191895,
      "learning_rate": 6.556702487140037e-05,
      "loss": 4.5566,
      "step": 37045
    },
    {
      "epoch": 0.7587082505682633,
      "grad_norm": 0.2895149290561676,
      "learning_rate": 6.551587291191899e-05,
      "loss": 4.6071,
      "step": 37050
    },
    {
      "epoch": 0.7588106403456679,
      "grad_norm": 0.29634541273117065,
      "learning_rate": 6.546473533769266e-05,
      "loss": 4.6792,
      "step": 37055
    },
    {
      "epoch": 0.7589130301230725,
      "grad_norm": 0.28362852334976196,
      "learning_rate": 6.541361215742862e-05,
      "loss": 4.6141,
      "step": 37060
    },
    {
      "epoch": 0.7590154199004772,
      "grad_norm": 0.27192139625549316,
      "learning_rate": 6.536250337983163e-05,
      "loss": 4.5812,
      "step": 37065
    },
    {
      "epoch": 0.7591178096778818,
      "grad_norm": 0.27279242873191833,
      "learning_rate": 6.531140901360421e-05,
      "loss": 4.594,
      "step": 37070
    },
    {
      "epoch": 0.7592201994552864,
      "grad_norm": 0.2664951980113983,
      "learning_rate": 6.526032906744613e-05,
      "loss": 4.6143,
      "step": 37075
    },
    {
      "epoch": 0.759322589232691,
      "grad_norm": 0.29349449276924133,
      "learning_rate": 6.520926355005505e-05,
      "loss": 4.6169,
      "step": 37080
    },
    {
      "epoch": 0.7594249790100956,
      "grad_norm": 0.2738834619522095,
      "learning_rate": 6.515821247012574e-05,
      "loss": 4.6126,
      "step": 37085
    },
    {
      "epoch": 0.7595273687875003,
      "grad_norm": 0.2886669933795929,
      "learning_rate": 6.510717583635089e-05,
      "loss": 4.6484,
      "step": 37090
    },
    {
      "epoch": 0.7596297585649049,
      "grad_norm": 0.2747144401073456,
      "learning_rate": 6.50561536574207e-05,
      "loss": 4.6175,
      "step": 37095
    },
    {
      "epoch": 0.7597321483423095,
      "grad_norm": 0.28018927574157715,
      "learning_rate": 6.500514594202258e-05,
      "loss": 4.5505,
      "step": 37100
    },
    {
      "epoch": 0.7598345381197141,
      "grad_norm": 0.3059290945529938,
      "learning_rate": 6.495415269884187e-05,
      "loss": 4.6257,
      "step": 37105
    },
    {
      "epoch": 0.7599369278971188,
      "grad_norm": 0.28672704100608826,
      "learning_rate": 6.490317393656118e-05,
      "loss": 4.6077,
      "step": 37110
    },
    {
      "epoch": 0.7600393176745234,
      "grad_norm": 0.277348130941391,
      "learning_rate": 6.485220966386074e-05,
      "loss": 4.6029,
      "step": 37115
    },
    {
      "epoch": 0.760141707451928,
      "grad_norm": 0.3029509484767914,
      "learning_rate": 6.480125988941845e-05,
      "loss": 4.6646,
      "step": 37120
    },
    {
      "epoch": 0.7602440972293326,
      "grad_norm": 0.26892364025115967,
      "learning_rate": 6.475032462190949e-05,
      "loss": 4.6111,
      "step": 37125
    },
    {
      "epoch": 0.7603464870067372,
      "grad_norm": 0.40442508459091187,
      "learning_rate": 6.469940387000674e-05,
      "loss": 4.6424,
      "step": 37130
    },
    {
      "epoch": 0.7604488767841419,
      "grad_norm": 0.2839796245098114,
      "learning_rate": 6.46484976423805e-05,
      "loss": 4.5256,
      "step": 37135
    },
    {
      "epoch": 0.7605512665615465,
      "grad_norm": 0.2990724742412567,
      "learning_rate": 6.459760594769879e-05,
      "loss": 4.5923,
      "step": 37140
    },
    {
      "epoch": 0.7606536563389511,
      "grad_norm": 0.28737035393714905,
      "learning_rate": 6.454672879462693e-05,
      "loss": 4.59,
      "step": 37145
    },
    {
      "epoch": 0.7607560461163557,
      "grad_norm": 0.28966018557548523,
      "learning_rate": 6.449586619182783e-05,
      "loss": 4.5735,
      "step": 37150
    },
    {
      "epoch": 0.7608584358937603,
      "grad_norm": 0.29240936040878296,
      "learning_rate": 6.444501814796205e-05,
      "loss": 4.6054,
      "step": 37155
    },
    {
      "epoch": 0.760960825671165,
      "grad_norm": 0.3128106892108917,
      "learning_rate": 6.439418467168754e-05,
      "loss": 4.5458,
      "step": 37160
    },
    {
      "epoch": 0.7610632154485696,
      "grad_norm": 0.2720398008823395,
      "learning_rate": 6.43433657716597e-05,
      "loss": 4.6539,
      "step": 37165
    },
    {
      "epoch": 0.7611656052259742,
      "grad_norm": 0.26985418796539307,
      "learning_rate": 6.429256145653175e-05,
      "loss": 4.632,
      "step": 37170
    },
    {
      "epoch": 0.7612679950033788,
      "grad_norm": 0.28882190585136414,
      "learning_rate": 6.424177173495401e-05,
      "loss": 4.5902,
      "step": 37175
    },
    {
      "epoch": 0.7613703847807834,
      "grad_norm": 0.2826944589614868,
      "learning_rate": 6.419099661557463e-05,
      "loss": 4.5867,
      "step": 37180
    },
    {
      "epoch": 0.7614727745581881,
      "grad_norm": 0.26971328258514404,
      "learning_rate": 6.414023610703928e-05,
      "loss": 4.5876,
      "step": 37185
    },
    {
      "epoch": 0.7615751643355927,
      "grad_norm": 0.28219759464263916,
      "learning_rate": 6.408949021799084e-05,
      "loss": 4.6054,
      "step": 37190
    },
    {
      "epoch": 0.7616775541129973,
      "grad_norm": 0.2821265757083893,
      "learning_rate": 6.403875895707003e-05,
      "loss": 4.6131,
      "step": 37195
    },
    {
      "epoch": 0.7617799438904019,
      "grad_norm": 0.2961079478263855,
      "learning_rate": 6.398804233291492e-05,
      "loss": 4.6034,
      "step": 37200
    },
    {
      "epoch": 0.7618823336678066,
      "grad_norm": 0.28093793988227844,
      "learning_rate": 6.393734035416105e-05,
      "loss": 4.5963,
      "step": 37205
    },
    {
      "epoch": 0.7619847234452112,
      "grad_norm": 0.2887776792049408,
      "learning_rate": 6.388665302944163e-05,
      "loss": 4.587,
      "step": 37210
    },
    {
      "epoch": 0.7620871132226159,
      "grad_norm": 0.2903097867965698,
      "learning_rate": 6.38359803673872e-05,
      "loss": 4.5828,
      "step": 37215
    },
    {
      "epoch": 0.7621895030000205,
      "grad_norm": 0.27850037813186646,
      "learning_rate": 6.3785322376626e-05,
      "loss": 4.5873,
      "step": 37220
    },
    {
      "epoch": 0.7622918927774252,
      "grad_norm": 0.28699085116386414,
      "learning_rate": 6.373467906578347e-05,
      "loss": 4.5973,
      "step": 37225
    },
    {
      "epoch": 0.7623942825548298,
      "grad_norm": 0.2638085186481476,
      "learning_rate": 6.368405044348283e-05,
      "loss": 4.6501,
      "step": 37230
    },
    {
      "epoch": 0.7624966723322344,
      "grad_norm": 0.286062628030777,
      "learning_rate": 6.36334365183448e-05,
      "loss": 4.6185,
      "step": 37235
    },
    {
      "epoch": 0.762599062109639,
      "grad_norm": 0.28829097747802734,
      "learning_rate": 6.35828372989873e-05,
      "loss": 4.6196,
      "step": 37240
    },
    {
      "epoch": 0.7627014518870436,
      "grad_norm": 0.29331281781196594,
      "learning_rate": 6.353225279402607e-05,
      "loss": 4.5655,
      "step": 37245
    },
    {
      "epoch": 0.7628038416644483,
      "grad_norm": 0.28903159499168396,
      "learning_rate": 6.34816830120742e-05,
      "loss": 4.5723,
      "step": 37250
    },
    {
      "epoch": 0.7629062314418529,
      "grad_norm": 0.29470354318618774,
      "learning_rate": 6.343112796174223e-05,
      "loss": 4.5886,
      "step": 37255
    },
    {
      "epoch": 0.7630086212192575,
      "grad_norm": 0.30720871686935425,
      "learning_rate": 6.338058765163838e-05,
      "loss": 4.5834,
      "step": 37260
    },
    {
      "epoch": 0.7631110109966621,
      "grad_norm": 0.2884748876094818,
      "learning_rate": 6.333006209036813e-05,
      "loss": 4.5836,
      "step": 37265
    },
    {
      "epoch": 0.7632134007740667,
      "grad_norm": 0.28455010056495667,
      "learning_rate": 6.32795512865346e-05,
      "loss": 4.5724,
      "step": 37270
    },
    {
      "epoch": 0.7633157905514714,
      "grad_norm": 0.2831284701824188,
      "learning_rate": 6.322905524873827e-05,
      "loss": 4.5781,
      "step": 37275
    },
    {
      "epoch": 0.763418180328876,
      "grad_norm": 0.30581367015838623,
      "learning_rate": 6.31785739855773e-05,
      "loss": 4.6296,
      "step": 37280
    },
    {
      "epoch": 0.7635205701062806,
      "grad_norm": 0.3325393497943878,
      "learning_rate": 6.312810750564718e-05,
      "loss": 4.5549,
      "step": 37285
    },
    {
      "epoch": 0.7636229598836852,
      "grad_norm": 0.26691895723342896,
      "learning_rate": 6.307765581754085e-05,
      "loss": 4.6191,
      "step": 37290
    },
    {
      "epoch": 0.7637253496610898,
      "grad_norm": 0.27299752831459045,
      "learning_rate": 6.30272189298489e-05,
      "loss": 4.5532,
      "step": 37295
    },
    {
      "epoch": 0.7638277394384945,
      "grad_norm": 0.2956436276435852,
      "learning_rate": 6.29767968511593e-05,
      "loss": 4.5685,
      "step": 37300
    },
    {
      "epoch": 0.7639301292158991,
      "grad_norm": 0.3262317478656769,
      "learning_rate": 6.292638959005741e-05,
      "loss": 4.6277,
      "step": 37305
    },
    {
      "epoch": 0.7640325189933037,
      "grad_norm": 0.2778327465057373,
      "learning_rate": 6.287599715512632e-05,
      "loss": 4.5751,
      "step": 37310
    },
    {
      "epoch": 0.7641349087707083,
      "grad_norm": 0.28934070467948914,
      "learning_rate": 6.282561955494623e-05,
      "loss": 4.6262,
      "step": 37315
    },
    {
      "epoch": 0.764237298548113,
      "grad_norm": 0.27689051628112793,
      "learning_rate": 6.277525679809518e-05,
      "loss": 4.6306,
      "step": 37320
    },
    {
      "epoch": 0.7643396883255176,
      "grad_norm": 0.2804756760597229,
      "learning_rate": 6.272490889314847e-05,
      "loss": 4.5508,
      "step": 37325
    },
    {
      "epoch": 0.7644420781029222,
      "grad_norm": 0.2770512104034424,
      "learning_rate": 6.267457584867887e-05,
      "loss": 4.5737,
      "step": 37330
    },
    {
      "epoch": 0.7645444678803268,
      "grad_norm": 0.27728188037872314,
      "learning_rate": 6.262425767325677e-05,
      "loss": 4.5765,
      "step": 37335
    },
    {
      "epoch": 0.7646468576577314,
      "grad_norm": 0.2826690077781677,
      "learning_rate": 6.257395437544988e-05,
      "loss": 4.6062,
      "step": 37340
    },
    {
      "epoch": 0.7647492474351361,
      "grad_norm": 0.2677403390407562,
      "learning_rate": 6.252366596382338e-05,
      "loss": 4.5737,
      "step": 37345
    },
    {
      "epoch": 0.7648516372125407,
      "grad_norm": 0.29870080947875977,
      "learning_rate": 6.247339244694007e-05,
      "loss": 4.6348,
      "step": 37350
    },
    {
      "epoch": 0.7649540269899453,
      "grad_norm": 0.27728724479675293,
      "learning_rate": 6.242313383336e-05,
      "loss": 4.5664,
      "step": 37355
    },
    {
      "epoch": 0.7650564167673499,
      "grad_norm": 0.27288126945495605,
      "learning_rate": 6.237289013164093e-05,
      "loss": 4.6004,
      "step": 37360
    },
    {
      "epoch": 0.7651588065447545,
      "grad_norm": 0.2839893102645874,
      "learning_rate": 6.232266135033775e-05,
      "loss": 4.5854,
      "step": 37365
    },
    {
      "epoch": 0.7652611963221592,
      "grad_norm": 0.3005150258541107,
      "learning_rate": 6.227244749800309e-05,
      "loss": 4.5727,
      "step": 37370
    },
    {
      "epoch": 0.7653635860995638,
      "grad_norm": 0.2766365706920624,
      "learning_rate": 6.222224858318706e-05,
      "loss": 4.5729,
      "step": 37375
    },
    {
      "epoch": 0.7654659758769684,
      "grad_norm": 0.28654855489730835,
      "learning_rate": 6.217206461443692e-05,
      "loss": 4.5589,
      "step": 37380
    },
    {
      "epoch": 0.765568365654373,
      "grad_norm": 0.2733350694179535,
      "learning_rate": 6.212189560029772e-05,
      "loss": 4.5948,
      "step": 37385
    },
    {
      "epoch": 0.7656707554317776,
      "grad_norm": 0.27922144532203674,
      "learning_rate": 6.207174154931177e-05,
      "loss": 4.574,
      "step": 37390
    },
    {
      "epoch": 0.7657731452091823,
      "grad_norm": 0.27702051401138306,
      "learning_rate": 6.202160247001883e-05,
      "loss": 4.6048,
      "step": 37395
    },
    {
      "epoch": 0.7658755349865869,
      "grad_norm": 0.2795740067958832,
      "learning_rate": 6.197147837095627e-05,
      "loss": 4.6568,
      "step": 37400
    },
    {
      "epoch": 0.7659779247639915,
      "grad_norm": 0.28041133284568787,
      "learning_rate": 6.192136926065876e-05,
      "loss": 4.5447,
      "step": 37405
    },
    {
      "epoch": 0.7660803145413961,
      "grad_norm": 0.33503907918930054,
      "learning_rate": 6.187127514765846e-05,
      "loss": 4.6238,
      "step": 37410
    },
    {
      "epoch": 0.7661827043188008,
      "grad_norm": 0.28736916184425354,
      "learning_rate": 6.182119604048493e-05,
      "loss": 4.6131,
      "step": 37415
    },
    {
      "epoch": 0.7662850940962055,
      "grad_norm": 0.2802198827266693,
      "learning_rate": 6.177113194766532e-05,
      "loss": 4.6347,
      "step": 37420
    },
    {
      "epoch": 0.7663874838736101,
      "grad_norm": 0.2767556607723236,
      "learning_rate": 6.172108287772408e-05,
      "loss": 4.593,
      "step": 37425
    },
    {
      "epoch": 0.7664898736510147,
      "grad_norm": 0.26829272508621216,
      "learning_rate": 6.16710488391831e-05,
      "loss": 4.5967,
      "step": 37430
    },
    {
      "epoch": 0.7665922634284194,
      "grad_norm": 0.27992019057273865,
      "learning_rate": 6.162102984056186e-05,
      "loss": 4.5953,
      "step": 37435
    },
    {
      "epoch": 0.766694653205824,
      "grad_norm": 0.2766593098640442,
      "learning_rate": 6.157102589037715e-05,
      "loss": 4.5804,
      "step": 37440
    },
    {
      "epoch": 0.7667970429832286,
      "grad_norm": 0.30000054836273193,
      "learning_rate": 6.152103699714314e-05,
      "loss": 4.6185,
      "step": 37445
    },
    {
      "epoch": 0.7668994327606332,
      "grad_norm": 0.30132678151130676,
      "learning_rate": 6.147106316937169e-05,
      "loss": 4.6012,
      "step": 37450
    },
    {
      "epoch": 0.7670018225380378,
      "grad_norm": 0.271884948015213,
      "learning_rate": 6.142110441557174e-05,
      "loss": 4.5928,
      "step": 37455
    },
    {
      "epoch": 0.7671042123154425,
      "grad_norm": 0.2828238606452942,
      "learning_rate": 6.137116074424997e-05,
      "loss": 4.6395,
      "step": 37460
    },
    {
      "epoch": 0.7672066020928471,
      "grad_norm": 0.2779662013053894,
      "learning_rate": 6.132123216391036e-05,
      "loss": 4.5751,
      "step": 37465
    },
    {
      "epoch": 0.7673089918702517,
      "grad_norm": 0.26808327436447144,
      "learning_rate": 6.127131868305426e-05,
      "loss": 4.5592,
      "step": 37470
    },
    {
      "epoch": 0.7674113816476563,
      "grad_norm": 0.2726622223854065,
      "learning_rate": 6.122142031018062e-05,
      "loss": 4.6192,
      "step": 37475
    },
    {
      "epoch": 0.7675137714250609,
      "grad_norm": 0.27267372608184814,
      "learning_rate": 6.117153705378571e-05,
      "loss": 4.5802,
      "step": 37480
    },
    {
      "epoch": 0.7676161612024656,
      "grad_norm": 0.2768978476524353,
      "learning_rate": 6.112166892236314e-05,
      "loss": 4.6243,
      "step": 37485
    },
    {
      "epoch": 0.7677185509798702,
      "grad_norm": 0.28463026881217957,
      "learning_rate": 6.107181592440414e-05,
      "loss": 4.652,
      "step": 37490
    },
    {
      "epoch": 0.7678209407572748,
      "grad_norm": 0.3004951775074005,
      "learning_rate": 6.10219780683972e-05,
      "loss": 4.5509,
      "step": 37495
    },
    {
      "epoch": 0.7679233305346794,
      "grad_norm": 0.3154648542404175,
      "learning_rate": 6.097215536282842e-05,
      "loss": 4.5509,
      "step": 37500
    },
    {
      "epoch": 0.768025720312084,
      "grad_norm": 0.2721574008464813,
      "learning_rate": 6.0922347816181e-05,
      "loss": 4.5981,
      "step": 37505
    },
    {
      "epoch": 0.7681281100894887,
      "grad_norm": 0.2939707040786743,
      "learning_rate": 6.087255543693591e-05,
      "loss": 4.62,
      "step": 37510
    },
    {
      "epoch": 0.7682304998668933,
      "grad_norm": 0.27639541029930115,
      "learning_rate": 6.082277823357134e-05,
      "loss": 4.6862,
      "step": 37515
    },
    {
      "epoch": 0.7683328896442979,
      "grad_norm": 0.29966050386428833,
      "learning_rate": 6.077301621456288e-05,
      "loss": 4.6113,
      "step": 37520
    },
    {
      "epoch": 0.7684352794217025,
      "grad_norm": 0.26800772547721863,
      "learning_rate": 6.072326938838367e-05,
      "loss": 4.5855,
      "step": 37525
    },
    {
      "epoch": 0.7685376691991072,
      "grad_norm": 0.267238587141037,
      "learning_rate": 6.0673537763504185e-05,
      "loss": 4.5581,
      "step": 37530
    },
    {
      "epoch": 0.7686400589765118,
      "grad_norm": 0.2806229889392853,
      "learning_rate": 6.062382134839222e-05,
      "loss": 4.6597,
      "step": 37535
    },
    {
      "epoch": 0.7687424487539164,
      "grad_norm": 0.2705599367618561,
      "learning_rate": 6.05741201515132e-05,
      "loss": 4.5987,
      "step": 37540
    },
    {
      "epoch": 0.768844838531321,
      "grad_norm": 0.26709672808647156,
      "learning_rate": 6.052443418132977e-05,
      "loss": 4.5802,
      "step": 37545
    },
    {
      "epoch": 0.7689472283087256,
      "grad_norm": 0.2724774181842804,
      "learning_rate": 6.047476344630203e-05,
      "loss": 4.5914,
      "step": 37550
    },
    {
      "epoch": 0.7690496180861303,
      "grad_norm": 0.29947328567504883,
      "learning_rate": 6.042510795488748e-05,
      "loss": 4.6055,
      "step": 37555
    },
    {
      "epoch": 0.7691520078635349,
      "grad_norm": 0.284551739692688,
      "learning_rate": 6.037546771554114e-05,
      "loss": 4.6055,
      "step": 37560
    },
    {
      "epoch": 0.7692543976409395,
      "grad_norm": 0.31512659788131714,
      "learning_rate": 6.032584273671526e-05,
      "loss": 4.5537,
      "step": 37565
    },
    {
      "epoch": 0.7693567874183441,
      "grad_norm": 0.2858293354511261,
      "learning_rate": 6.027623302685953e-05,
      "loss": 4.67,
      "step": 37570
    },
    {
      "epoch": 0.7694591771957487,
      "grad_norm": 0.2723812758922577,
      "learning_rate": 6.022663859442119e-05,
      "loss": 4.5688,
      "step": 37575
    },
    {
      "epoch": 0.7695615669731534,
      "grad_norm": 0.274245947599411,
      "learning_rate": 6.017705944784472e-05,
      "loss": 4.5995,
      "step": 37580
    },
    {
      "epoch": 0.769663956750558,
      "grad_norm": 0.2815904915332794,
      "learning_rate": 6.0127495595571975e-05,
      "loss": 4.5514,
      "step": 37585
    },
    {
      "epoch": 0.7697663465279626,
      "grad_norm": 0.2978804409503937,
      "learning_rate": 6.007794704604246e-05,
      "loss": 4.6603,
      "step": 37590
    },
    {
      "epoch": 0.7698687363053672,
      "grad_norm": 0.2692318558692932,
      "learning_rate": 6.002841380769266e-05,
      "loss": 4.5663,
      "step": 37595
    },
    {
      "epoch": 0.7699711260827719,
      "grad_norm": 0.3634311854839325,
      "learning_rate": 5.9978895888956825e-05,
      "loss": 4.5738,
      "step": 37600
    },
    {
      "epoch": 0.7700735158601765,
      "grad_norm": 0.2816818058490753,
      "learning_rate": 5.992939329826644e-05,
      "loss": 4.5572,
      "step": 37605
    },
    {
      "epoch": 0.7701759056375811,
      "grad_norm": 0.27900898456573486,
      "learning_rate": 5.987990604405033e-05,
      "loss": 4.5772,
      "step": 37610
    },
    {
      "epoch": 0.7702782954149857,
      "grad_norm": 0.2727174758911133,
      "learning_rate": 5.983043413473484e-05,
      "loss": 4.5825,
      "step": 37615
    },
    {
      "epoch": 0.7703806851923904,
      "grad_norm": 0.3469656705856323,
      "learning_rate": 5.978097757874364e-05,
      "loss": 4.5801,
      "step": 37620
    },
    {
      "epoch": 0.7704830749697951,
      "grad_norm": 0.27901020646095276,
      "learning_rate": 5.9731536384497705e-05,
      "loss": 4.601,
      "step": 37625
    },
    {
      "epoch": 0.7705854647471997,
      "grad_norm": 0.28966450691223145,
      "learning_rate": 5.968211056041557e-05,
      "loss": 4.6858,
      "step": 37630
    },
    {
      "epoch": 0.7706878545246043,
      "grad_norm": 0.2710847854614258,
      "learning_rate": 5.9632700114912944e-05,
      "loss": 4.6181,
      "step": 37635
    },
    {
      "epoch": 0.7707902443020089,
      "grad_norm": 0.8204988241195679,
      "learning_rate": 5.95833050564032e-05,
      "loss": 4.5649,
      "step": 37640
    },
    {
      "epoch": 0.7708926340794136,
      "grad_norm": 0.2956174314022064,
      "learning_rate": 5.9533925393296714e-05,
      "loss": 4.5958,
      "step": 37645
    },
    {
      "epoch": 0.7709950238568182,
      "grad_norm": 0.2778283655643463,
      "learning_rate": 5.948456113400158e-05,
      "loss": 4.6055,
      "step": 37650
    },
    {
      "epoch": 0.7710974136342228,
      "grad_norm": 0.269569456577301,
      "learning_rate": 5.943521228692308e-05,
      "loss": 4.5891,
      "step": 37655
    },
    {
      "epoch": 0.7711998034116274,
      "grad_norm": 0.28458067774772644,
      "learning_rate": 5.9385878860463905e-05,
      "loss": 4.5533,
      "step": 37660
    },
    {
      "epoch": 0.771302193189032,
      "grad_norm": 0.3094825744628906,
      "learning_rate": 5.93365608630242e-05,
      "loss": 4.6104,
      "step": 37665
    },
    {
      "epoch": 0.7714045829664367,
      "grad_norm": 0.27909228205680847,
      "learning_rate": 5.928725830300143e-05,
      "loss": 4.6095,
      "step": 37670
    },
    {
      "epoch": 0.7715069727438413,
      "grad_norm": 0.27436867356300354,
      "learning_rate": 5.92379711887903e-05,
      "loss": 4.5679,
      "step": 37675
    },
    {
      "epoch": 0.7716093625212459,
      "grad_norm": 0.2822236120700836,
      "learning_rate": 5.918869952878319e-05,
      "loss": 4.6314,
      "step": 37680
    },
    {
      "epoch": 0.7717117522986505,
      "grad_norm": 0.293186217546463,
      "learning_rate": 5.913944333136955e-05,
      "loss": 4.55,
      "step": 37685
    },
    {
      "epoch": 0.7718141420760551,
      "grad_norm": 0.29767438769340515,
      "learning_rate": 5.909020260493636e-05,
      "loss": 4.5905,
      "step": 37690
    },
    {
      "epoch": 0.7719165318534598,
      "grad_norm": 0.2744883894920349,
      "learning_rate": 5.904097735786786e-05,
      "loss": 4.6283,
      "step": 37695
    },
    {
      "epoch": 0.7720189216308644,
      "grad_norm": 0.2894776463508606,
      "learning_rate": 5.8991767598545806e-05,
      "loss": 4.5789,
      "step": 37700
    },
    {
      "epoch": 0.772121311408269,
      "grad_norm": 0.2848042845726013,
      "learning_rate": 5.8942573335349194e-05,
      "loss": 4.5971,
      "step": 37705
    },
    {
      "epoch": 0.7722237011856736,
      "grad_norm": 0.2997800409793854,
      "learning_rate": 5.889339457665434e-05,
      "loss": 4.6336,
      "step": 37710
    },
    {
      "epoch": 0.7723260909630782,
      "grad_norm": 0.2878294587135315,
      "learning_rate": 5.8844231330835124e-05,
      "loss": 4.6161,
      "step": 37715
    },
    {
      "epoch": 0.7724284807404829,
      "grad_norm": 0.27903154492378235,
      "learning_rate": 5.8795083606262594e-05,
      "loss": 4.5961,
      "step": 37720
    },
    {
      "epoch": 0.7725308705178875,
      "grad_norm": 0.276852011680603,
      "learning_rate": 5.874595141130514e-05,
      "loss": 4.6124,
      "step": 37725
    },
    {
      "epoch": 0.7726332602952921,
      "grad_norm": 0.28702512383461,
      "learning_rate": 5.869683475432877e-05,
      "loss": 4.6696,
      "step": 37730
    },
    {
      "epoch": 0.7727356500726967,
      "grad_norm": 0.3925803601741791,
      "learning_rate": 5.8647733643696455e-05,
      "loss": 4.5229,
      "step": 37735
    },
    {
      "epoch": 0.7728380398501014,
      "grad_norm": 0.2754385769367218,
      "learning_rate": 5.859864808776886e-05,
      "loss": 4.5714,
      "step": 37740
    },
    {
      "epoch": 0.772940429627506,
      "grad_norm": 0.29182150959968567,
      "learning_rate": 5.8549578094903813e-05,
      "loss": 4.589,
      "step": 37745
    },
    {
      "epoch": 0.7730428194049106,
      "grad_norm": 0.28724372386932373,
      "learning_rate": 5.850052367345649e-05,
      "loss": 4.5774,
      "step": 37750
    },
    {
      "epoch": 0.7731452091823152,
      "grad_norm": 0.2842564582824707,
      "learning_rate": 5.8451484831779616e-05,
      "loss": 4.5768,
      "step": 37755
    },
    {
      "epoch": 0.7732475989597198,
      "grad_norm": 0.276254266500473,
      "learning_rate": 5.840246157822301e-05,
      "loss": 4.6367,
      "step": 37760
    },
    {
      "epoch": 0.7733499887371245,
      "grad_norm": 0.2792101204395294,
      "learning_rate": 5.835345392113391e-05,
      "loss": 4.5904,
      "step": 37765
    },
    {
      "epoch": 0.7734523785145291,
      "grad_norm": 0.29593244194984436,
      "learning_rate": 5.8304461868857074e-05,
      "loss": 4.6085,
      "step": 37770
    },
    {
      "epoch": 0.7735547682919337,
      "grad_norm": 0.27402353286743164,
      "learning_rate": 5.82554854297343e-05,
      "loss": 4.6298,
      "step": 37775
    },
    {
      "epoch": 0.7736571580693383,
      "grad_norm": 0.2819097638130188,
      "learning_rate": 5.820652461210508e-05,
      "loss": 4.6513,
      "step": 37780
    },
    {
      "epoch": 0.773759547846743,
      "grad_norm": 0.2651239037513733,
      "learning_rate": 5.815757942430584e-05,
      "loss": 4.5607,
      "step": 37785
    },
    {
      "epoch": 0.7738619376241476,
      "grad_norm": 0.3204714059829712,
      "learning_rate": 5.8108649874670706e-05,
      "loss": 4.6206,
      "step": 37790
    },
    {
      "epoch": 0.7739643274015522,
      "grad_norm": 0.30024945735931396,
      "learning_rate": 5.805973597153096e-05,
      "loss": 4.6077,
      "step": 37795
    },
    {
      "epoch": 0.7740667171789568,
      "grad_norm": 0.27434980869293213,
      "learning_rate": 5.801083772321518e-05,
      "loss": 4.585,
      "step": 37800
    },
    {
      "epoch": 0.7741691069563614,
      "grad_norm": 0.2731095254421234,
      "learning_rate": 5.796195513804948e-05,
      "loss": 4.5724,
      "step": 37805
    },
    {
      "epoch": 0.774271496733766,
      "grad_norm": 0.2779126465320587,
      "learning_rate": 5.791308822435712e-05,
      "loss": 4.522,
      "step": 37810
    },
    {
      "epoch": 0.7743738865111707,
      "grad_norm": 0.26512452960014343,
      "learning_rate": 5.7864236990458704e-05,
      "loss": 4.5672,
      "step": 37815
    },
    {
      "epoch": 0.7744762762885753,
      "grad_norm": 0.271696537733078,
      "learning_rate": 5.781540144467229e-05,
      "loss": 4.564,
      "step": 37820
    },
    {
      "epoch": 0.77457866606598,
      "grad_norm": 0.2706857919692993,
      "learning_rate": 5.776658159531318e-05,
      "loss": 4.5892,
      "step": 37825
    },
    {
      "epoch": 0.7746810558433846,
      "grad_norm": 0.27617236971855164,
      "learning_rate": 5.7717777450694e-05,
      "loss": 4.5907,
      "step": 37830
    },
    {
      "epoch": 0.7747834456207893,
      "grad_norm": 0.278952419757843,
      "learning_rate": 5.766898901912464e-05,
      "loss": 4.6563,
      "step": 37835
    },
    {
      "epoch": 0.7748858353981939,
      "grad_norm": 0.29718565940856934,
      "learning_rate": 5.762021630891251e-05,
      "loss": 4.577,
      "step": 37840
    },
    {
      "epoch": 0.7749882251755985,
      "grad_norm": 0.268237829208374,
      "learning_rate": 5.7571459328362165e-05,
      "loss": 4.5341,
      "step": 37845
    },
    {
      "epoch": 0.7750906149530031,
      "grad_norm": 0.2789965271949768,
      "learning_rate": 5.7522718085775496e-05,
      "loss": 4.5685,
      "step": 37850
    },
    {
      "epoch": 0.7751930047304078,
      "grad_norm": 0.2974662184715271,
      "learning_rate": 5.7473992589451865e-05,
      "loss": 4.6499,
      "step": 37855
    },
    {
      "epoch": 0.7752953945078124,
      "grad_norm": 0.2781980633735657,
      "learning_rate": 5.7425282847687756e-05,
      "loss": 4.6085,
      "step": 37860
    },
    {
      "epoch": 0.775397784285217,
      "grad_norm": 0.3048999309539795,
      "learning_rate": 5.737658886877704e-05,
      "loss": 4.5934,
      "step": 37865
    },
    {
      "epoch": 0.7755001740626216,
      "grad_norm": 0.28093934059143066,
      "learning_rate": 5.732791066101107e-05,
      "loss": 4.5681,
      "step": 37870
    },
    {
      "epoch": 0.7756025638400262,
      "grad_norm": 0.3381670415401459,
      "learning_rate": 5.727924823267817e-05,
      "loss": 4.6332,
      "step": 37875
    },
    {
      "epoch": 0.7757049536174309,
      "grad_norm": 0.26587992906570435,
      "learning_rate": 5.7230601592064304e-05,
      "loss": 4.5393,
      "step": 37880
    },
    {
      "epoch": 0.7758073433948355,
      "grad_norm": 0.2888360917568207,
      "learning_rate": 5.718197074745259e-05,
      "loss": 4.5999,
      "step": 37885
    },
    {
      "epoch": 0.7759097331722401,
      "grad_norm": 0.2756347954273224,
      "learning_rate": 5.713335570712342e-05,
      "loss": 4.5961,
      "step": 37890
    },
    {
      "epoch": 0.7760121229496447,
      "grad_norm": 0.29758137464523315,
      "learning_rate": 5.7084756479354673e-05,
      "loss": 4.6282,
      "step": 37895
    },
    {
      "epoch": 0.7761145127270493,
      "grad_norm": 0.2744895815849304,
      "learning_rate": 5.703617307242136e-05,
      "loss": 4.6118,
      "step": 37900
    },
    {
      "epoch": 0.776216902504454,
      "grad_norm": 0.2868340313434601,
      "learning_rate": 5.698760549459581e-05,
      "loss": 4.5726,
      "step": 37905
    },
    {
      "epoch": 0.7763192922818586,
      "grad_norm": 0.2860718071460724,
      "learning_rate": 5.69390537541478e-05,
      "loss": 4.5842,
      "step": 37910
    },
    {
      "epoch": 0.7764216820592632,
      "grad_norm": 0.2746017575263977,
      "learning_rate": 5.689051785934423e-05,
      "loss": 4.6116,
      "step": 37915
    },
    {
      "epoch": 0.7765240718366678,
      "grad_norm": 0.2836153209209442,
      "learning_rate": 5.6841997818449537e-05,
      "loss": 4.5946,
      "step": 37920
    },
    {
      "epoch": 0.7766264616140724,
      "grad_norm": 0.36896389722824097,
      "learning_rate": 5.67934936397251e-05,
      "loss": 4.5661,
      "step": 37925
    },
    {
      "epoch": 0.7767288513914771,
      "grad_norm": 0.28288668394088745,
      "learning_rate": 5.6745005331429986e-05,
      "loss": 4.574,
      "step": 37930
    },
    {
      "epoch": 0.7768312411688817,
      "grad_norm": 0.2924422025680542,
      "learning_rate": 5.669653290182029e-05,
      "loss": 4.6238,
      "step": 37935
    },
    {
      "epoch": 0.7769336309462863,
      "grad_norm": 0.2679120600223541,
      "learning_rate": 5.6648076359149476e-05,
      "loss": 4.557,
      "step": 37940
    },
    {
      "epoch": 0.7770360207236909,
      "grad_norm": 0.29370808601379395,
      "learning_rate": 5.6599635711668404e-05,
      "loss": 4.616,
      "step": 37945
    },
    {
      "epoch": 0.7771384105010956,
      "grad_norm": 0.2717418968677521,
      "learning_rate": 5.6551210967625105e-05,
      "loss": 4.5977,
      "step": 37950
    },
    {
      "epoch": 0.7772408002785002,
      "grad_norm": 0.26935046911239624,
      "learning_rate": 5.650280213526489e-05,
      "loss": 4.6466,
      "step": 37955
    },
    {
      "epoch": 0.7773431900559048,
      "grad_norm": 0.27534303069114685,
      "learning_rate": 5.6454409222830507e-05,
      "loss": 4.6109,
      "step": 37960
    },
    {
      "epoch": 0.7774455798333094,
      "grad_norm": 0.28345656394958496,
      "learning_rate": 5.640603223856185e-05,
      "loss": 4.5577,
      "step": 37965
    },
    {
      "epoch": 0.777547969610714,
      "grad_norm": 0.27641671895980835,
      "learning_rate": 5.635767119069615e-05,
      "loss": 4.5759,
      "step": 37970
    },
    {
      "epoch": 0.7776503593881187,
      "grad_norm": 0.2733546197414398,
      "learning_rate": 5.630932608746789e-05,
      "loss": 4.6053,
      "step": 37975
    },
    {
      "epoch": 0.7777527491655233,
      "grad_norm": 0.27268344163894653,
      "learning_rate": 5.626099693710894e-05,
      "loss": 4.5986,
      "step": 37980
    },
    {
      "epoch": 0.7778551389429279,
      "grad_norm": 0.29251590371131897,
      "learning_rate": 5.621268374784837e-05,
      "loss": 4.5748,
      "step": 37985
    },
    {
      "epoch": 0.7779575287203325,
      "grad_norm": 0.2936033010482788,
      "learning_rate": 5.6164386527912473e-05,
      "loss": 4.6025,
      "step": 37990
    },
    {
      "epoch": 0.7780599184977371,
      "grad_norm": 0.2928719222545624,
      "learning_rate": 5.611610528552508e-05,
      "loss": 4.6442,
      "step": 37995
    },
    {
      "epoch": 0.7781623082751418,
      "grad_norm": 0.3127742409706116,
      "learning_rate": 5.606784002890688e-05,
      "loss": 4.4863,
      "step": 38000
    },
    {
      "epoch": 0.7781623082751418,
      "eval_loss": 4.6109232902526855,
      "eval_runtime": 28.7515,
      "eval_samples_per_second": 20.242,
      "eval_steps_per_second": 2.539,
      "step": 38000
    },
    {
      "epoch": 0.7782646980525464,
      "grad_norm": 0.28875091671943665,
      "learning_rate": 5.60195907662762e-05,
      "loss": 4.5884,
      "step": 38005
    },
    {
      "epoch": 0.778367087829951,
      "grad_norm": 0.2909495234489441,
      "learning_rate": 5.5971357505848625e-05,
      "loss": 4.5766,
      "step": 38010
    },
    {
      "epoch": 0.7784694776073556,
      "grad_norm": 0.28629270195961,
      "learning_rate": 5.5923140255836696e-05,
      "loss": 4.609,
      "step": 38015
    },
    {
      "epoch": 0.7785718673847603,
      "grad_norm": 0.28371039032936096,
      "learning_rate": 5.5874939024450625e-05,
      "loss": 4.562,
      "step": 38020
    },
    {
      "epoch": 0.778674257162165,
      "grad_norm": 0.2701385021209717,
      "learning_rate": 5.5826753819897654e-05,
      "loss": 4.5909,
      "step": 38025
    },
    {
      "epoch": 0.7787766469395696,
      "grad_norm": 0.2845456302165985,
      "learning_rate": 5.5778584650382307e-05,
      "loss": 4.5905,
      "step": 38030
    },
    {
      "epoch": 0.7788790367169742,
      "grad_norm": 0.30409011244773865,
      "learning_rate": 5.573043152410652e-05,
      "loss": 4.5325,
      "step": 38035
    },
    {
      "epoch": 0.7789814264943788,
      "grad_norm": 0.3092602491378784,
      "learning_rate": 5.568229444926937e-05,
      "loss": 4.5918,
      "step": 38040
    },
    {
      "epoch": 0.7790838162717835,
      "grad_norm": 0.2709414064884186,
      "learning_rate": 5.563417343406718e-05,
      "loss": 4.5563,
      "step": 38045
    },
    {
      "epoch": 0.7791862060491881,
      "grad_norm": 0.3534092903137207,
      "learning_rate": 5.5586068486693705e-05,
      "loss": 4.6015,
      "step": 38050
    },
    {
      "epoch": 0.7792885958265927,
      "grad_norm": 0.29504942893981934,
      "learning_rate": 5.553797961533977e-05,
      "loss": 4.5999,
      "step": 38055
    },
    {
      "epoch": 0.7793909856039973,
      "grad_norm": 0.31223994493484497,
      "learning_rate": 5.5489906828193656e-05,
      "loss": 4.6144,
      "step": 38060
    },
    {
      "epoch": 0.779493375381402,
      "grad_norm": 0.3386547863483429,
      "learning_rate": 5.5441850133440634e-05,
      "loss": 4.5956,
      "step": 38065
    },
    {
      "epoch": 0.7795957651588066,
      "grad_norm": 0.27085545659065247,
      "learning_rate": 5.5393809539263535e-05,
      "loss": 4.5194,
      "step": 38070
    },
    {
      "epoch": 0.7796981549362112,
      "grad_norm": 0.5431325435638428,
      "learning_rate": 5.534578505384226e-05,
      "loss": 4.5828,
      "step": 38075
    },
    {
      "epoch": 0.7798005447136158,
      "grad_norm": 0.2880275845527649,
      "learning_rate": 5.529777668535398e-05,
      "loss": 4.5893,
      "step": 38080
    },
    {
      "epoch": 0.7799029344910204,
      "grad_norm": 0.27586859464645386,
      "learning_rate": 5.5249784441973244e-05,
      "loss": 4.5867,
      "step": 38085
    },
    {
      "epoch": 0.7800053242684251,
      "grad_norm": 0.30714014172554016,
      "learning_rate": 5.5201808331871736e-05,
      "loss": 4.5511,
      "step": 38090
    },
    {
      "epoch": 0.7801077140458297,
      "grad_norm": 0.28004249930381775,
      "learning_rate": 5.515384836321838e-05,
      "loss": 4.6185,
      "step": 38095
    },
    {
      "epoch": 0.7802101038232343,
      "grad_norm": 0.2809518277645111,
      "learning_rate": 5.5105904544179496e-05,
      "loss": 4.5796,
      "step": 38100
    },
    {
      "epoch": 0.7803124936006389,
      "grad_norm": 0.2761389911174774,
      "learning_rate": 5.5057976882918516e-05,
      "loss": 4.5321,
      "step": 38105
    },
    {
      "epoch": 0.7804148833780435,
      "grad_norm": 0.3077124059200287,
      "learning_rate": 5.501006538759615e-05,
      "loss": 4.5659,
      "step": 38110
    },
    {
      "epoch": 0.7805172731554482,
      "grad_norm": 0.2878948748111725,
      "learning_rate": 5.496217006637032e-05,
      "loss": 4.5707,
      "step": 38115
    },
    {
      "epoch": 0.7806196629328528,
      "grad_norm": 0.2937699854373932,
      "learning_rate": 5.491429092739636e-05,
      "loss": 4.5778,
      "step": 38120
    },
    {
      "epoch": 0.7807220527102574,
      "grad_norm": 0.2876644730567932,
      "learning_rate": 5.486642797882667e-05,
      "loss": 4.6051,
      "step": 38125
    },
    {
      "epoch": 0.780824442487662,
      "grad_norm": 0.31219735741615295,
      "learning_rate": 5.4818581228810906e-05,
      "loss": 4.585,
      "step": 38130
    },
    {
      "epoch": 0.7809268322650667,
      "grad_norm": 0.26896148920059204,
      "learning_rate": 5.477075068549617e-05,
      "loss": 4.6168,
      "step": 38135
    },
    {
      "epoch": 0.7810292220424713,
      "grad_norm": 0.334077924489975,
      "learning_rate": 5.472293635702644e-05,
      "loss": 4.6099,
      "step": 38140
    },
    {
      "epoch": 0.7811316118198759,
      "grad_norm": 0.2796894907951355,
      "learning_rate": 5.467513825154325e-05,
      "loss": 4.5843,
      "step": 38145
    },
    {
      "epoch": 0.7812340015972805,
      "grad_norm": 0.3577290177345276,
      "learning_rate": 5.462735637718535e-05,
      "loss": 4.6231,
      "step": 38150
    },
    {
      "epoch": 0.7813363913746851,
      "grad_norm": 0.27947860956192017,
      "learning_rate": 5.4579590742088466e-05,
      "loss": 4.6125,
      "step": 38155
    },
    {
      "epoch": 0.7814387811520898,
      "grad_norm": 0.2753649950027466,
      "learning_rate": 5.453184135438587e-05,
      "loss": 4.6266,
      "step": 38160
    },
    {
      "epoch": 0.7815411709294944,
      "grad_norm": 0.3409067988395691,
      "learning_rate": 5.448410822220786e-05,
      "loss": 4.5962,
      "step": 38165
    },
    {
      "epoch": 0.781643560706899,
      "grad_norm": 0.2706255316734314,
      "learning_rate": 5.443639135368203e-05,
      "loss": 4.5975,
      "step": 38170
    },
    {
      "epoch": 0.7817459504843036,
      "grad_norm": 0.2951542139053345,
      "learning_rate": 5.438869075693329e-05,
      "loss": 4.5826,
      "step": 38175
    },
    {
      "epoch": 0.7818483402617082,
      "grad_norm": 0.2736496031284332,
      "learning_rate": 5.434100644008365e-05,
      "loss": 4.5533,
      "step": 38180
    },
    {
      "epoch": 0.7819507300391129,
      "grad_norm": 0.270050048828125,
      "learning_rate": 5.4293338411252387e-05,
      "loss": 4.5591,
      "step": 38185
    },
    {
      "epoch": 0.7820531198165175,
      "grad_norm": 0.28652334213256836,
      "learning_rate": 5.424568667855598e-05,
      "loss": 4.5555,
      "step": 38190
    },
    {
      "epoch": 0.7821555095939221,
      "grad_norm": 0.27525264024734497,
      "learning_rate": 5.4198051250108214e-05,
      "loss": 4.6588,
      "step": 38195
    },
    {
      "epoch": 0.7822578993713267,
      "grad_norm": 0.2802639603614807,
      "learning_rate": 5.415043213402018e-05,
      "loss": 4.6168,
      "step": 38200
    },
    {
      "epoch": 0.7823602891487313,
      "grad_norm": 0.27300724387168884,
      "learning_rate": 5.410282933839983e-05,
      "loss": 4.5991,
      "step": 38205
    },
    {
      "epoch": 0.782462678926136,
      "grad_norm": 0.27656131982803345,
      "learning_rate": 5.405524287135275e-05,
      "loss": 4.5723,
      "step": 38210
    },
    {
      "epoch": 0.7825650687035406,
      "grad_norm": 0.30277910828590393,
      "learning_rate": 5.400767274098149e-05,
      "loss": 4.5386,
      "step": 38215
    },
    {
      "epoch": 0.7826674584809452,
      "grad_norm": 0.28365784883499146,
      "learning_rate": 5.396011895538587e-05,
      "loss": 4.5923,
      "step": 38220
    },
    {
      "epoch": 0.7827698482583498,
      "grad_norm": 0.2815352976322174,
      "learning_rate": 5.391258152266304e-05,
      "loss": 4.498,
      "step": 38225
    },
    {
      "epoch": 0.7828722380357546,
      "grad_norm": 0.2815379202365875,
      "learning_rate": 5.386506045090724e-05,
      "loss": 4.6176,
      "step": 38230
    },
    {
      "epoch": 0.7829746278131592,
      "grad_norm": 0.2661249339580536,
      "learning_rate": 5.381755574820989e-05,
      "loss": 4.5613,
      "step": 38235
    },
    {
      "epoch": 0.7830770175905638,
      "grad_norm": 0.30869239568710327,
      "learning_rate": 5.377006742265984e-05,
      "loss": 4.5774,
      "step": 38240
    },
    {
      "epoch": 0.7831794073679684,
      "grad_norm": 0.26999416947364807,
      "learning_rate": 5.3722595482342925e-05,
      "loss": 4.5696,
      "step": 38245
    },
    {
      "epoch": 0.783281797145373,
      "grad_norm": 0.2677248418331146,
      "learning_rate": 5.367513993534226e-05,
      "loss": 4.6001,
      "step": 38250
    },
    {
      "epoch": 0.7833841869227777,
      "grad_norm": 0.2759122848510742,
      "learning_rate": 5.3627700789738156e-05,
      "loss": 4.6219,
      "step": 38255
    },
    {
      "epoch": 0.7834865767001823,
      "grad_norm": 0.26571130752563477,
      "learning_rate": 5.358027805360826e-05,
      "loss": 4.6068,
      "step": 38260
    },
    {
      "epoch": 0.7835889664775869,
      "grad_norm": 0.2919250726699829,
      "learning_rate": 5.353287173502726e-05,
      "loss": 4.6113,
      "step": 38265
    },
    {
      "epoch": 0.7836913562549915,
      "grad_norm": 0.2928933799266815,
      "learning_rate": 5.348548184206705e-05,
      "loss": 4.598,
      "step": 38270
    },
    {
      "epoch": 0.7837937460323962,
      "grad_norm": 0.2709066867828369,
      "learning_rate": 5.3438108382796945e-05,
      "loss": 4.5881,
      "step": 38275
    },
    {
      "epoch": 0.7838961358098008,
      "grad_norm": 0.30528467893600464,
      "learning_rate": 5.339075136528312e-05,
      "loss": 4.6173,
      "step": 38280
    },
    {
      "epoch": 0.7839985255872054,
      "grad_norm": 0.2768768072128296,
      "learning_rate": 5.33434107975892e-05,
      "loss": 4.5767,
      "step": 38285
    },
    {
      "epoch": 0.78410091536461,
      "grad_norm": 0.27764803171157837,
      "learning_rate": 5.3296086687776084e-05,
      "loss": 4.6202,
      "step": 38290
    },
    {
      "epoch": 0.7842033051420146,
      "grad_norm": 0.27811387181282043,
      "learning_rate": 5.32487790439015e-05,
      "loss": 4.6258,
      "step": 38295
    },
    {
      "epoch": 0.7843056949194193,
      "grad_norm": 0.29696446657180786,
      "learning_rate": 5.320148787402077e-05,
      "loss": 4.6243,
      "step": 38300
    },
    {
      "epoch": 0.7844080846968239,
      "grad_norm": 0.29483506083488464,
      "learning_rate": 5.3154213186186165e-05,
      "loss": 4.5898,
      "step": 38305
    },
    {
      "epoch": 0.7845104744742285,
      "grad_norm": 0.2779998183250427,
      "learning_rate": 5.3106954988447204e-05,
      "loss": 4.5817,
      "step": 38310
    },
    {
      "epoch": 0.7846128642516331,
      "grad_norm": 0.2914763391017914,
      "learning_rate": 5.305971328885071e-05,
      "loss": 4.5334,
      "step": 38315
    },
    {
      "epoch": 0.7847152540290377,
      "grad_norm": 0.2841431796550751,
      "learning_rate": 5.301248809544053e-05,
      "loss": 4.5977,
      "step": 38320
    },
    {
      "epoch": 0.7848176438064424,
      "grad_norm": 0.26590022444725037,
      "learning_rate": 5.296527941625782e-05,
      "loss": 4.5799,
      "step": 38325
    },
    {
      "epoch": 0.784920033583847,
      "grad_norm": 0.280025452375412,
      "learning_rate": 5.2918087259340816e-05,
      "loss": 4.5617,
      "step": 38330
    },
    {
      "epoch": 0.7850224233612516,
      "grad_norm": 0.2960115373134613,
      "learning_rate": 5.2870911632725045e-05,
      "loss": 4.4566,
      "step": 38335
    },
    {
      "epoch": 0.7851248131386562,
      "grad_norm": 0.3256969749927521,
      "learning_rate": 5.2823752544443304e-05,
      "loss": 4.5568,
      "step": 38340
    },
    {
      "epoch": 0.7852272029160609,
      "grad_norm": 0.3043464720249176,
      "learning_rate": 5.277661000252523e-05,
      "loss": 4.5841,
      "step": 38345
    },
    {
      "epoch": 0.7853295926934655,
      "grad_norm": 0.2751842439174652,
      "learning_rate": 5.2729484014998006e-05,
      "loss": 4.5798,
      "step": 38350
    },
    {
      "epoch": 0.7854319824708701,
      "grad_norm": 0.28036338090896606,
      "learning_rate": 5.268237458988583e-05,
      "loss": 4.5693,
      "step": 38355
    },
    {
      "epoch": 0.7855343722482747,
      "grad_norm": 0.27092334628105164,
      "learning_rate": 5.263528173521004e-05,
      "loss": 4.6061,
      "step": 38360
    },
    {
      "epoch": 0.7856367620256793,
      "grad_norm": 0.33851486444473267,
      "learning_rate": 5.2588205458989305e-05,
      "loss": 4.5907,
      "step": 38365
    },
    {
      "epoch": 0.785739151803084,
      "grad_norm": 0.2805909514427185,
      "learning_rate": 5.2541145769239355e-05,
      "loss": 4.5984,
      "step": 38370
    },
    {
      "epoch": 0.7858415415804886,
      "grad_norm": 0.29815682768821716,
      "learning_rate": 5.249410267397309e-05,
      "loss": 4.5704,
      "step": 38375
    },
    {
      "epoch": 0.7859439313578932,
      "grad_norm": 0.27646151185035706,
      "learning_rate": 5.244707618120061e-05,
      "loss": 4.5422,
      "step": 38380
    },
    {
      "epoch": 0.7860463211352978,
      "grad_norm": 0.32369154691696167,
      "learning_rate": 5.2400066298929264e-05,
      "loss": 4.5315,
      "step": 38385
    },
    {
      "epoch": 0.7861487109127024,
      "grad_norm": 0.28201964497566223,
      "learning_rate": 5.235307303516348e-05,
      "loss": 4.6324,
      "step": 38390
    },
    {
      "epoch": 0.7862511006901071,
      "grad_norm": 0.27669161558151245,
      "learning_rate": 5.230609639790479e-05,
      "loss": 4.5939,
      "step": 38395
    },
    {
      "epoch": 0.7863534904675117,
      "grad_norm": 0.28544601798057556,
      "learning_rate": 5.225913639515212e-05,
      "loss": 4.6243,
      "step": 38400
    },
    {
      "epoch": 0.7864558802449163,
      "grad_norm": 0.3115270137786865,
      "learning_rate": 5.2212193034901365e-05,
      "loss": 4.5518,
      "step": 38405
    },
    {
      "epoch": 0.7865582700223209,
      "grad_norm": 0.2983291745185852,
      "learning_rate": 5.21652663251456e-05,
      "loss": 4.6791,
      "step": 38410
    },
    {
      "epoch": 0.7866606597997255,
      "grad_norm": 0.8530471324920654,
      "learning_rate": 5.211835627387529e-05,
      "loss": 4.566,
      "step": 38415
    },
    {
      "epoch": 0.7867630495771302,
      "grad_norm": 0.2956152856349945,
      "learning_rate": 5.207146288907765e-05,
      "loss": 4.5612,
      "step": 38420
    },
    {
      "epoch": 0.7868654393545348,
      "grad_norm": 0.2825814187526703,
      "learning_rate": 5.20245861787374e-05,
      "loss": 4.5784,
      "step": 38425
    },
    {
      "epoch": 0.7869678291319395,
      "grad_norm": 0.2654380798339844,
      "learning_rate": 5.1977726150836456e-05,
      "loss": 4.5618,
      "step": 38430
    },
    {
      "epoch": 0.7870702189093441,
      "grad_norm": 0.2694794833660126,
      "learning_rate": 5.1930882813353506e-05,
      "loss": 4.6147,
      "step": 38435
    },
    {
      "epoch": 0.7871726086867488,
      "grad_norm": 0.2700869143009186,
      "learning_rate": 5.1884056174264825e-05,
      "loss": 4.581,
      "step": 38440
    },
    {
      "epoch": 0.7872749984641534,
      "grad_norm": 0.2823755145072937,
      "learning_rate": 5.183724624154362e-05,
      "loss": 4.6208,
      "step": 38445
    },
    {
      "epoch": 0.787377388241558,
      "grad_norm": 0.30310189723968506,
      "learning_rate": 5.179045302316021e-05,
      "loss": 4.5524,
      "step": 38450
    },
    {
      "epoch": 0.7874797780189626,
      "grad_norm": 0.2830992043018341,
      "learning_rate": 5.174367652708228e-05,
      "loss": 4.6022,
      "step": 38455
    },
    {
      "epoch": 0.7875821677963672,
      "grad_norm": 0.27573373913764954,
      "learning_rate": 5.16969167612745e-05,
      "loss": 4.5758,
      "step": 38460
    },
    {
      "epoch": 0.7876845575737719,
      "grad_norm": 0.2764032781124115,
      "learning_rate": 5.1650173733698714e-05,
      "loss": 4.5994,
      "step": 38465
    },
    {
      "epoch": 0.7877869473511765,
      "grad_norm": 0.2729256749153137,
      "learning_rate": 5.160344745231391e-05,
      "loss": 4.5756,
      "step": 38470
    },
    {
      "epoch": 0.7878893371285811,
      "grad_norm": 0.277967244386673,
      "learning_rate": 5.155673792507628e-05,
      "loss": 4.4947,
      "step": 38475
    },
    {
      "epoch": 0.7879917269059857,
      "grad_norm": 0.28639402985572815,
      "learning_rate": 5.151004515993924e-05,
      "loss": 4.5977,
      "step": 38480
    },
    {
      "epoch": 0.7880941166833904,
      "grad_norm": 0.27312996983528137,
      "learning_rate": 5.1463369164853063e-05,
      "loss": 4.5738,
      "step": 38485
    },
    {
      "epoch": 0.788196506460795,
      "grad_norm": 0.2720889747142792,
      "learning_rate": 5.141670994776546e-05,
      "loss": 4.6397,
      "step": 38490
    },
    {
      "epoch": 0.7882988962381996,
      "grad_norm": 0.2668980360031128,
      "learning_rate": 5.1370067516621164e-05,
      "loss": 4.5876,
      "step": 38495
    },
    {
      "epoch": 0.7884012860156042,
      "grad_norm": 0.27054527401924133,
      "learning_rate": 5.1323441879361995e-05,
      "loss": 4.5783,
      "step": 38500
    },
    {
      "epoch": 0.7885036757930088,
      "grad_norm": 0.28997063636779785,
      "learning_rate": 5.1276833043927075e-05,
      "loss": 4.5805,
      "step": 38505
    },
    {
      "epoch": 0.7886060655704135,
      "grad_norm": 0.30233728885650635,
      "learning_rate": 5.123024101825254e-05,
      "loss": 4.6095,
      "step": 38510
    },
    {
      "epoch": 0.7887084553478181,
      "grad_norm": 0.4892871677875519,
      "learning_rate": 5.1183665810271657e-05,
      "loss": 4.4724,
      "step": 38515
    },
    {
      "epoch": 0.7888108451252227,
      "grad_norm": 0.27788451313972473,
      "learning_rate": 5.113710742791484e-05,
      "loss": 4.5916,
      "step": 38520
    },
    {
      "epoch": 0.7889132349026273,
      "grad_norm": 0.27443966269493103,
      "learning_rate": 5.1090565879109766e-05,
      "loss": 4.5886,
      "step": 38525
    },
    {
      "epoch": 0.789015624680032,
      "grad_norm": 0.2806726396083832,
      "learning_rate": 5.104404117178108e-05,
      "loss": 4.5716,
      "step": 38530
    },
    {
      "epoch": 0.7891180144574366,
      "grad_norm": 0.330605149269104,
      "learning_rate": 5.099753331385057e-05,
      "loss": 4.5924,
      "step": 38535
    },
    {
      "epoch": 0.7892204042348412,
      "grad_norm": 0.2898256778717041,
      "learning_rate": 5.095104231323732e-05,
      "loss": 4.5542,
      "step": 38540
    },
    {
      "epoch": 0.7893227940122458,
      "grad_norm": 0.28494715690612793,
      "learning_rate": 5.090456817785739e-05,
      "loss": 4.5786,
      "step": 38545
    },
    {
      "epoch": 0.7894251837896504,
      "grad_norm": 0.2826581299304962,
      "learning_rate": 5.085811091562393e-05,
      "loss": 4.559,
      "step": 38550
    },
    {
      "epoch": 0.789527573567055,
      "grad_norm": 0.2861868441104889,
      "learning_rate": 5.081167053444747e-05,
      "loss": 4.55,
      "step": 38555
    },
    {
      "epoch": 0.7896299633444597,
      "grad_norm": 0.27087080478668213,
      "learning_rate": 5.076524704223529e-05,
      "loss": 4.5422,
      "step": 38560
    },
    {
      "epoch": 0.7897323531218643,
      "grad_norm": 0.29426097869873047,
      "learning_rate": 5.0718840446892126e-05,
      "loss": 4.5025,
      "step": 38565
    },
    {
      "epoch": 0.7898347428992689,
      "grad_norm": 0.9471340775489807,
      "learning_rate": 5.067245075631968e-05,
      "loss": 4.4886,
      "step": 38570
    },
    {
      "epoch": 0.7899371326766735,
      "grad_norm": 0.2938225567340851,
      "learning_rate": 5.0626077978416746e-05,
      "loss": 4.578,
      "step": 38575
    },
    {
      "epoch": 0.7900395224540782,
      "grad_norm": 0.2911272644996643,
      "learning_rate": 5.0579722121079395e-05,
      "loss": 4.5978,
      "step": 38580
    },
    {
      "epoch": 0.7901419122314828,
      "grad_norm": 0.27577346563339233,
      "learning_rate": 5.053338319220068e-05,
      "loss": 4.6174,
      "step": 38585
    },
    {
      "epoch": 0.7902443020088874,
      "grad_norm": 0.31076574325561523,
      "learning_rate": 5.0487061199670724e-05,
      "loss": 4.5837,
      "step": 38590
    },
    {
      "epoch": 0.790346691786292,
      "grad_norm": 0.27288109064102173,
      "learning_rate": 5.0440756151376985e-05,
      "loss": 4.5775,
      "step": 38595
    },
    {
      "epoch": 0.7904490815636966,
      "grad_norm": 0.3731415569782257,
      "learning_rate": 5.039446805520385e-05,
      "loss": 4.6132,
      "step": 38600
    },
    {
      "epoch": 0.7905514713411013,
      "grad_norm": 0.2759488523006439,
      "learning_rate": 5.034819691903283e-05,
      "loss": 4.5818,
      "step": 38605
    },
    {
      "epoch": 0.7906538611185059,
      "grad_norm": 0.27843356132507324,
      "learning_rate": 5.0301942750742596e-05,
      "loss": 4.5705,
      "step": 38610
    },
    {
      "epoch": 0.7907562508959105,
      "grad_norm": 0.29055437445640564,
      "learning_rate": 5.025570555820893e-05,
      "loss": 4.6136,
      "step": 38615
    },
    {
      "epoch": 0.7908586406733151,
      "grad_norm": 0.2942349314689636,
      "learning_rate": 5.0209485349304854e-05,
      "loss": 4.6066,
      "step": 38620
    },
    {
      "epoch": 0.7909610304507197,
      "grad_norm": 0.28809309005737305,
      "learning_rate": 5.016328213190013e-05,
      "loss": 4.629,
      "step": 38625
    },
    {
      "epoch": 0.7910634202281244,
      "grad_norm": 0.27980896830558777,
      "learning_rate": 5.011709591386202e-05,
      "loss": 4.5645,
      "step": 38630
    },
    {
      "epoch": 0.7911658100055291,
      "grad_norm": 0.2762549817562103,
      "learning_rate": 5.007092670305467e-05,
      "loss": 4.5757,
      "step": 38635
    },
    {
      "epoch": 0.7912681997829337,
      "grad_norm": 0.2708359956741333,
      "learning_rate": 5.0024774507339354e-05,
      "loss": 4.5934,
      "step": 38640
    },
    {
      "epoch": 0.7913705895603383,
      "grad_norm": 0.29552069306373596,
      "learning_rate": 4.997863933457459e-05,
      "loss": 4.5822,
      "step": 38645
    },
    {
      "epoch": 0.791472979337743,
      "grad_norm": 0.2964145243167877,
      "learning_rate": 4.993252119261582e-05,
      "loss": 4.5853,
      "step": 38650
    },
    {
      "epoch": 0.7915753691151476,
      "grad_norm": 0.28075456619262695,
      "learning_rate": 4.988642008931567e-05,
      "loss": 4.5389,
      "step": 38655
    },
    {
      "epoch": 0.7916777588925522,
      "grad_norm": 0.2771109342575073,
      "learning_rate": 4.984033603252379e-05,
      "loss": 4.6293,
      "step": 38660
    },
    {
      "epoch": 0.7917801486699568,
      "grad_norm": 0.2782790958881378,
      "learning_rate": 4.9794269030087126e-05,
      "loss": 4.6077,
      "step": 38665
    },
    {
      "epoch": 0.7918825384473615,
      "grad_norm": 0.28569403290748596,
      "learning_rate": 4.974821908984948e-05,
      "loss": 4.6277,
      "step": 38670
    },
    {
      "epoch": 0.7919849282247661,
      "grad_norm": 0.28869330883026123,
      "learning_rate": 4.970218621965185e-05,
      "loss": 4.5884,
      "step": 38675
    },
    {
      "epoch": 0.7920873180021707,
      "grad_norm": 0.28534212708473206,
      "learning_rate": 4.965617042733241e-05,
      "loss": 4.5845,
      "step": 38680
    },
    {
      "epoch": 0.7921897077795753,
      "grad_norm": 0.28274044394493103,
      "learning_rate": 4.9610171720726297e-05,
      "loss": 4.621,
      "step": 38685
    },
    {
      "epoch": 0.7922920975569799,
      "grad_norm": 0.28722789883613586,
      "learning_rate": 4.956419010766575e-05,
      "loss": 4.5866,
      "step": 38690
    },
    {
      "epoch": 0.7923944873343846,
      "grad_norm": 0.28945234417915344,
      "learning_rate": 4.9518225595980274e-05,
      "loss": 4.5916,
      "step": 38695
    },
    {
      "epoch": 0.7924968771117892,
      "grad_norm": 0.27766236662864685,
      "learning_rate": 4.9472278193496115e-05,
      "loss": 4.6023,
      "step": 38700
    },
    {
      "epoch": 0.7925992668891938,
      "grad_norm": 0.2779901921749115,
      "learning_rate": 4.9426347908036985e-05,
      "loss": 4.5972,
      "step": 38705
    },
    {
      "epoch": 0.7927016566665984,
      "grad_norm": 0.2771102488040924,
      "learning_rate": 4.938043474742345e-05,
      "loss": 4.5655,
      "step": 38710
    },
    {
      "epoch": 0.792804046444003,
      "grad_norm": 0.26912617683410645,
      "learning_rate": 4.9334538719473184e-05,
      "loss": 4.6776,
      "step": 38715
    },
    {
      "epoch": 0.7929064362214077,
      "grad_norm": 0.2799225449562073,
      "learning_rate": 4.928865983200107e-05,
      "loss": 4.5719,
      "step": 38720
    },
    {
      "epoch": 0.7930088259988123,
      "grad_norm": 0.2933421730995178,
      "learning_rate": 4.924279809281893e-05,
      "loss": 4.5874,
      "step": 38725
    },
    {
      "epoch": 0.7931112157762169,
      "grad_norm": 0.27239277958869934,
      "learning_rate": 4.919695350973567e-05,
      "loss": 4.6052,
      "step": 38730
    },
    {
      "epoch": 0.7932136055536215,
      "grad_norm": 0.30010247230529785,
      "learning_rate": 4.915112609055743e-05,
      "loss": 4.5632,
      "step": 38735
    },
    {
      "epoch": 0.7933159953310261,
      "grad_norm": 0.4915987253189087,
      "learning_rate": 4.9105315843087285e-05,
      "loss": 4.6218,
      "step": 38740
    },
    {
      "epoch": 0.7934183851084308,
      "grad_norm": 0.2945227026939392,
      "learning_rate": 4.905952277512541e-05,
      "loss": 4.5622,
      "step": 38745
    },
    {
      "epoch": 0.7935207748858354,
      "grad_norm": 0.27611690759658813,
      "learning_rate": 4.9013746894469e-05,
      "loss": 4.5864,
      "step": 38750
    },
    {
      "epoch": 0.79362316466324,
      "grad_norm": 0.2813268303871155,
      "learning_rate": 4.8967988208912514e-05,
      "loss": 4.6249,
      "step": 38755
    },
    {
      "epoch": 0.7937255544406446,
      "grad_norm": 0.2851206362247467,
      "learning_rate": 4.892224672624732e-05,
      "loss": 4.5687,
      "step": 38760
    },
    {
      "epoch": 0.7938279442180493,
      "grad_norm": 0.29030922055244446,
      "learning_rate": 4.887652245426182e-05,
      "loss": 4.5937,
      "step": 38765
    },
    {
      "epoch": 0.7939303339954539,
      "grad_norm": 0.3038567006587982,
      "learning_rate": 4.883081540074169e-05,
      "loss": 4.5905,
      "step": 38770
    },
    {
      "epoch": 0.7940327237728585,
      "grad_norm": 0.27633538842201233,
      "learning_rate": 4.878512557346948e-05,
      "loss": 4.6112,
      "step": 38775
    },
    {
      "epoch": 0.7941351135502631,
      "grad_norm": 0.2828822135925293,
      "learning_rate": 4.873945298022481e-05,
      "loss": 4.6193,
      "step": 38780
    },
    {
      "epoch": 0.7942375033276677,
      "grad_norm": 0.2662680447101593,
      "learning_rate": 4.869379762878457e-05,
      "loss": 4.5674,
      "step": 38785
    },
    {
      "epoch": 0.7943398931050724,
      "grad_norm": 0.2869718074798584,
      "learning_rate": 4.8648159526922494e-05,
      "loss": 4.6003,
      "step": 38790
    },
    {
      "epoch": 0.794442282882477,
      "grad_norm": 0.2725258469581604,
      "learning_rate": 4.860253868240946e-05,
      "loss": 4.6024,
      "step": 38795
    },
    {
      "epoch": 0.7945446726598816,
      "grad_norm": 0.36530521512031555,
      "learning_rate": 4.8556935103013374e-05,
      "loss": 4.6455,
      "step": 38800
    },
    {
      "epoch": 0.7946470624372862,
      "grad_norm": 0.27146273851394653,
      "learning_rate": 4.85113487964993e-05,
      "loss": 4.5913,
      "step": 38805
    },
    {
      "epoch": 0.7947494522146908,
      "grad_norm": 0.2952762842178345,
      "learning_rate": 4.8465779770629303e-05,
      "loss": 4.5957,
      "step": 38810
    },
    {
      "epoch": 0.7948518419920955,
      "grad_norm": 0.278006911277771,
      "learning_rate": 4.8420228033162395e-05,
      "loss": 4.6793,
      "step": 38815
    },
    {
      "epoch": 0.7949542317695001,
      "grad_norm": 0.2903681695461273,
      "learning_rate": 4.837469359185486e-05,
      "loss": 4.5519,
      "step": 38820
    },
    {
      "epoch": 0.7950566215469047,
      "grad_norm": 0.301890105009079,
      "learning_rate": 4.832917645445989e-05,
      "loss": 4.6172,
      "step": 38825
    },
    {
      "epoch": 0.7951590113243093,
      "grad_norm": 0.2739580571651459,
      "learning_rate": 4.828367662872769e-05,
      "loss": 4.5555,
      "step": 38830
    },
    {
      "epoch": 0.795261401101714,
      "grad_norm": 0.2762344777584076,
      "learning_rate": 4.8238194122405767e-05,
      "loss": 4.5947,
      "step": 38835
    },
    {
      "epoch": 0.7953637908791187,
      "grad_norm": 0.2905920147895813,
      "learning_rate": 4.81927289432383e-05,
      "loss": 4.589,
      "step": 38840
    },
    {
      "epoch": 0.7954661806565233,
      "grad_norm": 0.29773831367492676,
      "learning_rate": 4.8147281098966856e-05,
      "loss": 4.5164,
      "step": 38845
    },
    {
      "epoch": 0.7955685704339279,
      "grad_norm": 0.27923330664634705,
      "learning_rate": 4.8101850597329875e-05,
      "loss": 4.5958,
      "step": 38850
    },
    {
      "epoch": 0.7956709602113325,
      "grad_norm": 0.2984638214111328,
      "learning_rate": 4.805643744606285e-05,
      "loss": 4.5721,
      "step": 38855
    },
    {
      "epoch": 0.7957733499887372,
      "grad_norm": 0.2867434024810791,
      "learning_rate": 4.801104165289843e-05,
      "loss": 4.6152,
      "step": 38860
    },
    {
      "epoch": 0.7958757397661418,
      "grad_norm": 0.2986920177936554,
      "learning_rate": 4.7965663225566195e-05,
      "loss": 4.593,
      "step": 38865
    },
    {
      "epoch": 0.7959781295435464,
      "grad_norm": 0.29490727186203003,
      "learning_rate": 4.792030217179275e-05,
      "loss": 4.6056,
      "step": 38870
    },
    {
      "epoch": 0.796080519320951,
      "grad_norm": 0.28355735540390015,
      "learning_rate": 4.787495849930191e-05,
      "loss": 4.5794,
      "step": 38875
    },
    {
      "epoch": 0.7961829090983557,
      "grad_norm": 0.2785748541355133,
      "learning_rate": 4.782963221581437e-05,
      "loss": 4.5824,
      "step": 38880
    },
    {
      "epoch": 0.7962852988757603,
      "grad_norm": 0.2727014720439911,
      "learning_rate": 4.778432332904789e-05,
      "loss": 4.5673,
      "step": 38885
    },
    {
      "epoch": 0.7963876886531649,
      "grad_norm": 0.2806575298309326,
      "learning_rate": 4.773903184671726e-05,
      "loss": 4.5598,
      "step": 38890
    },
    {
      "epoch": 0.7964900784305695,
      "grad_norm": 0.2848622798919678,
      "learning_rate": 4.7693757776534446e-05,
      "loss": 4.6432,
      "step": 38895
    },
    {
      "epoch": 0.7965924682079741,
      "grad_norm": 0.28101709485054016,
      "learning_rate": 4.764850112620826e-05,
      "loss": 4.6285,
      "step": 38900
    },
    {
      "epoch": 0.7966948579853788,
      "grad_norm": 0.26877561211586,
      "learning_rate": 4.7603261903444615e-05,
      "loss": 4.5875,
      "step": 38905
    },
    {
      "epoch": 0.7967972477627834,
      "grad_norm": 0.3063620626926422,
      "learning_rate": 4.7558040115946536e-05,
      "loss": 4.5648,
      "step": 38910
    },
    {
      "epoch": 0.796899637540188,
      "grad_norm": 0.28085240721702576,
      "learning_rate": 4.751283577141398e-05,
      "loss": 4.6174,
      "step": 38915
    },
    {
      "epoch": 0.7970020273175926,
      "grad_norm": 0.2840745747089386,
      "learning_rate": 4.7467648877543925e-05,
      "loss": 4.6076,
      "step": 38920
    },
    {
      "epoch": 0.7971044170949972,
      "grad_norm": 0.2984735071659088,
      "learning_rate": 4.742247944203051e-05,
      "loss": 4.6254,
      "step": 38925
    },
    {
      "epoch": 0.7972068068724019,
      "grad_norm": 0.29564350843429565,
      "learning_rate": 4.7377327472564755e-05,
      "loss": 4.5739,
      "step": 38930
    },
    {
      "epoch": 0.7973091966498065,
      "grad_norm": 0.2938145101070404,
      "learning_rate": 4.733219297683477e-05,
      "loss": 4.5883,
      "step": 38935
    },
    {
      "epoch": 0.7974115864272111,
      "grad_norm": 0.28416699171066284,
      "learning_rate": 4.7287075962525634e-05,
      "loss": 4.569,
      "step": 38940
    },
    {
      "epoch": 0.7975139762046157,
      "grad_norm": 0.48297443985939026,
      "learning_rate": 4.724197643731958e-05,
      "loss": 4.5492,
      "step": 38945
    },
    {
      "epoch": 0.7976163659820203,
      "grad_norm": 0.32836616039276123,
      "learning_rate": 4.719689440889576e-05,
      "loss": 4.5837,
      "step": 38950
    },
    {
      "epoch": 0.797718755759425,
      "grad_norm": 0.2783101797103882,
      "learning_rate": 4.7151829884930296e-05,
      "loss": 4.5946,
      "step": 38955
    },
    {
      "epoch": 0.7978211455368296,
      "grad_norm": 0.27545347809791565,
      "learning_rate": 4.7106782873096495e-05,
      "loss": 4.5771,
      "step": 38960
    },
    {
      "epoch": 0.7979235353142342,
      "grad_norm": 0.2808891832828522,
      "learning_rate": 4.706175338106457e-05,
      "loss": 4.5738,
      "step": 38965
    },
    {
      "epoch": 0.7980259250916388,
      "grad_norm": 0.2797003984451294,
      "learning_rate": 4.7016741416501664e-05,
      "loss": 4.6184,
      "step": 38970
    },
    {
      "epoch": 0.7981283148690435,
      "grad_norm": 0.2799162268638611,
      "learning_rate": 4.697174698707224e-05,
      "loss": 4.5623,
      "step": 38975
    },
    {
      "epoch": 0.7982307046464481,
      "grad_norm": 0.2852189540863037,
      "learning_rate": 4.692677010043737e-05,
      "loss": 4.5475,
      "step": 38980
    },
    {
      "epoch": 0.7983330944238527,
      "grad_norm": 0.2748004198074341,
      "learning_rate": 4.688181076425547e-05,
      "loss": 4.5872,
      "step": 38985
    },
    {
      "epoch": 0.7984354842012573,
      "grad_norm": 0.2709660828113556,
      "learning_rate": 4.68368689861818e-05,
      "loss": 4.6041,
      "step": 38990
    },
    {
      "epoch": 0.7985378739786619,
      "grad_norm": 0.29953452944755554,
      "learning_rate": 4.6791944773868635e-05,
      "loss": 4.5579,
      "step": 38995
    },
    {
      "epoch": 0.7986402637560666,
      "grad_norm": 0.2670140564441681,
      "learning_rate": 4.674703813496539e-05,
      "loss": 4.6503,
      "step": 39000
    },
    {
      "epoch": 0.7986402637560666,
      "eval_loss": 4.60409688949585,
      "eval_runtime": 17.1316,
      "eval_samples_per_second": 33.972,
      "eval_steps_per_second": 4.261,
      "step": 39000
    },
    {
      "epoch": 0.7987426535334712,
      "grad_norm": 0.2708469331264496,
      "learning_rate": 4.670214907711835e-05,
      "loss": 4.5807,
      "step": 39005
    },
    {
      "epoch": 0.7988450433108758,
      "grad_norm": 0.2768626809120178,
      "learning_rate": 4.6657277607970775e-05,
      "loss": 4.5605,
      "step": 39010
    },
    {
      "epoch": 0.7989474330882804,
      "grad_norm": 0.28735220432281494,
      "learning_rate": 4.661242373516315e-05,
      "loss": 4.5613,
      "step": 39015
    },
    {
      "epoch": 0.799049822865685,
      "grad_norm": 0.29037731885910034,
      "learning_rate": 4.6567587466332724e-05,
      "loss": 4.5707,
      "step": 39020
    },
    {
      "epoch": 0.7991522126430897,
      "grad_norm": 0.27839553356170654,
      "learning_rate": 4.652276880911387e-05,
      "loss": 4.6282,
      "step": 39025
    },
    {
      "epoch": 0.7992546024204943,
      "grad_norm": 0.28246358036994934,
      "learning_rate": 4.647796777113788e-05,
      "loss": 4.6188,
      "step": 39030
    },
    {
      "epoch": 0.7993569921978989,
      "grad_norm": 0.27705833315849304,
      "learning_rate": 4.64331843600332e-05,
      "loss": 4.6162,
      "step": 39035
    },
    {
      "epoch": 0.7994593819753036,
      "grad_norm": 0.2875292897224426,
      "learning_rate": 4.638841858342514e-05,
      "loss": 4.5495,
      "step": 39040
    },
    {
      "epoch": 0.7995617717527083,
      "grad_norm": 0.2829298973083496,
      "learning_rate": 4.634367044893598e-05,
      "loss": 4.5837,
      "step": 39045
    },
    {
      "epoch": 0.7996641615301129,
      "grad_norm": 0.32871323823928833,
      "learning_rate": 4.6298939964185164e-05,
      "loss": 4.5854,
      "step": 39050
    },
    {
      "epoch": 0.7997665513075175,
      "grad_norm": 0.2664298415184021,
      "learning_rate": 4.6254227136788986e-05,
      "loss": 4.6725,
      "step": 39055
    },
    {
      "epoch": 0.7998689410849221,
      "grad_norm": 0.2779654264450073,
      "learning_rate": 4.6209531974360705e-05,
      "loss": 4.6213,
      "step": 39060
    },
    {
      "epoch": 0.7999713308623267,
      "grad_norm": 0.2852494716644287,
      "learning_rate": 4.6164854484510764e-05,
      "loss": 4.6053,
      "step": 39065
    },
    {
      "epoch": 0.8000737206397314,
      "grad_norm": 0.3127801716327667,
      "learning_rate": 4.6120194674846426e-05,
      "loss": 4.5472,
      "step": 39070
    },
    {
      "epoch": 0.800176110417136,
      "grad_norm": 0.2792738676071167,
      "learning_rate": 4.607555255297196e-05,
      "loss": 4.5388,
      "step": 39075
    },
    {
      "epoch": 0.8002785001945406,
      "grad_norm": 0.2929307520389557,
      "learning_rate": 4.603092812648862e-05,
      "loss": 4.6292,
      "step": 39080
    },
    {
      "epoch": 0.8003808899719452,
      "grad_norm": 0.29168641567230225,
      "learning_rate": 4.5986321402994805e-05,
      "loss": 4.5284,
      "step": 39085
    },
    {
      "epoch": 0.8004832797493499,
      "grad_norm": 0.270711749792099,
      "learning_rate": 4.594173239008572e-05,
      "loss": 4.5815,
      "step": 39090
    },
    {
      "epoch": 0.8005856695267545,
      "grad_norm": 0.28868192434310913,
      "learning_rate": 4.5897161095353526e-05,
      "loss": 4.5659,
      "step": 39095
    },
    {
      "epoch": 0.8006880593041591,
      "grad_norm": 0.27453774213790894,
      "learning_rate": 4.58526075263876e-05,
      "loss": 4.5569,
      "step": 39100
    },
    {
      "epoch": 0.8007904490815637,
      "grad_norm": 0.29395386576652527,
      "learning_rate": 4.5808071690774064e-05,
      "loss": 4.5804,
      "step": 39105
    },
    {
      "epoch": 0.8008928388589683,
      "grad_norm": 0.27555522322654724,
      "learning_rate": 4.5763553596096104e-05,
      "loss": 4.6182,
      "step": 39110
    },
    {
      "epoch": 0.800995228636373,
      "grad_norm": 0.2816535532474518,
      "learning_rate": 4.5719053249934e-05,
      "loss": 4.5691,
      "step": 39115
    },
    {
      "epoch": 0.8010976184137776,
      "grad_norm": 0.28808650374412537,
      "learning_rate": 4.5674570659864717e-05,
      "loss": 4.6323,
      "step": 39120
    },
    {
      "epoch": 0.8012000081911822,
      "grad_norm": 0.28865253925323486,
      "learning_rate": 4.5630105833462526e-05,
      "loss": 4.5761,
      "step": 39125
    },
    {
      "epoch": 0.8013023979685868,
      "grad_norm": 0.30832570791244507,
      "learning_rate": 4.5585658778298485e-05,
      "loss": 4.6301,
      "step": 39130
    },
    {
      "epoch": 0.8014047877459914,
      "grad_norm": 0.290213018655777,
      "learning_rate": 4.554122950194062e-05,
      "loss": 4.6544,
      "step": 39135
    },
    {
      "epoch": 0.8015071775233961,
      "grad_norm": 0.2868783473968506,
      "learning_rate": 4.549681801195408e-05,
      "loss": 4.6037,
      "step": 39140
    },
    {
      "epoch": 0.8016095673008007,
      "grad_norm": 0.2926163375377655,
      "learning_rate": 4.545242431590081e-05,
      "loss": 4.6423,
      "step": 39145
    },
    {
      "epoch": 0.8017119570782053,
      "grad_norm": 0.2848108112812042,
      "learning_rate": 4.5408048421339796e-05,
      "loss": 4.5587,
      "step": 39150
    },
    {
      "epoch": 0.8018143468556099,
      "grad_norm": 0.35969194769859314,
      "learning_rate": 4.536369033582705e-05,
      "loss": 4.6215,
      "step": 39155
    },
    {
      "epoch": 0.8019167366330145,
      "grad_norm": 0.36385461688041687,
      "learning_rate": 4.531935006691548e-05,
      "loss": 4.5802,
      "step": 39160
    },
    {
      "epoch": 0.8020191264104192,
      "grad_norm": 0.2822389304637909,
      "learning_rate": 4.5275027622154966e-05,
      "loss": 4.6143,
      "step": 39165
    },
    {
      "epoch": 0.8021215161878238,
      "grad_norm": 0.2893722951412201,
      "learning_rate": 4.5230723009092326e-05,
      "loss": 4.5839,
      "step": 39170
    },
    {
      "epoch": 0.8022239059652284,
      "grad_norm": 0.28926926851272583,
      "learning_rate": 4.518643623527145e-05,
      "loss": 4.5928,
      "step": 39175
    },
    {
      "epoch": 0.802326295742633,
      "grad_norm": 0.27141159772872925,
      "learning_rate": 4.51421673082331e-05,
      "loss": 4.5484,
      "step": 39180
    },
    {
      "epoch": 0.8024286855200377,
      "grad_norm": 0.27838727831840515,
      "learning_rate": 4.509791623551498e-05,
      "loss": 4.6451,
      "step": 39185
    },
    {
      "epoch": 0.8025310752974423,
      "grad_norm": 0.2750069200992584,
      "learning_rate": 4.5053683024651856e-05,
      "loss": 4.5899,
      "step": 39190
    },
    {
      "epoch": 0.8026334650748469,
      "grad_norm": 0.30682578682899475,
      "learning_rate": 4.500946768317538e-05,
      "loss": 4.5936,
      "step": 39195
    },
    {
      "epoch": 0.8027358548522515,
      "grad_norm": 0.31645169854164124,
      "learning_rate": 4.496527021861411e-05,
      "loss": 4.5895,
      "step": 39200
    },
    {
      "epoch": 0.8028382446296561,
      "grad_norm": 0.2702699303627014,
      "learning_rate": 4.49210906384937e-05,
      "loss": 4.6033,
      "step": 39205
    },
    {
      "epoch": 0.8029406344070608,
      "grad_norm": 0.2715344727039337,
      "learning_rate": 4.487692895033666e-05,
      "loss": 4.5896,
      "step": 39210
    },
    {
      "epoch": 0.8030430241844654,
      "grad_norm": 0.27582502365112305,
      "learning_rate": 4.483278516166249e-05,
      "loss": 4.6032,
      "step": 39215
    },
    {
      "epoch": 0.80314541396187,
      "grad_norm": 0.27882251143455505,
      "learning_rate": 4.4788659279987535e-05,
      "loss": 4.6138,
      "step": 39220
    },
    {
      "epoch": 0.8032478037392746,
      "grad_norm": 0.2906031906604767,
      "learning_rate": 4.47445513128253e-05,
      "loss": 4.5029,
      "step": 39225
    },
    {
      "epoch": 0.8033501935166792,
      "grad_norm": 0.2661421000957489,
      "learning_rate": 4.4700461267686084e-05,
      "loss": 4.5761,
      "step": 39230
    },
    {
      "epoch": 0.8034525832940839,
      "grad_norm": 0.2729162871837616,
      "learning_rate": 4.4656389152077107e-05,
      "loss": 4.5717,
      "step": 39235
    },
    {
      "epoch": 0.8035549730714885,
      "grad_norm": 0.28579139709472656,
      "learning_rate": 4.461233497350278e-05,
      "loss": 4.6001,
      "step": 39240
    },
    {
      "epoch": 0.8036573628488932,
      "grad_norm": 0.2768882215023041,
      "learning_rate": 4.456829873946405e-05,
      "loss": 4.5897,
      "step": 39245
    },
    {
      "epoch": 0.8037597526262978,
      "grad_norm": 0.29882630705833435,
      "learning_rate": 4.452428045745915e-05,
      "loss": 4.5495,
      "step": 39250
    },
    {
      "epoch": 0.8038621424037025,
      "grad_norm": 0.284458726644516,
      "learning_rate": 4.448028013498325e-05,
      "loss": 4.5268,
      "step": 39255
    },
    {
      "epoch": 0.8039645321811071,
      "grad_norm": 0.27446749806404114,
      "learning_rate": 4.4436297779528174e-05,
      "loss": 4.5373,
      "step": 39260
    },
    {
      "epoch": 0.8040669219585117,
      "grad_norm": 0.27560141682624817,
      "learning_rate": 4.439233339858301e-05,
      "loss": 4.6377,
      "step": 39265
    },
    {
      "epoch": 0.8041693117359163,
      "grad_norm": 0.33306679129600525,
      "learning_rate": 4.43483869996336e-05,
      "loss": 4.5824,
      "step": 39270
    },
    {
      "epoch": 0.804271701513321,
      "grad_norm": 0.2933988869190216,
      "learning_rate": 4.43044585901627e-05,
      "loss": 4.5762,
      "step": 39275
    },
    {
      "epoch": 0.8043740912907256,
      "grad_norm": 0.28053733706474304,
      "learning_rate": 4.4260548177650204e-05,
      "loss": 4.5857,
      "step": 39280
    },
    {
      "epoch": 0.8044764810681302,
      "grad_norm": 0.32424768805503845,
      "learning_rate": 4.4216655769572757e-05,
      "loss": 4.554,
      "step": 39285
    },
    {
      "epoch": 0.8045788708455348,
      "grad_norm": 0.2807777523994446,
      "learning_rate": 4.417278137340392e-05,
      "loss": 4.6263,
      "step": 39290
    },
    {
      "epoch": 0.8046812606229394,
      "grad_norm": 0.29028016328811646,
      "learning_rate": 4.412892499661439e-05,
      "loss": 4.6218,
      "step": 39295
    },
    {
      "epoch": 0.804783650400344,
      "grad_norm": 0.2970632314682007,
      "learning_rate": 4.408508664667159e-05,
      "loss": 4.5771,
      "step": 39300
    },
    {
      "epoch": 0.8048860401777487,
      "grad_norm": 0.2803395390510559,
      "learning_rate": 4.4041266331039986e-05,
      "loss": 4.6243,
      "step": 39305
    },
    {
      "epoch": 0.8049884299551533,
      "grad_norm": 0.2990550696849823,
      "learning_rate": 4.3997464057180845e-05,
      "loss": 4.5418,
      "step": 39310
    },
    {
      "epoch": 0.8050908197325579,
      "grad_norm": 0.2797633409500122,
      "learning_rate": 4.395367983255259e-05,
      "loss": 4.5788,
      "step": 39315
    },
    {
      "epoch": 0.8051932095099625,
      "grad_norm": 0.26845577359199524,
      "learning_rate": 4.3909913664610355e-05,
      "loss": 4.562,
      "step": 39320
    },
    {
      "epoch": 0.8052955992873672,
      "grad_norm": 0.27030444145202637,
      "learning_rate": 4.386616556080625e-05,
      "loss": 4.5199,
      "step": 39325
    },
    {
      "epoch": 0.8053979890647718,
      "grad_norm": 0.2664375901222229,
      "learning_rate": 4.382243552858944e-05,
      "loss": 4.495,
      "step": 39330
    },
    {
      "epoch": 0.8055003788421764,
      "grad_norm": 0.31012603640556335,
      "learning_rate": 4.377872357540583e-05,
      "loss": 4.6348,
      "step": 39335
    },
    {
      "epoch": 0.805602768619581,
      "grad_norm": 0.2901231050491333,
      "learning_rate": 4.3735029708698325e-05,
      "loss": 4.6027,
      "step": 39340
    },
    {
      "epoch": 0.8057051583969856,
      "grad_norm": 0.2725827991962433,
      "learning_rate": 4.369135393590682e-05,
      "loss": 4.5934,
      "step": 39345
    },
    {
      "epoch": 0.8058075481743903,
      "grad_norm": 0.2755502164363861,
      "learning_rate": 4.364769626446803e-05,
      "loss": 4.6253,
      "step": 39350
    },
    {
      "epoch": 0.8059099379517949,
      "grad_norm": 0.3057175278663635,
      "learning_rate": 4.360405670181559e-05,
      "loss": 4.5737,
      "step": 39355
    },
    {
      "epoch": 0.8060123277291995,
      "grad_norm": 0.2964734137058258,
      "learning_rate": 4.356043525538008e-05,
      "loss": 4.5851,
      "step": 39360
    },
    {
      "epoch": 0.8061147175066041,
      "grad_norm": 0.2873198091983795,
      "learning_rate": 4.3516831932589056e-05,
      "loss": 4.6087,
      "step": 39365
    },
    {
      "epoch": 0.8062171072840087,
      "grad_norm": 0.2713853716850281,
      "learning_rate": 4.347324674086691e-05,
      "loss": 4.5689,
      "step": 39370
    },
    {
      "epoch": 0.8063194970614134,
      "grad_norm": 0.27605798840522766,
      "learning_rate": 4.34296796876349e-05,
      "loss": 4.605,
      "step": 39375
    },
    {
      "epoch": 0.806421886838818,
      "grad_norm": 0.27485141158103943,
      "learning_rate": 4.33861307803114e-05,
      "loss": 4.5899,
      "step": 39380
    },
    {
      "epoch": 0.8065242766162226,
      "grad_norm": 0.28419703245162964,
      "learning_rate": 4.3342600026311385e-05,
      "loss": 4.5744,
      "step": 39385
    },
    {
      "epoch": 0.8066266663936272,
      "grad_norm": 0.27581843733787537,
      "learning_rate": 4.329908743304699e-05,
      "loss": 4.5795,
      "step": 39390
    },
    {
      "epoch": 0.8067290561710319,
      "grad_norm": 0.2819613814353943,
      "learning_rate": 4.3255593007927273e-05,
      "loss": 4.6254,
      "step": 39395
    },
    {
      "epoch": 0.8068314459484365,
      "grad_norm": 0.3055405020713806,
      "learning_rate": 4.321211675835793e-05,
      "loss": 4.5892,
      "step": 39400
    },
    {
      "epoch": 0.8069338357258411,
      "grad_norm": 0.32575732469558716,
      "learning_rate": 4.316865869174184e-05,
      "loss": 4.5819,
      "step": 39405
    },
    {
      "epoch": 0.8070362255032457,
      "grad_norm": 0.3296785354614258,
      "learning_rate": 4.312521881547868e-05,
      "loss": 4.5235,
      "step": 39410
    },
    {
      "epoch": 0.8071386152806503,
      "grad_norm": 0.30570748448371887,
      "learning_rate": 4.308179713696497e-05,
      "loss": 4.5757,
      "step": 39415
    },
    {
      "epoch": 0.807241005058055,
      "grad_norm": 0.2826966345310211,
      "learning_rate": 4.303839366359426e-05,
      "loss": 4.5665,
      "step": 39420
    },
    {
      "epoch": 0.8073433948354596,
      "grad_norm": 0.2718033790588379,
      "learning_rate": 4.2995008402756915e-05,
      "loss": 4.578,
      "step": 39425
    },
    {
      "epoch": 0.8074457846128642,
      "grad_norm": 0.2811250686645508,
      "learning_rate": 4.295164136184022e-05,
      "loss": 4.6362,
      "step": 39430
    },
    {
      "epoch": 0.8075481743902688,
      "grad_norm": 0.3076514005661011,
      "learning_rate": 4.290829254822828e-05,
      "loss": 4.5803,
      "step": 39435
    },
    {
      "epoch": 0.8076505641676734,
      "grad_norm": 0.2890389859676361,
      "learning_rate": 4.286496196930228e-05,
      "loss": 4.5506,
      "step": 39440
    },
    {
      "epoch": 0.8077529539450782,
      "grad_norm": 0.28918203711509705,
      "learning_rate": 4.282164963244014e-05,
      "loss": 4.5758,
      "step": 39445
    },
    {
      "epoch": 0.8078553437224828,
      "grad_norm": 0.31487032771110535,
      "learning_rate": 4.2778355545016696e-05,
      "loss": 4.6039,
      "step": 39450
    },
    {
      "epoch": 0.8079577334998874,
      "grad_norm": 0.27409568428993225,
      "learning_rate": 4.273507971440376e-05,
      "loss": 4.5724,
      "step": 39455
    },
    {
      "epoch": 0.808060123277292,
      "grad_norm": 0.297097772359848,
      "learning_rate": 4.2691822147969985e-05,
      "loss": 4.5531,
      "step": 39460
    },
    {
      "epoch": 0.8081625130546967,
      "grad_norm": 0.2799051105976105,
      "learning_rate": 4.264858285308083e-05,
      "loss": 4.5839,
      "step": 39465
    },
    {
      "epoch": 0.8082649028321013,
      "grad_norm": 0.2736682891845703,
      "learning_rate": 4.260536183709882e-05,
      "loss": 4.5616,
      "step": 39470
    },
    {
      "epoch": 0.8083672926095059,
      "grad_norm": 0.282229483127594,
      "learning_rate": 4.256215910738322e-05,
      "loss": 4.5735,
      "step": 39475
    },
    {
      "epoch": 0.8084696823869105,
      "grad_norm": 0.2849828600883484,
      "learning_rate": 4.251897467129019e-05,
      "loss": 4.5599,
      "step": 39480
    },
    {
      "epoch": 0.8085720721643151,
      "grad_norm": 0.36352962255477905,
      "learning_rate": 4.247580853617291e-05,
      "loss": 4.5819,
      "step": 39485
    },
    {
      "epoch": 0.8086744619417198,
      "grad_norm": 0.29873931407928467,
      "learning_rate": 4.243266070938132e-05,
      "loss": 4.5764,
      "step": 39490
    },
    {
      "epoch": 0.8087768517191244,
      "grad_norm": 0.28227362036705017,
      "learning_rate": 4.238953119826225e-05,
      "loss": 4.5845,
      "step": 39495
    },
    {
      "epoch": 0.808879241496529,
      "grad_norm": 0.3054857552051544,
      "learning_rate": 4.234642001015939e-05,
      "loss": 4.6181,
      "step": 39500
    },
    {
      "epoch": 0.8089816312739336,
      "grad_norm": 0.2842195928096771,
      "learning_rate": 4.230332715241346e-05,
      "loss": 4.6409,
      "step": 39505
    },
    {
      "epoch": 0.8090840210513383,
      "grad_norm": 0.2869653105735779,
      "learning_rate": 4.226025263236192e-05,
      "loss": 4.5387,
      "step": 39510
    },
    {
      "epoch": 0.8091864108287429,
      "grad_norm": 0.2856428921222687,
      "learning_rate": 4.221719645733905e-05,
      "loss": 4.5624,
      "step": 39515
    },
    {
      "epoch": 0.8092888006061475,
      "grad_norm": 0.2736533284187317,
      "learning_rate": 4.217415863467627e-05,
      "loss": 4.6077,
      "step": 39520
    },
    {
      "epoch": 0.8093911903835521,
      "grad_norm": 0.284796804189682,
      "learning_rate": 4.213113917170152e-05,
      "loss": 4.6018,
      "step": 39525
    },
    {
      "epoch": 0.8094935801609567,
      "grad_norm": 0.27927646040916443,
      "learning_rate": 4.208813807573989e-05,
      "loss": 4.569,
      "step": 39530
    },
    {
      "epoch": 0.8095959699383614,
      "grad_norm": 0.2721383571624756,
      "learning_rate": 4.20451553541133e-05,
      "loss": 4.5113,
      "step": 39535
    },
    {
      "epoch": 0.809698359715766,
      "grad_norm": 0.2773188054561615,
      "learning_rate": 4.2002191014140366e-05,
      "loss": 4.582,
      "step": 39540
    },
    {
      "epoch": 0.8098007494931706,
      "grad_norm": 0.33200955390930176,
      "learning_rate": 4.19592450631368e-05,
      "loss": 4.5949,
      "step": 39545
    },
    {
      "epoch": 0.8099031392705752,
      "grad_norm": 0.28897106647491455,
      "learning_rate": 4.1916317508415044e-05,
      "loss": 4.4903,
      "step": 39550
    },
    {
      "epoch": 0.8100055290479798,
      "grad_norm": 0.2980147898197174,
      "learning_rate": 4.1873408357284406e-05,
      "loss": 4.5857,
      "step": 39555
    },
    {
      "epoch": 0.8101079188253845,
      "grad_norm": 0.28293874859809875,
      "learning_rate": 4.1830517617051175e-05,
      "loss": 4.5404,
      "step": 39560
    },
    {
      "epoch": 0.8102103086027891,
      "grad_norm": 0.28592419624328613,
      "learning_rate": 4.178764529501839e-05,
      "loss": 4.615,
      "step": 39565
    },
    {
      "epoch": 0.8103126983801937,
      "grad_norm": 0.2894458770751953,
      "learning_rate": 4.1744791398486016e-05,
      "loss": 4.5859,
      "step": 39570
    },
    {
      "epoch": 0.8104150881575983,
      "grad_norm": 0.2806173861026764,
      "learning_rate": 4.170195593475078e-05,
      "loss": 4.6081,
      "step": 39575
    },
    {
      "epoch": 0.810517477935003,
      "grad_norm": 0.35036706924438477,
      "learning_rate": 4.165913891110646e-05,
      "loss": 4.6226,
      "step": 39580
    },
    {
      "epoch": 0.8106198677124076,
      "grad_norm": 0.28919950127601624,
      "learning_rate": 4.161634033484352e-05,
      "loss": 4.6167,
      "step": 39585
    },
    {
      "epoch": 0.8107222574898122,
      "grad_norm": 0.2950119376182556,
      "learning_rate": 4.157356021324932e-05,
      "loss": 4.6162,
      "step": 39590
    },
    {
      "epoch": 0.8108246472672168,
      "grad_norm": 0.2698840796947479,
      "learning_rate": 4.1530798553608205e-05,
      "loss": 4.6037,
      "step": 39595
    },
    {
      "epoch": 0.8109270370446214,
      "grad_norm": 0.3377280831336975,
      "learning_rate": 4.14880553632012e-05,
      "loss": 4.5657,
      "step": 39600
    },
    {
      "epoch": 0.8110294268220261,
      "grad_norm": 0.2888646125793457,
      "learning_rate": 4.1445330649306214e-05,
      "loss": 4.5684,
      "step": 39605
    },
    {
      "epoch": 0.8111318165994307,
      "grad_norm": 0.2981690764427185,
      "learning_rate": 4.140262441919818e-05,
      "loss": 4.5822,
      "step": 39610
    },
    {
      "epoch": 0.8112342063768353,
      "grad_norm": 0.2968234717845917,
      "learning_rate": 4.1359936680148675e-05,
      "loss": 4.586,
      "step": 39615
    },
    {
      "epoch": 0.8113365961542399,
      "grad_norm": 0.3319108486175537,
      "learning_rate": 4.131726743942623e-05,
      "loss": 4.559,
      "step": 39620
    },
    {
      "epoch": 0.8114389859316445,
      "grad_norm": 0.28827497363090515,
      "learning_rate": 4.127461670429617e-05,
      "loss": 4.6009,
      "step": 39625
    },
    {
      "epoch": 0.8115413757090492,
      "grad_norm": 0.26755833625793457,
      "learning_rate": 4.12319844820208e-05,
      "loss": 4.5762,
      "step": 39630
    },
    {
      "epoch": 0.8116437654864538,
      "grad_norm": 0.27354004979133606,
      "learning_rate": 4.1189370779859115e-05,
      "loss": 4.5482,
      "step": 39635
    },
    {
      "epoch": 0.8117461552638584,
      "grad_norm": 0.2695753872394562,
      "learning_rate": 4.1146775605067004e-05,
      "loss": 4.56,
      "step": 39640
    },
    {
      "epoch": 0.811848545041263,
      "grad_norm": 0.28312042355537415,
      "learning_rate": 4.1104198964897283e-05,
      "loss": 4.5647,
      "step": 39645
    },
    {
      "epoch": 0.8119509348186678,
      "grad_norm": 0.27489081025123596,
      "learning_rate": 4.106164086659953e-05,
      "loss": 4.6066,
      "step": 39650
    },
    {
      "epoch": 0.8120533245960724,
      "grad_norm": 0.2789163291454315,
      "learning_rate": 4.1019101317420106e-05,
      "loss": 4.616,
      "step": 39655
    },
    {
      "epoch": 0.812155714373477,
      "grad_norm": 0.2964979410171509,
      "learning_rate": 4.097658032460246e-05,
      "loss": 4.63,
      "step": 39660
    },
    {
      "epoch": 0.8122581041508816,
      "grad_norm": 0.29779788851737976,
      "learning_rate": 4.0934077895386516e-05,
      "loss": 4.594,
      "step": 39665
    },
    {
      "epoch": 0.8123604939282862,
      "grad_norm": 0.271930068731308,
      "learning_rate": 4.0891594037009326e-05,
      "loss": 4.5532,
      "step": 39670
    },
    {
      "epoch": 0.8124628837056909,
      "grad_norm": 0.272431343793869,
      "learning_rate": 4.084912875670478e-05,
      "loss": 4.6512,
      "step": 39675
    },
    {
      "epoch": 0.8125652734830955,
      "grad_norm": 0.2750728726387024,
      "learning_rate": 4.0806682061703336e-05,
      "loss": 4.5432,
      "step": 39680
    },
    {
      "epoch": 0.8126676632605001,
      "grad_norm": 0.2824062705039978,
      "learning_rate": 4.076425395923259e-05,
      "loss": 4.584,
      "step": 39685
    },
    {
      "epoch": 0.8127700530379047,
      "grad_norm": 0.28390562534332275,
      "learning_rate": 4.072184445651681e-05,
      "loss": 4.5837,
      "step": 39690
    },
    {
      "epoch": 0.8128724428153093,
      "grad_norm": 0.27793005108833313,
      "learning_rate": 4.067945356077709e-05,
      "loss": 4.5361,
      "step": 39695
    },
    {
      "epoch": 0.812974832592714,
      "grad_norm": 0.30021142959594727,
      "learning_rate": 4.063708127923149e-05,
      "loss": 4.6323,
      "step": 39700
    },
    {
      "epoch": 0.8130772223701186,
      "grad_norm": 1.442164659500122,
      "learning_rate": 4.0594727619094755e-05,
      "loss": 4.5769,
      "step": 39705
    },
    {
      "epoch": 0.8131796121475232,
      "grad_norm": 0.29034852981567383,
      "learning_rate": 4.055239258757852e-05,
      "loss": 4.5869,
      "step": 39710
    },
    {
      "epoch": 0.8132820019249278,
      "grad_norm": 0.2804703116416931,
      "learning_rate": 4.051007619189119e-05,
      "loss": 4.6193,
      "step": 39715
    },
    {
      "epoch": 0.8133843917023325,
      "grad_norm": 0.28716444969177246,
      "learning_rate": 4.046777843923817e-05,
      "loss": 4.5762,
      "step": 39720
    },
    {
      "epoch": 0.8134867814797371,
      "grad_norm": 0.2703050673007965,
      "learning_rate": 4.0425499336821484e-05,
      "loss": 4.5967,
      "step": 39725
    },
    {
      "epoch": 0.8135891712571417,
      "grad_norm": 0.2838965356349945,
      "learning_rate": 4.038323889184004e-05,
      "loss": 4.5966,
      "step": 39730
    },
    {
      "epoch": 0.8136915610345463,
      "grad_norm": 0.28783389925956726,
      "learning_rate": 4.034099711148968e-05,
      "loss": 4.5273,
      "step": 39735
    },
    {
      "epoch": 0.8137939508119509,
      "grad_norm": 0.2864513695240021,
      "learning_rate": 4.0298774002962945e-05,
      "loss": 4.5948,
      "step": 39740
    },
    {
      "epoch": 0.8138963405893556,
      "grad_norm": 0.3301258087158203,
      "learning_rate": 4.0256569573449185e-05,
      "loss": 4.6003,
      "step": 39745
    },
    {
      "epoch": 0.8139987303667602,
      "grad_norm": 0.2918735444545746,
      "learning_rate": 4.0214383830134734e-05,
      "loss": 4.5969,
      "step": 39750
    },
    {
      "epoch": 0.8141011201441648,
      "grad_norm": 0.2782702147960663,
      "learning_rate": 4.0172216780202534e-05,
      "loss": 4.6031,
      "step": 39755
    },
    {
      "epoch": 0.8142035099215694,
      "grad_norm": 0.29124337434768677,
      "learning_rate": 4.013006843083249e-05,
      "loss": 4.5927,
      "step": 39760
    },
    {
      "epoch": 0.814305899698974,
      "grad_norm": 0.2667020559310913,
      "learning_rate": 4.008793878920119e-05,
      "loss": 4.5262,
      "step": 39765
    },
    {
      "epoch": 0.8144082894763787,
      "grad_norm": 0.28633174300193787,
      "learning_rate": 4.0045827862482234e-05,
      "loss": 4.4901,
      "step": 39770
    },
    {
      "epoch": 0.8145106792537833,
      "grad_norm": 0.27973079681396484,
      "learning_rate": 4.000373565784586e-05,
      "loss": 4.566,
      "step": 39775
    },
    {
      "epoch": 0.8146130690311879,
      "grad_norm": 0.27879223227500916,
      "learning_rate": 3.996166218245915e-05,
      "loss": 4.5679,
      "step": 39780
    },
    {
      "epoch": 0.8147154588085925,
      "grad_norm": 0.300350159406662,
      "learning_rate": 3.9919607443486084e-05,
      "loss": 4.6046,
      "step": 39785
    },
    {
      "epoch": 0.8148178485859972,
      "grad_norm": 0.27650022506713867,
      "learning_rate": 3.9877571448087395e-05,
      "loss": 4.5557,
      "step": 39790
    },
    {
      "epoch": 0.8149202383634018,
      "grad_norm": 0.2842326760292053,
      "learning_rate": 3.983555420342054e-05,
      "loss": 4.5814,
      "step": 39795
    },
    {
      "epoch": 0.8150226281408064,
      "grad_norm": 0.2776239216327667,
      "learning_rate": 3.979355571664002e-05,
      "loss": 4.6139,
      "step": 39800
    },
    {
      "epoch": 0.815125017918211,
      "grad_norm": 0.28626564145088196,
      "learning_rate": 3.975157599489682e-05,
      "loss": 4.5889,
      "step": 39805
    },
    {
      "epoch": 0.8152274076956156,
      "grad_norm": 0.2799535095691681,
      "learning_rate": 3.970961504533899e-05,
      "loss": 4.5765,
      "step": 39810
    },
    {
      "epoch": 0.8153297974730203,
      "grad_norm": 0.30687370896339417,
      "learning_rate": 3.9667672875111286e-05,
      "loss": 4.5923,
      "step": 39815
    },
    {
      "epoch": 0.8154321872504249,
      "grad_norm": 0.29484468698501587,
      "learning_rate": 3.962574949135521e-05,
      "loss": 4.5662,
      "step": 39820
    },
    {
      "epoch": 0.8155345770278295,
      "grad_norm": 0.33060598373413086,
      "learning_rate": 3.9583844901209214e-05,
      "loss": 4.5596,
      "step": 39825
    },
    {
      "epoch": 0.8156369668052341,
      "grad_norm": 0.2842893600463867,
      "learning_rate": 3.9541959111808433e-05,
      "loss": 4.5925,
      "step": 39830
    },
    {
      "epoch": 0.8157393565826387,
      "grad_norm": 0.2787351906299591,
      "learning_rate": 3.95000921302848e-05,
      "loss": 4.5817,
      "step": 39835
    },
    {
      "epoch": 0.8158417463600434,
      "grad_norm": 0.3185434639453888,
      "learning_rate": 3.94582439637671e-05,
      "loss": 4.5792,
      "step": 39840
    },
    {
      "epoch": 0.815944136137448,
      "grad_norm": 0.27078425884246826,
      "learning_rate": 3.941641461938094e-05,
      "loss": 4.5839,
      "step": 39845
    },
    {
      "epoch": 0.8160465259148527,
      "grad_norm": 0.2745194733142853,
      "learning_rate": 3.937460410424859e-05,
      "loss": 4.5662,
      "step": 39850
    },
    {
      "epoch": 0.8161489156922573,
      "grad_norm": 0.2862422466278076,
      "learning_rate": 3.933281242548921e-05,
      "loss": 4.629,
      "step": 39855
    },
    {
      "epoch": 0.816251305469662,
      "grad_norm": 0.26588085293769836,
      "learning_rate": 3.929103959021879e-05,
      "loss": 4.5757,
      "step": 39860
    },
    {
      "epoch": 0.8163536952470666,
      "grad_norm": 0.29439637064933777,
      "learning_rate": 3.9249285605550044e-05,
      "loss": 4.5791,
      "step": 39865
    },
    {
      "epoch": 0.8164560850244712,
      "grad_norm": 0.2926248610019684,
      "learning_rate": 3.9207550478592416e-05,
      "loss": 4.6052,
      "step": 39870
    },
    {
      "epoch": 0.8165584748018758,
      "grad_norm": 0.27602916955947876,
      "learning_rate": 3.916583421645235e-05,
      "loss": 4.5794,
      "step": 39875
    },
    {
      "epoch": 0.8166608645792804,
      "grad_norm": 0.28453734517097473,
      "learning_rate": 3.912413682623287e-05,
      "loss": 4.5508,
      "step": 39880
    },
    {
      "epoch": 0.8167632543566851,
      "grad_norm": 0.2766074538230896,
      "learning_rate": 3.908245831503383e-05,
      "loss": 4.5613,
      "step": 39885
    },
    {
      "epoch": 0.8168656441340897,
      "grad_norm": 0.27316877245903015,
      "learning_rate": 3.904079868995197e-05,
      "loss": 4.5547,
      "step": 39890
    },
    {
      "epoch": 0.8169680339114943,
      "grad_norm": 0.29508209228515625,
      "learning_rate": 3.899915795808072e-05,
      "loss": 4.6335,
      "step": 39895
    },
    {
      "epoch": 0.8170704236888989,
      "grad_norm": 0.28743311762809753,
      "learning_rate": 3.895753612651032e-05,
      "loss": 4.5782,
      "step": 39900
    },
    {
      "epoch": 0.8171728134663035,
      "grad_norm": 0.3072720468044281,
      "learning_rate": 3.891593320232772e-05,
      "loss": 4.6246,
      "step": 39905
    },
    {
      "epoch": 0.8172752032437082,
      "grad_norm": 0.2878606915473938,
      "learning_rate": 3.8874349192616826e-05,
      "loss": 4.6083,
      "step": 39910
    },
    {
      "epoch": 0.8173775930211128,
      "grad_norm": 0.27286794781684875,
      "learning_rate": 3.883278410445819e-05,
      "loss": 4.6387,
      "step": 39915
    },
    {
      "epoch": 0.8174799827985174,
      "grad_norm": 0.27541738748550415,
      "learning_rate": 3.879123794492909e-05,
      "loss": 4.5475,
      "step": 39920
    },
    {
      "epoch": 0.817582372575922,
      "grad_norm": 3.955681800842285,
      "learning_rate": 3.874971072110377e-05,
      "loss": 4.5906,
      "step": 39925
    },
    {
      "epoch": 0.8176847623533267,
      "grad_norm": 0.2810393273830414,
      "learning_rate": 3.870820244005311e-05,
      "loss": 4.5912,
      "step": 39930
    },
    {
      "epoch": 0.8177871521307313,
      "grad_norm": 0.318101167678833,
      "learning_rate": 3.866671310884472e-05,
      "loss": 4.5904,
      "step": 39935
    },
    {
      "epoch": 0.8178895419081359,
      "grad_norm": 0.30847010016441345,
      "learning_rate": 3.86252427345432e-05,
      "loss": 4.6128,
      "step": 39940
    },
    {
      "epoch": 0.8179919316855405,
      "grad_norm": 0.2958615720272064,
      "learning_rate": 3.858379132420962e-05,
      "loss": 4.5609,
      "step": 39945
    },
    {
      "epoch": 0.8180943214629451,
      "grad_norm": 0.27650511264801025,
      "learning_rate": 3.8542358884902093e-05,
      "loss": 4.6257,
      "step": 39950
    },
    {
      "epoch": 0.8181967112403498,
      "grad_norm": 0.27521902322769165,
      "learning_rate": 3.850094542367537e-05,
      "loss": 4.6032,
      "step": 39955
    },
    {
      "epoch": 0.8182991010177544,
      "grad_norm": 0.29434481263160706,
      "learning_rate": 3.8459550947580925e-05,
      "loss": 4.577,
      "step": 39960
    },
    {
      "epoch": 0.818401490795159,
      "grad_norm": 0.3031090795993805,
      "learning_rate": 3.841817546366716e-05,
      "loss": 4.6189,
      "step": 39965
    },
    {
      "epoch": 0.8185038805725636,
      "grad_norm": 0.2746244966983795,
      "learning_rate": 3.837681897897912e-05,
      "loss": 4.6039,
      "step": 39970
    },
    {
      "epoch": 0.8186062703499682,
      "grad_norm": 0.28014394640922546,
      "learning_rate": 3.833548150055858e-05,
      "loss": 4.5713,
      "step": 39975
    },
    {
      "epoch": 0.8187086601273729,
      "grad_norm": 0.27805647253990173,
      "learning_rate": 3.829416303544423e-05,
      "loss": 4.5126,
      "step": 39980
    },
    {
      "epoch": 0.8188110499047775,
      "grad_norm": 0.3021852672100067,
      "learning_rate": 3.82528635906714e-05,
      "loss": 4.6194,
      "step": 39985
    },
    {
      "epoch": 0.8189134396821821,
      "grad_norm": 0.2743503451347351,
      "learning_rate": 3.8211583173272216e-05,
      "loss": 4.5727,
      "step": 39990
    },
    {
      "epoch": 0.8190158294595867,
      "grad_norm": 0.28668925166130066,
      "learning_rate": 3.817032179027551e-05,
      "loss": 4.598,
      "step": 39995
    },
    {
      "epoch": 0.8191182192369914,
      "grad_norm": 0.2746945023536682,
      "learning_rate": 3.812907944870703e-05,
      "loss": 4.6093,
      "step": 40000
    },
    {
      "epoch": 0.8191182192369914,
      "eval_loss": 4.598438262939453,
      "eval_runtime": 17.1327,
      "eval_samples_per_second": 33.97,
      "eval_steps_per_second": 4.261,
      "step": 40000
    },
    {
      "epoch": 0.819220609014396,
      "grad_norm": 0.3382457196712494,
      "learning_rate": 3.8087856155589135e-05,
      "loss": 4.6227,
      "step": 40005
    },
    {
      "epoch": 0.8193229987918006,
      "grad_norm": 0.2767484486103058,
      "learning_rate": 3.8046651917940923e-05,
      "loss": 4.5871,
      "step": 40010
    },
    {
      "epoch": 0.8194253885692052,
      "grad_norm": 0.30373379588127136,
      "learning_rate": 3.800546674277842e-05,
      "loss": 4.4867,
      "step": 40015
    },
    {
      "epoch": 0.8195277783466098,
      "grad_norm": 0.28333204984664917,
      "learning_rate": 3.796430063711423e-05,
      "loss": 4.5592,
      "step": 40020
    },
    {
      "epoch": 0.8196301681240145,
      "grad_norm": 0.3391062915325165,
      "learning_rate": 3.792315360795774e-05,
      "loss": 4.553,
      "step": 40025
    },
    {
      "epoch": 0.8197325579014191,
      "grad_norm": 0.8202936053276062,
      "learning_rate": 3.7882025662315223e-05,
      "loss": 4.6518,
      "step": 40030
    },
    {
      "epoch": 0.8198349476788237,
      "grad_norm": 0.2817802131175995,
      "learning_rate": 3.784091680718954e-05,
      "loss": 4.5637,
      "step": 40035
    },
    {
      "epoch": 0.8199373374562283,
      "grad_norm": 0.2807997465133667,
      "learning_rate": 3.779982704958038e-05,
      "loss": 4.5998,
      "step": 40040
    },
    {
      "epoch": 0.8200397272336329,
      "grad_norm": 0.28021693229675293,
      "learning_rate": 3.7758756396484116e-05,
      "loss": 4.5725,
      "step": 40045
    },
    {
      "epoch": 0.8201421170110376,
      "grad_norm": 0.296805202960968,
      "learning_rate": 3.771770485489401e-05,
      "loss": 4.6079,
      "step": 40050
    },
    {
      "epoch": 0.8202445067884423,
      "grad_norm": 0.2737254500389099,
      "learning_rate": 3.767667243179991e-05,
      "loss": 4.6077,
      "step": 40055
    },
    {
      "epoch": 0.8203468965658469,
      "grad_norm": 0.3445216417312622,
      "learning_rate": 3.763565913418848e-05,
      "loss": 4.5476,
      "step": 40060
    },
    {
      "epoch": 0.8204492863432515,
      "grad_norm": 0.2809492349624634,
      "learning_rate": 3.759466496904316e-05,
      "loss": 4.6055,
      "step": 40065
    },
    {
      "epoch": 0.8205516761206562,
      "grad_norm": 0.2805745601654053,
      "learning_rate": 3.7553689943344086e-05,
      "loss": 4.5754,
      "step": 40070
    },
    {
      "epoch": 0.8206540658980608,
      "grad_norm": 0.27043336629867554,
      "learning_rate": 3.751273406406807e-05,
      "loss": 4.6063,
      "step": 40075
    },
    {
      "epoch": 0.8207564556754654,
      "grad_norm": 0.28540995717048645,
      "learning_rate": 3.747179733818892e-05,
      "loss": 4.5984,
      "step": 40080
    },
    {
      "epoch": 0.82085884545287,
      "grad_norm": 0.2703198492527008,
      "learning_rate": 3.743087977267678e-05,
      "loss": 4.6234,
      "step": 40085
    },
    {
      "epoch": 0.8209612352302746,
      "grad_norm": 0.28417128324508667,
      "learning_rate": 3.73899813744989e-05,
      "loss": 4.5485,
      "step": 40090
    },
    {
      "epoch": 0.8210636250076793,
      "grad_norm": 0.2963154911994934,
      "learning_rate": 3.734910215061907e-05,
      "loss": 4.5689,
      "step": 40095
    },
    {
      "epoch": 0.8211660147850839,
      "grad_norm": 0.2891373634338379,
      "learning_rate": 3.730824210799784e-05,
      "loss": 4.5853,
      "step": 40100
    },
    {
      "epoch": 0.8212684045624885,
      "grad_norm": 0.28095632791519165,
      "learning_rate": 3.726740125359257e-05,
      "loss": 4.6735,
      "step": 40105
    },
    {
      "epoch": 0.8213707943398931,
      "grad_norm": 0.2856515645980835,
      "learning_rate": 3.722657959435731e-05,
      "loss": 4.6088,
      "step": 40110
    },
    {
      "epoch": 0.8214731841172978,
      "grad_norm": 0.28138381242752075,
      "learning_rate": 3.718577713724274e-05,
      "loss": 4.6019,
      "step": 40115
    },
    {
      "epoch": 0.8215755738947024,
      "grad_norm": 0.2722416818141937,
      "learning_rate": 3.714499388919647e-05,
      "loss": 4.5733,
      "step": 40120
    },
    {
      "epoch": 0.821677963672107,
      "grad_norm": 0.2986077070236206,
      "learning_rate": 3.7104229857162705e-05,
      "loss": 4.6036,
      "step": 40125
    },
    {
      "epoch": 0.8217803534495116,
      "grad_norm": 0.2788534164428711,
      "learning_rate": 3.706348504808236e-05,
      "loss": 4.6201,
      "step": 40130
    },
    {
      "epoch": 0.8218827432269162,
      "grad_norm": 0.2822936177253723,
      "learning_rate": 3.7022759468893135e-05,
      "loss": 4.58,
      "step": 40135
    },
    {
      "epoch": 0.8219851330043209,
      "grad_norm": 0.28002282977104187,
      "learning_rate": 3.6982053126529495e-05,
      "loss": 4.6483,
      "step": 40140
    },
    {
      "epoch": 0.8220875227817255,
      "grad_norm": 0.28126972913742065,
      "learning_rate": 3.6941366027922535e-05,
      "loss": 4.5892,
      "step": 40145
    },
    {
      "epoch": 0.8221899125591301,
      "grad_norm": 0.2815869450569153,
      "learning_rate": 3.690069818000009e-05,
      "loss": 4.5944,
      "step": 40150
    },
    {
      "epoch": 0.8222923023365347,
      "grad_norm": 0.27708324790000916,
      "learning_rate": 3.6860049589686804e-05,
      "loss": 4.5795,
      "step": 40155
    },
    {
      "epoch": 0.8223946921139393,
      "grad_norm": 0.28458738327026367,
      "learning_rate": 3.6819420263903974e-05,
      "loss": 4.5768,
      "step": 40160
    },
    {
      "epoch": 0.822497081891344,
      "grad_norm": 0.27662721276283264,
      "learning_rate": 3.6778810209569546e-05,
      "loss": 4.6496,
      "step": 40165
    },
    {
      "epoch": 0.8225994716687486,
      "grad_norm": 0.2832746207714081,
      "learning_rate": 3.6738219433598373e-05,
      "loss": 4.6161,
      "step": 40170
    },
    {
      "epoch": 0.8227018614461532,
      "grad_norm": 0.28657278418540955,
      "learning_rate": 3.6697647942901865e-05,
      "loss": 4.5579,
      "step": 40175
    },
    {
      "epoch": 0.8228042512235578,
      "grad_norm": 0.30123379826545715,
      "learning_rate": 3.6657095744388185e-05,
      "loss": 4.5708,
      "step": 40180
    },
    {
      "epoch": 0.8229066410009624,
      "grad_norm": 0.2788209021091461,
      "learning_rate": 3.661656284496219e-05,
      "loss": 4.5561,
      "step": 40185
    },
    {
      "epoch": 0.8230090307783671,
      "grad_norm": 0.2665553092956543,
      "learning_rate": 3.657604925152558e-05,
      "loss": 4.5606,
      "step": 40190
    },
    {
      "epoch": 0.8231114205557717,
      "grad_norm": 0.2775776982307434,
      "learning_rate": 3.6535554970976626e-05,
      "loss": 4.59,
      "step": 40195
    },
    {
      "epoch": 0.8232138103331763,
      "grad_norm": 0.28505730628967285,
      "learning_rate": 3.64950800102103e-05,
      "loss": 4.5835,
      "step": 40200
    },
    {
      "epoch": 0.8233162001105809,
      "grad_norm": 0.28103506565093994,
      "learning_rate": 3.645462437611845e-05,
      "loss": 4.6265,
      "step": 40205
    },
    {
      "epoch": 0.8234185898879856,
      "grad_norm": 0.282204270362854,
      "learning_rate": 3.641418807558946e-05,
      "loss": 4.5734,
      "step": 40210
    },
    {
      "epoch": 0.8235209796653902,
      "grad_norm": 0.278434693813324,
      "learning_rate": 3.637377111550847e-05,
      "loss": 4.5609,
      "step": 40215
    },
    {
      "epoch": 0.8236233694427948,
      "grad_norm": 0.28934016823768616,
      "learning_rate": 3.633337350275745e-05,
      "loss": 4.5745,
      "step": 40220
    },
    {
      "epoch": 0.8237257592201994,
      "grad_norm": 0.27980026602745056,
      "learning_rate": 3.6292995244214804e-05,
      "loss": 4.6031,
      "step": 40225
    },
    {
      "epoch": 0.823828148997604,
      "grad_norm": 0.29846495389938354,
      "learning_rate": 3.6252636346755954e-05,
      "loss": 4.5965,
      "step": 40230
    },
    {
      "epoch": 0.8239305387750087,
      "grad_norm": 0.29432082176208496,
      "learning_rate": 3.6212296817252795e-05,
      "loss": 4.5886,
      "step": 40235
    },
    {
      "epoch": 0.8240329285524133,
      "grad_norm": 0.2681111693382263,
      "learning_rate": 3.6171976662574e-05,
      "loss": 4.5495,
      "step": 40240
    },
    {
      "epoch": 0.8241353183298179,
      "grad_norm": 0.27589935064315796,
      "learning_rate": 3.613167588958502e-05,
      "loss": 4.527,
      "step": 40245
    },
    {
      "epoch": 0.8242377081072225,
      "grad_norm": 0.3165608048439026,
      "learning_rate": 3.609139450514791e-05,
      "loss": 4.5943,
      "step": 40250
    },
    {
      "epoch": 0.8243400978846273,
      "grad_norm": 0.2817896604537964,
      "learning_rate": 3.605113251612137e-05,
      "loss": 4.5478,
      "step": 40255
    },
    {
      "epoch": 0.8244424876620319,
      "grad_norm": 0.294752299785614,
      "learning_rate": 3.6010889929361e-05,
      "loss": 4.5586,
      "step": 40260
    },
    {
      "epoch": 0.8245448774394365,
      "grad_norm": 0.28963619470596313,
      "learning_rate": 3.59706667517189e-05,
      "loss": 4.5754,
      "step": 40265
    },
    {
      "epoch": 0.8246472672168411,
      "grad_norm": 0.2832202911376953,
      "learning_rate": 3.593046299004397e-05,
      "loss": 4.6675,
      "step": 40270
    },
    {
      "epoch": 0.8247496569942457,
      "grad_norm": 0.291576087474823,
      "learning_rate": 3.58902786511817e-05,
      "loss": 4.5963,
      "step": 40275
    },
    {
      "epoch": 0.8248520467716504,
      "grad_norm": 0.2860012650489807,
      "learning_rate": 3.5850113741974454e-05,
      "loss": 4.5391,
      "step": 40280
    },
    {
      "epoch": 0.824954436549055,
      "grad_norm": 0.29083725810050964,
      "learning_rate": 3.580996826926113e-05,
      "loss": 4.5844,
      "step": 40285
    },
    {
      "epoch": 0.8250568263264596,
      "grad_norm": 0.26927658915519714,
      "learning_rate": 3.5769842239877304e-05,
      "loss": 4.6215,
      "step": 40290
    },
    {
      "epoch": 0.8251592161038642,
      "grad_norm": 0.27664250135421753,
      "learning_rate": 3.572973566065541e-05,
      "loss": 4.5786,
      "step": 40295
    },
    {
      "epoch": 0.8252616058812688,
      "grad_norm": 0.28033411502838135,
      "learning_rate": 3.568964853842442e-05,
      "loss": 4.5612,
      "step": 40300
    },
    {
      "epoch": 0.8253639956586735,
      "grad_norm": 0.2839049696922302,
      "learning_rate": 3.564958088000997e-05,
      "loss": 4.5493,
      "step": 40305
    },
    {
      "epoch": 0.8254663854360781,
      "grad_norm": 0.3106590509414673,
      "learning_rate": 3.560953269223457e-05,
      "loss": 4.3958,
      "step": 40310
    },
    {
      "epoch": 0.8255687752134827,
      "grad_norm": 0.2771055996417999,
      "learning_rate": 3.556950398191722e-05,
      "loss": 4.6082,
      "step": 40315
    },
    {
      "epoch": 0.8256711649908873,
      "grad_norm": 0.28918710350990295,
      "learning_rate": 3.552949475587368e-05,
      "loss": 4.5915,
      "step": 40320
    },
    {
      "epoch": 0.825773554768292,
      "grad_norm": 0.28916171193122864,
      "learning_rate": 3.548950502091637e-05,
      "loss": 4.5727,
      "step": 40325
    },
    {
      "epoch": 0.8258759445456966,
      "grad_norm": 0.27592727541923523,
      "learning_rate": 3.544953478385445e-05,
      "loss": 4.5593,
      "step": 40330
    },
    {
      "epoch": 0.8259783343231012,
      "grad_norm": 0.27757173776626587,
      "learning_rate": 3.5409584051493725e-05,
      "loss": 4.5804,
      "step": 40335
    },
    {
      "epoch": 0.8260807241005058,
      "grad_norm": 0.27432388067245483,
      "learning_rate": 3.536965283063661e-05,
      "loss": 4.5884,
      "step": 40340
    },
    {
      "epoch": 0.8261831138779104,
      "grad_norm": 0.28754088282585144,
      "learning_rate": 3.532974112808233e-05,
      "loss": 4.5547,
      "step": 40345
    },
    {
      "epoch": 0.8262855036553151,
      "grad_norm": 0.3026539981365204,
      "learning_rate": 3.528984895062672e-05,
      "loss": 4.6021,
      "step": 40350
    },
    {
      "epoch": 0.8263878934327197,
      "grad_norm": 0.28035303950309753,
      "learning_rate": 3.5249976305062186e-05,
      "loss": 4.5682,
      "step": 40355
    },
    {
      "epoch": 0.8264902832101243,
      "grad_norm": 0.270597368478775,
      "learning_rate": 3.521012319817809e-05,
      "loss": 4.5731,
      "step": 40360
    },
    {
      "epoch": 0.8265926729875289,
      "grad_norm": 0.28226950764656067,
      "learning_rate": 3.517028963676008e-05,
      "loss": 4.5483,
      "step": 40365
    },
    {
      "epoch": 0.8266950627649335,
      "grad_norm": 0.28886494040489197,
      "learning_rate": 3.513047562759085e-05,
      "loss": 4.5326,
      "step": 40370
    },
    {
      "epoch": 0.8267974525423382,
      "grad_norm": 0.29873228073120117,
      "learning_rate": 3.509068117744953e-05,
      "loss": 4.5491,
      "step": 40375
    },
    {
      "epoch": 0.8268998423197428,
      "grad_norm": 0.2729114890098572,
      "learning_rate": 3.505090629311194e-05,
      "loss": 4.6089,
      "step": 40380
    },
    {
      "epoch": 0.8270022320971474,
      "grad_norm": 0.33923688530921936,
      "learning_rate": 3.5011150981350713e-05,
      "loss": 4.5897,
      "step": 40385
    },
    {
      "epoch": 0.827104621874552,
      "grad_norm": 0.3152208626270294,
      "learning_rate": 3.497141524893502e-05,
      "loss": 4.6187,
      "step": 40390
    },
    {
      "epoch": 0.8272070116519566,
      "grad_norm": 0.28245314955711365,
      "learning_rate": 3.4931699102630686e-05,
      "loss": 4.5578,
      "step": 40395
    },
    {
      "epoch": 0.8273094014293613,
      "grad_norm": 0.29945963621139526,
      "learning_rate": 3.4892002549200316e-05,
      "loss": 4.6068,
      "step": 40400
    },
    {
      "epoch": 0.8274117912067659,
      "grad_norm": 0.2750268280506134,
      "learning_rate": 3.4852325595403074e-05,
      "loss": 4.566,
      "step": 40405
    },
    {
      "epoch": 0.8275141809841705,
      "grad_norm": 0.31208157539367676,
      "learning_rate": 3.481266824799481e-05,
      "loss": 4.4874,
      "step": 40410
    },
    {
      "epoch": 0.8276165707615751,
      "grad_norm": 0.30593588948249817,
      "learning_rate": 3.477303051372805e-05,
      "loss": 4.5458,
      "step": 40415
    },
    {
      "epoch": 0.8277189605389798,
      "grad_norm": 0.3171047270298004,
      "learning_rate": 3.473341239935201e-05,
      "loss": 4.6186,
      "step": 40420
    },
    {
      "epoch": 0.8278213503163844,
      "grad_norm": 0.279586523771286,
      "learning_rate": 3.46938139116125e-05,
      "loss": 4.5988,
      "step": 40425
    },
    {
      "epoch": 0.827923740093789,
      "grad_norm": 0.2898420989513397,
      "learning_rate": 3.4654235057252014e-05,
      "loss": 4.6204,
      "step": 40430
    },
    {
      "epoch": 0.8280261298711936,
      "grad_norm": 0.28708988428115845,
      "learning_rate": 3.4614675843009796e-05,
      "loss": 4.576,
      "step": 40435
    },
    {
      "epoch": 0.8281285196485982,
      "grad_norm": 0.29046791791915894,
      "learning_rate": 3.4575136275621525e-05,
      "loss": 4.5745,
      "step": 40440
    },
    {
      "epoch": 0.8282309094260029,
      "grad_norm": 0.2747562825679779,
      "learning_rate": 3.4535616361819724e-05,
      "loss": 4.6188,
      "step": 40445
    },
    {
      "epoch": 0.8283332992034075,
      "grad_norm": 0.29619061946868896,
      "learning_rate": 3.449611610833358e-05,
      "loss": 4.6506,
      "step": 40450
    },
    {
      "epoch": 0.8284356889808121,
      "grad_norm": 0.30195343494415283,
      "learning_rate": 3.445663552188882e-05,
      "loss": 4.5114,
      "step": 40455
    },
    {
      "epoch": 0.8285380787582168,
      "grad_norm": 0.29632195830345154,
      "learning_rate": 3.4417174609207856e-05,
      "loss": 4.5947,
      "step": 40460
    },
    {
      "epoch": 0.8286404685356215,
      "grad_norm": 0.2991366684436798,
      "learning_rate": 3.437773337700976e-05,
      "loss": 4.5803,
      "step": 40465
    },
    {
      "epoch": 0.8287428583130261,
      "grad_norm": 0.2884520888328552,
      "learning_rate": 3.4338311832010296e-05,
      "loss": 4.5416,
      "step": 40470
    },
    {
      "epoch": 0.8288452480904307,
      "grad_norm": 0.27997058629989624,
      "learning_rate": 3.429890998092182e-05,
      "loss": 4.5782,
      "step": 40475
    },
    {
      "epoch": 0.8289476378678353,
      "grad_norm": 0.2756667137145996,
      "learning_rate": 3.4259527830453294e-05,
      "loss": 4.5538,
      "step": 40480
    },
    {
      "epoch": 0.8290500276452399,
      "grad_norm": 0.2803933620452881,
      "learning_rate": 3.422016538731049e-05,
      "loss": 4.5603,
      "step": 40485
    },
    {
      "epoch": 0.8291524174226446,
      "grad_norm": 0.2907135784626007,
      "learning_rate": 3.418082265819566e-05,
      "loss": 4.5366,
      "step": 40490
    },
    {
      "epoch": 0.8292548072000492,
      "grad_norm": 0.48615598678588867,
      "learning_rate": 3.414149964980772e-05,
      "loss": 4.5828,
      "step": 40495
    },
    {
      "epoch": 0.8293571969774538,
      "grad_norm": 0.27886176109313965,
      "learning_rate": 3.410219636884239e-05,
      "loss": 4.5782,
      "step": 40500
    },
    {
      "epoch": 0.8294595867548584,
      "grad_norm": 0.28133606910705566,
      "learning_rate": 3.4062912821991734e-05,
      "loss": 4.577,
      "step": 40505
    },
    {
      "epoch": 0.829561976532263,
      "grad_norm": 0.2881931960582733,
      "learning_rate": 3.4023649015944756e-05,
      "loss": 4.5106,
      "step": 40510
    },
    {
      "epoch": 0.8296643663096677,
      "grad_norm": 0.2737469971179962,
      "learning_rate": 3.398440495738692e-05,
      "loss": 4.5582,
      "step": 40515
    },
    {
      "epoch": 0.8297667560870723,
      "grad_norm": 0.2830897271633148,
      "learning_rate": 3.3945180653000366e-05,
      "loss": 4.5933,
      "step": 40520
    },
    {
      "epoch": 0.8298691458644769,
      "grad_norm": 0.2840888798236847,
      "learning_rate": 3.3905976109463936e-05,
      "loss": 4.5815,
      "step": 40525
    },
    {
      "epoch": 0.8299715356418815,
      "grad_norm": 0.293977290391922,
      "learning_rate": 3.386679133345302e-05,
      "loss": 4.5645,
      "step": 40530
    },
    {
      "epoch": 0.8300739254192862,
      "grad_norm": 0.2870458960533142,
      "learning_rate": 3.3827626331639635e-05,
      "loss": 4.583,
      "step": 40535
    },
    {
      "epoch": 0.8301763151966908,
      "grad_norm": 0.284534752368927,
      "learning_rate": 3.378848111069256e-05,
      "loss": 4.5753,
      "step": 40540
    },
    {
      "epoch": 0.8302787049740954,
      "grad_norm": 0.2755497097969055,
      "learning_rate": 3.3749355677277085e-05,
      "loss": 4.5738,
      "step": 40545
    },
    {
      "epoch": 0.8303810947515,
      "grad_norm": 0.2797011435031891,
      "learning_rate": 3.371025003805514e-05,
      "loss": 4.6116,
      "step": 40550
    },
    {
      "epoch": 0.8304834845289046,
      "grad_norm": 0.2987256944179535,
      "learning_rate": 3.367116419968527e-05,
      "loss": 4.5602,
      "step": 40555
    },
    {
      "epoch": 0.8305858743063093,
      "grad_norm": 0.27878180146217346,
      "learning_rate": 3.363209816882279e-05,
      "loss": 4.5695,
      "step": 40560
    },
    {
      "epoch": 0.8306882640837139,
      "grad_norm": 0.2940962612628937,
      "learning_rate": 3.359305195211949e-05,
      "loss": 4.6228,
      "step": 40565
    },
    {
      "epoch": 0.8307906538611185,
      "grad_norm": 0.280834436416626,
      "learning_rate": 3.355402555622378e-05,
      "loss": 4.5861,
      "step": 40570
    },
    {
      "epoch": 0.8308930436385231,
      "grad_norm": 0.2941542863845825,
      "learning_rate": 3.3515018987780884e-05,
      "loss": 4.5604,
      "step": 40575
    },
    {
      "epoch": 0.8309954334159277,
      "grad_norm": 0.2844966650009155,
      "learning_rate": 3.3476032253432356e-05,
      "loss": 4.5558,
      "step": 40580
    },
    {
      "epoch": 0.8310978231933324,
      "grad_norm": 0.3021025061607361,
      "learning_rate": 3.343706535981662e-05,
      "loss": 4.5832,
      "step": 40585
    },
    {
      "epoch": 0.831200212970737,
      "grad_norm": 0.27956724166870117,
      "learning_rate": 3.3398118313568647e-05,
      "loss": 4.5957,
      "step": 40590
    },
    {
      "epoch": 0.8313026027481416,
      "grad_norm": 0.30196988582611084,
      "learning_rate": 3.335919112132e-05,
      "loss": 4.5439,
      "step": 40595
    },
    {
      "epoch": 0.8314049925255462,
      "grad_norm": 0.28697851300239563,
      "learning_rate": 3.332028378969888e-05,
      "loss": 4.5951,
      "step": 40600
    },
    {
      "epoch": 0.8315073823029508,
      "grad_norm": 0.303102046251297,
      "learning_rate": 3.328139632533006e-05,
      "loss": 4.537,
      "step": 40605
    },
    {
      "epoch": 0.8316097720803555,
      "grad_norm": 0.2824726998806,
      "learning_rate": 3.324252873483505e-05,
      "loss": 4.5667,
      "step": 40610
    },
    {
      "epoch": 0.8317121618577601,
      "grad_norm": 0.2867865562438965,
      "learning_rate": 3.320368102483185e-05,
      "loss": 4.5588,
      "step": 40615
    },
    {
      "epoch": 0.8318145516351647,
      "grad_norm": 0.29350194334983826,
      "learning_rate": 3.3164853201935104e-05,
      "loss": 4.6302,
      "step": 40620
    },
    {
      "epoch": 0.8319169414125693,
      "grad_norm": 0.28072482347488403,
      "learning_rate": 3.3126045272756194e-05,
      "loss": 4.5966,
      "step": 40625
    },
    {
      "epoch": 0.832019331189974,
      "grad_norm": 0.31003057956695557,
      "learning_rate": 3.308725724390286e-05,
      "loss": 4.5576,
      "step": 40630
    },
    {
      "epoch": 0.8321217209673786,
      "grad_norm": 0.3015013337135315,
      "learning_rate": 3.3048489121979694e-05,
      "loss": 4.6012,
      "step": 40635
    },
    {
      "epoch": 0.8322241107447832,
      "grad_norm": 0.2813338339328766,
      "learning_rate": 3.300974091358788e-05,
      "loss": 4.6031,
      "step": 40640
    },
    {
      "epoch": 0.8323265005221878,
      "grad_norm": 0.2762916386127472,
      "learning_rate": 3.2971012625324987e-05,
      "loss": 4.5773,
      "step": 40645
    },
    {
      "epoch": 0.8324288902995924,
      "grad_norm": 0.28494009375572205,
      "learning_rate": 3.293230426378546e-05,
      "loss": 4.6118,
      "step": 40650
    },
    {
      "epoch": 0.8325312800769971,
      "grad_norm": 0.27528828382492065,
      "learning_rate": 3.2893615835560205e-05,
      "loss": 4.5894,
      "step": 40655
    },
    {
      "epoch": 0.8326336698544017,
      "grad_norm": 0.2866365909576416,
      "learning_rate": 3.285494734723671e-05,
      "loss": 4.5539,
      "step": 40660
    },
    {
      "epoch": 0.8327360596318064,
      "grad_norm": 0.2842416763305664,
      "learning_rate": 3.281629880539921e-05,
      "loss": 4.593,
      "step": 40665
    },
    {
      "epoch": 0.832838449409211,
      "grad_norm": 0.2805674076080322,
      "learning_rate": 3.277767021662844e-05,
      "loss": 4.5866,
      "step": 40670
    },
    {
      "epoch": 0.8329408391866157,
      "grad_norm": 0.27999147772789,
      "learning_rate": 3.273906158750168e-05,
      "loss": 4.6231,
      "step": 40675
    },
    {
      "epoch": 0.8330432289640203,
      "grad_norm": 0.2912527918815613,
      "learning_rate": 3.270047292459298e-05,
      "loss": 4.5316,
      "step": 40680
    },
    {
      "epoch": 0.8331456187414249,
      "grad_norm": 0.2727186977863312,
      "learning_rate": 3.2661904234472866e-05,
      "loss": 4.5816,
      "step": 40685
    },
    {
      "epoch": 0.8332480085188295,
      "grad_norm": 0.27656301856040955,
      "learning_rate": 3.262335552370849e-05,
      "loss": 4.5678,
      "step": 40690
    },
    {
      "epoch": 0.8333503982962341,
      "grad_norm": 0.27214816212654114,
      "learning_rate": 3.258482679886356e-05,
      "loss": 4.596,
      "step": 40695
    },
    {
      "epoch": 0.8334527880736388,
      "grad_norm": 0.30073586106300354,
      "learning_rate": 3.2546318066498523e-05,
      "loss": 4.5546,
      "step": 40700
    },
    {
      "epoch": 0.8335551778510434,
      "grad_norm": 0.282539963722229,
      "learning_rate": 3.250782933317025e-05,
      "loss": 4.6335,
      "step": 40705
    },
    {
      "epoch": 0.833657567628448,
      "grad_norm": 0.28320568799972534,
      "learning_rate": 3.246936060543228e-05,
      "loss": 4.6191,
      "step": 40710
    },
    {
      "epoch": 0.8337599574058526,
      "grad_norm": 0.2768017649650574,
      "learning_rate": 3.243091188983485e-05,
      "loss": 4.5991,
      "step": 40715
    },
    {
      "epoch": 0.8338623471832572,
      "grad_norm": 0.27154484391212463,
      "learning_rate": 3.2392483192924546e-05,
      "loss": 4.6122,
      "step": 40720
    },
    {
      "epoch": 0.8339647369606619,
      "grad_norm": 0.2970268130302429,
      "learning_rate": 3.2354074521244734e-05,
      "loss": 4.5865,
      "step": 40725
    },
    {
      "epoch": 0.8340671267380665,
      "grad_norm": 0.27358826994895935,
      "learning_rate": 3.2315685881335395e-05,
      "loss": 4.5937,
      "step": 40730
    },
    {
      "epoch": 0.8341695165154711,
      "grad_norm": 0.31513655185699463,
      "learning_rate": 3.227731727973297e-05,
      "loss": 4.5818,
      "step": 40735
    },
    {
      "epoch": 0.8342719062928757,
      "grad_norm": 0.2948060929775238,
      "learning_rate": 3.223896872297055e-05,
      "loss": 4.6303,
      "step": 40740
    },
    {
      "epoch": 0.8343742960702804,
      "grad_norm": 0.29236993193626404,
      "learning_rate": 3.220064021757777e-05,
      "loss": 4.5999,
      "step": 40745
    },
    {
      "epoch": 0.834476685847685,
      "grad_norm": 0.2867458760738373,
      "learning_rate": 3.216233177008097e-05,
      "loss": 4.5985,
      "step": 40750
    },
    {
      "epoch": 0.8345790756250896,
      "grad_norm": 0.2854442000389099,
      "learning_rate": 3.2124043387002954e-05,
      "loss": 4.6083,
      "step": 40755
    },
    {
      "epoch": 0.8346814654024942,
      "grad_norm": 0.2821553647518158,
      "learning_rate": 3.2085775074863105e-05,
      "loss": 4.5944,
      "step": 40760
    },
    {
      "epoch": 0.8347838551798988,
      "grad_norm": 0.2813735604286194,
      "learning_rate": 3.204752684017756e-05,
      "loss": 4.5255,
      "step": 40765
    },
    {
      "epoch": 0.8348862449573035,
      "grad_norm": 0.2835007905960083,
      "learning_rate": 3.2009298689458744e-05,
      "loss": 4.558,
      "step": 40770
    },
    {
      "epoch": 0.8349886347347081,
      "grad_norm": 0.28022801876068115,
      "learning_rate": 3.197109062921591e-05,
      "loss": 4.5757,
      "step": 40775
    },
    {
      "epoch": 0.8350910245121127,
      "grad_norm": 0.2893160283565521,
      "learning_rate": 3.1932902665954894e-05,
      "loss": 4.5803,
      "step": 40780
    },
    {
      "epoch": 0.8351934142895173,
      "grad_norm": 0.2787182033061981,
      "learning_rate": 3.189473480617786e-05,
      "loss": 4.6025,
      "step": 40785
    },
    {
      "epoch": 0.8352958040669219,
      "grad_norm": 0.27996236085891724,
      "learning_rate": 3.1856587056383846e-05,
      "loss": 4.5655,
      "step": 40790
    },
    {
      "epoch": 0.8353981938443266,
      "grad_norm": 0.27595436573028564,
      "learning_rate": 3.1818459423068266e-05,
      "loss": 4.5905,
      "step": 40795
    },
    {
      "epoch": 0.8355005836217312,
      "grad_norm": 0.2880684733390808,
      "learning_rate": 3.1780351912723155e-05,
      "loss": 4.6407,
      "step": 40800
    },
    {
      "epoch": 0.8356029733991358,
      "grad_norm": 0.2865196764469147,
      "learning_rate": 3.174226453183725e-05,
      "loss": 4.5965,
      "step": 40805
    },
    {
      "epoch": 0.8357053631765404,
      "grad_norm": 0.27217257022857666,
      "learning_rate": 3.170419728689565e-05,
      "loss": 4.5921,
      "step": 40810
    },
    {
      "epoch": 0.835807752953945,
      "grad_norm": 0.26777732372283936,
      "learning_rate": 3.166615018438018e-05,
      "loss": 4.5762,
      "step": 40815
    },
    {
      "epoch": 0.8359101427313497,
      "grad_norm": 0.2771703600883484,
      "learning_rate": 3.162812323076914e-05,
      "loss": 4.5757,
      "step": 40820
    },
    {
      "epoch": 0.8360125325087543,
      "grad_norm": 0.2957483232021332,
      "learning_rate": 3.1590116432537495e-05,
      "loss": 4.5597,
      "step": 40825
    },
    {
      "epoch": 0.8361149222861589,
      "grad_norm": 0.2982672154903412,
      "learning_rate": 3.1552129796156725e-05,
      "loss": 4.6019,
      "step": 40830
    },
    {
      "epoch": 0.8362173120635635,
      "grad_norm": 0.28175172209739685,
      "learning_rate": 3.15141633280948e-05,
      "loss": 4.5474,
      "step": 40835
    },
    {
      "epoch": 0.8363197018409682,
      "grad_norm": 0.2830298840999603,
      "learning_rate": 3.147621703481644e-05,
      "loss": 4.6469,
      "step": 40840
    },
    {
      "epoch": 0.8364220916183728,
      "grad_norm": 0.40411484241485596,
      "learning_rate": 3.1438290922782764e-05,
      "loss": 4.6078,
      "step": 40845
    },
    {
      "epoch": 0.8365244813957774,
      "grad_norm": 0.2745141088962555,
      "learning_rate": 3.1400384998451506e-05,
      "loss": 4.5984,
      "step": 40850
    },
    {
      "epoch": 0.836626871173182,
      "grad_norm": 0.28638893365859985,
      "learning_rate": 3.136249926827707e-05,
      "loss": 4.5967,
      "step": 40855
    },
    {
      "epoch": 0.8367292609505866,
      "grad_norm": 0.30151182413101196,
      "learning_rate": 3.1324633738710164e-05,
      "loss": 4.5752,
      "step": 40860
    },
    {
      "epoch": 0.8368316507279914,
      "grad_norm": 0.276595801115036,
      "learning_rate": 3.12867884161983e-05,
      "loss": 4.552,
      "step": 40865
    },
    {
      "epoch": 0.836934040505396,
      "grad_norm": 0.2695423364639282,
      "learning_rate": 3.124896330718551e-05,
      "loss": 4.5706,
      "step": 40870
    },
    {
      "epoch": 0.8370364302828006,
      "grad_norm": 0.27079302072525024,
      "learning_rate": 3.121115841811229e-05,
      "loss": 4.5541,
      "step": 40875
    },
    {
      "epoch": 0.8371388200602052,
      "grad_norm": 0.2798822522163391,
      "learning_rate": 3.117337375541575e-05,
      "loss": 4.5892,
      "step": 40880
    },
    {
      "epoch": 0.8372412098376099,
      "grad_norm": 0.28381332755088806,
      "learning_rate": 3.11356093255295e-05,
      "loss": 4.5229,
      "step": 40885
    },
    {
      "epoch": 0.8373435996150145,
      "grad_norm": 0.31425201892852783,
      "learning_rate": 3.1097865134883834e-05,
      "loss": 4.5647,
      "step": 40890
    },
    {
      "epoch": 0.8374459893924191,
      "grad_norm": 0.30924391746520996,
      "learning_rate": 3.1060141189905474e-05,
      "loss": 4.5454,
      "step": 40895
    },
    {
      "epoch": 0.8375483791698237,
      "grad_norm": 0.27807021141052246,
      "learning_rate": 3.102243749701772e-05,
      "loss": 4.601,
      "step": 40900
    },
    {
      "epoch": 0.8376507689472283,
      "grad_norm": 0.3153919279575348,
      "learning_rate": 3.0984754062640556e-05,
      "loss": 4.52,
      "step": 40905
    },
    {
      "epoch": 0.837753158724633,
      "grad_norm": 0.2784497141838074,
      "learning_rate": 3.094709089319024e-05,
      "loss": 4.5393,
      "step": 40910
    },
    {
      "epoch": 0.8378555485020376,
      "grad_norm": 0.28070005774497986,
      "learning_rate": 3.090944799507982e-05,
      "loss": 4.5848,
      "step": 40915
    },
    {
      "epoch": 0.8379579382794422,
      "grad_norm": 0.34741532802581787,
      "learning_rate": 3.087182537471889e-05,
      "loss": 4.6055,
      "step": 40920
    },
    {
      "epoch": 0.8380603280568468,
      "grad_norm": 0.2964925467967987,
      "learning_rate": 3.083422303851337e-05,
      "loss": 4.5471,
      "step": 40925
    },
    {
      "epoch": 0.8381627178342514,
      "grad_norm": 0.2837379276752472,
      "learning_rate": 3.0796640992866e-05,
      "loss": 4.6261,
      "step": 40930
    },
    {
      "epoch": 0.8382651076116561,
      "grad_norm": 0.2933388650417328,
      "learning_rate": 3.075907924417587e-05,
      "loss": 4.5777,
      "step": 40935
    },
    {
      "epoch": 0.8383674973890607,
      "grad_norm": 0.27644672989845276,
      "learning_rate": 3.072153779883868e-05,
      "loss": 4.5548,
      "step": 40940
    },
    {
      "epoch": 0.8384698871664653,
      "grad_norm": 0.279789000749588,
      "learning_rate": 3.068401666324672e-05,
      "loss": 4.5693,
      "step": 40945
    },
    {
      "epoch": 0.8385722769438699,
      "grad_norm": 0.28126752376556396,
      "learning_rate": 3.0646515843788763e-05,
      "loss": 4.5984,
      "step": 40950
    },
    {
      "epoch": 0.8386746667212746,
      "grad_norm": 0.28015634417533875,
      "learning_rate": 3.060903534685013e-05,
      "loss": 4.5702,
      "step": 40955
    },
    {
      "epoch": 0.8387770564986792,
      "grad_norm": 0.2922801077365875,
      "learning_rate": 3.057157517881263e-05,
      "loss": 4.6126,
      "step": 40960
    },
    {
      "epoch": 0.8388794462760838,
      "grad_norm": 0.27999451756477356,
      "learning_rate": 3.0534135346054774e-05,
      "loss": 4.5665,
      "step": 40965
    },
    {
      "epoch": 0.8389818360534884,
      "grad_norm": 0.2854362428188324,
      "learning_rate": 3.049671585495146e-05,
      "loss": 4.5704,
      "step": 40970
    },
    {
      "epoch": 0.839084225830893,
      "grad_norm": 0.27575409412384033,
      "learning_rate": 3.045931671187411e-05,
      "loss": 4.5876,
      "step": 40975
    },
    {
      "epoch": 0.8391866156082977,
      "grad_norm": 0.31625238060951233,
      "learning_rate": 3.042193792319083e-05,
      "loss": 4.6041,
      "step": 40980
    },
    {
      "epoch": 0.8392890053857023,
      "grad_norm": 0.27901962399482727,
      "learning_rate": 3.0384579495266147e-05,
      "loss": 4.5466,
      "step": 40985
    },
    {
      "epoch": 0.8393913951631069,
      "grad_norm": 0.29163533449172974,
      "learning_rate": 3.034724143446109e-05,
      "loss": 4.5922,
      "step": 40990
    },
    {
      "epoch": 0.8394937849405115,
      "grad_norm": 0.29959243535995483,
      "learning_rate": 3.0309923747133386e-05,
      "loss": 4.5846,
      "step": 40995
    },
    {
      "epoch": 0.8395961747179161,
      "grad_norm": 0.2757844626903534,
      "learning_rate": 3.027262643963702e-05,
      "loss": 4.552,
      "step": 41000
    },
    {
      "epoch": 0.8395961747179161,
      "eval_loss": 4.593181133270264,
      "eval_runtime": 17.1222,
      "eval_samples_per_second": 33.991,
      "eval_steps_per_second": 4.263,
      "step": 41000
    },
    {
      "epoch": 0.8396985644953208,
      "grad_norm": 0.28662437200546265,
      "learning_rate": 3.02353495183228e-05,
      "loss": 4.6171,
      "step": 41005
    },
    {
      "epoch": 0.8398009542727254,
      "grad_norm": 0.2877452075481415,
      "learning_rate": 3.0198092989537858e-05,
      "loss": 4.5878,
      "step": 41010
    },
    {
      "epoch": 0.83990334405013,
      "grad_norm": 0.28487837314605713,
      "learning_rate": 3.0160856859625977e-05,
      "loss": 4.5274,
      "step": 41015
    },
    {
      "epoch": 0.8400057338275346,
      "grad_norm": 0.2777809798717499,
      "learning_rate": 3.0123641134927395e-05,
      "loss": 4.5628,
      "step": 41020
    },
    {
      "epoch": 0.8401081236049392,
      "grad_norm": 0.2962719798088074,
      "learning_rate": 3.0086445821778863e-05,
      "loss": 4.5668,
      "step": 41025
    },
    {
      "epoch": 0.8402105133823439,
      "grad_norm": 0.31961870193481445,
      "learning_rate": 3.0049270926513754e-05,
      "loss": 4.5609,
      "step": 41030
    },
    {
      "epoch": 0.8403129031597485,
      "grad_norm": 0.27768102288246155,
      "learning_rate": 3.001211645546186e-05,
      "loss": 4.6029,
      "step": 41035
    },
    {
      "epoch": 0.8404152929371531,
      "grad_norm": 0.2697317600250244,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5762,
      "step": 41040
    },
    {
      "epoch": 0.8405176827145577,
      "grad_norm": 0.28922295570373535,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5458,
      "step": 41045
    },
    {
      "epoch": 0.8406200724919624,
      "grad_norm": 0.2773067355155945,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6026,
      "step": 41050
    },
    {
      "epoch": 0.840722462269367,
      "grad_norm": 0.27529552578926086,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6204,
      "step": 41055
    },
    {
      "epoch": 0.8408248520467716,
      "grad_norm": 0.2800615131855011,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5814,
      "step": 41060
    },
    {
      "epoch": 0.8409272418241762,
      "grad_norm": 0.2822313606739044,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5486,
      "step": 41065
    },
    {
      "epoch": 0.841029631601581,
      "grad_norm": 0.27693304419517517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5574,
      "step": 41070
    },
    {
      "epoch": 0.8411320213789856,
      "grad_norm": 0.28031983971595764,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5619,
      "step": 41075
    },
    {
      "epoch": 0.8412344111563902,
      "grad_norm": 0.29966622591018677,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5669,
      "step": 41080
    },
    {
      "epoch": 0.8413368009337948,
      "grad_norm": 0.2828528583049774,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6466,
      "step": 41085
    },
    {
      "epoch": 0.8414391907111994,
      "grad_norm": 0.28675830364227295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5915,
      "step": 41090
    },
    {
      "epoch": 0.8415415804886041,
      "grad_norm": 0.2805325984954834,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5625,
      "step": 41095
    },
    {
      "epoch": 0.8416439702660087,
      "grad_norm": 0.2904501259326935,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5674,
      "step": 41100
    },
    {
      "epoch": 0.8417463600434133,
      "grad_norm": 0.28026166558265686,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.601,
      "step": 41105
    },
    {
      "epoch": 0.8418487498208179,
      "grad_norm": 0.2790822684764862,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6097,
      "step": 41110
    },
    {
      "epoch": 0.8419511395982225,
      "grad_norm": 0.27744176983833313,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5876,
      "step": 41115
    },
    {
      "epoch": 0.8420535293756272,
      "grad_norm": 0.3159765899181366,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5175,
      "step": 41120
    },
    {
      "epoch": 0.8421559191530318,
      "grad_norm": 0.27997395396232605,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5976,
      "step": 41125
    },
    {
      "epoch": 0.8422583089304364,
      "grad_norm": 0.28126683831214905,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6137,
      "step": 41130
    },
    {
      "epoch": 0.842360698707841,
      "grad_norm": 0.28357672691345215,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.586,
      "step": 41135
    },
    {
      "epoch": 0.8424630884852456,
      "grad_norm": 0.2679639160633087,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5666,
      "step": 41140
    },
    {
      "epoch": 0.8425654782626503,
      "grad_norm": 0.2936253249645233,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5809,
      "step": 41145
    },
    {
      "epoch": 0.8426678680400549,
      "grad_norm": 0.297695130109787,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5803,
      "step": 41150
    },
    {
      "epoch": 0.8427702578174595,
      "grad_norm": 0.2765631675720215,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5503,
      "step": 41155
    },
    {
      "epoch": 0.8428726475948641,
      "grad_norm": 0.318791925907135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5819,
      "step": 41160
    },
    {
      "epoch": 0.8429750373722688,
      "grad_norm": 0.29362010955810547,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5931,
      "step": 41165
    },
    {
      "epoch": 0.8430774271496734,
      "grad_norm": 0.2787388563156128,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5473,
      "step": 41170
    },
    {
      "epoch": 0.843179816927078,
      "grad_norm": 0.3351594805717468,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.577,
      "step": 41175
    },
    {
      "epoch": 0.8432822067044826,
      "grad_norm": 0.3098401725292206,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5853,
      "step": 41180
    },
    {
      "epoch": 0.8433845964818872,
      "grad_norm": 0.302744060754776,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5882,
      "step": 41185
    },
    {
      "epoch": 0.8434869862592919,
      "grad_norm": 0.27730438113212585,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6287,
      "step": 41190
    },
    {
      "epoch": 0.8435893760366965,
      "grad_norm": 0.2852722406387329,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5574,
      "step": 41195
    },
    {
      "epoch": 0.8436917658141011,
      "grad_norm": 0.2758505344390869,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.568,
      "step": 41200
    },
    {
      "epoch": 0.8437941555915057,
      "grad_norm": 0.28374865651130676,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5608,
      "step": 41205
    },
    {
      "epoch": 0.8438965453689103,
      "grad_norm": 0.2999465763568878,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5649,
      "step": 41210
    },
    {
      "epoch": 0.843998935146315,
      "grad_norm": 0.28205832839012146,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5096,
      "step": 41215
    },
    {
      "epoch": 0.8441013249237196,
      "grad_norm": 0.27388060092926025,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5995,
      "step": 41220
    },
    {
      "epoch": 0.8442037147011242,
      "grad_norm": 0.2652662694454193,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.453,
      "step": 41225
    },
    {
      "epoch": 0.8443061044785288,
      "grad_norm": 0.30840641260147095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.526,
      "step": 41230
    },
    {
      "epoch": 0.8444084942559335,
      "grad_norm": 0.2902662754058838,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5417,
      "step": 41235
    },
    {
      "epoch": 0.8445108840333381,
      "grad_norm": 0.27841314673423767,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5865,
      "step": 41240
    },
    {
      "epoch": 0.8446132738107427,
      "grad_norm": 0.2912626266479492,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6389,
      "step": 41245
    },
    {
      "epoch": 0.8447156635881473,
      "grad_norm": 0.2787330150604248,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5385,
      "step": 41250
    },
    {
      "epoch": 0.8448180533655519,
      "grad_norm": 0.2781839072704315,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6031,
      "step": 41255
    },
    {
      "epoch": 0.8449204431429566,
      "grad_norm": 0.3461420238018036,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5803,
      "step": 41260
    },
    {
      "epoch": 0.8450228329203612,
      "grad_norm": 0.282930850982666,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5916,
      "step": 41265
    },
    {
      "epoch": 0.8451252226977659,
      "grad_norm": 0.2960691452026367,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5519,
      "step": 41270
    },
    {
      "epoch": 0.8452276124751705,
      "grad_norm": 0.28614863753318787,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 41275
    },
    {
      "epoch": 0.8453300022525752,
      "grad_norm": 0.2875410318374634,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5659,
      "step": 41280
    },
    {
      "epoch": 0.8454323920299798,
      "grad_norm": 0.6949137449264526,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 41285
    },
    {
      "epoch": 0.8455347818073844,
      "grad_norm": 0.2943025529384613,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6048,
      "step": 41290
    },
    {
      "epoch": 0.845637171584789,
      "grad_norm": 0.2923896312713623,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.585,
      "step": 41295
    },
    {
      "epoch": 0.8457395613621936,
      "grad_norm": 0.30523452162742615,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5578,
      "step": 41300
    },
    {
      "epoch": 0.8458419511395983,
      "grad_norm": 0.2870691418647766,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6085,
      "step": 41305
    },
    {
      "epoch": 0.8459443409170029,
      "grad_norm": 0.2939387559890747,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5408,
      "step": 41310
    },
    {
      "epoch": 0.8460467306944075,
      "grad_norm": 0.26975417137145996,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5448,
      "step": 41315
    },
    {
      "epoch": 0.8461491204718121,
      "grad_norm": 0.2728458344936371,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6098,
      "step": 41320
    },
    {
      "epoch": 0.8462515102492167,
      "grad_norm": 0.27852603793144226,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5419,
      "step": 41325
    },
    {
      "epoch": 0.8463539000266214,
      "grad_norm": 0.2822387218475342,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5798,
      "step": 41330
    },
    {
      "epoch": 0.846456289804026,
      "grad_norm": 0.28445008397102356,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5386,
      "step": 41335
    },
    {
      "epoch": 0.8465586795814306,
      "grad_norm": 0.2744186520576477,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5464,
      "step": 41340
    },
    {
      "epoch": 0.8466610693588352,
      "grad_norm": 0.2751639783382416,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5761,
      "step": 41345
    },
    {
      "epoch": 0.8467634591362398,
      "grad_norm": 0.31351083517074585,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 41350
    },
    {
      "epoch": 0.8468658489136445,
      "grad_norm": 0.286384642124176,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5635,
      "step": 41355
    },
    {
      "epoch": 0.8469682386910491,
      "grad_norm": 0.28469738364219666,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6158,
      "step": 41360
    },
    {
      "epoch": 0.8470706284684537,
      "grad_norm": 0.2857683300971985,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5918,
      "step": 41365
    },
    {
      "epoch": 0.8471730182458583,
      "grad_norm": 0.3154352307319641,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5592,
      "step": 41370
    },
    {
      "epoch": 0.847275408023263,
      "grad_norm": 0.2844149172306061,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5721,
      "step": 41375
    },
    {
      "epoch": 0.8473777978006676,
      "grad_norm": 0.2833268642425537,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5631,
      "step": 41380
    },
    {
      "epoch": 0.8474801875780722,
      "grad_norm": 0.3229454457759857,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6033,
      "step": 41385
    },
    {
      "epoch": 0.8475825773554768,
      "grad_norm": 0.28634709119796753,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6129,
      "step": 41390
    },
    {
      "epoch": 0.8476849671328814,
      "grad_norm": 0.3008575141429901,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5326,
      "step": 41395
    },
    {
      "epoch": 0.8477873569102861,
      "grad_norm": 0.332033634185791,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5979,
      "step": 41400
    },
    {
      "epoch": 0.8478897466876907,
      "grad_norm": 0.2948870360851288,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5302,
      "step": 41405
    },
    {
      "epoch": 0.8479921364650953,
      "grad_norm": 0.2760053873062134,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6197,
      "step": 41410
    },
    {
      "epoch": 0.8480945262424999,
      "grad_norm": 0.2818705141544342,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5919,
      "step": 41415
    },
    {
      "epoch": 0.8481969160199045,
      "grad_norm": 0.30404141545295715,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5394,
      "step": 41420
    },
    {
      "epoch": 0.8482993057973092,
      "grad_norm": 0.2797127962112427,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4302,
      "step": 41425
    },
    {
      "epoch": 0.8484016955747138,
      "grad_norm": 0.29585498571395874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5359,
      "step": 41430
    },
    {
      "epoch": 0.8485040853521184,
      "grad_norm": 0.2758622467517853,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5738,
      "step": 41435
    },
    {
      "epoch": 0.848606475129523,
      "grad_norm": 0.27927926182746887,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5562,
      "step": 41440
    },
    {
      "epoch": 0.8487088649069277,
      "grad_norm": 0.28380414843559265,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5316,
      "step": 41445
    },
    {
      "epoch": 0.8488112546843323,
      "grad_norm": 0.2843684256076813,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6323,
      "step": 41450
    },
    {
      "epoch": 0.8489136444617369,
      "grad_norm": 0.27560290694236755,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 41455
    },
    {
      "epoch": 0.8490160342391415,
      "grad_norm": 0.2876559793949127,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5247,
      "step": 41460
    },
    {
      "epoch": 0.8491184240165461,
      "grad_norm": 0.31301239132881165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5129,
      "step": 41465
    },
    {
      "epoch": 0.8492208137939508,
      "grad_norm": 0.2861793637275696,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.587,
      "step": 41470
    },
    {
      "epoch": 0.8493232035713555,
      "grad_norm": 0.28453120589256287,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5889,
      "step": 41475
    },
    {
      "epoch": 0.8494255933487601,
      "grad_norm": 0.3102613091468811,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6361,
      "step": 41480
    },
    {
      "epoch": 0.8495279831261647,
      "grad_norm": 0.28760915994644165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6013,
      "step": 41485
    },
    {
      "epoch": 0.8496303729035694,
      "grad_norm": 0.28489428758621216,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5335,
      "step": 41490
    },
    {
      "epoch": 0.849732762680974,
      "grad_norm": 0.28293073177337646,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5785,
      "step": 41495
    },
    {
      "epoch": 0.8498351524583786,
      "grad_norm": 0.277610719203949,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5531,
      "step": 41500
    },
    {
      "epoch": 0.8499375422357832,
      "grad_norm": 0.2881927788257599,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6054,
      "step": 41505
    },
    {
      "epoch": 0.8500399320131878,
      "grad_norm": 0.28182360529899597,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.597,
      "step": 41510
    },
    {
      "epoch": 0.8501423217905925,
      "grad_norm": 0.29651278257369995,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5848,
      "step": 41515
    },
    {
      "epoch": 0.8502447115679971,
      "grad_norm": 0.2813228368759155,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5891,
      "step": 41520
    },
    {
      "epoch": 0.8503471013454017,
      "grad_norm": 0.27777281403541565,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6135,
      "step": 41525
    },
    {
      "epoch": 0.8504494911228063,
      "grad_norm": 0.26647210121154785,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6196,
      "step": 41530
    },
    {
      "epoch": 0.8505518809002109,
      "grad_norm": 0.2821933627128601,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5883,
      "step": 41535
    },
    {
      "epoch": 0.8506542706776156,
      "grad_norm": 0.2846284806728363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5561,
      "step": 41540
    },
    {
      "epoch": 0.8507566604550202,
      "grad_norm": 0.2869037985801697,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6337,
      "step": 41545
    },
    {
      "epoch": 0.8508590502324248,
      "grad_norm": 0.2855621874332428,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5948,
      "step": 41550
    },
    {
      "epoch": 0.8509614400098294,
      "grad_norm": 0.2860376834869385,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5101,
      "step": 41555
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 0.3692401051521301,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5895,
      "step": 41560
    },
    {
      "epoch": 0.8511662195646387,
      "grad_norm": 0.29354408383369446,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5458,
      "step": 41565
    },
    {
      "epoch": 0.8512686093420433,
      "grad_norm": 0.2744632065296173,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6093,
      "step": 41570
    },
    {
      "epoch": 0.8513709991194479,
      "grad_norm": 0.27942997217178345,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5687,
      "step": 41575
    },
    {
      "epoch": 0.8514733888968525,
      "grad_norm": 0.3143038749694824,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6032,
      "step": 41580
    },
    {
      "epoch": 0.8515757786742572,
      "grad_norm": 0.2794727087020874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5768,
      "step": 41585
    },
    {
      "epoch": 0.8516781684516618,
      "grad_norm": 0.2896329164505005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5505,
      "step": 41590
    },
    {
      "epoch": 0.8517805582290664,
      "grad_norm": 0.29437339305877686,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 41595
    },
    {
      "epoch": 0.851882948006471,
      "grad_norm": 0.30380910634994507,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4836,
      "step": 41600
    },
    {
      "epoch": 0.8519853377838756,
      "grad_norm": 0.28047603368759155,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6065,
      "step": 41605
    },
    {
      "epoch": 0.8520877275612803,
      "grad_norm": 0.29028409719467163,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5723,
      "step": 41610
    },
    {
      "epoch": 0.8521901173386849,
      "grad_norm": 0.27676650881767273,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6149,
      "step": 41615
    },
    {
      "epoch": 0.8522925071160895,
      "grad_norm": 0.28937408328056335,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5606,
      "step": 41620
    },
    {
      "epoch": 0.8523948968934941,
      "grad_norm": 0.2824362516403198,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5846,
      "step": 41625
    },
    {
      "epoch": 0.8524972866708987,
      "grad_norm": 0.2866837978363037,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6187,
      "step": 41630
    },
    {
      "epoch": 0.8525996764483034,
      "grad_norm": 0.28847113251686096,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5751,
      "step": 41635
    },
    {
      "epoch": 0.852702066225708,
      "grad_norm": 0.28669634461402893,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5576,
      "step": 41640
    },
    {
      "epoch": 0.8528044560031126,
      "grad_norm": 0.2918095290660858,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5455,
      "step": 41645
    },
    {
      "epoch": 0.8529068457805172,
      "grad_norm": 0.27304890751838684,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6047,
      "step": 41650
    },
    {
      "epoch": 0.8530092355579219,
      "grad_norm": 0.2830185294151306,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5711,
      "step": 41655
    },
    {
      "epoch": 0.8531116253353265,
      "grad_norm": 0.2840002775192261,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.548,
      "step": 41660
    },
    {
      "epoch": 0.8532140151127311,
      "grad_norm": 0.29644858837127686,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6533,
      "step": 41665
    },
    {
      "epoch": 0.8533164048901357,
      "grad_norm": 0.27836039662361145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5634,
      "step": 41670
    },
    {
      "epoch": 0.8534187946675404,
      "grad_norm": 0.28981637954711914,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5704,
      "step": 41675
    },
    {
      "epoch": 0.8535211844449451,
      "grad_norm": 0.2824649512767792,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5704,
      "step": 41680
    },
    {
      "epoch": 0.8536235742223497,
      "grad_norm": 0.27254143357276917,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.528,
      "step": 41685
    },
    {
      "epoch": 0.8537259639997543,
      "grad_norm": 0.32417652010917664,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.596,
      "step": 41690
    },
    {
      "epoch": 0.8538283537771589,
      "grad_norm": 0.27936333417892456,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5343,
      "step": 41695
    },
    {
      "epoch": 0.8539307435545636,
      "grad_norm": 0.302987664937973,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5512,
      "step": 41700
    },
    {
      "epoch": 0.8540331333319682,
      "grad_norm": 0.28835564851760864,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6041,
      "step": 41705
    },
    {
      "epoch": 0.8541355231093728,
      "grad_norm": 0.29047226905822754,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5264,
      "step": 41710
    },
    {
      "epoch": 0.8542379128867774,
      "grad_norm": 0.2981569170951843,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5758,
      "step": 41715
    },
    {
      "epoch": 0.854340302664182,
      "grad_norm": 0.330504834651947,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5811,
      "step": 41720
    },
    {
      "epoch": 0.8544426924415867,
      "grad_norm": 0.28930309414863586,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5309,
      "step": 41725
    },
    {
      "epoch": 0.8545450822189913,
      "grad_norm": 0.2907178997993469,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5726,
      "step": 41730
    },
    {
      "epoch": 0.8546474719963959,
      "grad_norm": 0.3091176152229309,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5667,
      "step": 41735
    },
    {
      "epoch": 0.8547498617738005,
      "grad_norm": 0.8943814635276794,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5795,
      "step": 41740
    },
    {
      "epoch": 0.8548522515512051,
      "grad_norm": 0.28994032740592957,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5899,
      "step": 41745
    },
    {
      "epoch": 0.8549546413286098,
      "grad_norm": 0.28370314836502075,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5931,
      "step": 41750
    },
    {
      "epoch": 0.8550570311060144,
      "grad_norm": 0.31428468227386475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6194,
      "step": 41755
    },
    {
      "epoch": 0.855159420883419,
      "grad_norm": 0.28025707602500916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6054,
      "step": 41760
    },
    {
      "epoch": 0.8552618106608236,
      "grad_norm": 0.2866079807281494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6121,
      "step": 41765
    },
    {
      "epoch": 0.8553642004382283,
      "grad_norm": 0.29860225319862366,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5499,
      "step": 41770
    },
    {
      "epoch": 0.8554665902156329,
      "grad_norm": 0.2892419695854187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5718,
      "step": 41775
    },
    {
      "epoch": 0.8555689799930375,
      "grad_norm": 0.28532499074935913,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5262,
      "step": 41780
    },
    {
      "epoch": 0.8556713697704421,
      "grad_norm": 0.2803010940551758,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5912,
      "step": 41785
    },
    {
      "epoch": 0.8557737595478467,
      "grad_norm": 0.2792489230632782,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5509,
      "step": 41790
    },
    {
      "epoch": 0.8558761493252514,
      "grad_norm": 0.2824473977088928,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5703,
      "step": 41795
    },
    {
      "epoch": 0.855978539102656,
      "grad_norm": 0.28354400396347046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5343,
      "step": 41800
    },
    {
      "epoch": 0.8560809288800606,
      "grad_norm": 0.2932973802089691,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5303,
      "step": 41805
    },
    {
      "epoch": 0.8561833186574652,
      "grad_norm": 0.2828560471534729,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5479,
      "step": 41810
    },
    {
      "epoch": 0.8562857084348698,
      "grad_norm": 0.2938644587993622,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5818,
      "step": 41815
    },
    {
      "epoch": 0.8563880982122745,
      "grad_norm": 0.2887813448905945,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5911,
      "step": 41820
    },
    {
      "epoch": 0.8564904879896791,
      "grad_norm": 0.29226839542388916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6075,
      "step": 41825
    },
    {
      "epoch": 0.8565928777670837,
      "grad_norm": 0.2809842824935913,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5584,
      "step": 41830
    },
    {
      "epoch": 0.8566952675444883,
      "grad_norm": 0.28355011343955994,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5645,
      "step": 41835
    },
    {
      "epoch": 0.856797657321893,
      "grad_norm": 0.2816533148288727,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5817,
      "step": 41840
    },
    {
      "epoch": 0.8569000470992976,
      "grad_norm": 0.2742515206336975,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5904,
      "step": 41845
    },
    {
      "epoch": 0.8570024368767022,
      "grad_norm": 0.27726268768310547,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5249,
      "step": 41850
    },
    {
      "epoch": 0.8571048266541068,
      "grad_norm": 0.2813457250595093,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5373,
      "step": 41855
    },
    {
      "epoch": 0.8572072164315114,
      "grad_norm": 0.32697102427482605,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5479,
      "step": 41860
    },
    {
      "epoch": 0.857309606208916,
      "grad_norm": 0.29770803451538086,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5563,
      "step": 41865
    },
    {
      "epoch": 0.8574119959863207,
      "grad_norm": 0.27651724219322205,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5707,
      "step": 41870
    },
    {
      "epoch": 0.8575143857637253,
      "grad_norm": 0.28034499287605286,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5569,
      "step": 41875
    },
    {
      "epoch": 0.85761677554113,
      "grad_norm": 0.2771777808666229,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5803,
      "step": 41880
    },
    {
      "epoch": 0.8577191653185346,
      "grad_norm": 0.2932230532169342,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5385,
      "step": 41885
    },
    {
      "epoch": 0.8578215550959393,
      "grad_norm": 0.32938718795776367,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6359,
      "step": 41890
    },
    {
      "epoch": 0.8579239448733439,
      "grad_norm": 0.28788164258003235,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5867,
      "step": 41895
    },
    {
      "epoch": 0.8580263346507485,
      "grad_norm": 0.3202422559261322,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5261,
      "step": 41900
    },
    {
      "epoch": 0.8581287244281531,
      "grad_norm": 0.286275178194046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5838,
      "step": 41905
    },
    {
      "epoch": 0.8582311142055578,
      "grad_norm": 0.29635751247406006,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5943,
      "step": 41910
    },
    {
      "epoch": 0.8583335039829624,
      "grad_norm": 0.29215869307518005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6463,
      "step": 41915
    },
    {
      "epoch": 0.858435893760367,
      "grad_norm": 0.28455254435539246,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5548,
      "step": 41920
    },
    {
      "epoch": 0.8585382835377716,
      "grad_norm": 0.2889324724674225,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5618,
      "step": 41925
    },
    {
      "epoch": 0.8586406733151762,
      "grad_norm": 0.2818892002105713,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5916,
      "step": 41930
    },
    {
      "epoch": 0.8587430630925809,
      "grad_norm": 0.2804669737815857,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6042,
      "step": 41935
    },
    {
      "epoch": 0.8588454528699855,
      "grad_norm": 0.28407761454582214,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5983,
      "step": 41940
    },
    {
      "epoch": 0.8589478426473901,
      "grad_norm": 0.30998361110687256,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5394,
      "step": 41945
    },
    {
      "epoch": 0.8590502324247947,
      "grad_norm": 0.2851797342300415,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5569,
      "step": 41950
    },
    {
      "epoch": 0.8591526222021993,
      "grad_norm": 0.4206336438655853,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5316,
      "step": 41955
    },
    {
      "epoch": 0.859255011979604,
      "grad_norm": 0.29635483026504517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5399,
      "step": 41960
    },
    {
      "epoch": 0.8593574017570086,
      "grad_norm": 0.2879798114299774,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5464,
      "step": 41965
    },
    {
      "epoch": 0.8594597915344132,
      "grad_norm": 0.3061964511871338,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5961,
      "step": 41970
    },
    {
      "epoch": 0.8595621813118178,
      "grad_norm": 0.29285728931427,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5452,
      "step": 41975
    },
    {
      "epoch": 0.8596645710892225,
      "grad_norm": 0.2791544198989868,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.59,
      "step": 41980
    },
    {
      "epoch": 0.8597669608666271,
      "grad_norm": 0.28416672348976135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6065,
      "step": 41985
    },
    {
      "epoch": 0.8598693506440317,
      "grad_norm": 0.28662338852882385,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5532,
      "step": 41990
    },
    {
      "epoch": 0.8599717404214363,
      "grad_norm": 0.29011231660842896,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6009,
      "step": 41995
    },
    {
      "epoch": 0.8600741301988409,
      "grad_norm": 0.29358547925949097,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5478,
      "step": 42000
    },
    {
      "epoch": 0.8600741301988409,
      "eval_loss": 4.590585231781006,
      "eval_runtime": 17.1325,
      "eval_samples_per_second": 33.971,
      "eval_steps_per_second": 4.261,
      "step": 42000
    },
    {
      "epoch": 0.8601765199762456,
      "grad_norm": 0.2929130792617798,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5863,
      "step": 42005
    },
    {
      "epoch": 0.8602789097536502,
      "grad_norm": 0.29090362787246704,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5614,
      "step": 42010
    },
    {
      "epoch": 0.8603812995310548,
      "grad_norm": 0.27907586097717285,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5621,
      "step": 42015
    },
    {
      "epoch": 0.8604836893084594,
      "grad_norm": 0.282423734664917,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.549,
      "step": 42020
    },
    {
      "epoch": 0.860586079085864,
      "grad_norm": 0.28453391790390015,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.536,
      "step": 42025
    },
    {
      "epoch": 0.8606884688632687,
      "grad_norm": 0.3082663416862488,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5478,
      "step": 42030
    },
    {
      "epoch": 0.8607908586406733,
      "grad_norm": 0.2766706049442291,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5169,
      "step": 42035
    },
    {
      "epoch": 0.8608932484180779,
      "grad_norm": 0.28100666403770447,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.543,
      "step": 42040
    },
    {
      "epoch": 0.8609956381954825,
      "grad_norm": 0.30033186078071594,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6223,
      "step": 42045
    },
    {
      "epoch": 0.8610980279728871,
      "grad_norm": 0.2801840603351593,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5235,
      "step": 42050
    },
    {
      "epoch": 0.8612004177502918,
      "grad_norm": 0.27475932240486145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.581,
      "step": 42055
    },
    {
      "epoch": 0.8613028075276964,
      "grad_norm": 0.28085681796073914,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5854,
      "step": 42060
    },
    {
      "epoch": 0.861405197305101,
      "grad_norm": 0.2858322262763977,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6135,
      "step": 42065
    },
    {
      "epoch": 0.8615075870825056,
      "grad_norm": 0.29247337579727173,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5652,
      "step": 42070
    },
    {
      "epoch": 0.8616099768599103,
      "grad_norm": 0.28009599447250366,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5821,
      "step": 42075
    },
    {
      "epoch": 0.861712366637315,
      "grad_norm": 0.2810896337032318,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.578,
      "step": 42080
    },
    {
      "epoch": 0.8618147564147196,
      "grad_norm": 0.3237174153327942,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5869,
      "step": 42085
    },
    {
      "epoch": 0.8619171461921242,
      "grad_norm": 0.2805618643760681,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6044,
      "step": 42090
    },
    {
      "epoch": 0.8620195359695288,
      "grad_norm": 0.2970050871372223,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5761,
      "step": 42095
    },
    {
      "epoch": 0.8621219257469335,
      "grad_norm": 0.29210126399993896,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5839,
      "step": 42100
    },
    {
      "epoch": 0.8622243155243381,
      "grad_norm": 0.27998438477516174,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5538,
      "step": 42105
    },
    {
      "epoch": 0.8623267053017427,
      "grad_norm": 0.2854512333869934,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6154,
      "step": 42110
    },
    {
      "epoch": 0.8624290950791473,
      "grad_norm": 0.36673498153686523,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5629,
      "step": 42115
    },
    {
      "epoch": 0.862531484856552,
      "grad_norm": 0.4028867781162262,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6183,
      "step": 42120
    },
    {
      "epoch": 0.8626338746339566,
      "grad_norm": 0.29421451687812805,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5932,
      "step": 42125
    },
    {
      "epoch": 0.8627362644113612,
      "grad_norm": 0.27536529302597046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5751,
      "step": 42130
    },
    {
      "epoch": 0.8628386541887658,
      "grad_norm": 0.28098100423812866,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6018,
      "step": 42135
    },
    {
      "epoch": 0.8629410439661704,
      "grad_norm": 0.27294838428497314,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6355,
      "step": 42140
    },
    {
      "epoch": 0.8630434337435751,
      "grad_norm": 0.273320734500885,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.595,
      "step": 42145
    },
    {
      "epoch": 0.8631458235209797,
      "grad_norm": 0.31179067492485046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5569,
      "step": 42150
    },
    {
      "epoch": 0.8632482132983843,
      "grad_norm": 0.2889743745326996,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4923,
      "step": 42155
    },
    {
      "epoch": 0.8633506030757889,
      "grad_norm": 0.27962151169776917,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5543,
      "step": 42160
    },
    {
      "epoch": 0.8634529928531935,
      "grad_norm": 0.2781238555908203,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5395,
      "step": 42165
    },
    {
      "epoch": 0.8635553826305982,
      "grad_norm": 0.27796876430511475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5305,
      "step": 42170
    },
    {
      "epoch": 0.8636577724080028,
      "grad_norm": 0.2779185473918915,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.572,
      "step": 42175
    },
    {
      "epoch": 0.8637601621854074,
      "grad_norm": 0.3074818253517151,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5777,
      "step": 42180
    },
    {
      "epoch": 0.863862551962812,
      "grad_norm": 0.2755137085914612,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5879,
      "step": 42185
    },
    {
      "epoch": 0.8639649417402167,
      "grad_norm": 0.3061232566833496,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.57,
      "step": 42190
    },
    {
      "epoch": 0.8640673315176213,
      "grad_norm": 0.27670255303382874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6003,
      "step": 42195
    },
    {
      "epoch": 0.8641697212950259,
      "grad_norm": 0.32295048236846924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.578,
      "step": 42200
    },
    {
      "epoch": 0.8642721110724305,
      "grad_norm": 0.2795508801937103,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5444,
      "step": 42205
    },
    {
      "epoch": 0.8643745008498351,
      "grad_norm": 0.28401806950569153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5526,
      "step": 42210
    },
    {
      "epoch": 0.8644768906272398,
      "grad_norm": 0.29551583528518677,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5639,
      "step": 42215
    },
    {
      "epoch": 0.8645792804046444,
      "grad_norm": 0.2854187786579132,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6054,
      "step": 42220
    },
    {
      "epoch": 0.864681670182049,
      "grad_norm": 0.2859326899051666,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 42225
    },
    {
      "epoch": 0.8647840599594536,
      "grad_norm": 0.2843199372291565,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5481,
      "step": 42230
    },
    {
      "epoch": 0.8648864497368582,
      "grad_norm": 0.2788691818714142,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6117,
      "step": 42235
    },
    {
      "epoch": 0.8649888395142629,
      "grad_norm": 0.29678836464881897,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5059,
      "step": 42240
    },
    {
      "epoch": 0.8650912292916675,
      "grad_norm": 0.277492880821228,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5295,
      "step": 42245
    },
    {
      "epoch": 0.8651936190690721,
      "grad_norm": 0.2821527421474457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5915,
      "step": 42250
    },
    {
      "epoch": 0.8652960088464767,
      "grad_norm": 0.28887367248535156,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5833,
      "step": 42255
    },
    {
      "epoch": 0.8653983986238813,
      "grad_norm": 0.28567564487457275,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5263,
      "step": 42260
    },
    {
      "epoch": 0.865500788401286,
      "grad_norm": 0.2992493212223053,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5533,
      "step": 42265
    },
    {
      "epoch": 0.8656031781786906,
      "grad_norm": 0.28368014097213745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6059,
      "step": 42270
    },
    {
      "epoch": 0.8657055679560952,
      "grad_norm": 0.27183762192726135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5358,
      "step": 42275
    },
    {
      "epoch": 0.8658079577334998,
      "grad_norm": 0.565661609172821,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5022,
      "step": 42280
    },
    {
      "epoch": 0.8659103475109046,
      "grad_norm": 0.28837451338768005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6116,
      "step": 42285
    },
    {
      "epoch": 0.8660127372883092,
      "grad_norm": 0.2835419476032257,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4694,
      "step": 42290
    },
    {
      "epoch": 0.8661151270657138,
      "grad_norm": 1.099507451057434,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6911,
      "step": 42295
    },
    {
      "epoch": 0.8662175168431184,
      "grad_norm": 0.2809179425239563,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5912,
      "step": 42300
    },
    {
      "epoch": 0.866319906620523,
      "grad_norm": 0.2871781587600708,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5526,
      "step": 42305
    },
    {
      "epoch": 0.8664222963979277,
      "grad_norm": 0.3089900016784668,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5706,
      "step": 42310
    },
    {
      "epoch": 0.8665246861753323,
      "grad_norm": 0.2921340763568878,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5614,
      "step": 42315
    },
    {
      "epoch": 0.8666270759527369,
      "grad_norm": 0.28606513142585754,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5368,
      "step": 42320
    },
    {
      "epoch": 0.8667294657301415,
      "grad_norm": 0.2628849744796753,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5403,
      "step": 42325
    },
    {
      "epoch": 0.8668318555075462,
      "grad_norm": 0.2877076268196106,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5371,
      "step": 42330
    },
    {
      "epoch": 0.8669342452849508,
      "grad_norm": 0.2788994014263153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5858,
      "step": 42335
    },
    {
      "epoch": 0.8670366350623554,
      "grad_norm": 0.27799710631370544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.566,
      "step": 42340
    },
    {
      "epoch": 0.86713902483976,
      "grad_norm": 0.4452689588069916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6125,
      "step": 42345
    },
    {
      "epoch": 0.8672414146171646,
      "grad_norm": 0.30955836176872253,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.555,
      "step": 42350
    },
    {
      "epoch": 0.8673438043945693,
      "grad_norm": 0.30161789059638977,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6057,
      "step": 42355
    },
    {
      "epoch": 0.8674461941719739,
      "grad_norm": 0.3116542100906372,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.545,
      "step": 42360
    },
    {
      "epoch": 0.8675485839493785,
      "grad_norm": 0.283415287733078,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.556,
      "step": 42365
    },
    {
      "epoch": 0.8676509737267831,
      "grad_norm": 0.3659295439720154,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5269,
      "step": 42370
    },
    {
      "epoch": 0.8677533635041877,
      "grad_norm": 0.3008820116519928,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5487,
      "step": 42375
    },
    {
      "epoch": 0.8678557532815924,
      "grad_norm": 0.29011234641075134,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6369,
      "step": 42380
    },
    {
      "epoch": 0.867958143058997,
      "grad_norm": 0.29220259189605713,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6019,
      "step": 42385
    },
    {
      "epoch": 0.8680605328364016,
      "grad_norm": 0.33202919363975525,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5812,
      "step": 42390
    },
    {
      "epoch": 0.8681629226138062,
      "grad_norm": 0.28137901425361633,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5701,
      "step": 42395
    },
    {
      "epoch": 0.8682653123912109,
      "grad_norm": 0.2826639711856842,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6094,
      "step": 42400
    },
    {
      "epoch": 0.8683677021686155,
      "grad_norm": 0.27954402565956116,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5638,
      "step": 42405
    },
    {
      "epoch": 0.8684700919460201,
      "grad_norm": 0.29488644003868103,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5684,
      "step": 42410
    },
    {
      "epoch": 0.8685724817234247,
      "grad_norm": 0.30218660831451416,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5535,
      "step": 42415
    },
    {
      "epoch": 0.8686748715008293,
      "grad_norm": 0.28542274236679077,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5591,
      "step": 42420
    },
    {
      "epoch": 0.868777261278234,
      "grad_norm": 0.289728045463562,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5724,
      "step": 42425
    },
    {
      "epoch": 0.8688796510556386,
      "grad_norm": 0.2904987037181854,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5675,
      "step": 42430
    },
    {
      "epoch": 0.8689820408330432,
      "grad_norm": 0.28901374340057373,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5979,
      "step": 42435
    },
    {
      "epoch": 0.8690844306104478,
      "grad_norm": 0.30808329582214355,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5183,
      "step": 42440
    },
    {
      "epoch": 0.8691868203878524,
      "grad_norm": 0.2810181975364685,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5318,
      "step": 42445
    },
    {
      "epoch": 0.8692892101652571,
      "grad_norm": 0.28387823700904846,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5458,
      "step": 42450
    },
    {
      "epoch": 0.8693915999426617,
      "grad_norm": 0.2959236800670624,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5628,
      "step": 42455
    },
    {
      "epoch": 0.8694939897200663,
      "grad_norm": 0.33226150274276733,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5292,
      "step": 42460
    },
    {
      "epoch": 0.8695963794974709,
      "grad_norm": 0.2962782382965088,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5998,
      "step": 42465
    },
    {
      "epoch": 0.8696987692748755,
      "grad_norm": 0.3628597557544708,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.633,
      "step": 42470
    },
    {
      "epoch": 0.8698011590522802,
      "grad_norm": 0.30368393659591675,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5538,
      "step": 42475
    },
    {
      "epoch": 0.8699035488296848,
      "grad_norm": 0.29624828696250916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5782,
      "step": 42480
    },
    {
      "epoch": 0.8700059386070894,
      "grad_norm": 0.2826397716999054,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5451,
      "step": 42485
    },
    {
      "epoch": 0.8701083283844941,
      "grad_norm": 0.2871544063091278,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5634,
      "step": 42490
    },
    {
      "epoch": 0.8702107181618988,
      "grad_norm": 0.28926998376846313,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5313,
      "step": 42495
    },
    {
      "epoch": 0.8703131079393034,
      "grad_norm": 0.2764125466346741,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5853,
      "step": 42500
    },
    {
      "epoch": 0.870415497716708,
      "grad_norm": 0.2877795398235321,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5678,
      "step": 42505
    },
    {
      "epoch": 0.8705178874941126,
      "grad_norm": 0.3008129298686981,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5419,
      "step": 42510
    },
    {
      "epoch": 0.8706202772715173,
      "grad_norm": 0.2770102322101593,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5813,
      "step": 42515
    },
    {
      "epoch": 0.8707226670489219,
      "grad_norm": 0.2859014570713043,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6064,
      "step": 42520
    },
    {
      "epoch": 0.8708250568263265,
      "grad_norm": 0.3001577854156494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5527,
      "step": 42525
    },
    {
      "epoch": 0.8709274466037311,
      "grad_norm": 0.28082939982414246,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5822,
      "step": 42530
    },
    {
      "epoch": 0.8710298363811357,
      "grad_norm": 0.2857016921043396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5967,
      "step": 42535
    },
    {
      "epoch": 0.8711322261585404,
      "grad_norm": 0.6677747368812561,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5275,
      "step": 42540
    },
    {
      "epoch": 0.871234615935945,
      "grad_norm": 0.29151827096939087,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5613,
      "step": 42545
    },
    {
      "epoch": 0.8713370057133496,
      "grad_norm": 0.29731205105781555,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5673,
      "step": 42550
    },
    {
      "epoch": 0.8714393954907542,
      "grad_norm": 0.4542264938354492,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5017,
      "step": 42555
    },
    {
      "epoch": 0.8715417852681588,
      "grad_norm": 0.33812379837036133,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4928,
      "step": 42560
    },
    {
      "epoch": 0.8716441750455635,
      "grad_norm": 0.30377838015556335,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5837,
      "step": 42565
    },
    {
      "epoch": 0.8717465648229681,
      "grad_norm": 0.2798381447792053,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6044,
      "step": 42570
    },
    {
      "epoch": 0.8718489546003727,
      "grad_norm": 0.2912675440311432,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5703,
      "step": 42575
    },
    {
      "epoch": 0.8719513443777773,
      "grad_norm": 0.2935977578163147,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5622,
      "step": 42580
    },
    {
      "epoch": 0.872053734155182,
      "grad_norm": 0.2936408519744873,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5041,
      "step": 42585
    },
    {
      "epoch": 0.8721561239325866,
      "grad_norm": 0.39489099383354187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5734,
      "step": 42590
    },
    {
      "epoch": 0.8722585137099912,
      "grad_norm": 0.2872966229915619,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 42595
    },
    {
      "epoch": 0.8723609034873958,
      "grad_norm": 0.29036393761634827,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5485,
      "step": 42600
    },
    {
      "epoch": 0.8724632932648004,
      "grad_norm": 0.28508612513542175,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5078,
      "step": 42605
    },
    {
      "epoch": 0.872565683042205,
      "grad_norm": 0.3030988276004791,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4962,
      "step": 42610
    },
    {
      "epoch": 0.8726680728196097,
      "grad_norm": 0.2963111102581024,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5934,
      "step": 42615
    },
    {
      "epoch": 0.8727704625970143,
      "grad_norm": 0.28452345728874207,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5454,
      "step": 42620
    },
    {
      "epoch": 0.8728728523744189,
      "grad_norm": 0.29631686210632324,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 42625
    },
    {
      "epoch": 0.8729752421518235,
      "grad_norm": 0.2943074703216553,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5386,
      "step": 42630
    },
    {
      "epoch": 0.8730776319292282,
      "grad_norm": 0.28320398926734924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.553,
      "step": 42635
    },
    {
      "epoch": 0.8731800217066328,
      "grad_norm": 0.29807671904563904,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5019,
      "step": 42640
    },
    {
      "epoch": 0.8732824114840374,
      "grad_norm": 0.2764509916305542,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5759,
      "step": 42645
    },
    {
      "epoch": 0.873384801261442,
      "grad_norm": 0.30844801664352417,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6115,
      "step": 42650
    },
    {
      "epoch": 0.8734871910388466,
      "grad_norm": 0.2824011445045471,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6278,
      "step": 42655
    },
    {
      "epoch": 0.8735895808162513,
      "grad_norm": 0.2877044677734375,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5788,
      "step": 42660
    },
    {
      "epoch": 0.8736919705936559,
      "grad_norm": 0.27674928307533264,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5951,
      "step": 42665
    },
    {
      "epoch": 0.8737943603710605,
      "grad_norm": 0.2907717823982239,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4565,
      "step": 42670
    },
    {
      "epoch": 0.8738967501484651,
      "grad_norm": 0.3075560927391052,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5205,
      "step": 42675
    },
    {
      "epoch": 0.8739991399258698,
      "grad_norm": 0.29234230518341064,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5986,
      "step": 42680
    },
    {
      "epoch": 0.8741015297032744,
      "grad_norm": 0.2991107702255249,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6258,
      "step": 42685
    },
    {
      "epoch": 0.8742039194806791,
      "grad_norm": 0.2882688343524933,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5573,
      "step": 42690
    },
    {
      "epoch": 0.8743063092580837,
      "grad_norm": 0.286185622215271,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5586,
      "step": 42695
    },
    {
      "epoch": 0.8744086990354883,
      "grad_norm": 0.3102838397026062,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.62,
      "step": 42700
    },
    {
      "epoch": 0.874511088812893,
      "grad_norm": 0.2933298945426941,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6344,
      "step": 42705
    },
    {
      "epoch": 0.8746134785902976,
      "grad_norm": 0.3276117146015167,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5119,
      "step": 42710
    },
    {
      "epoch": 0.8747158683677022,
      "grad_norm": 0.33533158898353577,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5592,
      "step": 42715
    },
    {
      "epoch": 0.8748182581451068,
      "grad_norm": 0.28675273060798645,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6168,
      "step": 42720
    },
    {
      "epoch": 0.8749206479225115,
      "grad_norm": 0.279441237449646,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5934,
      "step": 42725
    },
    {
      "epoch": 0.8750230376999161,
      "grad_norm": 0.3046245574951172,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5823,
      "step": 42730
    },
    {
      "epoch": 0.8751254274773207,
      "grad_norm": 0.30070775747299194,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5309,
      "step": 42735
    },
    {
      "epoch": 0.8752278172547253,
      "grad_norm": 0.27930042147636414,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5464,
      "step": 42740
    },
    {
      "epoch": 0.8753302070321299,
      "grad_norm": 0.2897210121154785,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6005,
      "step": 42745
    },
    {
      "epoch": 0.8754325968095346,
      "grad_norm": 0.3091546893119812,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5816,
      "step": 42750
    },
    {
      "epoch": 0.8755349865869392,
      "grad_norm": 0.3109922409057617,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5946,
      "step": 42755
    },
    {
      "epoch": 0.8756373763643438,
      "grad_norm": 0.29840153455734253,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6148,
      "step": 42760
    },
    {
      "epoch": 0.8757397661417484,
      "grad_norm": 0.2970777750015259,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5784,
      "step": 42765
    },
    {
      "epoch": 0.875842155919153,
      "grad_norm": 0.2896730899810791,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5952,
      "step": 42770
    },
    {
      "epoch": 0.8759445456965577,
      "grad_norm": 0.29766690731048584,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5608,
      "step": 42775
    },
    {
      "epoch": 0.8760469354739623,
      "grad_norm": 0.2844490110874176,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5695,
      "step": 42780
    },
    {
      "epoch": 0.8761493252513669,
      "grad_norm": 0.2948978543281555,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.573,
      "step": 42785
    },
    {
      "epoch": 0.8762517150287715,
      "grad_norm": 0.3014620840549469,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6025,
      "step": 42790
    },
    {
      "epoch": 0.8763541048061761,
      "grad_norm": 0.2828783094882965,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5536,
      "step": 42795
    },
    {
      "epoch": 0.8764564945835808,
      "grad_norm": 0.2940693795681,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5767,
      "step": 42800
    },
    {
      "epoch": 0.8765588843609854,
      "grad_norm": 0.29585936665534973,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5989,
      "step": 42805
    },
    {
      "epoch": 0.87666127413839,
      "grad_norm": 0.2852897346019745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5354,
      "step": 42810
    },
    {
      "epoch": 0.8767636639157946,
      "grad_norm": 0.28615662455558777,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5227,
      "step": 42815
    },
    {
      "epoch": 0.8768660536931993,
      "grad_norm": 0.29597118496894836,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5613,
      "step": 42820
    },
    {
      "epoch": 0.8769684434706039,
      "grad_norm": 0.28099918365478516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5753,
      "step": 42825
    },
    {
      "epoch": 0.8770708332480085,
      "grad_norm": 0.29338228702545166,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5536,
      "step": 42830
    },
    {
      "epoch": 0.8771732230254131,
      "grad_norm": 0.28326815366744995,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 42835
    },
    {
      "epoch": 0.8772756128028177,
      "grad_norm": 0.2854740023612976,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6029,
      "step": 42840
    },
    {
      "epoch": 0.8773780025802224,
      "grad_norm": 0.2841622531414032,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.595,
      "step": 42845
    },
    {
      "epoch": 0.877480392357627,
      "grad_norm": 0.28582340478897095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5594,
      "step": 42850
    },
    {
      "epoch": 0.8775827821350316,
      "grad_norm": 0.3178538382053375,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.58,
      "step": 42855
    },
    {
      "epoch": 0.8776851719124362,
      "grad_norm": 0.28321757912635803,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6036,
      "step": 42860
    },
    {
      "epoch": 0.8777875616898408,
      "grad_norm": 0.28525102138519287,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5574,
      "step": 42865
    },
    {
      "epoch": 0.8778899514672455,
      "grad_norm": 0.29060178995132446,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5759,
      "step": 42870
    },
    {
      "epoch": 0.8779923412446501,
      "grad_norm": 0.30368462204933167,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5436,
      "step": 42875
    },
    {
      "epoch": 0.8780947310220547,
      "grad_norm": 0.30066636204719543,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.559,
      "step": 42880
    },
    {
      "epoch": 0.8781971207994593,
      "grad_norm": 0.41088446974754333,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5113,
      "step": 42885
    },
    {
      "epoch": 0.878299510576864,
      "grad_norm": 0.3057084083557129,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5543,
      "step": 42890
    },
    {
      "epoch": 0.8784019003542687,
      "grad_norm": 0.2873361110687256,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5912,
      "step": 42895
    },
    {
      "epoch": 0.8785042901316733,
      "grad_norm": 0.2893359065055847,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6089,
      "step": 42900
    },
    {
      "epoch": 0.8786066799090779,
      "grad_norm": 0.30454859137535095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5792,
      "step": 42905
    },
    {
      "epoch": 0.8787090696864825,
      "grad_norm": 0.3234281539916992,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6122,
      "step": 42910
    },
    {
      "epoch": 0.8788114594638872,
      "grad_norm": 0.307363897562027,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5924,
      "step": 42915
    },
    {
      "epoch": 0.8789138492412918,
      "grad_norm": 0.2866811752319336,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5603,
      "step": 42920
    },
    {
      "epoch": 0.8790162390186964,
      "grad_norm": 0.29870229959487915,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5697,
      "step": 42925
    },
    {
      "epoch": 0.879118628796101,
      "grad_norm": 0.29656752943992615,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5812,
      "step": 42930
    },
    {
      "epoch": 0.8792210185735057,
      "grad_norm": 0.30561700463294983,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5504,
      "step": 42935
    },
    {
      "epoch": 0.8793234083509103,
      "grad_norm": 0.32332366704940796,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5786,
      "step": 42940
    },
    {
      "epoch": 0.8794257981283149,
      "grad_norm": 0.28673917055130005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5428,
      "step": 42945
    },
    {
      "epoch": 0.8795281879057195,
      "grad_norm": 0.28213217854499817,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5288,
      "step": 42950
    },
    {
      "epoch": 0.8796305776831241,
      "grad_norm": 0.29143473505973816,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5779,
      "step": 42955
    },
    {
      "epoch": 0.8797329674605288,
      "grad_norm": 0.2868902087211609,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5497,
      "step": 42960
    },
    {
      "epoch": 0.8798353572379334,
      "grad_norm": 0.3348364233970642,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5077,
      "step": 42965
    },
    {
      "epoch": 0.879937747015338,
      "grad_norm": 0.27767911553382874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4875,
      "step": 42970
    },
    {
      "epoch": 0.8800401367927426,
      "grad_norm": 0.28155142068862915,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5819,
      "step": 42975
    },
    {
      "epoch": 0.8801425265701472,
      "grad_norm": 0.27674272656440735,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5774,
      "step": 42980
    },
    {
      "epoch": 0.8802449163475519,
      "grad_norm": 0.29031631350517273,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5194,
      "step": 42985
    },
    {
      "epoch": 0.8803473061249565,
      "grad_norm": 0.28039461374282837,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6035,
      "step": 42990
    },
    {
      "epoch": 0.8804496959023611,
      "grad_norm": 0.29909491539001465,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 42995
    },
    {
      "epoch": 0.8805520856797657,
      "grad_norm": 0.2924463748931885,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5262,
      "step": 43000
    },
    {
      "epoch": 0.8805520856797657,
      "eval_loss": 4.588317394256592,
      "eval_runtime": 17.1268,
      "eval_samples_per_second": 33.982,
      "eval_steps_per_second": 4.262,
      "step": 43000
    },
    {
      "epoch": 0.8806544754571703,
      "grad_norm": 0.29849857091903687,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5815,
      "step": 43005
    },
    {
      "epoch": 0.880756865234575,
      "grad_norm": 0.30793818831443787,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5941,
      "step": 43010
    },
    {
      "epoch": 0.8808592550119796,
      "grad_norm": 0.28552597761154175,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5661,
      "step": 43015
    },
    {
      "epoch": 0.8809616447893842,
      "grad_norm": 0.2882366180419922,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5673,
      "step": 43020
    },
    {
      "epoch": 0.8810640345667888,
      "grad_norm": 0.295576810836792,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.568,
      "step": 43025
    },
    {
      "epoch": 0.8811664243441935,
      "grad_norm": 0.28542160987854004,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5625,
      "step": 43030
    },
    {
      "epoch": 0.8812688141215981,
      "grad_norm": 0.8199988603591919,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6262,
      "step": 43035
    },
    {
      "epoch": 0.8813712038990027,
      "grad_norm": 0.30435624718666077,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5742,
      "step": 43040
    },
    {
      "epoch": 0.8814735936764073,
      "grad_norm": 0.29068076610565186,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5668,
      "step": 43045
    },
    {
      "epoch": 0.8815759834538119,
      "grad_norm": 0.3287653625011444,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5645,
      "step": 43050
    },
    {
      "epoch": 0.8816783732312166,
      "grad_norm": 0.2941044270992279,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5611,
      "step": 43055
    },
    {
      "epoch": 0.8817807630086212,
      "grad_norm": 0.27592143416404724,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5598,
      "step": 43060
    },
    {
      "epoch": 0.8818831527860258,
      "grad_norm": 0.304882675409317,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5648,
      "step": 43065
    },
    {
      "epoch": 0.8819855425634304,
      "grad_norm": 0.29764097929000854,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5748,
      "step": 43070
    },
    {
      "epoch": 0.882087932340835,
      "grad_norm": 0.28902390599250793,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 43075
    },
    {
      "epoch": 0.8821903221182397,
      "grad_norm": 0.27456074953079224,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.543,
      "step": 43080
    },
    {
      "epoch": 0.8822927118956443,
      "grad_norm": 0.2862781882286072,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5668,
      "step": 43085
    },
    {
      "epoch": 0.8823951016730489,
      "grad_norm": 0.27993398904800415,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6255,
      "step": 43090
    },
    {
      "epoch": 0.8824974914504536,
      "grad_norm": 0.29803958535194397,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5789,
      "step": 43095
    },
    {
      "epoch": 0.8825998812278583,
      "grad_norm": 0.28480470180511475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5901,
      "step": 43100
    },
    {
      "epoch": 0.8827022710052629,
      "grad_norm": 0.29084596037864685,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5822,
      "step": 43105
    },
    {
      "epoch": 0.8828046607826675,
      "grad_norm": 0.28622016310691833,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5038,
      "step": 43110
    },
    {
      "epoch": 0.8829070505600721,
      "grad_norm": 0.2887270450592041,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5189,
      "step": 43115
    },
    {
      "epoch": 0.8830094403374767,
      "grad_norm": 0.2953267991542816,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5724,
      "step": 43120
    },
    {
      "epoch": 0.8831118301148814,
      "grad_norm": 0.28226035833358765,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5794,
      "step": 43125
    },
    {
      "epoch": 0.883214219892286,
      "grad_norm": 0.29001089930534363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.7161,
      "step": 43130
    },
    {
      "epoch": 0.8833166096696906,
      "grad_norm": 0.3066718876361847,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5183,
      "step": 43135
    },
    {
      "epoch": 0.8834189994470952,
      "grad_norm": 0.2722647786140442,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5257,
      "step": 43140
    },
    {
      "epoch": 0.8835213892244999,
      "grad_norm": 0.3023661971092224,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5622,
      "step": 43145
    },
    {
      "epoch": 0.8836237790019045,
      "grad_norm": 0.347561776638031,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5337,
      "step": 43150
    },
    {
      "epoch": 0.8837261687793091,
      "grad_norm": 0.2886524200439453,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5653,
      "step": 43155
    },
    {
      "epoch": 0.8838285585567137,
      "grad_norm": 0.2908353805541992,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5652,
      "step": 43160
    },
    {
      "epoch": 0.8839309483341183,
      "grad_norm": 0.299456924200058,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5631,
      "step": 43165
    },
    {
      "epoch": 0.884033338111523,
      "grad_norm": 0.2918829321861267,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5485,
      "step": 43170
    },
    {
      "epoch": 0.8841357278889276,
      "grad_norm": 0.3255084455013275,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5075,
      "step": 43175
    },
    {
      "epoch": 0.8842381176663322,
      "grad_norm": 0.29269468784332275,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6193,
      "step": 43180
    },
    {
      "epoch": 0.8843405074437368,
      "grad_norm": 0.30426040291786194,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5508,
      "step": 43185
    },
    {
      "epoch": 0.8844428972211414,
      "grad_norm": 0.3047192394733429,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5607,
      "step": 43190
    },
    {
      "epoch": 0.8845452869985461,
      "grad_norm": 0.30153313279151917,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5032,
      "step": 43195
    },
    {
      "epoch": 0.8846476767759507,
      "grad_norm": 0.2783660292625427,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5909,
      "step": 43200
    },
    {
      "epoch": 0.8847500665533553,
      "grad_norm": 0.3032394349575043,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5839,
      "step": 43205
    },
    {
      "epoch": 0.8848524563307599,
      "grad_norm": 0.28079283237457275,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.596,
      "step": 43210
    },
    {
      "epoch": 0.8849548461081646,
      "grad_norm": 0.3117208778858185,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5706,
      "step": 43215
    },
    {
      "epoch": 0.8850572358855692,
      "grad_norm": 0.29995861649513245,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5058,
      "step": 43220
    },
    {
      "epoch": 0.8851596256629738,
      "grad_norm": 0.29243379831314087,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5553,
      "step": 43225
    },
    {
      "epoch": 0.8852620154403784,
      "grad_norm": 0.28652480244636536,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5708,
      "step": 43230
    },
    {
      "epoch": 0.885364405217783,
      "grad_norm": 0.2879696488380432,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5917,
      "step": 43235
    },
    {
      "epoch": 0.8854667949951877,
      "grad_norm": 0.3314764201641083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5605,
      "step": 43240
    },
    {
      "epoch": 0.8855691847725923,
      "grad_norm": 0.3095373213291168,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5641,
      "step": 43245
    },
    {
      "epoch": 0.8856715745499969,
      "grad_norm": 0.30055418610572815,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5265,
      "step": 43250
    },
    {
      "epoch": 0.8857739643274015,
      "grad_norm": 0.3154790997505188,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6022,
      "step": 43255
    },
    {
      "epoch": 0.8858763541048061,
      "grad_norm": 0.28868356347084045,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5839,
      "step": 43260
    },
    {
      "epoch": 0.8859787438822108,
      "grad_norm": 0.2998608648777008,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5308,
      "step": 43265
    },
    {
      "epoch": 0.8860811336596154,
      "grad_norm": 0.2835061550140381,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5232,
      "step": 43270
    },
    {
      "epoch": 0.88618352343702,
      "grad_norm": 0.2820570468902588,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6076,
      "step": 43275
    },
    {
      "epoch": 0.8862859132144246,
      "grad_norm": 0.28567567467689514,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6502,
      "step": 43280
    },
    {
      "epoch": 0.8863883029918292,
      "grad_norm": 0.2970583438873291,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5814,
      "step": 43285
    },
    {
      "epoch": 0.8864906927692339,
      "grad_norm": 0.2917497456073761,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5495,
      "step": 43290
    },
    {
      "epoch": 0.8865930825466385,
      "grad_norm": 0.30360355973243713,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5851,
      "step": 43295
    },
    {
      "epoch": 0.8866954723240432,
      "grad_norm": 0.2801843583583832,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5801,
      "step": 43300
    },
    {
      "epoch": 0.8867978621014478,
      "grad_norm": 0.3223794102668762,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.555,
      "step": 43305
    },
    {
      "epoch": 0.8869002518788525,
      "grad_norm": 0.3030499219894409,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6037,
      "step": 43310
    },
    {
      "epoch": 0.8870026416562571,
      "grad_norm": 0.2885530889034271,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5635,
      "step": 43315
    },
    {
      "epoch": 0.8871050314336617,
      "grad_norm": 0.3388378620147705,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5617,
      "step": 43320
    },
    {
      "epoch": 0.8872074212110663,
      "grad_norm": 0.28313079476356506,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5497,
      "step": 43325
    },
    {
      "epoch": 0.887309810988471,
      "grad_norm": 0.27532637119293213,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5404,
      "step": 43330
    },
    {
      "epoch": 0.8874122007658756,
      "grad_norm": 0.2831867039203644,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5259,
      "step": 43335
    },
    {
      "epoch": 0.8875145905432802,
      "grad_norm": 0.28463199734687805,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5789,
      "step": 43340
    },
    {
      "epoch": 0.8876169803206848,
      "grad_norm": 0.3134804666042328,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5373,
      "step": 43345
    },
    {
      "epoch": 0.8877193700980894,
      "grad_norm": 0.3044542074203491,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5625,
      "step": 43350
    },
    {
      "epoch": 0.887821759875494,
      "grad_norm": 0.2829276919364929,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5932,
      "step": 43355
    },
    {
      "epoch": 0.8879241496528987,
      "grad_norm": 0.2904761731624603,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5612,
      "step": 43360
    },
    {
      "epoch": 0.8880265394303033,
      "grad_norm": 0.28154832124710083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5256,
      "step": 43365
    },
    {
      "epoch": 0.8881289292077079,
      "grad_norm": 0.2937324345111847,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5595,
      "step": 43370
    },
    {
      "epoch": 0.8882313189851125,
      "grad_norm": 0.30040445923805237,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5505,
      "step": 43375
    },
    {
      "epoch": 0.8883337087625172,
      "grad_norm": 0.29687491059303284,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.622,
      "step": 43380
    },
    {
      "epoch": 0.8884360985399218,
      "grad_norm": 0.3083922863006592,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5899,
      "step": 43385
    },
    {
      "epoch": 0.8885384883173264,
      "grad_norm": 0.2865968942642212,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6159,
      "step": 43390
    },
    {
      "epoch": 0.888640878094731,
      "grad_norm": 0.2884815037250519,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5902,
      "step": 43395
    },
    {
      "epoch": 0.8887432678721356,
      "grad_norm": 0.28484782576560974,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.635,
      "step": 43400
    },
    {
      "epoch": 0.8888456576495403,
      "grad_norm": 0.2970082461833954,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5632,
      "step": 43405
    },
    {
      "epoch": 0.8889480474269449,
      "grad_norm": 0.2938386797904968,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5672,
      "step": 43410
    },
    {
      "epoch": 0.8890504372043495,
      "grad_norm": 0.28469425439834595,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5242,
      "step": 43415
    },
    {
      "epoch": 0.8891528269817541,
      "grad_norm": 0.364021360874176,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5575,
      "step": 43420
    },
    {
      "epoch": 0.8892552167591588,
      "grad_norm": 0.2875620126724243,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5952,
      "step": 43425
    },
    {
      "epoch": 0.8893576065365634,
      "grad_norm": 0.29636284708976746,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5787,
      "step": 43430
    },
    {
      "epoch": 0.889459996313968,
      "grad_norm": 0.2957055866718292,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6116,
      "step": 43435
    },
    {
      "epoch": 0.8895623860913726,
      "grad_norm": 0.2801075279712677,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5848,
      "step": 43440
    },
    {
      "epoch": 0.8896647758687772,
      "grad_norm": 0.3111990988254547,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6566,
      "step": 43445
    },
    {
      "epoch": 0.8897671656461819,
      "grad_norm": 0.295200914144516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5115,
      "step": 43450
    },
    {
      "epoch": 0.8898695554235865,
      "grad_norm": 0.2847495675086975,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4961,
      "step": 43455
    },
    {
      "epoch": 0.8899719452009911,
      "grad_norm": 0.3003501892089844,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6224,
      "step": 43460
    },
    {
      "epoch": 0.8900743349783957,
      "grad_norm": 0.29068243503570557,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5787,
      "step": 43465
    },
    {
      "epoch": 0.8901767247558003,
      "grad_norm": 0.28233209252357483,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.58,
      "step": 43470
    },
    {
      "epoch": 0.890279114533205,
      "grad_norm": 0.2820115387439728,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5867,
      "step": 43475
    },
    {
      "epoch": 0.8903815043106096,
      "grad_norm": 0.28251320123672485,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4561,
      "step": 43480
    },
    {
      "epoch": 0.8904838940880142,
      "grad_norm": 0.27587589621543884,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5677,
      "step": 43485
    },
    {
      "epoch": 0.8905862838654188,
      "grad_norm": 0.2913191020488739,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5786,
      "step": 43490
    },
    {
      "epoch": 0.8906886736428234,
      "grad_norm": 0.29503539204597473,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.587,
      "step": 43495
    },
    {
      "epoch": 0.8907910634202282,
      "grad_norm": 0.3059171736240387,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5826,
      "step": 43500
    },
    {
      "epoch": 0.8908934531976328,
      "grad_norm": 0.29003840684890747,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6057,
      "step": 43505
    },
    {
      "epoch": 0.8909958429750374,
      "grad_norm": 0.29289379715919495,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5023,
      "step": 43510
    },
    {
      "epoch": 0.891098232752442,
      "grad_norm": 0.29439595341682434,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5855,
      "step": 43515
    },
    {
      "epoch": 0.8912006225298467,
      "grad_norm": 0.2912348508834839,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5522,
      "step": 43520
    },
    {
      "epoch": 0.8913030123072513,
      "grad_norm": 0.27954715490341187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6042,
      "step": 43525
    },
    {
      "epoch": 0.8914054020846559,
      "grad_norm": 0.28138938546180725,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5829,
      "step": 43530
    },
    {
      "epoch": 0.8915077918620605,
      "grad_norm": 0.28235381841659546,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.549,
      "step": 43535
    },
    {
      "epoch": 0.8916101816394651,
      "grad_norm": 0.8361275792121887,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.572,
      "step": 43540
    },
    {
      "epoch": 0.8917125714168698,
      "grad_norm": 0.29305335879325867,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6044,
      "step": 43545
    },
    {
      "epoch": 0.8918149611942744,
      "grad_norm": 0.2782655358314514,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5625,
      "step": 43550
    },
    {
      "epoch": 0.891917350971679,
      "grad_norm": 0.2837987542152405,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5642,
      "step": 43555
    },
    {
      "epoch": 0.8920197407490836,
      "grad_norm": 0.2914084196090698,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5829,
      "step": 43560
    },
    {
      "epoch": 0.8921221305264883,
      "grad_norm": 0.28752848505973816,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5515,
      "step": 43565
    },
    {
      "epoch": 0.8922245203038929,
      "grad_norm": 0.30271926522254944,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5679,
      "step": 43570
    },
    {
      "epoch": 0.8923269100812975,
      "grad_norm": 0.27412354946136475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5685,
      "step": 43575
    },
    {
      "epoch": 0.8924292998587021,
      "grad_norm": 0.284602552652359,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6077,
      "step": 43580
    },
    {
      "epoch": 0.8925316896361067,
      "grad_norm": 0.3156874179840088,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5639,
      "step": 43585
    },
    {
      "epoch": 0.8926340794135114,
      "grad_norm": 0.3039139211177826,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5724,
      "step": 43590
    },
    {
      "epoch": 0.892736469190916,
      "grad_norm": 0.2927379608154297,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6056,
      "step": 43595
    },
    {
      "epoch": 0.8928388589683206,
      "grad_norm": 0.2834281623363495,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5665,
      "step": 43600
    },
    {
      "epoch": 0.8929412487457252,
      "grad_norm": 0.2961020767688751,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5478,
      "step": 43605
    },
    {
      "epoch": 0.8930436385231298,
      "grad_norm": 0.28263378143310547,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.551,
      "step": 43610
    },
    {
      "epoch": 0.8931460283005345,
      "grad_norm": 0.2814967930316925,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5584,
      "step": 43615
    },
    {
      "epoch": 0.8932484180779391,
      "grad_norm": 0.2874368131160736,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5555,
      "step": 43620
    },
    {
      "epoch": 0.8933508078553437,
      "grad_norm": 0.3158700168132782,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5803,
      "step": 43625
    },
    {
      "epoch": 0.8934531976327483,
      "grad_norm": 0.3642716109752655,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6043,
      "step": 43630
    },
    {
      "epoch": 0.893555587410153,
      "grad_norm": 0.29824936389923096,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5823,
      "step": 43635
    },
    {
      "epoch": 0.8936579771875576,
      "grad_norm": 0.3525303304195404,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5454,
      "step": 43640
    },
    {
      "epoch": 0.8937603669649622,
      "grad_norm": 0.2897406816482544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5495,
      "step": 43645
    },
    {
      "epoch": 0.8938627567423668,
      "grad_norm": 0.3116193413734436,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5517,
      "step": 43650
    },
    {
      "epoch": 0.8939651465197714,
      "grad_norm": 0.29880356788635254,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5423,
      "step": 43655
    },
    {
      "epoch": 0.8940675362971761,
      "grad_norm": 0.2957082688808441,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.555,
      "step": 43660
    },
    {
      "epoch": 0.8941699260745807,
      "grad_norm": 0.28723859786987305,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5485,
      "step": 43665
    },
    {
      "epoch": 0.8942723158519853,
      "grad_norm": 0.2814950942993164,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5718,
      "step": 43670
    },
    {
      "epoch": 0.8943747056293899,
      "grad_norm": 0.4035419225692749,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5666,
      "step": 43675
    },
    {
      "epoch": 0.8944770954067945,
      "grad_norm": 0.2928883731365204,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5292,
      "step": 43680
    },
    {
      "epoch": 0.8945794851841992,
      "grad_norm": 0.3120344281196594,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5156,
      "step": 43685
    },
    {
      "epoch": 0.8946818749616038,
      "grad_norm": 0.29340067505836487,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5921,
      "step": 43690
    },
    {
      "epoch": 0.8947842647390084,
      "grad_norm": 0.286952942609787,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5884,
      "step": 43695
    },
    {
      "epoch": 0.894886654516413,
      "grad_norm": 0.2862982749938965,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6195,
      "step": 43700
    },
    {
      "epoch": 0.8949890442938178,
      "grad_norm": 0.33828821778297424,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5335,
      "step": 43705
    },
    {
      "epoch": 0.8950914340712224,
      "grad_norm": 0.3000277280807495,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5795,
      "step": 43710
    },
    {
      "epoch": 0.895193823848627,
      "grad_norm": 0.2789219915866852,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 43715
    },
    {
      "epoch": 0.8952962136260316,
      "grad_norm": 0.2951725423336029,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5604,
      "step": 43720
    },
    {
      "epoch": 0.8953986034034362,
      "grad_norm": 0.31653720140457153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5404,
      "step": 43725
    },
    {
      "epoch": 0.8955009931808409,
      "grad_norm": 0.2837054133415222,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5111,
      "step": 43730
    },
    {
      "epoch": 0.8956033829582455,
      "grad_norm": 0.287752628326416,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5963,
      "step": 43735
    },
    {
      "epoch": 0.8957057727356501,
      "grad_norm": 0.30167561769485474,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5534,
      "step": 43740
    },
    {
      "epoch": 0.8958081625130547,
      "grad_norm": 0.2922893762588501,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5727,
      "step": 43745
    },
    {
      "epoch": 0.8959105522904593,
      "grad_norm": 0.3056415021419525,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6226,
      "step": 43750
    },
    {
      "epoch": 0.896012942067864,
      "grad_norm": 0.27673646807670593,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5942,
      "step": 43755
    },
    {
      "epoch": 0.8961153318452686,
      "grad_norm": 0.29115694761276245,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.594,
      "step": 43760
    },
    {
      "epoch": 0.8962177216226732,
      "grad_norm": 0.2822927236557007,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6041,
      "step": 43765
    },
    {
      "epoch": 0.8963201114000778,
      "grad_norm": 0.28792211413383484,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5621,
      "step": 43770
    },
    {
      "epoch": 0.8964225011774825,
      "grad_norm": 0.27710363268852234,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5044,
      "step": 43775
    },
    {
      "epoch": 0.8965248909548871,
      "grad_norm": 0.2837262749671936,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5753,
      "step": 43780
    },
    {
      "epoch": 0.8966272807322917,
      "grad_norm": 0.28804516792297363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5626,
      "step": 43785
    },
    {
      "epoch": 0.8967296705096963,
      "grad_norm": 0.3009595572948456,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6036,
      "step": 43790
    },
    {
      "epoch": 0.8968320602871009,
      "grad_norm": 0.29668760299682617,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5692,
      "step": 43795
    },
    {
      "epoch": 0.8969344500645056,
      "grad_norm": 0.28489741683006287,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5532,
      "step": 43800
    },
    {
      "epoch": 0.8970368398419102,
      "grad_norm": 0.2997526526451111,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5727,
      "step": 43805
    },
    {
      "epoch": 0.8971392296193148,
      "grad_norm": 0.3091844618320465,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5011,
      "step": 43810
    },
    {
      "epoch": 0.8972416193967194,
      "grad_norm": 0.36383694410324097,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.573,
      "step": 43815
    },
    {
      "epoch": 0.897344009174124,
      "grad_norm": 0.28524136543273926,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5614,
      "step": 43820
    },
    {
      "epoch": 0.8974463989515287,
      "grad_norm": 0.2953522503376007,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6166,
      "step": 43825
    },
    {
      "epoch": 0.8975487887289333,
      "grad_norm": 0.28813883662223816,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.589,
      "step": 43830
    },
    {
      "epoch": 0.8976511785063379,
      "grad_norm": 0.28438302874565125,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.595,
      "step": 43835
    },
    {
      "epoch": 0.8977535682837425,
      "grad_norm": 0.2952619194984436,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5704,
      "step": 43840
    },
    {
      "epoch": 0.8978559580611472,
      "grad_norm": 0.31650814414024353,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5711,
      "step": 43845
    },
    {
      "epoch": 0.8979583478385518,
      "grad_norm": 0.29960745573043823,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5576,
      "step": 43850
    },
    {
      "epoch": 0.8980607376159564,
      "grad_norm": 0.2964019179344177,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 43855
    },
    {
      "epoch": 0.898163127393361,
      "grad_norm": 0.29202255606651306,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5914,
      "step": 43860
    },
    {
      "epoch": 0.8982655171707656,
      "grad_norm": 0.28947746753692627,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5512,
      "step": 43865
    },
    {
      "epoch": 0.8983679069481703,
      "grad_norm": 0.28201887011528015,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5778,
      "step": 43870
    },
    {
      "epoch": 0.8984702967255749,
      "grad_norm": 0.28833359479904175,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5778,
      "step": 43875
    },
    {
      "epoch": 0.8985726865029795,
      "grad_norm": 0.31362542510032654,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5621,
      "step": 43880
    },
    {
      "epoch": 0.8986750762803841,
      "grad_norm": 0.31593671441078186,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5782,
      "step": 43885
    },
    {
      "epoch": 0.8987774660577887,
      "grad_norm": 0.3144204616546631,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5675,
      "step": 43890
    },
    {
      "epoch": 0.8988798558351934,
      "grad_norm": 0.2929304540157318,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5404,
      "step": 43895
    },
    {
      "epoch": 0.898982245612598,
      "grad_norm": 0.3831968903541565,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5263,
      "step": 43900
    },
    {
      "epoch": 0.8990846353900026,
      "grad_norm": 0.2977750599384308,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5645,
      "step": 43905
    },
    {
      "epoch": 0.8991870251674073,
      "grad_norm": 0.30486249923706055,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5733,
      "step": 43910
    },
    {
      "epoch": 0.899289414944812,
      "grad_norm": 0.2831956446170807,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 43915
    },
    {
      "epoch": 0.8993918047222166,
      "grad_norm": 0.29849597811698914,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5467,
      "step": 43920
    },
    {
      "epoch": 0.8994941944996212,
      "grad_norm": 1.5628610849380493,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5232,
      "step": 43925
    },
    {
      "epoch": 0.8995965842770258,
      "grad_norm": 0.2881596088409424,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5868,
      "step": 43930
    },
    {
      "epoch": 0.8996989740544304,
      "grad_norm": 0.30083224177360535,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5806,
      "step": 43935
    },
    {
      "epoch": 0.8998013638318351,
      "grad_norm": 0.3666914999485016,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5509,
      "step": 43940
    },
    {
      "epoch": 0.8999037536092397,
      "grad_norm": 0.3085542619228363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5729,
      "step": 43945
    },
    {
      "epoch": 0.9000061433866443,
      "grad_norm": 0.3142866790294647,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5119,
      "step": 43950
    },
    {
      "epoch": 0.9001085331640489,
      "grad_norm": 0.29957863688468933,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 43955
    },
    {
      "epoch": 0.9002109229414536,
      "grad_norm": 0.2904525399208069,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5366,
      "step": 43960
    },
    {
      "epoch": 0.9003133127188582,
      "grad_norm": 0.3167794644832611,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5689,
      "step": 43965
    },
    {
      "epoch": 0.9004157024962628,
      "grad_norm": 0.28093332052230835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5652,
      "step": 43970
    },
    {
      "epoch": 0.9005180922736674,
      "grad_norm": 0.288825660943985,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6403,
      "step": 43975
    },
    {
      "epoch": 0.900620482051072,
      "grad_norm": 0.2897427976131439,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5656,
      "step": 43980
    },
    {
      "epoch": 0.9007228718284767,
      "grad_norm": 0.2949637472629547,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.564,
      "step": 43985
    },
    {
      "epoch": 0.9008252616058813,
      "grad_norm": 0.29060909152030945,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5662,
      "step": 43990
    },
    {
      "epoch": 0.9009276513832859,
      "grad_norm": 0.2961573302745819,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5381,
      "step": 43995
    },
    {
      "epoch": 0.9010300411606905,
      "grad_norm": 0.2899945080280304,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5682,
      "step": 44000
    },
    {
      "epoch": 0.9010300411606905,
      "eval_loss": 4.586471080780029,
      "eval_runtime": 17.1248,
      "eval_samples_per_second": 33.986,
      "eval_steps_per_second": 4.263,
      "step": 44000
    },
    {
      "epoch": 0.9011324309380951,
      "grad_norm": 0.2913335859775543,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.571,
      "step": 44005
    },
    {
      "epoch": 0.9012348207154998,
      "grad_norm": 0.29117926955223083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4865,
      "step": 44010
    },
    {
      "epoch": 0.9013372104929044,
      "grad_norm": 0.2909631133079529,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.561,
      "step": 44015
    },
    {
      "epoch": 0.901439600270309,
      "grad_norm": 0.2952113449573517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5701,
      "step": 44020
    },
    {
      "epoch": 0.9015419900477136,
      "grad_norm": 0.31003907322883606,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5731,
      "step": 44025
    },
    {
      "epoch": 0.9016443798251182,
      "grad_norm": 0.29012972116470337,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5748,
      "step": 44030
    },
    {
      "epoch": 0.9017467696025229,
      "grad_norm": 0.2859537899494171,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5642,
      "step": 44035
    },
    {
      "epoch": 0.9018491593799275,
      "grad_norm": 0.32267364859580994,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.611,
      "step": 44040
    },
    {
      "epoch": 0.9019515491573321,
      "grad_norm": 0.31911522150039673,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5812,
      "step": 44045
    },
    {
      "epoch": 0.9020539389347367,
      "grad_norm": 0.29209789633750916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6097,
      "step": 44050
    },
    {
      "epoch": 0.9021563287121414,
      "grad_norm": 0.28505492210388184,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6094,
      "step": 44055
    },
    {
      "epoch": 0.902258718489546,
      "grad_norm": 0.3423404097557068,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5438,
      "step": 44060
    },
    {
      "epoch": 0.9023611082669506,
      "grad_norm": 0.29405495524406433,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6121,
      "step": 44065
    },
    {
      "epoch": 0.9024634980443552,
      "grad_norm": 0.3212548792362213,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.615,
      "step": 44070
    },
    {
      "epoch": 0.9025658878217598,
      "grad_norm": 0.2923244535923004,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.579,
      "step": 44075
    },
    {
      "epoch": 0.9026682775991645,
      "grad_norm": 0.2975698709487915,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6111,
      "step": 44080
    },
    {
      "epoch": 0.9027706673765691,
      "grad_norm": 0.3004777431488037,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5691,
      "step": 44085
    },
    {
      "epoch": 0.9028730571539737,
      "grad_norm": 0.2993500828742981,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6132,
      "step": 44090
    },
    {
      "epoch": 0.9029754469313783,
      "grad_norm": 0.2951490879058838,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6079,
      "step": 44095
    },
    {
      "epoch": 0.9030778367087829,
      "grad_norm": 0.29161161184310913,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5427,
      "step": 44100
    },
    {
      "epoch": 0.9031802264861876,
      "grad_norm": 0.2915920317173004,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5934,
      "step": 44105
    },
    {
      "epoch": 0.9032826162635923,
      "grad_norm": 0.2810094654560089,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5154,
      "step": 44110
    },
    {
      "epoch": 0.9033850060409969,
      "grad_norm": 0.29250696301460266,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5773,
      "step": 44115
    },
    {
      "epoch": 0.9034873958184015,
      "grad_norm": 0.29555365443229675,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.558,
      "step": 44120
    },
    {
      "epoch": 0.9035897855958062,
      "grad_norm": 0.2818824350833893,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5882,
      "step": 44125
    },
    {
      "epoch": 0.9036921753732108,
      "grad_norm": 0.2955513000488281,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5514,
      "step": 44130
    },
    {
      "epoch": 0.9037945651506154,
      "grad_norm": 0.29016879200935364,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5746,
      "step": 44135
    },
    {
      "epoch": 0.90389695492802,
      "grad_norm": 0.30701562762260437,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5606,
      "step": 44140
    },
    {
      "epoch": 0.9039993447054246,
      "grad_norm": 0.30402061343193054,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5813,
      "step": 44145
    },
    {
      "epoch": 0.9041017344828293,
      "grad_norm": 0.3075377643108368,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5684,
      "step": 44150
    },
    {
      "epoch": 0.9042041242602339,
      "grad_norm": 0.28911861777305603,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6001,
      "step": 44155
    },
    {
      "epoch": 0.9043065140376385,
      "grad_norm": 0.2968486547470093,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5897,
      "step": 44160
    },
    {
      "epoch": 0.9044089038150431,
      "grad_norm": 0.29539307951927185,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5797,
      "step": 44165
    },
    {
      "epoch": 0.9045112935924478,
      "grad_norm": 0.287286639213562,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5892,
      "step": 44170
    },
    {
      "epoch": 0.9046136833698524,
      "grad_norm": 0.296395868062973,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5772,
      "step": 44175
    },
    {
      "epoch": 0.904716073147257,
      "grad_norm": 0.29325658082962036,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5902,
      "step": 44180
    },
    {
      "epoch": 0.9048184629246616,
      "grad_norm": 0.2802959978580475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5655,
      "step": 44185
    },
    {
      "epoch": 0.9049208527020662,
      "grad_norm": 0.2855745851993561,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.576,
      "step": 44190
    },
    {
      "epoch": 0.9050232424794709,
      "grad_norm": 0.31628891825675964,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 44195
    },
    {
      "epoch": 0.9051256322568755,
      "grad_norm": 0.29802003502845764,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4967,
      "step": 44200
    },
    {
      "epoch": 0.9052280220342801,
      "grad_norm": 0.2871008813381195,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5601,
      "step": 44205
    },
    {
      "epoch": 0.9053304118116847,
      "grad_norm": 0.3069486916065216,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5297,
      "step": 44210
    },
    {
      "epoch": 0.9054328015890893,
      "grad_norm": 0.30202245712280273,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6304,
      "step": 44215
    },
    {
      "epoch": 0.905535191366494,
      "grad_norm": 0.2802700400352478,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.538,
      "step": 44220
    },
    {
      "epoch": 0.9056375811438986,
      "grad_norm": 0.2830926477909088,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5441,
      "step": 44225
    },
    {
      "epoch": 0.9057399709213032,
      "grad_norm": 0.2824036478996277,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5481,
      "step": 44230
    },
    {
      "epoch": 0.9058423606987078,
      "grad_norm": 0.2798176109790802,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 44235
    },
    {
      "epoch": 0.9059447504761124,
      "grad_norm": 0.29679057002067566,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6041,
      "step": 44240
    },
    {
      "epoch": 0.9060471402535171,
      "grad_norm": 0.30267518758773804,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5215,
      "step": 44245
    },
    {
      "epoch": 0.9061495300309217,
      "grad_norm": 0.2759597897529602,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5094,
      "step": 44250
    },
    {
      "epoch": 0.9062519198083263,
      "grad_norm": 0.2725431025028229,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5103,
      "step": 44255
    },
    {
      "epoch": 0.9063543095857309,
      "grad_norm": 0.3026810586452484,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.592,
      "step": 44260
    },
    {
      "epoch": 0.9064566993631356,
      "grad_norm": 0.2777859568595886,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5875,
      "step": 44265
    },
    {
      "epoch": 0.9065590891405402,
      "grad_norm": 0.318081796169281,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5309,
      "step": 44270
    },
    {
      "epoch": 0.9066614789179448,
      "grad_norm": 0.2968437671661377,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6267,
      "step": 44275
    },
    {
      "epoch": 0.9067638686953494,
      "grad_norm": 0.30005747079849243,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5652,
      "step": 44280
    },
    {
      "epoch": 0.906866258472754,
      "grad_norm": 0.2993433177471161,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.587,
      "step": 44285
    },
    {
      "epoch": 0.9069686482501587,
      "grad_norm": 0.30756816267967224,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5544,
      "step": 44290
    },
    {
      "epoch": 0.9070710380275633,
      "grad_norm": 0.29073426127433777,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6004,
      "step": 44295
    },
    {
      "epoch": 0.9071734278049679,
      "grad_norm": 0.3133041560649872,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5797,
      "step": 44300
    },
    {
      "epoch": 0.9072758175823725,
      "grad_norm": 0.28762057423591614,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.584,
      "step": 44305
    },
    {
      "epoch": 0.9073782073597771,
      "grad_norm": 0.31225842237472534,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5958,
      "step": 44310
    },
    {
      "epoch": 0.9074805971371819,
      "grad_norm": 0.29675835371017456,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.587,
      "step": 44315
    },
    {
      "epoch": 0.9075829869145865,
      "grad_norm": 0.2963467836380005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.578,
      "step": 44320
    },
    {
      "epoch": 0.9076853766919911,
      "grad_norm": 0.28992506861686707,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5554,
      "step": 44325
    },
    {
      "epoch": 0.9077877664693957,
      "grad_norm": 0.3521689474582672,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5715,
      "step": 44330
    },
    {
      "epoch": 0.9078901562468004,
      "grad_norm": 0.29133668541908264,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5617,
      "step": 44335
    },
    {
      "epoch": 0.907992546024205,
      "grad_norm": 0.2901957631111145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5878,
      "step": 44340
    },
    {
      "epoch": 0.9080949358016096,
      "grad_norm": 0.28737345337867737,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5934,
      "step": 44345
    },
    {
      "epoch": 0.9081973255790142,
      "grad_norm": 0.30114200711250305,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5603,
      "step": 44350
    },
    {
      "epoch": 0.9082997153564188,
      "grad_norm": 0.28318148851394653,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6067,
      "step": 44355
    },
    {
      "epoch": 0.9084021051338235,
      "grad_norm": 0.29417601227760315,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.569,
      "step": 44360
    },
    {
      "epoch": 0.9085044949112281,
      "grad_norm": 0.3162309229373932,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5249,
      "step": 44365
    },
    {
      "epoch": 0.9086068846886327,
      "grad_norm": 0.29203394055366516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5939,
      "step": 44370
    },
    {
      "epoch": 0.9087092744660373,
      "grad_norm": 0.2932101786136627,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5592,
      "step": 44375
    },
    {
      "epoch": 0.908811664243442,
      "grad_norm": 0.30883273482322693,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5787,
      "step": 44380
    },
    {
      "epoch": 0.9089140540208466,
      "grad_norm": 0.29501113295555115,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5564,
      "step": 44385
    },
    {
      "epoch": 0.9090164437982512,
      "grad_norm": 0.2937309145927429,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5598,
      "step": 44390
    },
    {
      "epoch": 0.9091188335756558,
      "grad_norm": 0.2975621223449707,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5995,
      "step": 44395
    },
    {
      "epoch": 0.9092212233530604,
      "grad_norm": 0.30689218640327454,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5718,
      "step": 44400
    },
    {
      "epoch": 0.9093236131304651,
      "grad_norm": 0.2986644208431244,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5933,
      "step": 44405
    },
    {
      "epoch": 0.9094260029078697,
      "grad_norm": 0.3601891100406647,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5599,
      "step": 44410
    },
    {
      "epoch": 0.9095283926852743,
      "grad_norm": 0.2950226068496704,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.526,
      "step": 44415
    },
    {
      "epoch": 0.9096307824626789,
      "grad_norm": 0.2891264259815216,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5576,
      "step": 44420
    },
    {
      "epoch": 0.9097331722400835,
      "grad_norm": 0.2955225706100464,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5698,
      "step": 44425
    },
    {
      "epoch": 0.9098355620174882,
      "grad_norm": 0.29458922147750854,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5549,
      "step": 44430
    },
    {
      "epoch": 0.9099379517948928,
      "grad_norm": 0.29744821786880493,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5999,
      "step": 44435
    },
    {
      "epoch": 0.9100403415722974,
      "grad_norm": 0.2866936922073364,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.2777,
      "step": 44440
    },
    {
      "epoch": 0.910142731349702,
      "grad_norm": 0.28462445735931396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5301,
      "step": 44445
    },
    {
      "epoch": 0.9102451211271066,
      "grad_norm": 0.30194762349128723,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5634,
      "step": 44450
    },
    {
      "epoch": 0.9103475109045113,
      "grad_norm": 0.32271724939346313,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5497,
      "step": 44455
    },
    {
      "epoch": 0.9104499006819159,
      "grad_norm": 0.2883274555206299,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5857,
      "step": 44460
    },
    {
      "epoch": 0.9105522904593205,
      "grad_norm": 0.29806938767433167,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5851,
      "step": 44465
    },
    {
      "epoch": 0.9106546802367251,
      "grad_norm": 0.29009348154067993,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5907,
      "step": 44470
    },
    {
      "epoch": 0.9107570700141298,
      "grad_norm": 0.28038156032562256,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5614,
      "step": 44475
    },
    {
      "epoch": 0.9108594597915344,
      "grad_norm": 0.29314082860946655,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 44480
    },
    {
      "epoch": 0.910961849568939,
      "grad_norm": 0.3111684024333954,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6112,
      "step": 44485
    },
    {
      "epoch": 0.9110642393463436,
      "grad_norm": 0.2904896140098572,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.576,
      "step": 44490
    },
    {
      "epoch": 0.9111666291237482,
      "grad_norm": 0.30856579542160034,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4689,
      "step": 44495
    },
    {
      "epoch": 0.9112690189011529,
      "grad_norm": 0.28661075234413147,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5718,
      "step": 44500
    },
    {
      "epoch": 0.9113714086785575,
      "grad_norm": 0.3105640709400177,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5617,
      "step": 44505
    },
    {
      "epoch": 0.9114737984559621,
      "grad_norm": 0.28419485688209534,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5509,
      "step": 44510
    },
    {
      "epoch": 0.9115761882333668,
      "grad_norm": 0.2916661202907562,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6011,
      "step": 44515
    },
    {
      "epoch": 0.9116785780107715,
      "grad_norm": 0.2916611135005951,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5583,
      "step": 44520
    },
    {
      "epoch": 0.9117809677881761,
      "grad_norm": 0.3471600115299225,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.636,
      "step": 44525
    },
    {
      "epoch": 0.9118833575655807,
      "grad_norm": 0.29736122488975525,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5856,
      "step": 44530
    },
    {
      "epoch": 0.9119857473429853,
      "grad_norm": 0.3281717896461487,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5828,
      "step": 44535
    },
    {
      "epoch": 0.9120881371203899,
      "grad_norm": 0.2958275079727173,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5746,
      "step": 44540
    },
    {
      "epoch": 0.9121905268977946,
      "grad_norm": 0.29329514503479004,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5412,
      "step": 44545
    },
    {
      "epoch": 0.9122929166751992,
      "grad_norm": 0.2900955080986023,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5821,
      "step": 44550
    },
    {
      "epoch": 0.9123953064526038,
      "grad_norm": 0.4106867015361786,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6349,
      "step": 44555
    },
    {
      "epoch": 0.9124976962300084,
      "grad_norm": 0.3019821047782898,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5892,
      "step": 44560
    },
    {
      "epoch": 0.912600086007413,
      "grad_norm": 0.2903803586959839,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5456,
      "step": 44565
    },
    {
      "epoch": 0.9127024757848177,
      "grad_norm": 0.28803229331970215,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5925,
      "step": 44570
    },
    {
      "epoch": 0.9128048655622223,
      "grad_norm": 0.27864280343055725,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.584,
      "step": 44575
    },
    {
      "epoch": 0.9129072553396269,
      "grad_norm": 0.28240007162094116,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5872,
      "step": 44580
    },
    {
      "epoch": 0.9130096451170315,
      "grad_norm": 0.28655725717544556,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5561,
      "step": 44585
    },
    {
      "epoch": 0.9131120348944362,
      "grad_norm": 0.3613794147968292,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5624,
      "step": 44590
    },
    {
      "epoch": 0.9132144246718408,
      "grad_norm": 0.27940067648887634,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6067,
      "step": 44595
    },
    {
      "epoch": 0.9133168144492454,
      "grad_norm": 0.291115403175354,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.605,
      "step": 44600
    },
    {
      "epoch": 0.91341920422665,
      "grad_norm": 0.30663806200027466,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6237,
      "step": 44605
    },
    {
      "epoch": 0.9135215940040546,
      "grad_norm": 0.292237251996994,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6276,
      "step": 44610
    },
    {
      "epoch": 0.9136239837814593,
      "grad_norm": 0.3186366558074951,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5705,
      "step": 44615
    },
    {
      "epoch": 0.9137263735588639,
      "grad_norm": 0.2985791265964508,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5121,
      "step": 44620
    },
    {
      "epoch": 0.9138287633362685,
      "grad_norm": 0.3115338981151581,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5789,
      "step": 44625
    },
    {
      "epoch": 0.9139311531136731,
      "grad_norm": 0.2941564917564392,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5967,
      "step": 44630
    },
    {
      "epoch": 0.9140335428910777,
      "grad_norm": 0.29198572039604187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5936,
      "step": 44635
    },
    {
      "epoch": 0.9141359326684824,
      "grad_norm": 0.28676462173461914,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5669,
      "step": 44640
    },
    {
      "epoch": 0.914238322445887,
      "grad_norm": 0.3021875321865082,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5505,
      "step": 44645
    },
    {
      "epoch": 0.9143407122232916,
      "grad_norm": 0.3352340757846832,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.581,
      "step": 44650
    },
    {
      "epoch": 0.9144431020006962,
      "grad_norm": 0.28663456439971924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5722,
      "step": 44655
    },
    {
      "epoch": 0.9145454917781008,
      "grad_norm": 0.29421597719192505,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5843,
      "step": 44660
    },
    {
      "epoch": 0.9146478815555055,
      "grad_norm": 0.2880072593688965,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5498,
      "step": 44665
    },
    {
      "epoch": 0.9147502713329101,
      "grad_norm": 0.28296324610710144,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5583,
      "step": 44670
    },
    {
      "epoch": 0.9148526611103147,
      "grad_norm": 0.2951556444168091,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5616,
      "step": 44675
    },
    {
      "epoch": 0.9149550508877193,
      "grad_norm": 0.27845102548599243,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5802,
      "step": 44680
    },
    {
      "epoch": 0.915057440665124,
      "grad_norm": 0.30597835779190063,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5567,
      "step": 44685
    },
    {
      "epoch": 0.9151598304425286,
      "grad_norm": 0.3044925034046173,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5787,
      "step": 44690
    },
    {
      "epoch": 0.9152622202199332,
      "grad_norm": 0.2863486111164093,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5216,
      "step": 44695
    },
    {
      "epoch": 0.9153646099973378,
      "grad_norm": 0.29713574051856995,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5699,
      "step": 44700
    },
    {
      "epoch": 0.9154669997747424,
      "grad_norm": 0.3571939766407013,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5255,
      "step": 44705
    },
    {
      "epoch": 0.9155693895521471,
      "grad_norm": 0.2816871702671051,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6053,
      "step": 44710
    },
    {
      "epoch": 0.9156717793295517,
      "grad_norm": 0.2798614203929901,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4954,
      "step": 44715
    },
    {
      "epoch": 0.9157741691069564,
      "grad_norm": 0.302320271730423,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5727,
      "step": 44720
    },
    {
      "epoch": 0.915876558884361,
      "grad_norm": 0.2995151877403259,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5294,
      "step": 44725
    },
    {
      "epoch": 0.9159789486617657,
      "grad_norm": 0.28221645951271057,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5591,
      "step": 44730
    },
    {
      "epoch": 0.9160813384391703,
      "grad_norm": 0.29791709780693054,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5143,
      "step": 44735
    },
    {
      "epoch": 0.9161837282165749,
      "grad_norm": 0.29077425599098206,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5638,
      "step": 44740
    },
    {
      "epoch": 0.9162861179939795,
      "grad_norm": 0.28820061683654785,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5873,
      "step": 44745
    },
    {
      "epoch": 0.9163885077713841,
      "grad_norm": 0.2995607256889343,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5914,
      "step": 44750
    },
    {
      "epoch": 0.9164908975487888,
      "grad_norm": 0.29834872484207153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5233,
      "step": 44755
    },
    {
      "epoch": 0.9165932873261934,
      "grad_norm": 0.28764039278030396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5648,
      "step": 44760
    },
    {
      "epoch": 0.916695677103598,
      "grad_norm": 0.32084712386131287,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5085,
      "step": 44765
    },
    {
      "epoch": 0.9167980668810026,
      "grad_norm": 0.3009834289550781,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5689,
      "step": 44770
    },
    {
      "epoch": 0.9169004566584072,
      "grad_norm": 0.2824249863624573,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.504,
      "step": 44775
    },
    {
      "epoch": 0.9170028464358119,
      "grad_norm": 0.2883903980255127,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5494,
      "step": 44780
    },
    {
      "epoch": 0.9171052362132165,
      "grad_norm": 0.3404322862625122,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5699,
      "step": 44785
    },
    {
      "epoch": 0.9172076259906211,
      "grad_norm": 0.31625035405158997,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5463,
      "step": 44790
    },
    {
      "epoch": 0.9173100157680257,
      "grad_norm": 0.2832892835140228,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5663,
      "step": 44795
    },
    {
      "epoch": 0.9174124055454304,
      "grad_norm": 0.38594958186149597,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5514,
      "step": 44800
    },
    {
      "epoch": 0.917514795322835,
      "grad_norm": 0.29314351081848145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5338,
      "step": 44805
    },
    {
      "epoch": 0.9176171851002396,
      "grad_norm": 0.28442466259002686,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5978,
      "step": 44810
    },
    {
      "epoch": 0.9177195748776442,
      "grad_norm": 0.3025929033756256,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5993,
      "step": 44815
    },
    {
      "epoch": 0.9178219646550488,
      "grad_norm": 0.29419243335723877,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5366,
      "step": 44820
    },
    {
      "epoch": 0.9179243544324535,
      "grad_norm": 0.2948008179664612,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6115,
      "step": 44825
    },
    {
      "epoch": 0.9180267442098581,
      "grad_norm": 0.2893778085708618,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 44830
    },
    {
      "epoch": 0.9181291339872627,
      "grad_norm": 0.3166704475879669,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5779,
      "step": 44835
    },
    {
      "epoch": 0.9182315237646673,
      "grad_norm": 0.2853911519050598,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6075,
      "step": 44840
    },
    {
      "epoch": 0.918333913542072,
      "grad_norm": 0.3583363890647888,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5763,
      "step": 44845
    },
    {
      "epoch": 0.9184363033194766,
      "grad_norm": 0.2975505292415619,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5851,
      "step": 44850
    },
    {
      "epoch": 0.9185386930968812,
      "grad_norm": 0.2748933434486389,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.551,
      "step": 44855
    },
    {
      "epoch": 0.9186410828742858,
      "grad_norm": 0.29540279507637024,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5765,
      "step": 44860
    },
    {
      "epoch": 0.9187434726516904,
      "grad_norm": 0.28963378071784973,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5538,
      "step": 44865
    },
    {
      "epoch": 0.918845862429095,
      "grad_norm": 0.329802006483078,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5686,
      "step": 44870
    },
    {
      "epoch": 0.9189482522064997,
      "grad_norm": 0.29359158873558044,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5927,
      "step": 44875
    },
    {
      "epoch": 0.9190506419839043,
      "grad_norm": 0.28925612568855286,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5471,
      "step": 44880
    },
    {
      "epoch": 0.9191530317613089,
      "grad_norm": 0.2977502644062042,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5916,
      "step": 44885
    },
    {
      "epoch": 0.9192554215387135,
      "grad_norm": 0.3059592843055725,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5924,
      "step": 44890
    },
    {
      "epoch": 0.9193578113161182,
      "grad_norm": 0.31695204973220825,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5645,
      "step": 44895
    },
    {
      "epoch": 0.9194602010935228,
      "grad_norm": 0.3907400369644165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5567,
      "step": 44900
    },
    {
      "epoch": 0.9195625908709274,
      "grad_norm": 0.2951764762401581,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6061,
      "step": 44905
    },
    {
      "epoch": 0.919664980648332,
      "grad_norm": 0.29211193323135376,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6376,
      "step": 44910
    },
    {
      "epoch": 0.9197673704257366,
      "grad_norm": 0.30369940400123596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.567,
      "step": 44915
    },
    {
      "epoch": 0.9198697602031414,
      "grad_norm": 0.292265385389328,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5784,
      "step": 44920
    },
    {
      "epoch": 0.919972149980546,
      "grad_norm": 0.44335368275642395,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5753,
      "step": 44925
    },
    {
      "epoch": 0.9200745397579506,
      "grad_norm": 0.3001779615879059,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5473,
      "step": 44930
    },
    {
      "epoch": 0.9201769295353552,
      "grad_norm": 0.2842656970024109,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5376,
      "step": 44935
    },
    {
      "epoch": 0.9202793193127599,
      "grad_norm": 0.29240307211875916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5703,
      "step": 44940
    },
    {
      "epoch": 0.9203817090901645,
      "grad_norm": 0.28944623470306396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.567,
      "step": 44945
    },
    {
      "epoch": 0.9204840988675691,
      "grad_norm": 0.2853572964668274,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6032,
      "step": 44950
    },
    {
      "epoch": 0.9205864886449737,
      "grad_norm": 0.2876571714878082,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5727,
      "step": 44955
    },
    {
      "epoch": 0.9206888784223783,
      "grad_norm": 0.2950356900691986,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5612,
      "step": 44960
    },
    {
      "epoch": 0.920791268199783,
      "grad_norm": 0.3180888891220093,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5039,
      "step": 44965
    },
    {
      "epoch": 0.9208936579771876,
      "grad_norm": 0.2985819876194,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5599,
      "step": 44970
    },
    {
      "epoch": 0.9209960477545922,
      "grad_norm": 0.3413555324077606,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5591,
      "step": 44975
    },
    {
      "epoch": 0.9210984375319968,
      "grad_norm": 0.6849969029426575,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5641,
      "step": 44980
    },
    {
      "epoch": 0.9212008273094014,
      "grad_norm": 0.3084063231945038,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 44985
    },
    {
      "epoch": 0.9213032170868061,
      "grad_norm": 0.3066835105419159,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5276,
      "step": 44990
    },
    {
      "epoch": 0.9214056068642107,
      "grad_norm": 0.3001596927642822,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5656,
      "step": 44995
    },
    {
      "epoch": 0.9215079966416153,
      "grad_norm": 0.2873099148273468,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5462,
      "step": 45000
    },
    {
      "epoch": 0.9215079966416153,
      "eval_loss": 4.58529806137085,
      "eval_runtime": 17.1325,
      "eval_samples_per_second": 33.971,
      "eval_steps_per_second": 4.261,
      "step": 45000
    },
    {
      "epoch": 0.9216103864190199,
      "grad_norm": 0.2973378896713257,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5837,
      "step": 45005
    },
    {
      "epoch": 0.9217127761964246,
      "grad_norm": 0.2963358759880066,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5435,
      "step": 45010
    },
    {
      "epoch": 0.9218151659738292,
      "grad_norm": 0.32997384667396545,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5147,
      "step": 45015
    },
    {
      "epoch": 0.9219175557512338,
      "grad_norm": 0.2832203507423401,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5975,
      "step": 45020
    },
    {
      "epoch": 0.9220199455286384,
      "grad_norm": 0.3515530228614807,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.643,
      "step": 45025
    },
    {
      "epoch": 0.922122335306043,
      "grad_norm": 0.3126188814640045,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6006,
      "step": 45030
    },
    {
      "epoch": 0.9222247250834477,
      "grad_norm": 0.3004818260669708,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6157,
      "step": 45035
    },
    {
      "epoch": 0.9223271148608523,
      "grad_norm": 0.2943071126937866,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5681,
      "step": 45040
    },
    {
      "epoch": 0.9224295046382569,
      "grad_norm": 0.2926907241344452,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5578,
      "step": 45045
    },
    {
      "epoch": 0.9225318944156615,
      "grad_norm": 0.2933685779571533,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5802,
      "step": 45050
    },
    {
      "epoch": 0.9226342841930661,
      "grad_norm": 0.29561647772789,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5653,
      "step": 45055
    },
    {
      "epoch": 0.9227366739704708,
      "grad_norm": 0.28530168533325195,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.501,
      "step": 45060
    },
    {
      "epoch": 0.9228390637478754,
      "grad_norm": 0.30260658264160156,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5781,
      "step": 45065
    },
    {
      "epoch": 0.92294145352528,
      "grad_norm": 0.29369866847991943,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5609,
      "step": 45070
    },
    {
      "epoch": 0.9230438433026846,
      "grad_norm": 0.2968003451824188,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5517,
      "step": 45075
    },
    {
      "epoch": 0.9231462330800893,
      "grad_norm": 0.2818485498428345,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5691,
      "step": 45080
    },
    {
      "epoch": 0.9232486228574939,
      "grad_norm": 0.29828980565071106,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5619,
      "step": 45085
    },
    {
      "epoch": 0.9233510126348985,
      "grad_norm": 0.289543092250824,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.589,
      "step": 45090
    },
    {
      "epoch": 0.9234534024123031,
      "grad_norm": 0.290693461894989,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5699,
      "step": 45095
    },
    {
      "epoch": 0.9235557921897077,
      "grad_norm": 0.3078167736530304,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6009,
      "step": 45100
    },
    {
      "epoch": 0.9236581819671124,
      "grad_norm": 0.28769147396087646,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5726,
      "step": 45105
    },
    {
      "epoch": 0.923760571744517,
      "grad_norm": 0.2848482131958008,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5611,
      "step": 45110
    },
    {
      "epoch": 0.9238629615219216,
      "grad_norm": 0.3099733293056488,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5544,
      "step": 45115
    },
    {
      "epoch": 0.9239653512993262,
      "grad_norm": 0.4073297083377838,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6083,
      "step": 45120
    },
    {
      "epoch": 0.924067741076731,
      "grad_norm": 0.3098052144050598,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5774,
      "step": 45125
    },
    {
      "epoch": 0.9241701308541356,
      "grad_norm": 0.29929396510124207,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5612,
      "step": 45130
    },
    {
      "epoch": 0.9242725206315402,
      "grad_norm": 0.3185664713382721,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6396,
      "step": 45135
    },
    {
      "epoch": 0.9243749104089448,
      "grad_norm": 0.29102253913879395,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5724,
      "step": 45140
    },
    {
      "epoch": 0.9244773001863494,
      "grad_norm": 0.2837943732738495,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5406,
      "step": 45145
    },
    {
      "epoch": 0.9245796899637541,
      "grad_norm": 0.2911560833454132,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6176,
      "step": 45150
    },
    {
      "epoch": 0.9246820797411587,
      "grad_norm": 0.2981114089488983,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5812,
      "step": 45155
    },
    {
      "epoch": 0.9247844695185633,
      "grad_norm": 0.30821493268013,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 45160
    },
    {
      "epoch": 0.9248868592959679,
      "grad_norm": 0.28796255588531494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5772,
      "step": 45165
    },
    {
      "epoch": 0.9249892490733725,
      "grad_norm": 0.28619998693466187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6008,
      "step": 45170
    },
    {
      "epoch": 0.9250916388507772,
      "grad_norm": 0.3595195412635803,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5806,
      "step": 45175
    },
    {
      "epoch": 0.9251940286281818,
      "grad_norm": 0.29030317068099976,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5276,
      "step": 45180
    },
    {
      "epoch": 0.9252964184055864,
      "grad_norm": 0.3279653489589691,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5719,
      "step": 45185
    },
    {
      "epoch": 0.925398808182991,
      "grad_norm": 0.2800334393978119,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5404,
      "step": 45190
    },
    {
      "epoch": 0.9255011979603956,
      "grad_norm": 0.2963065505027771,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5914,
      "step": 45195
    },
    {
      "epoch": 0.9256035877378003,
      "grad_norm": 0.29062843322753906,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5451,
      "step": 45200
    },
    {
      "epoch": 0.9257059775152049,
      "grad_norm": 0.2946456968784332,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5691,
      "step": 45205
    },
    {
      "epoch": 0.9258083672926095,
      "grad_norm": 0.30529794096946716,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5734,
      "step": 45210
    },
    {
      "epoch": 0.9259107570700141,
      "grad_norm": 0.287738174200058,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.553,
      "step": 45215
    },
    {
      "epoch": 0.9260131468474188,
      "grad_norm": 0.29744094610214233,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5858,
      "step": 45220
    },
    {
      "epoch": 0.9261155366248234,
      "grad_norm": 0.3122273087501526,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5747,
      "step": 45225
    },
    {
      "epoch": 0.926217926402228,
      "grad_norm": 0.2894415259361267,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6024,
      "step": 45230
    },
    {
      "epoch": 0.9263203161796326,
      "grad_norm": 0.29068198800086975,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5874,
      "step": 45235
    },
    {
      "epoch": 0.9264227059570372,
      "grad_norm": 0.29694217443466187,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5712,
      "step": 45240
    },
    {
      "epoch": 0.9265250957344419,
      "grad_norm": 0.28193652629852295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5896,
      "step": 45245
    },
    {
      "epoch": 0.9266274855118465,
      "grad_norm": 0.29648712277412415,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6129,
      "step": 45250
    },
    {
      "epoch": 0.9267298752892511,
      "grad_norm": 0.2974899709224701,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5979,
      "step": 45255
    },
    {
      "epoch": 0.9268322650666557,
      "grad_norm": 0.3110671937465668,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5784,
      "step": 45260
    },
    {
      "epoch": 0.9269346548440603,
      "grad_norm": 0.3000732362270355,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5646,
      "step": 45265
    },
    {
      "epoch": 0.927037044621465,
      "grad_norm": 0.28198617696762085,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5945,
      "step": 45270
    },
    {
      "epoch": 0.9271394343988696,
      "grad_norm": 0.28437891602516174,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.574,
      "step": 45275
    },
    {
      "epoch": 0.9272418241762742,
      "grad_norm": 0.2919110953807831,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5663,
      "step": 45280
    },
    {
      "epoch": 0.9273442139536788,
      "grad_norm": 0.2821648418903351,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6059,
      "step": 45285
    },
    {
      "epoch": 0.9274466037310835,
      "grad_norm": 0.27970513701438904,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5912,
      "step": 45290
    },
    {
      "epoch": 0.9275489935084881,
      "grad_norm": 0.28932279348373413,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5852,
      "step": 45295
    },
    {
      "epoch": 0.9276513832858927,
      "grad_norm": 0.28567251563072205,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5788,
      "step": 45300
    },
    {
      "epoch": 0.9277537730632973,
      "grad_norm": 0.297882080078125,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5693,
      "step": 45305
    },
    {
      "epoch": 0.9278561628407019,
      "grad_norm": 0.2826113700866699,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5739,
      "step": 45310
    },
    {
      "epoch": 0.9279585526181066,
      "grad_norm": 0.2921213209629059,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5849,
      "step": 45315
    },
    {
      "epoch": 0.9280609423955112,
      "grad_norm": 0.29327523708343506,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 45320
    },
    {
      "epoch": 0.9281633321729159,
      "grad_norm": 0.3010656535625458,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5771,
      "step": 45325
    },
    {
      "epoch": 0.9282657219503205,
      "grad_norm": 0.28777647018432617,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5675,
      "step": 45330
    },
    {
      "epoch": 0.9283681117277252,
      "grad_norm": 0.29613229632377625,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5744,
      "step": 45335
    },
    {
      "epoch": 0.9284705015051298,
      "grad_norm": 0.3020172715187073,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5726,
      "step": 45340
    },
    {
      "epoch": 0.9285728912825344,
      "grad_norm": 0.31469112634658813,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5326,
      "step": 45345
    },
    {
      "epoch": 0.928675281059939,
      "grad_norm": 0.29810044169425964,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.568,
      "step": 45350
    },
    {
      "epoch": 0.9287776708373436,
      "grad_norm": 0.28736141324043274,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5611,
      "step": 45355
    },
    {
      "epoch": 0.9288800606147483,
      "grad_norm": 0.3133838176727295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5245,
      "step": 45360
    },
    {
      "epoch": 0.9289824503921529,
      "grad_norm": 0.3475625514984131,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5655,
      "step": 45365
    },
    {
      "epoch": 0.9290848401695575,
      "grad_norm": 0.30536213517189026,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5825,
      "step": 45370
    },
    {
      "epoch": 0.9291872299469621,
      "grad_norm": 0.29482272267341614,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6484,
      "step": 45375
    },
    {
      "epoch": 0.9292896197243667,
      "grad_norm": 0.3003271818161011,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6236,
      "step": 45380
    },
    {
      "epoch": 0.9293920095017714,
      "grad_norm": 0.31987905502319336,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5396,
      "step": 45385
    },
    {
      "epoch": 0.929494399279176,
      "grad_norm": 0.3040323555469513,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5648,
      "step": 45390
    },
    {
      "epoch": 0.9295967890565806,
      "grad_norm": 0.2893180847167969,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5472,
      "step": 45395
    },
    {
      "epoch": 0.9296991788339852,
      "grad_norm": 0.29543089866638184,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.536,
      "step": 45400
    },
    {
      "epoch": 0.9298015686113899,
      "grad_norm": 0.307453453540802,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5594,
      "step": 45405
    },
    {
      "epoch": 0.9299039583887945,
      "grad_norm": 0.3002399206161499,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5713,
      "step": 45410
    },
    {
      "epoch": 0.9300063481661991,
      "grad_norm": 0.29182058572769165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5828,
      "step": 45415
    },
    {
      "epoch": 0.9301087379436037,
      "grad_norm": 0.3115497827529907,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5522,
      "step": 45420
    },
    {
      "epoch": 0.9302111277210083,
      "grad_norm": 0.303641676902771,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5316,
      "step": 45425
    },
    {
      "epoch": 0.930313517498413,
      "grad_norm": 0.29586124420166016,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5839,
      "step": 45430
    },
    {
      "epoch": 0.9304159072758176,
      "grad_norm": 0.29437705874443054,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6212,
      "step": 45435
    },
    {
      "epoch": 0.9305182970532222,
      "grad_norm": 0.29417547583580017,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5583,
      "step": 45440
    },
    {
      "epoch": 0.9306206868306268,
      "grad_norm": 0.30192485451698303,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5494,
      "step": 45445
    },
    {
      "epoch": 0.9307230766080314,
      "grad_norm": 0.3112773001194,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5531,
      "step": 45450
    },
    {
      "epoch": 0.9308254663854361,
      "grad_norm": 0.2834312915802002,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5793,
      "step": 45455
    },
    {
      "epoch": 0.9309278561628407,
      "grad_norm": 0.2876500189304352,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5761,
      "step": 45460
    },
    {
      "epoch": 0.9310302459402453,
      "grad_norm": 0.2978384494781494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5379,
      "step": 45465
    },
    {
      "epoch": 0.9311326357176499,
      "grad_norm": 0.28771841526031494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5884,
      "step": 45470
    },
    {
      "epoch": 0.9312350254950545,
      "grad_norm": 0.2919498383998871,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5736,
      "step": 45475
    },
    {
      "epoch": 0.9313374152724592,
      "grad_norm": 0.2844051122665405,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5499,
      "step": 45480
    },
    {
      "epoch": 0.9314398050498638,
      "grad_norm": 0.2960943281650543,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5758,
      "step": 45485
    },
    {
      "epoch": 0.9315421948272684,
      "grad_norm": 0.2949587106704712,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5149,
      "step": 45490
    },
    {
      "epoch": 0.931644584604673,
      "grad_norm": 0.30094870924949646,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6182,
      "step": 45495
    },
    {
      "epoch": 0.9317469743820777,
      "grad_norm": 0.31304439902305603,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5844,
      "step": 45500
    },
    {
      "epoch": 0.9318493641594823,
      "grad_norm": 0.29755863547325134,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.572,
      "step": 45505
    },
    {
      "epoch": 0.9319517539368869,
      "grad_norm": 0.3069724142551422,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5895,
      "step": 45510
    },
    {
      "epoch": 0.9320541437142915,
      "grad_norm": 0.2978571653366089,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5374,
      "step": 45515
    },
    {
      "epoch": 0.9321565334916961,
      "grad_norm": 0.2921341359615326,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4502,
      "step": 45520
    },
    {
      "epoch": 0.9322589232691008,
      "grad_norm": 0.28570982813835144,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.497,
      "step": 45525
    },
    {
      "epoch": 0.9323613130465055,
      "grad_norm": 0.2988256812095642,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6021,
      "step": 45530
    },
    {
      "epoch": 0.9324637028239101,
      "grad_norm": 0.28905680775642395,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6271,
      "step": 45535
    },
    {
      "epoch": 0.9325660926013147,
      "grad_norm": 0.29460835456848145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5773,
      "step": 45540
    },
    {
      "epoch": 0.9326684823787194,
      "grad_norm": 0.2926810383796692,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5584,
      "step": 45545
    },
    {
      "epoch": 0.932770872156124,
      "grad_norm": 0.3007657527923584,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 45550
    },
    {
      "epoch": 0.9328732619335286,
      "grad_norm": 0.28152361512184143,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5201,
      "step": 45555
    },
    {
      "epoch": 0.9329756517109332,
      "grad_norm": 0.2795524299144745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5287,
      "step": 45560
    },
    {
      "epoch": 0.9330780414883378,
      "grad_norm": 0.3183031678199768,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6047,
      "step": 45565
    },
    {
      "epoch": 0.9331804312657425,
      "grad_norm": 0.3116456866264343,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5388,
      "step": 45570
    },
    {
      "epoch": 0.9332828210431471,
      "grad_norm": 0.3290521800518036,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.557,
      "step": 45575
    },
    {
      "epoch": 0.9333852108205517,
      "grad_norm": 0.30255621671676636,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6539,
      "step": 45580
    },
    {
      "epoch": 0.9334876005979563,
      "grad_norm": 0.2903241515159607,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5398,
      "step": 45585
    },
    {
      "epoch": 0.933589990375361,
      "grad_norm": 0.3150928318500519,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.577,
      "step": 45590
    },
    {
      "epoch": 0.9336923801527656,
      "grad_norm": 0.3061113655567169,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5662,
      "step": 45595
    },
    {
      "epoch": 0.9337947699301702,
      "grad_norm": 0.2998338043689728,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.527,
      "step": 45600
    },
    {
      "epoch": 0.9338971597075748,
      "grad_norm": 0.3258943259716034,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6084,
      "step": 45605
    },
    {
      "epoch": 0.9339995494849794,
      "grad_norm": 0.2946352958679199,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5985,
      "step": 45610
    },
    {
      "epoch": 0.934101939262384,
      "grad_norm": 0.2906305193901062,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5695,
      "step": 45615
    },
    {
      "epoch": 0.9342043290397887,
      "grad_norm": 0.309285968542099,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5823,
      "step": 45620
    },
    {
      "epoch": 0.9343067188171933,
      "grad_norm": 0.2940753102302551,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5745,
      "step": 45625
    },
    {
      "epoch": 0.9344091085945979,
      "grad_norm": 0.29665306210517883,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5608,
      "step": 45630
    },
    {
      "epoch": 0.9345114983720025,
      "grad_norm": 0.28580552339553833,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5819,
      "step": 45635
    },
    {
      "epoch": 0.9346138881494072,
      "grad_norm": 0.30722177028656006,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5603,
      "step": 45640
    },
    {
      "epoch": 0.9347162779268118,
      "grad_norm": 0.2918175160884857,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6299,
      "step": 45645
    },
    {
      "epoch": 0.9348186677042164,
      "grad_norm": 0.2978382110595703,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5489,
      "step": 45650
    },
    {
      "epoch": 0.934921057481621,
      "grad_norm": 0.36765149235725403,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6238,
      "step": 45655
    },
    {
      "epoch": 0.9350234472590256,
      "grad_norm": 0.29678916931152344,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5749,
      "step": 45660
    },
    {
      "epoch": 0.9351258370364303,
      "grad_norm": 0.3510517179965973,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 45665
    },
    {
      "epoch": 0.9352282268138349,
      "grad_norm": 0.3230513334274292,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.554,
      "step": 45670
    },
    {
      "epoch": 0.9353306165912395,
      "grad_norm": 0.3000127375125885,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.573,
      "step": 45675
    },
    {
      "epoch": 0.9354330063686441,
      "grad_norm": 0.30339938402175903,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5225,
      "step": 45680
    },
    {
      "epoch": 0.9355353961460487,
      "grad_norm": 0.32276660203933716,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5064,
      "step": 45685
    },
    {
      "epoch": 0.9356377859234534,
      "grad_norm": 0.3067030906677246,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5368,
      "step": 45690
    },
    {
      "epoch": 0.935740175700858,
      "grad_norm": 0.2950015962123871,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 45695
    },
    {
      "epoch": 0.9358425654782626,
      "grad_norm": 0.2833322286605835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6006,
      "step": 45700
    },
    {
      "epoch": 0.9359449552556672,
      "grad_norm": 0.3211897909641266,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.528,
      "step": 45705
    },
    {
      "epoch": 0.9360473450330719,
      "grad_norm": 0.3024519383907318,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5693,
      "step": 45710
    },
    {
      "epoch": 0.9361497348104765,
      "grad_norm": 0.2962220311164856,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5626,
      "step": 45715
    },
    {
      "epoch": 0.9362521245878811,
      "grad_norm": 0.29894790053367615,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5809,
      "step": 45720
    },
    {
      "epoch": 0.9363545143652857,
      "grad_norm": 0.2957574725151062,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6018,
      "step": 45725
    },
    {
      "epoch": 0.9364569041426903,
      "grad_norm": 0.30089783668518066,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5585,
      "step": 45730
    },
    {
      "epoch": 0.9365592939200951,
      "grad_norm": 0.2974870204925537,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5591,
      "step": 45735
    },
    {
      "epoch": 0.9366616836974997,
      "grad_norm": 0.29053524136543274,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4857,
      "step": 45740
    },
    {
      "epoch": 0.9367640734749043,
      "grad_norm": 0.29384881258010864,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5834,
      "step": 45745
    },
    {
      "epoch": 0.9368664632523089,
      "grad_norm": 0.29201626777648926,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5219,
      "step": 45750
    },
    {
      "epoch": 0.9369688530297136,
      "grad_norm": 0.2854151129722595,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.561,
      "step": 45755
    },
    {
      "epoch": 0.9370712428071182,
      "grad_norm": 0.2902468740940094,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5425,
      "step": 45760
    },
    {
      "epoch": 0.9371736325845228,
      "grad_norm": 0.2796638011932373,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5831,
      "step": 45765
    },
    {
      "epoch": 0.9372760223619274,
      "grad_norm": 0.2874743938446045,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5176,
      "step": 45770
    },
    {
      "epoch": 0.937378412139332,
      "grad_norm": 0.2998960018157959,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5575,
      "step": 45775
    },
    {
      "epoch": 0.9374808019167367,
      "grad_norm": 0.30951982736587524,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.597,
      "step": 45780
    },
    {
      "epoch": 0.9375831916941413,
      "grad_norm": 0.2959824502468109,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5971,
      "step": 45785
    },
    {
      "epoch": 0.9376855814715459,
      "grad_norm": 0.3059476613998413,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 45790
    },
    {
      "epoch": 0.9377879712489505,
      "grad_norm": 0.30178895592689514,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5713,
      "step": 45795
    },
    {
      "epoch": 0.9378903610263551,
      "grad_norm": 0.285638689994812,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6093,
      "step": 45800
    },
    {
      "epoch": 0.9379927508037598,
      "grad_norm": 0.3198152184486389,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5846,
      "step": 45805
    },
    {
      "epoch": 0.9380951405811644,
      "grad_norm": 0.30595332384109497,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5842,
      "step": 45810
    },
    {
      "epoch": 0.938197530358569,
      "grad_norm": 0.293392151594162,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5139,
      "step": 45815
    },
    {
      "epoch": 0.9382999201359736,
      "grad_norm": 0.433502733707428,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6002,
      "step": 45820
    },
    {
      "epoch": 0.9384023099133783,
      "grad_norm": 0.2840341031551361,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5529,
      "step": 45825
    },
    {
      "epoch": 0.9385046996907829,
      "grad_norm": 0.295367032289505,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5641,
      "step": 45830
    },
    {
      "epoch": 0.9386070894681875,
      "grad_norm": 0.28799545764923096,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5524,
      "step": 45835
    },
    {
      "epoch": 0.9387094792455921,
      "grad_norm": 0.2966258227825165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6065,
      "step": 45840
    },
    {
      "epoch": 0.9388118690229967,
      "grad_norm": 0.29658636450767517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5539,
      "step": 45845
    },
    {
      "epoch": 0.9389142588004014,
      "grad_norm": 0.30021369457244873,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5022,
      "step": 45850
    },
    {
      "epoch": 0.939016648577806,
      "grad_norm": 0.38482242822647095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5447,
      "step": 45855
    },
    {
      "epoch": 0.9391190383552106,
      "grad_norm": 0.2887100577354431,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5726,
      "step": 45860
    },
    {
      "epoch": 0.9392214281326152,
      "grad_norm": 0.29952242970466614,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.532,
      "step": 45865
    },
    {
      "epoch": 0.9393238179100198,
      "grad_norm": 0.3206510543823242,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.557,
      "step": 45870
    },
    {
      "epoch": 0.9394262076874245,
      "grad_norm": 0.3024630546569824,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5641,
      "step": 45875
    },
    {
      "epoch": 0.9395285974648291,
      "grad_norm": 0.3287646770477295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6189,
      "step": 45880
    },
    {
      "epoch": 0.9396309872422337,
      "grad_norm": 0.30085039138793945,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5405,
      "step": 45885
    },
    {
      "epoch": 0.9397333770196383,
      "grad_norm": 0.3252756595611572,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5482,
      "step": 45890
    },
    {
      "epoch": 0.939835766797043,
      "grad_norm": 0.29676443338394165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5869,
      "step": 45895
    },
    {
      "epoch": 0.9399381565744476,
      "grad_norm": 0.3045036196708679,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5434,
      "step": 45900
    },
    {
      "epoch": 0.9400405463518522,
      "grad_norm": 0.28315460681915283,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5877,
      "step": 45905
    },
    {
      "epoch": 0.9401429361292568,
      "grad_norm": 0.3238038420677185,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.638,
      "step": 45910
    },
    {
      "epoch": 0.9402453259066614,
      "grad_norm": 0.32335028052330017,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5703,
      "step": 45915
    },
    {
      "epoch": 0.940347715684066,
      "grad_norm": 0.28923776745796204,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5442,
      "step": 45920
    },
    {
      "epoch": 0.9404501054614707,
      "grad_norm": 0.2948213517665863,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5963,
      "step": 45925
    },
    {
      "epoch": 0.9405524952388753,
      "grad_norm": 0.30786654353141785,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5651,
      "step": 45930
    },
    {
      "epoch": 0.94065488501628,
      "grad_norm": 0.2849527597427368,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5857,
      "step": 45935
    },
    {
      "epoch": 0.9407572747936847,
      "grad_norm": 0.3040143549442291,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5946,
      "step": 45940
    },
    {
      "epoch": 0.9408596645710893,
      "grad_norm": 0.3182886242866516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.596,
      "step": 45945
    },
    {
      "epoch": 0.9409620543484939,
      "grad_norm": 0.288325697183609,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5533,
      "step": 45950
    },
    {
      "epoch": 0.9410644441258985,
      "grad_norm": 0.3175342082977295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 45955
    },
    {
      "epoch": 0.9411668339033031,
      "grad_norm": 0.28953540325164795,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5448,
      "step": 45960
    },
    {
      "epoch": 0.9412692236807078,
      "grad_norm": 0.30405816435813904,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.58,
      "step": 45965
    },
    {
      "epoch": 0.9413716134581124,
      "grad_norm": 0.2976589500904083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5448,
      "step": 45970
    },
    {
      "epoch": 0.941474003235517,
      "grad_norm": 0.3012731075286865,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5902,
      "step": 45975
    },
    {
      "epoch": 0.9415763930129216,
      "grad_norm": 0.29623860120773315,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5609,
      "step": 45980
    },
    {
      "epoch": 0.9416787827903262,
      "grad_norm": 0.31744128465652466,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5766,
      "step": 45985
    },
    {
      "epoch": 0.9417811725677309,
      "grad_norm": 0.30372464656829834,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4996,
      "step": 45990
    },
    {
      "epoch": 0.9418835623451355,
      "grad_norm": 0.29425516724586487,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5925,
      "step": 45995
    },
    {
      "epoch": 0.9419859521225401,
      "grad_norm": 0.3015255928039551,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.614,
      "step": 46000
    },
    {
      "epoch": 0.9419859521225401,
      "eval_loss": 4.583684921264648,
      "eval_runtime": 17.1326,
      "eval_samples_per_second": 33.97,
      "eval_steps_per_second": 4.261,
      "step": 46000
    },
    {
      "epoch": 0.9420883418999447,
      "grad_norm": 0.3247185945510864,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5689,
      "step": 46005
    },
    {
      "epoch": 0.9421907316773493,
      "grad_norm": 0.31341496109962463,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5098,
      "step": 46010
    },
    {
      "epoch": 0.942293121454754,
      "grad_norm": 0.3268587589263916,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6441,
      "step": 46015
    },
    {
      "epoch": 0.9423955112321586,
      "grad_norm": 0.30855366587638855,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 46020
    },
    {
      "epoch": 0.9424979010095632,
      "grad_norm": 0.2909298837184906,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5305,
      "step": 46025
    },
    {
      "epoch": 0.9426002907869678,
      "grad_norm": 0.34957823157310486,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5302,
      "step": 46030
    },
    {
      "epoch": 0.9427026805643725,
      "grad_norm": 0.3359202742576599,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5863,
      "step": 46035
    },
    {
      "epoch": 0.9428050703417771,
      "grad_norm": 0.290401428937912,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5969,
      "step": 46040
    },
    {
      "epoch": 0.9429074601191817,
      "grad_norm": 0.3260645270347595,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5317,
      "step": 46045
    },
    {
      "epoch": 0.9430098498965863,
      "grad_norm": 0.3104172646999359,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5733,
      "step": 46050
    },
    {
      "epoch": 0.9431122396739909,
      "grad_norm": 0.29455071687698364,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5827,
      "step": 46055
    },
    {
      "epoch": 0.9432146294513956,
      "grad_norm": 0.3253496289253235,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5354,
      "step": 46060
    },
    {
      "epoch": 0.9433170192288002,
      "grad_norm": 0.2910245656967163,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5558,
      "step": 46065
    },
    {
      "epoch": 0.9434194090062048,
      "grad_norm": 0.28741756081581116,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5549,
      "step": 46070
    },
    {
      "epoch": 0.9435217987836094,
      "grad_norm": 0.323881059885025,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5806,
      "step": 46075
    },
    {
      "epoch": 0.943624188561014,
      "grad_norm": 0.3421439528465271,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.59,
      "step": 46080
    },
    {
      "epoch": 0.9437265783384187,
      "grad_norm": 0.2928997576236725,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.575,
      "step": 46085
    },
    {
      "epoch": 0.9438289681158233,
      "grad_norm": 0.29498398303985596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6085,
      "step": 46090
    },
    {
      "epoch": 0.9439313578932279,
      "grad_norm": 0.2876294255256653,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5466,
      "step": 46095
    },
    {
      "epoch": 0.9440337476706325,
      "grad_norm": 0.28106021881103516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5525,
      "step": 46100
    },
    {
      "epoch": 0.9441361374480371,
      "grad_norm": 0.2980225384235382,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5594,
      "step": 46105
    },
    {
      "epoch": 0.9442385272254418,
      "grad_norm": 0.29388806223869324,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4186,
      "step": 46110
    },
    {
      "epoch": 0.9443409170028464,
      "grad_norm": 0.3009973168373108,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5734,
      "step": 46115
    },
    {
      "epoch": 0.944443306780251,
      "grad_norm": 0.2896288335323334,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5197,
      "step": 46120
    },
    {
      "epoch": 0.9445456965576556,
      "grad_norm": 0.2991890609264374,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4904,
      "step": 46125
    },
    {
      "epoch": 0.9446480863350603,
      "grad_norm": 0.28481045365333557,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5626,
      "step": 46130
    },
    {
      "epoch": 0.9447504761124649,
      "grad_norm": 0.30722033977508545,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5853,
      "step": 46135
    },
    {
      "epoch": 0.9448528658898696,
      "grad_norm": 0.2884630262851715,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5902,
      "step": 46140
    },
    {
      "epoch": 0.9449552556672742,
      "grad_norm": 0.2827579975128174,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5697,
      "step": 46145
    },
    {
      "epoch": 0.9450576454446789,
      "grad_norm": 0.30568844079971313,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5744,
      "step": 46150
    },
    {
      "epoch": 0.9451600352220835,
      "grad_norm": 0.3058178126811981,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5761,
      "step": 46155
    },
    {
      "epoch": 0.9452624249994881,
      "grad_norm": 0.30323293805122375,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5446,
      "step": 46160
    },
    {
      "epoch": 0.9453648147768927,
      "grad_norm": 0.297662615776062,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5679,
      "step": 46165
    },
    {
      "epoch": 0.9454672045542973,
      "grad_norm": 0.2870453894138336,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.555,
      "step": 46170
    },
    {
      "epoch": 0.945569594331702,
      "grad_norm": 0.28932371735572815,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5938,
      "step": 46175
    },
    {
      "epoch": 0.9456719841091066,
      "grad_norm": 0.30089452862739563,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6057,
      "step": 46180
    },
    {
      "epoch": 0.9457743738865112,
      "grad_norm": 0.32060471177101135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5608,
      "step": 46185
    },
    {
      "epoch": 0.9458767636639158,
      "grad_norm": 0.28784120082855225,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5696,
      "step": 46190
    },
    {
      "epoch": 0.9459791534413204,
      "grad_norm": 0.2971588671207428,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6289,
      "step": 46195
    },
    {
      "epoch": 0.9460815432187251,
      "grad_norm": 0.3049401044845581,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6121,
      "step": 46200
    },
    {
      "epoch": 0.9461839329961297,
      "grad_norm": 0.3207658529281616,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5777,
      "step": 46205
    },
    {
      "epoch": 0.9462863227735343,
      "grad_norm": 0.2914637327194214,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5864,
      "step": 46210
    },
    {
      "epoch": 0.9463887125509389,
      "grad_norm": 0.2939217686653137,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5182,
      "step": 46215
    },
    {
      "epoch": 0.9464911023283435,
      "grad_norm": 0.2861398458480835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5119,
      "step": 46220
    },
    {
      "epoch": 0.9465934921057482,
      "grad_norm": 0.30429330468177795,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5959,
      "step": 46225
    },
    {
      "epoch": 0.9466958818831528,
      "grad_norm": 0.35886383056640625,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5833,
      "step": 46230
    },
    {
      "epoch": 0.9467982716605574,
      "grad_norm": 0.291823148727417,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.515,
      "step": 46235
    },
    {
      "epoch": 0.946900661437962,
      "grad_norm": 0.2943173944950104,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5744,
      "step": 46240
    },
    {
      "epoch": 0.9470030512153667,
      "grad_norm": 0.4758606553077698,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.547,
      "step": 46245
    },
    {
      "epoch": 0.9471054409927713,
      "grad_norm": 0.372763067483902,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4849,
      "step": 46250
    },
    {
      "epoch": 0.9472078307701759,
      "grad_norm": 0.28862863779067993,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4833,
      "step": 46255
    },
    {
      "epoch": 0.9473102205475805,
      "grad_norm": 0.2902795970439911,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5262,
      "step": 46260
    },
    {
      "epoch": 0.9474126103249851,
      "grad_norm": 0.2977970540523529,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5321,
      "step": 46265
    },
    {
      "epoch": 0.9475150001023898,
      "grad_norm": 0.4046977162361145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5846,
      "step": 46270
    },
    {
      "epoch": 0.9476173898797944,
      "grad_norm": 0.29380640387535095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5819,
      "step": 46275
    },
    {
      "epoch": 0.947719779657199,
      "grad_norm": 0.3022974729537964,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5197,
      "step": 46280
    },
    {
      "epoch": 0.9478221694346036,
      "grad_norm": 0.29053810238838196,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5742,
      "step": 46285
    },
    {
      "epoch": 0.9479245592120082,
      "grad_norm": 0.27897557616233826,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5777,
      "step": 46290
    },
    {
      "epoch": 0.9480269489894129,
      "grad_norm": 0.28883323073387146,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5788,
      "step": 46295
    },
    {
      "epoch": 0.9481293387668175,
      "grad_norm": 0.31002146005630493,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5711,
      "step": 46300
    },
    {
      "epoch": 0.9482317285442221,
      "grad_norm": 0.3166657090187073,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5532,
      "step": 46305
    },
    {
      "epoch": 0.9483341183216267,
      "grad_norm": 0.3122865855693817,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5489,
      "step": 46310
    },
    {
      "epoch": 0.9484365080990314,
      "grad_norm": 0.2848620116710663,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5409,
      "step": 46315
    },
    {
      "epoch": 0.948538897876436,
      "grad_norm": 0.3065323531627655,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5402,
      "step": 46320
    },
    {
      "epoch": 0.9486412876538406,
      "grad_norm": 0.2915879786014557,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5876,
      "step": 46325
    },
    {
      "epoch": 0.9487436774312452,
      "grad_norm": 0.3044731616973877,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4693,
      "step": 46330
    },
    {
      "epoch": 0.9488460672086498,
      "grad_norm": 0.33324384689331055,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5919,
      "step": 46335
    },
    {
      "epoch": 0.9489484569860546,
      "grad_norm": 0.2969648838043213,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5418,
      "step": 46340
    },
    {
      "epoch": 0.9490508467634592,
      "grad_norm": 0.2988201081752777,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5901,
      "step": 46345
    },
    {
      "epoch": 0.9491532365408638,
      "grad_norm": 0.3024013936519623,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.605,
      "step": 46350
    },
    {
      "epoch": 0.9492556263182684,
      "grad_norm": 0.302666574716568,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5439,
      "step": 46355
    },
    {
      "epoch": 0.949358016095673,
      "grad_norm": 0.3279913663864136,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5416,
      "step": 46360
    },
    {
      "epoch": 0.9494604058730777,
      "grad_norm": 0.3035995066165924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5195,
      "step": 46365
    },
    {
      "epoch": 0.9495627956504823,
      "grad_norm": 0.30347394943237305,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5328,
      "step": 46370
    },
    {
      "epoch": 0.9496651854278869,
      "grad_norm": 0.3168480098247528,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.484,
      "step": 46375
    },
    {
      "epoch": 0.9497675752052915,
      "grad_norm": 0.30056774616241455,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5872,
      "step": 46380
    },
    {
      "epoch": 0.9498699649826962,
      "grad_norm": 0.29635190963745117,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5843,
      "step": 46385
    },
    {
      "epoch": 0.9499723547601008,
      "grad_norm": 0.28613218665122986,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5818,
      "step": 46390
    },
    {
      "epoch": 0.9500747445375054,
      "grad_norm": 0.35271692276000977,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5918,
      "step": 46395
    },
    {
      "epoch": 0.95017713431491,
      "grad_norm": 0.2879866659641266,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5044,
      "step": 46400
    },
    {
      "epoch": 0.9502795240923146,
      "grad_norm": 0.2910770773887634,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.553,
      "step": 46405
    },
    {
      "epoch": 0.9503819138697193,
      "grad_norm": 0.2987903356552124,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5389,
      "step": 46410
    },
    {
      "epoch": 0.9504843036471239,
      "grad_norm": 0.28610172867774963,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5486,
      "step": 46415
    },
    {
      "epoch": 0.9505866934245285,
      "grad_norm": 0.3817905783653259,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5868,
      "step": 46420
    },
    {
      "epoch": 0.9506890832019331,
      "grad_norm": 0.31912466883659363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5107,
      "step": 46425
    },
    {
      "epoch": 0.9507914729793377,
      "grad_norm": 0.2898521423339844,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6252,
      "step": 46430
    },
    {
      "epoch": 0.9508938627567424,
      "grad_norm": 0.29013702273368835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5944,
      "step": 46435
    },
    {
      "epoch": 0.950996252534147,
      "grad_norm": 0.29249125719070435,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5531,
      "step": 46440
    },
    {
      "epoch": 0.9510986423115516,
      "grad_norm": 0.284308522939682,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5684,
      "step": 46445
    },
    {
      "epoch": 0.9512010320889562,
      "grad_norm": 0.30925360321998596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5651,
      "step": 46450
    },
    {
      "epoch": 0.9513034218663609,
      "grad_norm": 0.29362016916275024,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5901,
      "step": 46455
    },
    {
      "epoch": 0.9514058116437655,
      "grad_norm": 0.3067672848701477,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5692,
      "step": 46460
    },
    {
      "epoch": 0.9515082014211701,
      "grad_norm": 0.29780399799346924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5946,
      "step": 46465
    },
    {
      "epoch": 0.9516105911985747,
      "grad_norm": 0.3147890865802765,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.55,
      "step": 46470
    },
    {
      "epoch": 0.9517129809759793,
      "grad_norm": 0.2933990955352783,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5795,
      "step": 46475
    },
    {
      "epoch": 0.951815370753384,
      "grad_norm": 0.3006732165813446,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5222,
      "step": 46480
    },
    {
      "epoch": 0.9519177605307886,
      "grad_norm": 0.31373563408851624,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5303,
      "step": 46485
    },
    {
      "epoch": 0.9520201503081932,
      "grad_norm": 0.3008566200733185,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5521,
      "step": 46490
    },
    {
      "epoch": 0.9521225400855978,
      "grad_norm": 0.2952241003513336,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5618,
      "step": 46495
    },
    {
      "epoch": 0.9522249298630024,
      "grad_norm": 0.29259973764419556,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5506,
      "step": 46500
    },
    {
      "epoch": 0.9523273196404071,
      "grad_norm": 0.29110756516456604,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5649,
      "step": 46505
    },
    {
      "epoch": 0.9524297094178117,
      "grad_norm": 0.2845367193222046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5585,
      "step": 46510
    },
    {
      "epoch": 0.9525320991952163,
      "grad_norm": 0.28852665424346924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.612,
      "step": 46515
    },
    {
      "epoch": 0.9526344889726209,
      "grad_norm": 0.291458398103714,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6069,
      "step": 46520
    },
    {
      "epoch": 0.9527368787500256,
      "grad_norm": 0.3163106441497803,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.558,
      "step": 46525
    },
    {
      "epoch": 0.9528392685274302,
      "grad_norm": 0.3323710560798645,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5538,
      "step": 46530
    },
    {
      "epoch": 0.9529416583048348,
      "grad_norm": 0.3018573224544525,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5715,
      "step": 46535
    },
    {
      "epoch": 0.9530440480822394,
      "grad_norm": 0.28915610909461975,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5374,
      "step": 46540
    },
    {
      "epoch": 0.9531464378596441,
      "grad_norm": 0.3006097972393036,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5792,
      "step": 46545
    },
    {
      "epoch": 0.9532488276370488,
      "grad_norm": 0.28617480397224426,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5852,
      "step": 46550
    },
    {
      "epoch": 0.9533512174144534,
      "grad_norm": 0.2876349091529846,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.558,
      "step": 46555
    },
    {
      "epoch": 0.953453607191858,
      "grad_norm": 0.29858532547950745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5807,
      "step": 46560
    },
    {
      "epoch": 0.9535559969692626,
      "grad_norm": 0.31634581089019775,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5456,
      "step": 46565
    },
    {
      "epoch": 0.9536583867466673,
      "grad_norm": 0.6948842406272888,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5797,
      "step": 46570
    },
    {
      "epoch": 0.9537607765240719,
      "grad_norm": 0.29457876086235046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5399,
      "step": 46575
    },
    {
      "epoch": 0.9538631663014765,
      "grad_norm": 0.29184049367904663,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.535,
      "step": 46580
    },
    {
      "epoch": 0.9539655560788811,
      "grad_norm": 0.29152342677116394,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6023,
      "step": 46585
    },
    {
      "epoch": 0.9540679458562857,
      "grad_norm": 0.3121853172779083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5459,
      "step": 46590
    },
    {
      "epoch": 0.9541703356336904,
      "grad_norm": 0.29832789301872253,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.573,
      "step": 46595
    },
    {
      "epoch": 0.954272725411095,
      "grad_norm": 0.29351404309272766,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5808,
      "step": 46600
    },
    {
      "epoch": 0.9543751151884996,
      "grad_norm": 0.29937365651130676,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5569,
      "step": 46605
    },
    {
      "epoch": 0.9544775049659042,
      "grad_norm": 0.2994340658187866,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5266,
      "step": 46610
    },
    {
      "epoch": 0.9545798947433088,
      "grad_norm": 0.29143932461738586,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5655,
      "step": 46615
    },
    {
      "epoch": 0.9546822845207135,
      "grad_norm": 0.28990811109542847,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5287,
      "step": 46620
    },
    {
      "epoch": 0.9547846742981181,
      "grad_norm": 0.286898672580719,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5411,
      "step": 46625
    },
    {
      "epoch": 0.9548870640755227,
      "grad_norm": 0.28920093178749084,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5699,
      "step": 46630
    },
    {
      "epoch": 0.9549894538529273,
      "grad_norm": 0.30701369047164917,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5505,
      "step": 46635
    },
    {
      "epoch": 0.955091843630332,
      "grad_norm": 0.2999051511287689,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6291,
      "step": 46640
    },
    {
      "epoch": 0.9551942334077366,
      "grad_norm": 0.30438244342803955,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5471,
      "step": 46645
    },
    {
      "epoch": 0.9552966231851412,
      "grad_norm": 0.30867093801498413,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5316,
      "step": 46650
    },
    {
      "epoch": 0.9553990129625458,
      "grad_norm": 0.29062962532043457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5951,
      "step": 46655
    },
    {
      "epoch": 0.9555014027399504,
      "grad_norm": 0.2844584286212921,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5841,
      "step": 46660
    },
    {
      "epoch": 0.955603792517355,
      "grad_norm": 0.28820452094078064,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5852,
      "step": 46665
    },
    {
      "epoch": 0.9557061822947597,
      "grad_norm": 0.3631156086921692,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5461,
      "step": 46670
    },
    {
      "epoch": 0.9558085720721643,
      "grad_norm": 0.3036566972732544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5925,
      "step": 46675
    },
    {
      "epoch": 0.9559109618495689,
      "grad_norm": 0.30913856625556946,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5955,
      "step": 46680
    },
    {
      "epoch": 0.9560133516269735,
      "grad_norm": 0.3142358660697937,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5883,
      "step": 46685
    },
    {
      "epoch": 0.9561157414043782,
      "grad_norm": 0.2955227196216583,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.607,
      "step": 46690
    },
    {
      "epoch": 0.9562181311817828,
      "grad_norm": 0.3719719648361206,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5709,
      "step": 46695
    },
    {
      "epoch": 0.9563205209591874,
      "grad_norm": 0.2889150381088257,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 46700
    },
    {
      "epoch": 0.956422910736592,
      "grad_norm": 0.3480525314807892,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5391,
      "step": 46705
    },
    {
      "epoch": 0.9565253005139966,
      "grad_norm": 0.2765401303768158,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4554,
      "step": 46710
    },
    {
      "epoch": 0.9566276902914013,
      "grad_norm": 0.2994229793548584,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6014,
      "step": 46715
    },
    {
      "epoch": 0.9567300800688059,
      "grad_norm": 0.2966138422489166,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.557,
      "step": 46720
    },
    {
      "epoch": 0.9568324698462105,
      "grad_norm": 0.3034304678440094,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5691,
      "step": 46725
    },
    {
      "epoch": 0.9569348596236151,
      "grad_norm": 0.30969443917274475,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5806,
      "step": 46730
    },
    {
      "epoch": 0.9570372494010198,
      "grad_norm": 0.3021763265132904,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6384,
      "step": 46735
    },
    {
      "epoch": 0.9571396391784244,
      "grad_norm": 0.29978522658348083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5576,
      "step": 46740
    },
    {
      "epoch": 0.9572420289558291,
      "grad_norm": 0.2940644919872284,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.548,
      "step": 46745
    },
    {
      "epoch": 0.9573444187332337,
      "grad_norm": 0.2972068190574646,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5625,
      "step": 46750
    },
    {
      "epoch": 0.9574468085106383,
      "grad_norm": 0.30514392256736755,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.51,
      "step": 46755
    },
    {
      "epoch": 0.957549198288043,
      "grad_norm": 0.31049951910972595,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6043,
      "step": 46760
    },
    {
      "epoch": 0.9576515880654476,
      "grad_norm": 0.28459200263023376,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5769,
      "step": 46765
    },
    {
      "epoch": 0.9577539778428522,
      "grad_norm": 0.31559404730796814,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5461,
      "step": 46770
    },
    {
      "epoch": 0.9578563676202568,
      "grad_norm": 0.30458056926727295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6753,
      "step": 46775
    },
    {
      "epoch": 0.9579587573976615,
      "grad_norm": 0.3196863532066345,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6074,
      "step": 46780
    },
    {
      "epoch": 0.9580611471750661,
      "grad_norm": 0.3843401372432709,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5677,
      "step": 46785
    },
    {
      "epoch": 0.9581635369524707,
      "grad_norm": 0.29096993803977966,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5799,
      "step": 46790
    },
    {
      "epoch": 0.9582659267298753,
      "grad_norm": 0.28865140676498413,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5081,
      "step": 46795
    },
    {
      "epoch": 0.9583683165072799,
      "grad_norm": 0.29790523648262024,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6181,
      "step": 46800
    },
    {
      "epoch": 0.9584707062846846,
      "grad_norm": 0.29815900325775146,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.533,
      "step": 46805
    },
    {
      "epoch": 0.9585730960620892,
      "grad_norm": 0.28969037532806396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5034,
      "step": 46810
    },
    {
      "epoch": 0.9586754858394938,
      "grad_norm": 0.2895985245704651,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.553,
      "step": 46815
    },
    {
      "epoch": 0.9587778756168984,
      "grad_norm": 0.2950066328048706,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5655,
      "step": 46820
    },
    {
      "epoch": 0.958880265394303,
      "grad_norm": 0.30235326290130615,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5666,
      "step": 46825
    },
    {
      "epoch": 0.9589826551717077,
      "grad_norm": 0.2880280613899231,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5212,
      "step": 46830
    },
    {
      "epoch": 0.9590850449491123,
      "grad_norm": 1.0736048221588135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6092,
      "step": 46835
    },
    {
      "epoch": 0.9591874347265169,
      "grad_norm": 0.3134042024612427,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4997,
      "step": 46840
    },
    {
      "epoch": 0.9592898245039215,
      "grad_norm": 0.32664966583251953,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5354,
      "step": 46845
    },
    {
      "epoch": 0.9593922142813261,
      "grad_norm": 0.2921341061592102,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5767,
      "step": 46850
    },
    {
      "epoch": 0.9594946040587308,
      "grad_norm": 0.29374033212661743,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.571,
      "step": 46855
    },
    {
      "epoch": 0.9595969938361354,
      "grad_norm": 0.3189680874347687,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6013,
      "step": 46860
    },
    {
      "epoch": 0.95969938361354,
      "grad_norm": 0.29350003600120544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5769,
      "step": 46865
    },
    {
      "epoch": 0.9598017733909446,
      "grad_norm": 0.2955159544944763,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5107,
      "step": 46870
    },
    {
      "epoch": 0.9599041631683493,
      "grad_norm": 0.2950364053249359,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5467,
      "step": 46875
    },
    {
      "epoch": 0.9600065529457539,
      "grad_norm": 0.2971443235874176,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5136,
      "step": 46880
    },
    {
      "epoch": 0.9601089427231585,
      "grad_norm": 0.30741390585899353,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.536,
      "step": 46885
    },
    {
      "epoch": 0.9602113325005631,
      "grad_norm": 0.286006897687912,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5613,
      "step": 46890
    },
    {
      "epoch": 0.9603137222779677,
      "grad_norm": 0.3020000755786896,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5845,
      "step": 46895
    },
    {
      "epoch": 0.9604161120553724,
      "grad_norm": 0.2949473261833191,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5681,
      "step": 46900
    },
    {
      "epoch": 0.960518501832777,
      "grad_norm": 0.2953351140022278,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 46905
    },
    {
      "epoch": 0.9606208916101816,
      "grad_norm": 0.34834355115890503,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5961,
      "step": 46910
    },
    {
      "epoch": 0.9607232813875862,
      "grad_norm": 0.3003140091896057,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5965,
      "step": 46915
    },
    {
      "epoch": 0.9608256711649908,
      "grad_norm": 0.2925507426261902,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5687,
      "step": 46920
    },
    {
      "epoch": 0.9609280609423955,
      "grad_norm": 0.2905387580394745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5204,
      "step": 46925
    },
    {
      "epoch": 0.9610304507198001,
      "grad_norm": 0.3015080988407135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5774,
      "step": 46930
    },
    {
      "epoch": 0.9611328404972047,
      "grad_norm": 0.29733699560165405,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5868,
      "step": 46935
    },
    {
      "epoch": 0.9612352302746093,
      "grad_norm": 0.2915636897087097,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5334,
      "step": 46940
    },
    {
      "epoch": 0.961337620052014,
      "grad_norm": 0.30577629804611206,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.547,
      "step": 46945
    },
    {
      "epoch": 0.9614400098294187,
      "grad_norm": 0.31387439370155334,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6124,
      "step": 46950
    },
    {
      "epoch": 0.9615423996068233,
      "grad_norm": 0.39872995018959045,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5962,
      "step": 46955
    },
    {
      "epoch": 0.9616447893842279,
      "grad_norm": 0.32546061277389526,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4662,
      "step": 46960
    },
    {
      "epoch": 0.9617471791616325,
      "grad_norm": 0.3042566478252411,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5145,
      "step": 46965
    },
    {
      "epoch": 0.9618495689390372,
      "grad_norm": 0.29946258664131165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5425,
      "step": 46970
    },
    {
      "epoch": 0.9619519587164418,
      "grad_norm": 0.29231250286102295,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6074,
      "step": 46975
    },
    {
      "epoch": 0.9620543484938464,
      "grad_norm": 0.29093030095100403,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5394,
      "step": 46980
    },
    {
      "epoch": 0.962156738271251,
      "grad_norm": 0.3020989000797272,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5101,
      "step": 46985
    },
    {
      "epoch": 0.9622591280486557,
      "grad_norm": 0.30913466215133667,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5595,
      "step": 46990
    },
    {
      "epoch": 0.9623615178260603,
      "grad_norm": 0.2933889925479889,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5866,
      "step": 46995
    },
    {
      "epoch": 0.9624639076034649,
      "grad_norm": 0.29119619727134705,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5282,
      "step": 47000
    },
    {
      "epoch": 0.9624639076034649,
      "eval_loss": 4.581810474395752,
      "eval_runtime": 17.1294,
      "eval_samples_per_second": 33.977,
      "eval_steps_per_second": 4.262,
      "step": 47000
    },
    {
      "epoch": 0.9625662973808695,
      "grad_norm": 0.2967391610145569,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5758,
      "step": 47005
    },
    {
      "epoch": 0.9626686871582741,
      "grad_norm": 0.2947872579097748,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5727,
      "step": 47010
    },
    {
      "epoch": 0.9627710769356788,
      "grad_norm": 0.29944783449172974,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5765,
      "step": 47015
    },
    {
      "epoch": 0.9628734667130834,
      "grad_norm": 0.3540767729282379,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5526,
      "step": 47020
    },
    {
      "epoch": 0.962975856490488,
      "grad_norm": 0.2941175699234009,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5368,
      "step": 47025
    },
    {
      "epoch": 0.9630782462678926,
      "grad_norm": 0.29408174753189087,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5939,
      "step": 47030
    },
    {
      "epoch": 0.9631806360452972,
      "grad_norm": 0.3078875243663788,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5505,
      "step": 47035
    },
    {
      "epoch": 0.9632830258227019,
      "grad_norm": 0.36331120133399963,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5808,
      "step": 47040
    },
    {
      "epoch": 0.9633854156001065,
      "grad_norm": 0.3171165883541107,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5744,
      "step": 47045
    },
    {
      "epoch": 0.9634878053775111,
      "grad_norm": 0.29271119832992554,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5762,
      "step": 47050
    },
    {
      "epoch": 0.9635901951549157,
      "grad_norm": 0.31483906507492065,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6228,
      "step": 47055
    },
    {
      "epoch": 0.9636925849323204,
      "grad_norm": 0.3047357499599457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5675,
      "step": 47060
    },
    {
      "epoch": 0.963794974709725,
      "grad_norm": 0.2942754328250885,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5684,
      "step": 47065
    },
    {
      "epoch": 0.9638973644871296,
      "grad_norm": 0.2990550994873047,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6079,
      "step": 47070
    },
    {
      "epoch": 0.9639997542645342,
      "grad_norm": 0.2954290211200714,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5651,
      "step": 47075
    },
    {
      "epoch": 0.9641021440419388,
      "grad_norm": 0.2928082048892975,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5688,
      "step": 47080
    },
    {
      "epoch": 0.9642045338193435,
      "grad_norm": 0.3556124269962311,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.57,
      "step": 47085
    },
    {
      "epoch": 0.9643069235967481,
      "grad_norm": 0.31088879704475403,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5054,
      "step": 47090
    },
    {
      "epoch": 0.9644093133741527,
      "grad_norm": 0.3035901188850403,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5674,
      "step": 47095
    },
    {
      "epoch": 0.9645117031515573,
      "grad_norm": 0.2908082604408264,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5235,
      "step": 47100
    },
    {
      "epoch": 0.9646140929289619,
      "grad_norm": 0.303617924451828,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5973,
      "step": 47105
    },
    {
      "epoch": 0.9647164827063666,
      "grad_norm": 0.2881530821323395,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5752,
      "step": 47110
    },
    {
      "epoch": 0.9648188724837712,
      "grad_norm": 0.29704049229621887,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5137,
      "step": 47115
    },
    {
      "epoch": 0.9649212622611758,
      "grad_norm": 0.29342731833457947,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5698,
      "step": 47120
    },
    {
      "epoch": 0.9650236520385804,
      "grad_norm": 0.29932162165641785,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6056,
      "step": 47125
    },
    {
      "epoch": 0.965126041815985,
      "grad_norm": 0.29408302903175354,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5952,
      "step": 47130
    },
    {
      "epoch": 0.9652284315933897,
      "grad_norm": 0.30963706970214844,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5404,
      "step": 47135
    },
    {
      "epoch": 0.9653308213707943,
      "grad_norm": 0.31054189801216125,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.587,
      "step": 47140
    },
    {
      "epoch": 0.9654332111481989,
      "grad_norm": 0.3025176227092743,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5372,
      "step": 47145
    },
    {
      "epoch": 0.9655356009256036,
      "grad_norm": 0.32492348551750183,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5947,
      "step": 47150
    },
    {
      "epoch": 0.9656379907030083,
      "grad_norm": 0.30467331409454346,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5264,
      "step": 47155
    },
    {
      "epoch": 0.9657403804804129,
      "grad_norm": 0.8612712621688843,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.546,
      "step": 47160
    },
    {
      "epoch": 0.9658427702578175,
      "grad_norm": 0.2954406440258026,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5654,
      "step": 47165
    },
    {
      "epoch": 0.9659451600352221,
      "grad_norm": 0.29305946826934814,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6023,
      "step": 47170
    },
    {
      "epoch": 0.9660475498126267,
      "grad_norm": 0.30675962567329407,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5784,
      "step": 47175
    },
    {
      "epoch": 0.9661499395900314,
      "grad_norm": 0.29173755645751953,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6345,
      "step": 47180
    },
    {
      "epoch": 0.966252329367436,
      "grad_norm": 0.35791778564453125,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5232,
      "step": 47185
    },
    {
      "epoch": 0.9663547191448406,
      "grad_norm": 0.30453601479530334,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4895,
      "step": 47190
    },
    {
      "epoch": 0.9664571089222452,
      "grad_norm": 0.2907402217388153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5722,
      "step": 47195
    },
    {
      "epoch": 0.9665594986996499,
      "grad_norm": 0.2840563654899597,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5382,
      "step": 47200
    },
    {
      "epoch": 0.9666618884770545,
      "grad_norm": 0.3044736087322235,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.559,
      "step": 47205
    },
    {
      "epoch": 0.9667642782544591,
      "grad_norm": 0.3105946183204651,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5034,
      "step": 47210
    },
    {
      "epoch": 0.9668666680318637,
      "grad_norm": 0.3129786550998688,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4847,
      "step": 47215
    },
    {
      "epoch": 0.9669690578092683,
      "grad_norm": 0.29703766107559204,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.614,
      "step": 47220
    },
    {
      "epoch": 0.967071447586673,
      "grad_norm": 0.30712464451789856,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5898,
      "step": 47225
    },
    {
      "epoch": 0.9671738373640776,
      "grad_norm": 0.2927931845188141,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5368,
      "step": 47230
    },
    {
      "epoch": 0.9672762271414822,
      "grad_norm": 0.29242026805877686,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5323,
      "step": 47235
    },
    {
      "epoch": 0.9673786169188868,
      "grad_norm": 0.29771849513053894,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5597,
      "step": 47240
    },
    {
      "epoch": 0.9674810066962914,
      "grad_norm": 0.30348271131515503,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5753,
      "step": 47245
    },
    {
      "epoch": 0.9675833964736961,
      "grad_norm": 0.3018856942653656,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5624,
      "step": 47250
    },
    {
      "epoch": 0.9676857862511007,
      "grad_norm": 0.3198246955871582,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5444,
      "step": 47255
    },
    {
      "epoch": 0.9677881760285053,
      "grad_norm": 0.30506032705307007,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6412,
      "step": 47260
    },
    {
      "epoch": 0.9678905658059099,
      "grad_norm": 0.3034096956253052,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5816,
      "step": 47265
    },
    {
      "epoch": 0.9679929555833146,
      "grad_norm": 0.30574262142181396,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5733,
      "step": 47270
    },
    {
      "epoch": 0.9680953453607192,
      "grad_norm": 0.3164641857147217,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5484,
      "step": 47275
    },
    {
      "epoch": 0.9681977351381238,
      "grad_norm": 0.29358360171318054,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5942,
      "step": 47280
    },
    {
      "epoch": 0.9683001249155284,
      "grad_norm": 0.28766998648643494,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5644,
      "step": 47285
    },
    {
      "epoch": 0.968402514692933,
      "grad_norm": 0.29751890897750854,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5859,
      "step": 47290
    },
    {
      "epoch": 0.9685049044703377,
      "grad_norm": 0.29219794273376465,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5898,
      "step": 47295
    },
    {
      "epoch": 0.9686072942477423,
      "grad_norm": 0.3122159540653229,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5746,
      "step": 47300
    },
    {
      "epoch": 0.9687096840251469,
      "grad_norm": 0.28809303045272827,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5835,
      "step": 47305
    },
    {
      "epoch": 0.9688120738025515,
      "grad_norm": 0.2955372631549835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5614,
      "step": 47310
    },
    {
      "epoch": 0.9689144635799561,
      "grad_norm": 0.3172515034675598,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6072,
      "step": 47315
    },
    {
      "epoch": 0.9690168533573608,
      "grad_norm": 0.32738354802131653,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5829,
      "step": 47320
    },
    {
      "epoch": 0.9691192431347654,
      "grad_norm": 0.3401321470737457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5463,
      "step": 47325
    },
    {
      "epoch": 0.96922163291217,
      "grad_norm": 0.30369314551353455,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.595,
      "step": 47330
    },
    {
      "epoch": 0.9693240226895746,
      "grad_norm": 0.30510661005973816,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5918,
      "step": 47335
    },
    {
      "epoch": 0.9694264124669792,
      "grad_norm": 0.30570051074028015,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.591,
      "step": 47340
    },
    {
      "epoch": 0.9695288022443839,
      "grad_norm": 0.31844812631607056,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6063,
      "step": 47345
    },
    {
      "epoch": 0.9696311920217885,
      "grad_norm": 0.3160530924797058,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.584,
      "step": 47350
    },
    {
      "epoch": 0.9697335817991932,
      "grad_norm": 0.29038333892822266,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5466,
      "step": 47355
    },
    {
      "epoch": 0.9698359715765978,
      "grad_norm": 0.3079160749912262,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.486,
      "step": 47360
    },
    {
      "epoch": 0.9699383613540025,
      "grad_norm": 0.2997553050518036,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6374,
      "step": 47365
    },
    {
      "epoch": 0.9700407511314071,
      "grad_norm": 0.31009504199028015,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5786,
      "step": 47370
    },
    {
      "epoch": 0.9701431409088117,
      "grad_norm": 0.2981540858745575,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5433,
      "step": 47375
    },
    {
      "epoch": 0.9702455306862163,
      "grad_norm": 0.3187463581562042,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5122,
      "step": 47380
    },
    {
      "epoch": 0.970347920463621,
      "grad_norm": 0.29447510838508606,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5587,
      "step": 47385
    },
    {
      "epoch": 0.9704503102410256,
      "grad_norm": 0.29934313893318176,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5436,
      "step": 47390
    },
    {
      "epoch": 0.9705527000184302,
      "grad_norm": 0.284311980009079,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5495,
      "step": 47395
    },
    {
      "epoch": 0.9706550897958348,
      "grad_norm": 0.3030093014240265,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5605,
      "step": 47400
    },
    {
      "epoch": 0.9707574795732394,
      "grad_norm": 0.3019144833087921,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5776,
      "step": 47405
    },
    {
      "epoch": 0.9708598693506441,
      "grad_norm": 0.32203003764152527,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5775,
      "step": 47410
    },
    {
      "epoch": 0.9709622591280487,
      "grad_norm": 0.3082998991012573,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5634,
      "step": 47415
    },
    {
      "epoch": 0.9710646489054533,
      "grad_norm": 0.3060787320137024,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5531,
      "step": 47420
    },
    {
      "epoch": 0.9711670386828579,
      "grad_norm": 0.29228317737579346,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5748,
      "step": 47425
    },
    {
      "epoch": 0.9712694284602625,
      "grad_norm": 0.31799471378326416,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.572,
      "step": 47430
    },
    {
      "epoch": 0.9713718182376672,
      "grad_norm": 0.3584194481372833,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5999,
      "step": 47435
    },
    {
      "epoch": 0.9714742080150718,
      "grad_norm": 0.31477686762809753,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5323,
      "step": 47440
    },
    {
      "epoch": 0.9715765977924764,
      "grad_norm": 0.30608507990837097,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5314,
      "step": 47445
    },
    {
      "epoch": 0.971678987569881,
      "grad_norm": 0.3086124062538147,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5698,
      "step": 47450
    },
    {
      "epoch": 0.9717813773472856,
      "grad_norm": 0.3038252294063568,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5568,
      "step": 47455
    },
    {
      "epoch": 0.9718837671246903,
      "grad_norm": 0.3878996670246124,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5692,
      "step": 47460
    },
    {
      "epoch": 0.9719861569020949,
      "grad_norm": 0.30116572976112366,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5706,
      "step": 47465
    },
    {
      "epoch": 0.9720885466794995,
      "grad_norm": 0.2941262722015381,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5973,
      "step": 47470
    },
    {
      "epoch": 0.9721909364569041,
      "grad_norm": 0.3013308346271515,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.527,
      "step": 47475
    },
    {
      "epoch": 0.9722933262343088,
      "grad_norm": 0.32226625084877014,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6295,
      "step": 47480
    },
    {
      "epoch": 0.9723957160117134,
      "grad_norm": 0.294238805770874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5321,
      "step": 47485
    },
    {
      "epoch": 0.972498105789118,
      "grad_norm": 0.30070731043815613,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5867,
      "step": 47490
    },
    {
      "epoch": 0.9726004955665226,
      "grad_norm": 0.32017284631729126,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5652,
      "step": 47495
    },
    {
      "epoch": 0.9727028853439272,
      "grad_norm": 0.3248218894004822,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5568,
      "step": 47500
    },
    {
      "epoch": 0.9728052751213319,
      "grad_norm": 0.3047269582748413,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6889,
      "step": 47505
    },
    {
      "epoch": 0.9729076648987365,
      "grad_norm": 0.29684582352638245,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5553,
      "step": 47510
    },
    {
      "epoch": 0.9730100546761411,
      "grad_norm": 0.3022248148918152,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5786,
      "step": 47515
    },
    {
      "epoch": 0.9731124444535457,
      "grad_norm": 0.296001672744751,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5842,
      "step": 47520
    },
    {
      "epoch": 0.9732148342309503,
      "grad_norm": 0.3527770936489105,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5337,
      "step": 47525
    },
    {
      "epoch": 0.973317224008355,
      "grad_norm": 0.31018128991127014,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5359,
      "step": 47530
    },
    {
      "epoch": 0.9734196137857596,
      "grad_norm": 0.29156750440597534,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5901,
      "step": 47535
    },
    {
      "epoch": 0.9735220035631642,
      "grad_norm": 0.3041020631790161,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5851,
      "step": 47540
    },
    {
      "epoch": 0.9736243933405688,
      "grad_norm": 0.30444443225860596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6148,
      "step": 47545
    },
    {
      "epoch": 0.9737267831179734,
      "grad_norm": 0.29533252120018005,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5348,
      "step": 47550
    },
    {
      "epoch": 0.9738291728953781,
      "grad_norm": 0.2916995882987976,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6002,
      "step": 47555
    },
    {
      "epoch": 0.9739315626727828,
      "grad_norm": 0.2841915190219879,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5402,
      "step": 47560
    },
    {
      "epoch": 0.9740339524501874,
      "grad_norm": 0.2930445075035095,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5564,
      "step": 47565
    },
    {
      "epoch": 0.974136342227592,
      "grad_norm": 0.29968494176864624,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 47570
    },
    {
      "epoch": 0.9742387320049967,
      "grad_norm": 0.30154654383659363,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6124,
      "step": 47575
    },
    {
      "epoch": 0.9743411217824013,
      "grad_norm": 0.3050684630870819,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4887,
      "step": 47580
    },
    {
      "epoch": 0.9744435115598059,
      "grad_norm": 0.3294299840927124,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5553,
      "step": 47585
    },
    {
      "epoch": 0.9745459013372105,
      "grad_norm": 0.29346132278442383,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5657,
      "step": 47590
    },
    {
      "epoch": 0.9746482911146152,
      "grad_norm": 0.30177071690559387,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5472,
      "step": 47595
    },
    {
      "epoch": 0.9747506808920198,
      "grad_norm": 0.29101189970970154,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5339,
      "step": 47600
    },
    {
      "epoch": 0.9748530706694244,
      "grad_norm": 0.2959204316139221,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5085,
      "step": 47605
    },
    {
      "epoch": 0.974955460446829,
      "grad_norm": 0.2905808091163635,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5638,
      "step": 47610
    },
    {
      "epoch": 0.9750578502242336,
      "grad_norm": 0.31405702233314514,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5099,
      "step": 47615
    },
    {
      "epoch": 0.9751602400016383,
      "grad_norm": 0.30506718158721924,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5412,
      "step": 47620
    },
    {
      "epoch": 0.9752626297790429,
      "grad_norm": 0.3007020056247711,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5642,
      "step": 47625
    },
    {
      "epoch": 0.9753650195564475,
      "grad_norm": 0.4340749979019165,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5624,
      "step": 47630
    },
    {
      "epoch": 0.9754674093338521,
      "grad_norm": 0.29517611861228943,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5556,
      "step": 47635
    },
    {
      "epoch": 0.9755697991112567,
      "grad_norm": 0.3016403615474701,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5887,
      "step": 47640
    },
    {
      "epoch": 0.9756721888886614,
      "grad_norm": 0.29522985219955444,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5627,
      "step": 47645
    },
    {
      "epoch": 0.975774578666066,
      "grad_norm": 0.2997603118419647,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5753,
      "step": 47650
    },
    {
      "epoch": 0.9758769684434706,
      "grad_norm": 0.29548588395118713,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5231,
      "step": 47655
    },
    {
      "epoch": 0.9759793582208752,
      "grad_norm": 0.2996750473976135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5269,
      "step": 47660
    },
    {
      "epoch": 0.9760817479982798,
      "grad_norm": 0.5803531408309937,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6113,
      "step": 47665
    },
    {
      "epoch": 0.9761841377756845,
      "grad_norm": 0.3171710968017578,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.505,
      "step": 47670
    },
    {
      "epoch": 0.9762865275530891,
      "grad_norm": 0.31973206996917725,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5346,
      "step": 47675
    },
    {
      "epoch": 0.9763889173304937,
      "grad_norm": 0.2989446222782135,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5907,
      "step": 47680
    },
    {
      "epoch": 0.9764913071078983,
      "grad_norm": 0.30542537569999695,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5917,
      "step": 47685
    },
    {
      "epoch": 0.976593696885303,
      "grad_norm": 0.2987099587917328,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5065,
      "step": 47690
    },
    {
      "epoch": 0.9766960866627076,
      "grad_norm": 0.7516525983810425,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5825,
      "step": 47695
    },
    {
      "epoch": 0.9767984764401122,
      "grad_norm": 0.31141313910484314,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5883,
      "step": 47700
    },
    {
      "epoch": 0.9769008662175168,
      "grad_norm": 0.3016454875469208,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5647,
      "step": 47705
    },
    {
      "epoch": 0.9770032559949214,
      "grad_norm": 0.2989034354686737,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5689,
      "step": 47710
    },
    {
      "epoch": 0.9771056457723261,
      "grad_norm": 0.30508163571357727,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5473,
      "step": 47715
    },
    {
      "epoch": 0.9772080355497307,
      "grad_norm": 0.2942759394645691,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5663,
      "step": 47720
    },
    {
      "epoch": 0.9773104253271353,
      "grad_norm": 0.29524552822113037,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.486,
      "step": 47725
    },
    {
      "epoch": 0.9774128151045399,
      "grad_norm": 0.29157382249832153,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5848,
      "step": 47730
    },
    {
      "epoch": 0.9775152048819445,
      "grad_norm": 0.30245131254196167,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5711,
      "step": 47735
    },
    {
      "epoch": 0.9776175946593492,
      "grad_norm": 0.2983751893043518,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.549,
      "step": 47740
    },
    {
      "epoch": 0.9777199844367538,
      "grad_norm": 0.3162797689437866,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5891,
      "step": 47745
    },
    {
      "epoch": 0.9778223742141584,
      "grad_norm": 0.2861679792404175,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5444,
      "step": 47750
    },
    {
      "epoch": 0.977924763991563,
      "grad_norm": 0.3068501353263855,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5623,
      "step": 47755
    },
    {
      "epoch": 0.9780271537689678,
      "grad_norm": 0.2970331907272339,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5897,
      "step": 47760
    },
    {
      "epoch": 0.9781295435463724,
      "grad_norm": 0.28139594197273254,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5435,
      "step": 47765
    },
    {
      "epoch": 0.978231933323777,
      "grad_norm": 0.31797337532043457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5054,
      "step": 47770
    },
    {
      "epoch": 0.9783343231011816,
      "grad_norm": 0.29420650005340576,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5333,
      "step": 47775
    },
    {
      "epoch": 0.9784367128785862,
      "grad_norm": 0.29384753108024597,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5278,
      "step": 47780
    },
    {
      "epoch": 0.9785391026559909,
      "grad_norm": 0.30456167459487915,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5511,
      "step": 47785
    },
    {
      "epoch": 0.9786414924333955,
      "grad_norm": 0.2956148386001587,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5611,
      "step": 47790
    },
    {
      "epoch": 0.9787438822108001,
      "grad_norm": 0.2943970561027527,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6011,
      "step": 47795
    },
    {
      "epoch": 0.9788462719882047,
      "grad_norm": 0.2998138666152954,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5352,
      "step": 47800
    },
    {
      "epoch": 0.9789486617656094,
      "grad_norm": 0.31079331040382385,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.3662,
      "step": 47805
    },
    {
      "epoch": 0.979051051543014,
      "grad_norm": 0.2993094325065613,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5344,
      "step": 47810
    },
    {
      "epoch": 0.9791534413204186,
      "grad_norm": 0.30916672945022583,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.547,
      "step": 47815
    },
    {
      "epoch": 0.9792558310978232,
      "grad_norm": 0.3025127351284027,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.597,
      "step": 47820
    },
    {
      "epoch": 0.9793582208752278,
      "grad_norm": 0.2969529628753662,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5992,
      "step": 47825
    },
    {
      "epoch": 0.9794606106526325,
      "grad_norm": 0.30451536178588867,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6101,
      "step": 47830
    },
    {
      "epoch": 0.9795630004300371,
      "grad_norm": 0.28235021233558655,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6024,
      "step": 47835
    },
    {
      "epoch": 0.9796653902074417,
      "grad_norm": 0.2974168658256531,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5382,
      "step": 47840
    },
    {
      "epoch": 0.9797677799848463,
      "grad_norm": 0.3060206174850464,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5494,
      "step": 47845
    },
    {
      "epoch": 0.9798701697622509,
      "grad_norm": 0.3062215745449066,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.528,
      "step": 47850
    },
    {
      "epoch": 0.9799725595396556,
      "grad_norm": 0.3104513883590698,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5892,
      "step": 47855
    },
    {
      "epoch": 0.9800749493170602,
      "grad_norm": 0.2982814610004425,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5664,
      "step": 47860
    },
    {
      "epoch": 0.9801773390944648,
      "grad_norm": 0.29843899607658386,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5298,
      "step": 47865
    },
    {
      "epoch": 0.9802797288718694,
      "grad_norm": 0.29165926575660706,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5778,
      "step": 47870
    },
    {
      "epoch": 0.980382118649274,
      "grad_norm": 0.292339563369751,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.548,
      "step": 47875
    },
    {
      "epoch": 0.9804845084266787,
      "grad_norm": 0.30853620171546936,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.553,
      "step": 47880
    },
    {
      "epoch": 0.9805868982040833,
      "grad_norm": 0.3019552528858185,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5485,
      "step": 47885
    },
    {
      "epoch": 0.9806892879814879,
      "grad_norm": 0.2836818993091583,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5667,
      "step": 47890
    },
    {
      "epoch": 0.9807916777588925,
      "grad_norm": 0.3047296702861786,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6309,
      "step": 47895
    },
    {
      "epoch": 0.9808940675362972,
      "grad_norm": 0.3227206766605377,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6047,
      "step": 47900
    },
    {
      "epoch": 0.9809964573137018,
      "grad_norm": 0.3000172972679138,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5939,
      "step": 47905
    },
    {
      "epoch": 0.9810988470911064,
      "grad_norm": 0.31060072779655457,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5729,
      "step": 47910
    },
    {
      "epoch": 0.981201236868511,
      "grad_norm": 0.30266073346138,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5825,
      "step": 47915
    },
    {
      "epoch": 0.9813036266459156,
      "grad_norm": 0.28458890318870544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5837,
      "step": 47920
    },
    {
      "epoch": 0.9814060164233203,
      "grad_norm": 0.4968605637550354,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5642,
      "step": 47925
    },
    {
      "epoch": 0.9815084062007249,
      "grad_norm": 0.29144537448883057,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5405,
      "step": 47930
    },
    {
      "epoch": 0.9816107959781295,
      "grad_norm": 0.2946797311306,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5616,
      "step": 47935
    },
    {
      "epoch": 0.9817131857555341,
      "grad_norm": 0.3165966272354126,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5244,
      "step": 47940
    },
    {
      "epoch": 0.9818155755329387,
      "grad_norm": 0.29605773091316223,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5738,
      "step": 47945
    },
    {
      "epoch": 0.9819179653103434,
      "grad_norm": 0.36024630069732666,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5367,
      "step": 47950
    },
    {
      "epoch": 0.982020355087748,
      "grad_norm": 0.30009180307388306,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6108,
      "step": 47955
    },
    {
      "epoch": 0.9821227448651526,
      "grad_norm": 0.3001779317855835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6437,
      "step": 47960
    },
    {
      "epoch": 0.9822251346425573,
      "grad_norm": 0.3078416883945465,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5431,
      "step": 47965
    },
    {
      "epoch": 0.982327524419962,
      "grad_norm": 0.29466357827186584,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5391,
      "step": 47970
    },
    {
      "epoch": 0.9824299141973666,
      "grad_norm": 0.3046253025531769,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5418,
      "step": 47975
    },
    {
      "epoch": 0.9825323039747712,
      "grad_norm": 0.31107428669929504,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6064,
      "step": 47980
    },
    {
      "epoch": 0.9826346937521758,
      "grad_norm": 0.30249983072280884,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5771,
      "step": 47985
    },
    {
      "epoch": 0.9827370835295804,
      "grad_norm": 0.29450955986976624,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5659,
      "step": 47990
    },
    {
      "epoch": 0.9828394733069851,
      "grad_norm": 0.2966823875904083,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5811,
      "step": 47995
    },
    {
      "epoch": 0.9829418630843897,
      "grad_norm": 0.3255231976509094,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5493,
      "step": 48000
    },
    {
      "epoch": 0.9829418630843897,
      "eval_loss": 4.58038854598999,
      "eval_runtime": 17.1261,
      "eval_samples_per_second": 33.983,
      "eval_steps_per_second": 4.263,
      "step": 48000
    },
    {
      "epoch": 0.9830442528617943,
      "grad_norm": 0.29559794068336487,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5788,
      "step": 48005
    },
    {
      "epoch": 0.9831466426391989,
      "grad_norm": 0.2873050570487976,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6047,
      "step": 48010
    },
    {
      "epoch": 0.9832490324166036,
      "grad_norm": 0.3836257755756378,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5261,
      "step": 48015
    },
    {
      "epoch": 0.9833514221940082,
      "grad_norm": 0.2874748110771179,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6257,
      "step": 48020
    },
    {
      "epoch": 0.9834538119714128,
      "grad_norm": 0.2962822914123535,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5635,
      "step": 48025
    },
    {
      "epoch": 0.9835562017488174,
      "grad_norm": 0.31296682357788086,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5616,
      "step": 48030
    },
    {
      "epoch": 0.983658591526222,
      "grad_norm": 0.3060377240180969,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4271,
      "step": 48035
    },
    {
      "epoch": 0.9837609813036267,
      "grad_norm": 0.3045669198036194,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5513,
      "step": 48040
    },
    {
      "epoch": 0.9838633710810313,
      "grad_norm": 0.2947128117084503,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5827,
      "step": 48045
    },
    {
      "epoch": 0.9839657608584359,
      "grad_norm": 0.34637144207954407,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5234,
      "step": 48050
    },
    {
      "epoch": 0.9840681506358405,
      "grad_norm": 0.3139941990375519,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4869,
      "step": 48055
    },
    {
      "epoch": 0.9841705404132451,
      "grad_norm": 0.30256378650665283,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.602,
      "step": 48060
    },
    {
      "epoch": 0.9842729301906498,
      "grad_norm": 0.3016190826892853,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4978,
      "step": 48065
    },
    {
      "epoch": 0.9843753199680544,
      "grad_norm": 0.30201447010040283,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5377,
      "step": 48070
    },
    {
      "epoch": 0.984477709745459,
      "grad_norm": 0.2965356111526489,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5965,
      "step": 48075
    },
    {
      "epoch": 0.9845800995228636,
      "grad_norm": 0.32482680678367615,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5602,
      "step": 48080
    },
    {
      "epoch": 0.9846824893002682,
      "grad_norm": 0.29524749517440796,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5763,
      "step": 48085
    },
    {
      "epoch": 0.9847848790776729,
      "grad_norm": 0.2916395366191864,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5982,
      "step": 48090
    },
    {
      "epoch": 0.9848872688550775,
      "grad_norm": 0.30824756622314453,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5612,
      "step": 48095
    },
    {
      "epoch": 0.9849896586324821,
      "grad_norm": 0.3042619228363037,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5865,
      "step": 48100
    },
    {
      "epoch": 0.9850920484098867,
      "grad_norm": 0.3027566373348236,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5936,
      "step": 48105
    },
    {
      "epoch": 0.9851944381872914,
      "grad_norm": 0.33060187101364136,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4758,
      "step": 48110
    },
    {
      "epoch": 0.985296827964696,
      "grad_norm": 0.29017114639282227,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5754,
      "step": 48115
    },
    {
      "epoch": 0.9853992177421006,
      "grad_norm": 0.3315523862838745,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5919,
      "step": 48120
    },
    {
      "epoch": 0.9855016075195052,
      "grad_norm": 0.3086949288845062,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6245,
      "step": 48125
    },
    {
      "epoch": 0.9856039972969098,
      "grad_norm": 0.3112105429172516,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5972,
      "step": 48130
    },
    {
      "epoch": 0.9857063870743145,
      "grad_norm": 0.3045254647731781,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.53,
      "step": 48135
    },
    {
      "epoch": 0.9858087768517191,
      "grad_norm": 0.29396480321884155,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5782,
      "step": 48140
    },
    {
      "epoch": 0.9859111666291237,
      "grad_norm": 0.29180625081062317,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5637,
      "step": 48145
    },
    {
      "epoch": 0.9860135564065283,
      "grad_norm": 0.29800793528556824,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5683,
      "step": 48150
    },
    {
      "epoch": 0.986115946183933,
      "grad_norm": 0.3194294273853302,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5066,
      "step": 48155
    },
    {
      "epoch": 0.9862183359613376,
      "grad_norm": 0.2990261912345886,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6123,
      "step": 48160
    },
    {
      "epoch": 0.9863207257387423,
      "grad_norm": 0.3095386326313019,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6088,
      "step": 48165
    },
    {
      "epoch": 0.9864231155161469,
      "grad_norm": 0.29120439291000366,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5251,
      "step": 48170
    },
    {
      "epoch": 0.9865255052935515,
      "grad_norm": 0.32928574085235596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4909,
      "step": 48175
    },
    {
      "epoch": 0.9866278950709562,
      "grad_norm": 0.2973651587963104,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.526,
      "step": 48180
    },
    {
      "epoch": 0.9867302848483608,
      "grad_norm": 0.41812607645988464,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5729,
      "step": 48185
    },
    {
      "epoch": 0.9868326746257654,
      "grad_norm": 0.30254292488098145,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5442,
      "step": 48190
    },
    {
      "epoch": 0.98693506440317,
      "grad_norm": 0.29239410161972046,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5999,
      "step": 48195
    },
    {
      "epoch": 0.9870374541805746,
      "grad_norm": 0.288949579000473,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5229,
      "step": 48200
    },
    {
      "epoch": 0.9871398439579793,
      "grad_norm": 0.2900617718696594,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5424,
      "step": 48205
    },
    {
      "epoch": 0.9872422337353839,
      "grad_norm": 0.31114351749420166,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5506,
      "step": 48210
    },
    {
      "epoch": 0.9873446235127885,
      "grad_norm": 0.29957515001296997,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5503,
      "step": 48215
    },
    {
      "epoch": 0.9874470132901931,
      "grad_norm": 0.2843896150588989,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5402,
      "step": 48220
    },
    {
      "epoch": 0.9875494030675978,
      "grad_norm": 0.3023429811000824,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5835,
      "step": 48225
    },
    {
      "epoch": 0.9876517928450024,
      "grad_norm": 0.29364222288131714,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5936,
      "step": 48230
    },
    {
      "epoch": 0.987754182622407,
      "grad_norm": 0.2951924800872803,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5296,
      "step": 48235
    },
    {
      "epoch": 0.9878565723998116,
      "grad_norm": 0.3002241551876068,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6014,
      "step": 48240
    },
    {
      "epoch": 0.9879589621772162,
      "grad_norm": 0.3228725492954254,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5675,
      "step": 48245
    },
    {
      "epoch": 0.9880613519546209,
      "grad_norm": 0.2973325550556183,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5701,
      "step": 48250
    },
    {
      "epoch": 0.9881637417320255,
      "grad_norm": 0.3340296149253845,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5867,
      "step": 48255
    },
    {
      "epoch": 0.9882661315094301,
      "grad_norm": 0.29857170581817627,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5791,
      "step": 48260
    },
    {
      "epoch": 0.9883685212868347,
      "grad_norm": 0.31115448474884033,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5804,
      "step": 48265
    },
    {
      "epoch": 0.9884709110642393,
      "grad_norm": 0.33123666048049927,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5663,
      "step": 48270
    },
    {
      "epoch": 0.988573300841644,
      "grad_norm": 0.30342525243759155,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5603,
      "step": 48275
    },
    {
      "epoch": 0.9886756906190486,
      "grad_norm": 0.34847524762153625,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5266,
      "step": 48280
    },
    {
      "epoch": 0.9887780803964532,
      "grad_norm": 0.29856789112091064,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5256,
      "step": 48285
    },
    {
      "epoch": 0.9888804701738578,
      "grad_norm": 0.2932136654853821,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5389,
      "step": 48290
    },
    {
      "epoch": 0.9889828599512624,
      "grad_norm": 0.3099711835384369,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5411,
      "step": 48295
    },
    {
      "epoch": 0.9890852497286671,
      "grad_norm": 0.2868886888027191,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5768,
      "step": 48300
    },
    {
      "epoch": 0.9891876395060717,
      "grad_norm": 0.2987963855266571,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5785,
      "step": 48305
    },
    {
      "epoch": 0.9892900292834763,
      "grad_norm": 0.306930810213089,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5523,
      "step": 48310
    },
    {
      "epoch": 0.9893924190608809,
      "grad_norm": 0.3057742714881897,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5812,
      "step": 48315
    },
    {
      "epoch": 0.9894948088382856,
      "grad_norm": 0.29376351833343506,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5839,
      "step": 48320
    },
    {
      "epoch": 0.9895971986156902,
      "grad_norm": 0.3674677610397339,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5878,
      "step": 48325
    },
    {
      "epoch": 0.9896995883930948,
      "grad_norm": 0.3014540374279022,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5718,
      "step": 48330
    },
    {
      "epoch": 0.9898019781704994,
      "grad_norm": 0.5672804713249207,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5069,
      "step": 48335
    },
    {
      "epoch": 0.989904367947904,
      "grad_norm": 0.29727086424827576,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5248,
      "step": 48340
    },
    {
      "epoch": 0.9900067577253087,
      "grad_norm": 0.29575619101524353,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6139,
      "step": 48345
    },
    {
      "epoch": 0.9901091475027133,
      "grad_norm": 0.30974772572517395,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5381,
      "step": 48350
    },
    {
      "epoch": 0.9902115372801179,
      "grad_norm": 0.29965290427207947,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6347,
      "step": 48355
    },
    {
      "epoch": 0.9903139270575225,
      "grad_norm": 0.33263689279556274,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5864,
      "step": 48360
    },
    {
      "epoch": 0.9904163168349271,
      "grad_norm": 0.2891058325767517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5297,
      "step": 48365
    },
    {
      "epoch": 0.9905187066123319,
      "grad_norm": 0.31554582715034485,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5698,
      "step": 48370
    },
    {
      "epoch": 0.9906210963897365,
      "grad_norm": 0.2902148962020874,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5561,
      "step": 48375
    },
    {
      "epoch": 0.9907234861671411,
      "grad_norm": 0.2999386787414551,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5651,
      "step": 48380
    },
    {
      "epoch": 0.9908258759445457,
      "grad_norm": 0.318154901266098,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5657,
      "step": 48385
    },
    {
      "epoch": 0.9909282657219504,
      "grad_norm": 0.29872027039527893,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6069,
      "step": 48390
    },
    {
      "epoch": 0.991030655499355,
      "grad_norm": 0.30033451318740845,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.596,
      "step": 48395
    },
    {
      "epoch": 0.9911330452767596,
      "grad_norm": 0.28618279099464417,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6149,
      "step": 48400
    },
    {
      "epoch": 0.9912354350541642,
      "grad_norm": 0.3052704334259033,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5962,
      "step": 48405
    },
    {
      "epoch": 0.9913378248315688,
      "grad_norm": 0.30052441358566284,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5889,
      "step": 48410
    },
    {
      "epoch": 0.9914402146089735,
      "grad_norm": 0.2918172776699066,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5396,
      "step": 48415
    },
    {
      "epoch": 0.9915426043863781,
      "grad_norm": 0.32806017994880676,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5853,
      "step": 48420
    },
    {
      "epoch": 0.9916449941637827,
      "grad_norm": 0.3079558312892914,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6285,
      "step": 48425
    },
    {
      "epoch": 0.9917473839411873,
      "grad_norm": 0.7797358632087708,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5935,
      "step": 48430
    },
    {
      "epoch": 0.991849773718592,
      "grad_norm": 0.30594924092292786,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5901,
      "step": 48435
    },
    {
      "epoch": 0.9919521634959966,
      "grad_norm": 0.299163818359375,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5498,
      "step": 48440
    },
    {
      "epoch": 0.9920545532734012,
      "grad_norm": 0.2934512197971344,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5368,
      "step": 48445
    },
    {
      "epoch": 0.9921569430508058,
      "grad_norm": 0.42306414246559143,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5747,
      "step": 48450
    },
    {
      "epoch": 0.9922593328282104,
      "grad_norm": 0.3244866132736206,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.571,
      "step": 48455
    },
    {
      "epoch": 0.9923617226056151,
      "grad_norm": 0.2958906888961792,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5693,
      "step": 48460
    },
    {
      "epoch": 0.9924641123830197,
      "grad_norm": 0.46438702940940857,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6219,
      "step": 48465
    },
    {
      "epoch": 0.9925665021604243,
      "grad_norm": 0.2966323792934418,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5287,
      "step": 48470
    },
    {
      "epoch": 0.9926688919378289,
      "grad_norm": 0.36979418992996216,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5192,
      "step": 48475
    },
    {
      "epoch": 0.9927712817152335,
      "grad_norm": 0.29866376519203186,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5937,
      "step": 48480
    },
    {
      "epoch": 0.9928736714926382,
      "grad_norm": 0.29135310649871826,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5857,
      "step": 48485
    },
    {
      "epoch": 0.9929760612700428,
      "grad_norm": 0.32972216606140137,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5914,
      "step": 48490
    },
    {
      "epoch": 0.9930784510474474,
      "grad_norm": 0.30958205461502075,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.589,
      "step": 48495
    },
    {
      "epoch": 0.993180840824852,
      "grad_norm": 0.29018616676330566,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5243,
      "step": 48500
    },
    {
      "epoch": 0.9932832306022567,
      "grad_norm": 0.3043172359466553,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5926,
      "step": 48505
    },
    {
      "epoch": 0.9933856203796613,
      "grad_norm": 0.3271157741546631,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5279,
      "step": 48510
    },
    {
      "epoch": 0.9934880101570659,
      "grad_norm": 0.30671265721321106,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.541,
      "step": 48515
    },
    {
      "epoch": 0.9935903999344705,
      "grad_norm": 0.2887033224105835,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5568,
      "step": 48520
    },
    {
      "epoch": 0.9936927897118751,
      "grad_norm": 0.30974188446998596,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5086,
      "step": 48525
    },
    {
      "epoch": 0.9937951794892798,
      "grad_norm": 0.29903706908226013,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5542,
      "step": 48530
    },
    {
      "epoch": 0.9938975692666844,
      "grad_norm": 0.3158770501613617,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5431,
      "step": 48535
    },
    {
      "epoch": 0.993999959044089,
      "grad_norm": 0.2940845489501953,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5795,
      "step": 48540
    },
    {
      "epoch": 0.9941023488214936,
      "grad_norm": 0.2977944612503052,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5854,
      "step": 48545
    },
    {
      "epoch": 0.9942047385988982,
      "grad_norm": 0.29025569558143616,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5308,
      "step": 48550
    },
    {
      "epoch": 0.9943071283763029,
      "grad_norm": 0.3226446807384491,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5761,
      "step": 48555
    },
    {
      "epoch": 0.9944095181537075,
      "grad_norm": 0.2970184087753296,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5451,
      "step": 48560
    },
    {
      "epoch": 0.9945119079311121,
      "grad_norm": 0.2929586172103882,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5765,
      "step": 48565
    },
    {
      "epoch": 0.9946142977085168,
      "grad_norm": 0.3146069049835205,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5987,
      "step": 48570
    },
    {
      "epoch": 0.9947166874859215,
      "grad_norm": 0.2933782935142517,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5457,
      "step": 48575
    },
    {
      "epoch": 0.9948190772633261,
      "grad_norm": 0.30704832077026367,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5707,
      "step": 48580
    },
    {
      "epoch": 0.9949214670407307,
      "grad_norm": 0.3101729154586792,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5618,
      "step": 48585
    },
    {
      "epoch": 0.9950238568181353,
      "grad_norm": 0.2928634285926819,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5858,
      "step": 48590
    },
    {
      "epoch": 0.9951262465955399,
      "grad_norm": 0.3027421832084656,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5273,
      "step": 48595
    },
    {
      "epoch": 0.9952286363729446,
      "grad_norm": 0.3238886594772339,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5646,
      "step": 48600
    },
    {
      "epoch": 0.9953310261503492,
      "grad_norm": 0.30195388197898865,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5403,
      "step": 48605
    },
    {
      "epoch": 0.9954334159277538,
      "grad_norm": 0.31386324763298035,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5859,
      "step": 48610
    },
    {
      "epoch": 0.9955358057051584,
      "grad_norm": 0.32625406980514526,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5813,
      "step": 48615
    },
    {
      "epoch": 0.995638195482563,
      "grad_norm": 0.3073594868183136,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5636,
      "step": 48620
    },
    {
      "epoch": 0.9957405852599677,
      "grad_norm": 0.30962932109832764,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5483,
      "step": 48625
    },
    {
      "epoch": 0.9958429750373723,
      "grad_norm": 0.3184210956096649,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5602,
      "step": 48630
    },
    {
      "epoch": 0.9959453648147769,
      "grad_norm": 0.29708781838417053,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5435,
      "step": 48635
    },
    {
      "epoch": 0.9960477545921815,
      "grad_norm": 0.3089878559112549,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5665,
      "step": 48640
    },
    {
      "epoch": 0.9961501443695862,
      "grad_norm": 0.30662375688552856,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5029,
      "step": 48645
    },
    {
      "epoch": 0.9962525341469908,
      "grad_norm": 0.2912473976612091,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5023,
      "step": 48650
    },
    {
      "epoch": 0.9963549239243954,
      "grad_norm": 0.29344433546066284,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5046,
      "step": 48655
    },
    {
      "epoch": 0.9964573137018,
      "grad_norm": 0.3064979016780853,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5615,
      "step": 48660
    },
    {
      "epoch": 0.9965597034792046,
      "grad_norm": 0.31767401099205017,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5763,
      "step": 48665
    },
    {
      "epoch": 0.9966620932566093,
      "grad_norm": 0.3110029101371765,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5906,
      "step": 48670
    },
    {
      "epoch": 0.9967644830340139,
      "grad_norm": 0.2901157736778259,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5029,
      "step": 48675
    },
    {
      "epoch": 0.9968668728114185,
      "grad_norm": 0.2966289818286896,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5575,
      "step": 48680
    },
    {
      "epoch": 0.9969692625888231,
      "grad_norm": 0.34209704399108887,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5737,
      "step": 48685
    },
    {
      "epoch": 0.9970716523662277,
      "grad_norm": 0.31367409229278564,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6228,
      "step": 48690
    },
    {
      "epoch": 0.9971740421436324,
      "grad_norm": 0.29235970973968506,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6071,
      "step": 48695
    },
    {
      "epoch": 0.997276431921037,
      "grad_norm": 0.3141907751560211,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5596,
      "step": 48700
    },
    {
      "epoch": 0.9973788216984416,
      "grad_norm": 0.3179910480976105,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.4919,
      "step": 48705
    },
    {
      "epoch": 0.9974812114758462,
      "grad_norm": 0.30955952405929565,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5668,
      "step": 48710
    },
    {
      "epoch": 0.9975836012532509,
      "grad_norm": 0.2875402271747589,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5615,
      "step": 48715
    },
    {
      "epoch": 0.9976859910306555,
      "grad_norm": 0.316728413105011,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6114,
      "step": 48720
    },
    {
      "epoch": 0.9977883808080601,
      "grad_norm": 0.2882649004459381,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5686,
      "step": 48725
    },
    {
      "epoch": 0.9978907705854647,
      "grad_norm": 0.31529128551483154,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5824,
      "step": 48730
    },
    {
      "epoch": 0.9979931603628693,
      "grad_norm": 0.29442816972732544,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5153,
      "step": 48735
    },
    {
      "epoch": 0.998095550140274,
      "grad_norm": 0.28840017318725586,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5885,
      "step": 48740
    },
    {
      "epoch": 0.9981979399176786,
      "grad_norm": 0.29690900444984436,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6002,
      "step": 48745
    },
    {
      "epoch": 0.9983003296950832,
      "grad_norm": 0.29731130599975586,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5881,
      "step": 48750
    },
    {
      "epoch": 0.9984027194724878,
      "grad_norm": 0.3224312663078308,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5788,
      "step": 48755
    },
    {
      "epoch": 0.9985051092498924,
      "grad_norm": 0.302958607673645,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5155,
      "step": 48760
    },
    {
      "epoch": 0.9986074990272971,
      "grad_norm": 0.29821866750717163,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5768,
      "step": 48765
    },
    {
      "epoch": 0.9987098888047017,
      "grad_norm": 0.29861757159233093,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6109,
      "step": 48770
    },
    {
      "epoch": 0.9988122785821064,
      "grad_norm": 0.31388548016548157,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5713,
      "step": 48775
    },
    {
      "epoch": 0.998914668359511,
      "grad_norm": 0.3197086751461029,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5466,
      "step": 48780
    },
    {
      "epoch": 0.9990170581369157,
      "grad_norm": 0.3147348165512085,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.558,
      "step": 48785
    },
    {
      "epoch": 0.9991194479143203,
      "grad_norm": 0.29904741048812866,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6132,
      "step": 48790
    },
    {
      "epoch": 0.9992218376917249,
      "grad_norm": 0.2876797318458557,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5397,
      "step": 48795
    },
    {
      "epoch": 0.9993242274691295,
      "grad_norm": 0.3055194616317749,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5512,
      "step": 48800
    },
    {
      "epoch": 0.9994266172465341,
      "grad_norm": 0.2971906363964081,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5158,
      "step": 48805
    },
    {
      "epoch": 0.9995290070239388,
      "grad_norm": 0.30163276195526123,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6007,
      "step": 48810
    },
    {
      "epoch": 0.9996313968013434,
      "grad_norm": 0.30284538865089417,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5838,
      "step": 48815
    },
    {
      "epoch": 0.999733786578748,
      "grad_norm": 0.2849685847759247,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5669,
      "step": 48820
    },
    {
      "epoch": 0.9998361763561526,
      "grad_norm": 0.29327917098999023,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.5324,
      "step": 48825
    },
    {
      "epoch": 0.9999385661335572,
      "grad_norm": 0.2930566370487213,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.6017,
      "step": 48830
    }
  ],
  "logging_steps": 5,
  "max_steps": 48833,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}