{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.672795134089024,
  "eval_steps": 500,
  "global_step": 4867,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013823610727121925,
      "grad_norm": 1.381127119064331,
      "learning_rate": 3.1095890410958907e-06,
      "loss": 0.6841,
      "step": 10
    },
    {
      "epoch": 0.002764722145424385,
      "grad_norm": 0.9868093729019165,
      "learning_rate": 4.3424657534246575e-06,
      "loss": 0.6482,
      "step": 20
    },
    {
      "epoch": 0.004147083218136578,
      "grad_norm": 0.9629099369049072,
      "learning_rate": 5.575342465753425e-06,
      "loss": 0.6245,
      "step": 30
    },
    {
      "epoch": 0.00552944429084877,
      "grad_norm": 0.6574251055717468,
      "learning_rate": 6.808219178082193e-06,
      "loss": 0.5794,
      "step": 40
    },
    {
      "epoch": 0.006911805363560962,
      "grad_norm": 0.7135911583900452,
      "learning_rate": 8.04109589041096e-06,
      "loss": 0.6106,
      "step": 50
    },
    {
      "epoch": 0.008294166436273155,
      "grad_norm": 0.6244402527809143,
      "learning_rate": 9.273972602739727e-06,
      "loss": 0.5564,
      "step": 60
    },
    {
      "epoch": 0.009676527508985348,
      "grad_norm": 0.7258371710777283,
      "learning_rate": 1.0506849315068496e-05,
      "loss": 0.566,
      "step": 70
    },
    {
      "epoch": 0.01105888858169754,
      "grad_norm": 0.7419959902763367,
      "learning_rate": 1.1739726027397262e-05,
      "loss": 0.5749,
      "step": 80
    },
    {
      "epoch": 0.012441249654409731,
      "grad_norm": 0.775385320186615,
      "learning_rate": 1.2972602739726028e-05,
      "loss": 0.5924,
      "step": 90
    },
    {
      "epoch": 0.013823610727121923,
      "grad_norm": 0.8478221297264099,
      "learning_rate": 1.4205479452054796e-05,
      "loss": 0.5869,
      "step": 100
    },
    {
      "epoch": 0.015205971799834116,
      "grad_norm": 0.7597885727882385,
      "learning_rate": 1.5438356164383563e-05,
      "loss": 0.5788,
      "step": 110
    },
    {
      "epoch": 0.01658833287254631,
      "grad_norm": 0.619677722454071,
      "learning_rate": 1.6671232876712333e-05,
      "loss": 0.5574,
      "step": 120
    },
    {
      "epoch": 0.017970693945258503,
      "grad_norm": 0.6591234803199768,
      "learning_rate": 1.7904109589041096e-05,
      "loss": 0.5594,
      "step": 130
    },
    {
      "epoch": 0.019353055017970695,
      "grad_norm": 0.7304580807685852,
      "learning_rate": 1.9136986301369865e-05,
      "loss": 0.5754,
      "step": 140
    },
    {
      "epoch": 0.020735416090682888,
      "grad_norm": 0.7558080554008484,
      "learning_rate": 2e-05,
      "loss": 0.5704,
      "step": 150
    },
    {
      "epoch": 0.02211777716339508,
      "grad_norm": 0.6936461329460144,
      "learning_rate": 2e-05,
      "loss": 0.5854,
      "step": 160
    },
    {
      "epoch": 0.023500138236107273,
      "grad_norm": 0.6672089695930481,
      "learning_rate": 2e-05,
      "loss": 0.5752,
      "step": 170
    },
    {
      "epoch": 0.024882499308819462,
      "grad_norm": 0.5092358589172363,
      "learning_rate": 2e-05,
      "loss": 0.5661,
      "step": 180
    },
    {
      "epoch": 0.026264860381531654,
      "grad_norm": 0.5904619693756104,
      "learning_rate": 2e-05,
      "loss": 0.5707,
      "step": 190
    },
    {
      "epoch": 0.027647221454243847,
      "grad_norm": 0.5448220372200012,
      "learning_rate": 2e-05,
      "loss": 0.5647,
      "step": 200
    },
    {
      "epoch": 0.02902958252695604,
      "grad_norm": 0.5694534778594971,
      "learning_rate": 2e-05,
      "loss": 0.5804,
      "step": 210
    },
    {
      "epoch": 0.030411943599668232,
      "grad_norm": 0.48752841353416443,
      "learning_rate": 2e-05,
      "loss": 0.5714,
      "step": 220
    },
    {
      "epoch": 0.031794304672380425,
      "grad_norm": 0.4754009544849396,
      "learning_rate": 2e-05,
      "loss": 0.5576,
      "step": 230
    },
    {
      "epoch": 0.03317666574509262,
      "grad_norm": 0.47850653529167175,
      "learning_rate": 2e-05,
      "loss": 0.5605,
      "step": 240
    },
    {
      "epoch": 0.03455902681780481,
      "grad_norm": 0.474916011095047,
      "learning_rate": 2e-05,
      "loss": 0.5566,
      "step": 250
    },
    {
      "epoch": 0.035941387890517006,
      "grad_norm": 0.525550901889801,
      "learning_rate": 2e-05,
      "loss": 0.5694,
      "step": 260
    },
    {
      "epoch": 0.037323748963229195,
      "grad_norm": 0.5116958022117615,
      "learning_rate": 2e-05,
      "loss": 0.5628,
      "step": 270
    },
    {
      "epoch": 0.03870611003594139,
      "grad_norm": 0.4546569585800171,
      "learning_rate": 2e-05,
      "loss": 0.5581,
      "step": 280
    },
    {
      "epoch": 0.04008847110865358,
      "grad_norm": 0.5056052803993225,
      "learning_rate": 2e-05,
      "loss": 0.5462,
      "step": 290
    },
    {
      "epoch": 0.041470832181365776,
      "grad_norm": 0.48135024309158325,
      "learning_rate": 2e-05,
      "loss": 0.5452,
      "step": 300
    },
    {
      "epoch": 0.042853193254077965,
      "grad_norm": 0.5011653304100037,
      "learning_rate": 2e-05,
      "loss": 0.5634,
      "step": 310
    },
    {
      "epoch": 0.04423555432679016,
      "grad_norm": 0.43440499901771545,
      "learning_rate": 2e-05,
      "loss": 0.5542,
      "step": 320
    },
    {
      "epoch": 0.04561791539950235,
      "grad_norm": 0.4600420296192169,
      "learning_rate": 2e-05,
      "loss": 0.5664,
      "step": 330
    },
    {
      "epoch": 0.047000276472214546,
      "grad_norm": 0.4818698465824127,
      "learning_rate": 2e-05,
      "loss": 0.5575,
      "step": 340
    },
    {
      "epoch": 0.048382637544926735,
      "grad_norm": 0.4674888551235199,
      "learning_rate": 2e-05,
      "loss": 0.5416,
      "step": 350
    },
    {
      "epoch": 0.049764998617638924,
      "grad_norm": 0.4194387197494507,
      "learning_rate": 2e-05,
      "loss": 0.5645,
      "step": 360
    },
    {
      "epoch": 0.05114735969035112,
      "grad_norm": 0.466195672750473,
      "learning_rate": 2e-05,
      "loss": 0.5674,
      "step": 370
    },
    {
      "epoch": 0.05252972076306331,
      "grad_norm": 0.39796894788742065,
      "learning_rate": 2e-05,
      "loss": 0.5401,
      "step": 380
    },
    {
      "epoch": 0.053912081835775505,
      "grad_norm": 0.4034675657749176,
      "learning_rate": 2e-05,
      "loss": 0.5465,
      "step": 390
    },
    {
      "epoch": 0.055294442908487694,
      "grad_norm": 0.4046485126018524,
      "learning_rate": 2e-05,
      "loss": 0.5564,
      "step": 400
    },
    {
      "epoch": 0.05667680398119989,
      "grad_norm": 0.4396536946296692,
      "learning_rate": 2e-05,
      "loss": 0.5616,
      "step": 410
    },
    {
      "epoch": 0.05805916505391208,
      "grad_norm": 0.4417417347431183,
      "learning_rate": 2e-05,
      "loss": 0.5439,
      "step": 420
    },
    {
      "epoch": 0.059441526126624275,
      "grad_norm": 0.41711825132369995,
      "learning_rate": 2e-05,
      "loss": 0.5583,
      "step": 430
    },
    {
      "epoch": 0.060823887199336464,
      "grad_norm": 0.41658106446266174,
      "learning_rate": 2e-05,
      "loss": 0.5707,
      "step": 440
    },
    {
      "epoch": 0.06220624827204866,
      "grad_norm": 0.43480992317199707,
      "learning_rate": 2e-05,
      "loss": 0.5635,
      "step": 450
    },
    {
      "epoch": 0.06358860934476085,
      "grad_norm": 0.47909724712371826,
      "learning_rate": 2e-05,
      "loss": 0.5436,
      "step": 460
    },
    {
      "epoch": 0.06497097041747305,
      "grad_norm": 0.42764002084732056,
      "learning_rate": 2e-05,
      "loss": 0.5533,
      "step": 470
    },
    {
      "epoch": 0.06635333149018524,
      "grad_norm": 0.4060353636741638,
      "learning_rate": 2e-05,
      "loss": 0.5554,
      "step": 480
    },
    {
      "epoch": 0.06773569256289742,
      "grad_norm": 0.3965570628643036,
      "learning_rate": 2e-05,
      "loss": 0.5371,
      "step": 490
    },
    {
      "epoch": 0.06911805363560962,
      "grad_norm": 0.7175000309944153,
      "learning_rate": 2e-05,
      "loss": 0.5228,
      "step": 500
    },
    {
      "epoch": 0.07050041470832182,
      "grad_norm": 0.4022802710533142,
      "learning_rate": 2e-05,
      "loss": 0.547,
      "step": 510
    },
    {
      "epoch": 0.07188277578103401,
      "grad_norm": 0.4243316054344177,
      "learning_rate": 2e-05,
      "loss": 0.5558,
      "step": 520
    },
    {
      "epoch": 0.0732651368537462,
      "grad_norm": 0.4347960352897644,
      "learning_rate": 2e-05,
      "loss": 0.539,
      "step": 530
    },
    {
      "epoch": 0.07464749792645839,
      "grad_norm": 0.4559546709060669,
      "learning_rate": 2e-05,
      "loss": 0.5436,
      "step": 540
    },
    {
      "epoch": 0.07602985899917059,
      "grad_norm": 0.43995845317840576,
      "learning_rate": 2e-05,
      "loss": 0.5693,
      "step": 550
    },
    {
      "epoch": 0.07741222007188278,
      "grad_norm": 0.4222787618637085,
      "learning_rate": 2e-05,
      "loss": 0.5645,
      "step": 560
    },
    {
      "epoch": 0.07879458114459496,
      "grad_norm": 0.45873314142227173,
      "learning_rate": 2e-05,
      "loss": 0.5422,
      "step": 570
    },
    {
      "epoch": 0.08017694221730716,
      "grad_norm": 0.7522889375686646,
      "learning_rate": 2e-05,
      "loss": 0.5686,
      "step": 580
    },
    {
      "epoch": 0.08155930329001936,
      "grad_norm": 0.4449799954891205,
      "learning_rate": 2e-05,
      "loss": 0.5509,
      "step": 590
    },
    {
      "epoch": 0.08294166436273155,
      "grad_norm": 0.40505728125572205,
      "learning_rate": 2e-05,
      "loss": 0.5559,
      "step": 600
    },
    {
      "epoch": 0.08432402543544373,
      "grad_norm": 0.3969815671443939,
      "learning_rate": 2e-05,
      "loss": 0.5675,
      "step": 610
    },
    {
      "epoch": 0.08570638650815593,
      "grad_norm": 0.4717716872692108,
      "learning_rate": 2e-05,
      "loss": 0.556,
      "step": 620
    },
    {
      "epoch": 0.08708874758086813,
      "grad_norm": 0.43605294823646545,
      "learning_rate": 2e-05,
      "loss": 0.5759,
      "step": 630
    },
    {
      "epoch": 0.08847110865358032,
      "grad_norm": 0.8605064749717712,
      "learning_rate": 2e-05,
      "loss": 0.5505,
      "step": 640
    },
    {
      "epoch": 0.0898534697262925,
      "grad_norm": 0.4309298098087311,
      "learning_rate": 2e-05,
      "loss": 0.5523,
      "step": 650
    },
    {
      "epoch": 0.0912358307990047,
      "grad_norm": 0.48024067282676697,
      "learning_rate": 2e-05,
      "loss": 0.5512,
      "step": 660
    },
    {
      "epoch": 0.0926181918717169,
      "grad_norm": 0.39635729789733887,
      "learning_rate": 2e-05,
      "loss": 0.5654,
      "step": 670
    },
    {
      "epoch": 0.09400055294442909,
      "grad_norm": 0.40529000759124756,
      "learning_rate": 2e-05,
      "loss": 0.5577,
      "step": 680
    },
    {
      "epoch": 0.09538291401714127,
      "grad_norm": 0.4051470160484314,
      "learning_rate": 2e-05,
      "loss": 0.555,
      "step": 690
    },
    {
      "epoch": 0.09676527508985347,
      "grad_norm": 0.42778125405311584,
      "learning_rate": 2e-05,
      "loss": 0.5572,
      "step": 700
    },
    {
      "epoch": 0.09814763616256567,
      "grad_norm": 0.41309046745300293,
      "learning_rate": 2e-05,
      "loss": 0.5357,
      "step": 710
    },
    {
      "epoch": 0.09952999723527785,
      "grad_norm": 0.43384742736816406,
      "learning_rate": 2e-05,
      "loss": 0.5584,
      "step": 720
    },
    {
      "epoch": 0.10091235830799004,
      "grad_norm": 0.4447322487831116,
      "learning_rate": 2e-05,
      "loss": 0.5293,
      "step": 730
    },
    {
      "epoch": 0.10229471938070224,
      "grad_norm": 0.4224463999271393,
      "learning_rate": 2e-05,
      "loss": 0.5583,
      "step": 740
    },
    {
      "epoch": 0.10367708045341444,
      "grad_norm": 0.4041193723678589,
      "learning_rate": 2e-05,
      "loss": 0.5486,
      "step": 750
    },
    {
      "epoch": 0.10505944152612662,
      "grad_norm": 0.4127112030982971,
      "learning_rate": 2e-05,
      "loss": 0.5574,
      "step": 760
    },
    {
      "epoch": 0.10644180259883881,
      "grad_norm": 0.38974669575691223,
      "learning_rate": 2e-05,
      "loss": 0.5399,
      "step": 770
    },
    {
      "epoch": 0.10782416367155101,
      "grad_norm": 0.37967780232429504,
      "learning_rate": 2e-05,
      "loss": 0.5304,
      "step": 780
    },
    {
      "epoch": 0.1092065247442632,
      "grad_norm": 0.3848550617694855,
      "learning_rate": 2e-05,
      "loss": 0.5477,
      "step": 790
    },
    {
      "epoch": 0.11058888581697539,
      "grad_norm": 0.42792809009552,
      "learning_rate": 2e-05,
      "loss": 0.543,
      "step": 800
    },
    {
      "epoch": 0.11197124688968758,
      "grad_norm": 0.35718613862991333,
      "learning_rate": 2e-05,
      "loss": 0.54,
      "step": 810
    },
    {
      "epoch": 0.11335360796239978,
      "grad_norm": 0.37314122915267944,
      "learning_rate": 2e-05,
      "loss": 0.5329,
      "step": 820
    },
    {
      "epoch": 0.11473596903511198,
      "grad_norm": 0.5386238694190979,
      "learning_rate": 2e-05,
      "loss": 0.5325,
      "step": 830
    },
    {
      "epoch": 0.11611833010782416,
      "grad_norm": 0.4082709550857544,
      "learning_rate": 2e-05,
      "loss": 0.5335,
      "step": 840
    },
    {
      "epoch": 0.11750069118053635,
      "grad_norm": 0.44811609387397766,
      "learning_rate": 2e-05,
      "loss": 0.5348,
      "step": 850
    },
    {
      "epoch": 0.11888305225324855,
      "grad_norm": 0.37818026542663574,
      "learning_rate": 2e-05,
      "loss": 0.5597,
      "step": 860
    },
    {
      "epoch": 0.12026541332596075,
      "grad_norm": 0.3747405707836151,
      "learning_rate": 2e-05,
      "loss": 0.5519,
      "step": 870
    },
    {
      "epoch": 0.12164777439867293,
      "grad_norm": 0.5200716853141785,
      "learning_rate": 2e-05,
      "loss": 0.5439,
      "step": 880
    },
    {
      "epoch": 0.12303013547138512,
      "grad_norm": 0.4514472782611847,
      "learning_rate": 2e-05,
      "loss": 0.5428,
      "step": 890
    },
    {
      "epoch": 0.12441249654409732,
      "grad_norm": 0.38784292340278625,
      "learning_rate": 2e-05,
      "loss": 0.5594,
      "step": 900
    },
    {
      "epoch": 0.1257948576168095,
      "grad_norm": 0.3769837021827698,
      "learning_rate": 2e-05,
      "loss": 0.5396,
      "step": 910
    },
    {
      "epoch": 0.1271772186895217,
      "grad_norm": 0.3987969756126404,
      "learning_rate": 2e-05,
      "loss": 0.5546,
      "step": 920
    },
    {
      "epoch": 0.1285595797622339,
      "grad_norm": 0.4501831829547882,
      "learning_rate": 2e-05,
      "loss": 0.5239,
      "step": 930
    },
    {
      "epoch": 0.1299419408349461,
      "grad_norm": 0.5386440753936768,
      "learning_rate": 2e-05,
      "loss": 0.5376,
      "step": 940
    },
    {
      "epoch": 0.1313243019076583,
      "grad_norm": 0.49837714433670044,
      "learning_rate": 2e-05,
      "loss": 0.5474,
      "step": 950
    },
    {
      "epoch": 0.13270666298037048,
      "grad_norm": 0.3861762285232544,
      "learning_rate": 2e-05,
      "loss": 0.5276,
      "step": 960
    },
    {
      "epoch": 0.13408902405308268,
      "grad_norm": 0.483341783285141,
      "learning_rate": 2e-05,
      "loss": 0.5315,
      "step": 970
    },
    {
      "epoch": 0.13547138512579485,
      "grad_norm": 0.36261555552482605,
      "learning_rate": 2e-05,
      "loss": 0.5432,
      "step": 980
    },
    {
      "epoch": 0.13685374619850704,
      "grad_norm": 0.38093864917755127,
      "learning_rate": 2e-05,
      "loss": 0.5618,
      "step": 990
    },
    {
      "epoch": 0.13823610727121924,
      "grad_norm": 0.3663526475429535,
      "learning_rate": 2e-05,
      "loss": 0.5298,
      "step": 1000
    },
    {
      "epoch": 0.13961846834393143,
      "grad_norm": 0.36675480008125305,
      "learning_rate": 2e-05,
      "loss": 0.5409,
      "step": 1010
    },
    {
      "epoch": 0.14100082941664363,
      "grad_norm": 0.4003055691719055,
      "learning_rate": 2e-05,
      "loss": 0.5609,
      "step": 1020
    },
    {
      "epoch": 0.14238319048935583,
      "grad_norm": 0.3676624596118927,
      "learning_rate": 2e-05,
      "loss": 0.554,
      "step": 1030
    },
    {
      "epoch": 0.14376555156206802,
      "grad_norm": 0.34587177634239197,
      "learning_rate": 2e-05,
      "loss": 0.5382,
      "step": 1040
    },
    {
      "epoch": 0.14514791263478022,
      "grad_norm": 0.3634945750236511,
      "learning_rate": 2e-05,
      "loss": 0.534,
      "step": 1050
    },
    {
      "epoch": 0.1465302737074924,
      "grad_norm": 0.36864402890205383,
      "learning_rate": 2e-05,
      "loss": 0.5362,
      "step": 1060
    },
    {
      "epoch": 0.14791263478020458,
      "grad_norm": 0.38066819310188293,
      "learning_rate": 2e-05,
      "loss": 0.549,
      "step": 1070
    },
    {
      "epoch": 0.14929499585291678,
      "grad_norm": 0.4147531986236572,
      "learning_rate": 2e-05,
      "loss": 0.5417,
      "step": 1080
    },
    {
      "epoch": 0.15067735692562897,
      "grad_norm": 0.38785669207572937,
      "learning_rate": 2e-05,
      "loss": 0.5189,
      "step": 1090
    },
    {
      "epoch": 0.15205971799834117,
      "grad_norm": 0.360170841217041,
      "learning_rate": 2e-05,
      "loss": 0.5557,
      "step": 1100
    },
    {
      "epoch": 0.15344207907105337,
      "grad_norm": 0.36952078342437744,
      "learning_rate": 2e-05,
      "loss": 0.5387,
      "step": 1110
    },
    {
      "epoch": 0.15482444014376556,
      "grad_norm": 0.4045839309692383,
      "learning_rate": 2e-05,
      "loss": 0.5256,
      "step": 1120
    },
    {
      "epoch": 0.15620680121647773,
      "grad_norm": 0.41408610343933105,
      "learning_rate": 2e-05,
      "loss": 0.5336,
      "step": 1130
    },
    {
      "epoch": 0.15758916228918993,
      "grad_norm": 0.3644621670246124,
      "learning_rate": 2e-05,
      "loss": 0.537,
      "step": 1140
    },
    {
      "epoch": 0.15897152336190212,
      "grad_norm": 0.38892823457717896,
      "learning_rate": 2e-05,
      "loss": 0.5348,
      "step": 1150
    },
    {
      "epoch": 0.16035388443461432,
      "grad_norm": 0.3780713975429535,
      "learning_rate": 2e-05,
      "loss": 0.5446,
      "step": 1160
    },
    {
      "epoch": 0.16173624550732651,
      "grad_norm": 0.3657492399215698,
      "learning_rate": 2e-05,
      "loss": 0.5567,
      "step": 1170
    },
    {
      "epoch": 0.1631186065800387,
      "grad_norm": 0.3945137858390808,
      "learning_rate": 2e-05,
      "loss": 0.5235,
      "step": 1180
    },
    {
      "epoch": 0.1645009676527509,
      "grad_norm": 0.3907359540462494,
      "learning_rate": 2e-05,
      "loss": 0.5296,
      "step": 1190
    },
    {
      "epoch": 0.1658833287254631,
      "grad_norm": 0.4258662760257721,
      "learning_rate": 2e-05,
      "loss": 0.534,
      "step": 1200
    },
    {
      "epoch": 0.16726568979817527,
      "grad_norm": 0.39014285802841187,
      "learning_rate": 2e-05,
      "loss": 0.5532,
      "step": 1210
    },
    {
      "epoch": 0.16864805087088747,
      "grad_norm": 0.3833202123641968,
      "learning_rate": 2e-05,
      "loss": 0.5323,
      "step": 1220
    },
    {
      "epoch": 0.17003041194359966,
      "grad_norm": 0.39712953567504883,
      "learning_rate": 2e-05,
      "loss": 0.5448,
      "step": 1230
    },
    {
      "epoch": 0.17141277301631186,
      "grad_norm": 0.3670214116573334,
      "learning_rate": 2e-05,
      "loss": 0.5319,
      "step": 1240
    },
    {
      "epoch": 0.17279513408902406,
      "grad_norm": 0.36940327286720276,
      "learning_rate": 2e-05,
      "loss": 0.5527,
      "step": 1250
    },
    {
      "epoch": 0.17417749516173625,
      "grad_norm": 0.38193532824516296,
      "learning_rate": 2e-05,
      "loss": 0.5555,
      "step": 1260
    },
    {
      "epoch": 0.17555985623444845,
      "grad_norm": 0.3885504901409149,
      "learning_rate": 2e-05,
      "loss": 0.5379,
      "step": 1270
    },
    {
      "epoch": 0.17694221730716064,
      "grad_norm": 0.36657124757766724,
      "learning_rate": 2e-05,
      "loss": 0.5284,
      "step": 1280
    },
    {
      "epoch": 0.1783245783798728,
      "grad_norm": 0.37429049611091614,
      "learning_rate": 2e-05,
      "loss": 0.543,
      "step": 1290
    },
    {
      "epoch": 0.179706939452585,
      "grad_norm": 0.35409459471702576,
      "learning_rate": 2e-05,
      "loss": 0.5094,
      "step": 1300
    },
    {
      "epoch": 0.1810893005252972,
      "grad_norm": 0.3809099495410919,
      "learning_rate": 2e-05,
      "loss": 0.5229,
      "step": 1310
    },
    {
      "epoch": 0.1824716615980094,
      "grad_norm": 0.36208510398864746,
      "learning_rate": 2e-05,
      "loss": 0.5434,
      "step": 1320
    },
    {
      "epoch": 0.1838540226707216,
      "grad_norm": 0.368488609790802,
      "learning_rate": 2e-05,
      "loss": 0.5401,
      "step": 1330
    },
    {
      "epoch": 0.1852363837434338,
      "grad_norm": 0.3437720537185669,
      "learning_rate": 2e-05,
      "loss": 0.5404,
      "step": 1340
    },
    {
      "epoch": 0.186618744816146,
      "grad_norm": 0.38139086961746216,
      "learning_rate": 2e-05,
      "loss": 0.5286,
      "step": 1350
    },
    {
      "epoch": 0.18800110588885818,
      "grad_norm": 0.3491038680076599,
      "learning_rate": 2e-05,
      "loss": 0.5297,
      "step": 1360
    },
    {
      "epoch": 0.18938346696157035,
      "grad_norm": 0.37059760093688965,
      "learning_rate": 2e-05,
      "loss": 0.5463,
      "step": 1370
    },
    {
      "epoch": 0.19076582803428255,
      "grad_norm": 0.39735284447669983,
      "learning_rate": 2e-05,
      "loss": 0.5111,
      "step": 1380
    },
    {
      "epoch": 0.19214818910699474,
      "grad_norm": 0.39142096042633057,
      "learning_rate": 2e-05,
      "loss": 0.5352,
      "step": 1390
    },
    {
      "epoch": 0.19353055017970694,
      "grad_norm": 0.3784864544868469,
      "learning_rate": 2e-05,
      "loss": 0.5219,
      "step": 1400
    },
    {
      "epoch": 0.19491291125241914,
      "grad_norm": 0.3911328911781311,
      "learning_rate": 2e-05,
      "loss": 0.538,
      "step": 1410
    },
    {
      "epoch": 0.19629527232513133,
      "grad_norm": 0.38907352089881897,
      "learning_rate": 2e-05,
      "loss": 0.5325,
      "step": 1420
    },
    {
      "epoch": 0.19767763339784353,
      "grad_norm": 0.3667750060558319,
      "learning_rate": 2e-05,
      "loss": 0.5367,
      "step": 1430
    },
    {
      "epoch": 0.1990599944705557,
      "grad_norm": 0.3971737325191498,
      "learning_rate": 2e-05,
      "loss": 0.5381,
      "step": 1440
    },
    {
      "epoch": 0.2004423555432679,
      "grad_norm": 0.3768478333950043,
      "learning_rate": 2e-05,
      "loss": 0.5297,
      "step": 1450
    },
    {
      "epoch": 0.2018247166159801,
      "grad_norm": 0.3777858018875122,
      "learning_rate": 2e-05,
      "loss": 0.4985,
      "step": 1460
    },
    {
      "epoch": 0.20320707768869228,
      "grad_norm": 0.819282054901123,
      "learning_rate": 2e-05,
      "loss": 0.5297,
      "step": 1470
    },
    {
      "epoch": 0.20458943876140448,
      "grad_norm": 0.3887805640697479,
      "learning_rate": 2e-05,
      "loss": 0.5325,
      "step": 1480
    },
    {
      "epoch": 0.20597179983411668,
      "grad_norm": 0.3612997829914093,
      "learning_rate": 2e-05,
      "loss": 0.5366,
      "step": 1490
    },
    {
      "epoch": 0.20735416090682887,
      "grad_norm": 0.38899514079093933,
      "learning_rate": 2e-05,
      "loss": 0.5206,
      "step": 1500
    },
    {
      "epoch": 0.20873652197954107,
      "grad_norm": 0.35962843894958496,
      "learning_rate": 2e-05,
      "loss": 0.5149,
      "step": 1510
    },
    {
      "epoch": 0.21011888305225324,
      "grad_norm": 0.3745998740196228,
      "learning_rate": 2e-05,
      "loss": 0.5246,
      "step": 1520
    },
    {
      "epoch": 0.21150124412496543,
      "grad_norm": 0.3898766338825226,
      "learning_rate": 2e-05,
      "loss": 0.5457,
      "step": 1530
    },
    {
      "epoch": 0.21288360519767763,
      "grad_norm": 0.3771664500236511,
      "learning_rate": 2e-05,
      "loss": 0.5234,
      "step": 1540
    },
    {
      "epoch": 0.21426596627038982,
      "grad_norm": 0.3765556812286377,
      "learning_rate": 2e-05,
      "loss": 0.5268,
      "step": 1550
    },
    {
      "epoch": 0.21564832734310202,
      "grad_norm": 0.3814949691295624,
      "learning_rate": 2e-05,
      "loss": 0.5065,
      "step": 1560
    },
    {
      "epoch": 0.21703068841581422,
      "grad_norm": 0.35859838128089905,
      "learning_rate": 2e-05,
      "loss": 0.5231,
      "step": 1570
    },
    {
      "epoch": 0.2184130494885264,
      "grad_norm": 0.37819817662239075,
      "learning_rate": 2e-05,
      "loss": 0.5231,
      "step": 1580
    },
    {
      "epoch": 0.2197954105612386,
      "grad_norm": 0.36414748430252075,
      "learning_rate": 2e-05,
      "loss": 0.5301,
      "step": 1590
    },
    {
      "epoch": 0.22117777163395078,
      "grad_norm": 0.37390682101249695,
      "learning_rate": 2e-05,
      "loss": 0.522,
      "step": 1600
    },
    {
      "epoch": 0.22256013270666297,
      "grad_norm": 0.3698122203350067,
      "learning_rate": 2e-05,
      "loss": 0.5412,
      "step": 1610
    },
    {
      "epoch": 0.22394249377937517,
      "grad_norm": 0.35356974601745605,
      "learning_rate": 2e-05,
      "loss": 0.5319,
      "step": 1620
    },
    {
      "epoch": 0.22532485485208736,
      "grad_norm": 0.3945378065109253,
      "learning_rate": 2e-05,
      "loss": 0.5266,
      "step": 1630
    },
    {
      "epoch": 0.22670721592479956,
      "grad_norm": 0.3743287920951843,
      "learning_rate": 2e-05,
      "loss": 0.5412,
      "step": 1640
    },
    {
      "epoch": 0.22808957699751176,
      "grad_norm": 0.40030521154403687,
      "learning_rate": 2e-05,
      "loss": 0.5509,
      "step": 1650
    },
    {
      "epoch": 0.22947193807022395,
      "grad_norm": 0.4056825637817383,
      "learning_rate": 2e-05,
      "loss": 0.5306,
      "step": 1660
    },
    {
      "epoch": 0.23085429914293615,
      "grad_norm": 0.4221413731575012,
      "learning_rate": 2e-05,
      "loss": 0.5321,
      "step": 1670
    },
    {
      "epoch": 0.23223666021564832,
      "grad_norm": 0.39226195216178894,
      "learning_rate": 2e-05,
      "loss": 0.5022,
      "step": 1680
    },
    {
      "epoch": 0.2336190212883605,
      "grad_norm": 0.39041945338249207,
      "learning_rate": 2e-05,
      "loss": 0.5264,
      "step": 1690
    },
    {
      "epoch": 0.2350013823610727,
      "grad_norm": 0.37408989667892456,
      "learning_rate": 2e-05,
      "loss": 0.5186,
      "step": 1700
    },
    {
      "epoch": 0.2363837434337849,
      "grad_norm": 0.3956291675567627,
      "learning_rate": 2e-05,
      "loss": 0.5369,
      "step": 1710
    },
    {
      "epoch": 0.2377661045064971,
      "grad_norm": 0.3822111487388611,
      "learning_rate": 2e-05,
      "loss": 0.5037,
      "step": 1720
    },
    {
      "epoch": 0.2391484655792093,
      "grad_norm": 0.34465575218200684,
      "learning_rate": 2e-05,
      "loss": 0.5222,
      "step": 1730
    },
    {
      "epoch": 0.2405308266519215,
      "grad_norm": 0.37024855613708496,
      "learning_rate": 2e-05,
      "loss": 0.5449,
      "step": 1740
    },
    {
      "epoch": 0.2419131877246337,
      "grad_norm": 0.4419766664505005,
      "learning_rate": 2e-05,
      "loss": 0.5403,
      "step": 1750
    },
    {
      "epoch": 0.24329554879734586,
      "grad_norm": 0.3621579706668854,
      "learning_rate": 2e-05,
      "loss": 0.5263,
      "step": 1760
    },
    {
      "epoch": 0.24467790987005805,
      "grad_norm": 0.3967221975326538,
      "learning_rate": 2e-05,
      "loss": 0.5206,
      "step": 1770
    },
    {
      "epoch": 0.24606027094277025,
      "grad_norm": 0.362562894821167,
      "learning_rate": 2e-05,
      "loss": 0.529,
      "step": 1780
    },
    {
      "epoch": 0.24744263201548244,
      "grad_norm": 0.44712892174720764,
      "learning_rate": 2e-05,
      "loss": 0.5185,
      "step": 1790
    },
    {
      "epoch": 0.24882499308819464,
      "grad_norm": 0.3693428337574005,
      "learning_rate": 2e-05,
      "loss": 0.5282,
      "step": 1800
    },
    {
      "epoch": 0.25020735416090684,
      "grad_norm": 0.34170734882354736,
      "learning_rate": 2e-05,
      "loss": 0.5161,
      "step": 1810
    },
    {
      "epoch": 0.251589715233619,
      "grad_norm": 0.555030345916748,
      "learning_rate": 2e-05,
      "loss": 0.5365,
      "step": 1820
    },
    {
      "epoch": 0.25297207630633123,
      "grad_norm": 0.35996249318122864,
      "learning_rate": 2e-05,
      "loss": 0.5203,
      "step": 1830
    },
    {
      "epoch": 0.2543544373790434,
      "grad_norm": 0.40620744228363037,
      "learning_rate": 2e-05,
      "loss": 0.5152,
      "step": 1840
    },
    {
      "epoch": 0.2557367984517556,
      "grad_norm": 0.38798853754997253,
      "learning_rate": 2e-05,
      "loss": 0.5511,
      "step": 1850
    },
    {
      "epoch": 0.2571191595244678,
      "grad_norm": 0.3893015682697296,
      "learning_rate": 2e-05,
      "loss": 0.5329,
      "step": 1860
    },
    {
      "epoch": 0.25850152059717996,
      "grad_norm": 0.3852402865886688,
      "learning_rate": 2e-05,
      "loss": 0.5297,
      "step": 1870
    },
    {
      "epoch": 0.2598838816698922,
      "grad_norm": 0.35694757103919983,
      "learning_rate": 2e-05,
      "loss": 0.5065,
      "step": 1880
    },
    {
      "epoch": 0.26126624274260435,
      "grad_norm": 0.40064889192581177,
      "learning_rate": 2e-05,
      "loss": 0.527,
      "step": 1890
    },
    {
      "epoch": 0.2626486038153166,
      "grad_norm": 0.39502424001693726,
      "learning_rate": 2e-05,
      "loss": 0.527,
      "step": 1900
    },
    {
      "epoch": 0.26403096488802874,
      "grad_norm": 0.36546483635902405,
      "learning_rate": 2e-05,
      "loss": 0.5185,
      "step": 1910
    },
    {
      "epoch": 0.26541332596074096,
      "grad_norm": 0.37893611192703247,
      "learning_rate": 2e-05,
      "loss": 0.53,
      "step": 1920
    },
    {
      "epoch": 0.26679568703345313,
      "grad_norm": 0.38221481442451477,
      "learning_rate": 2e-05,
      "loss": 0.5153,
      "step": 1930
    },
    {
      "epoch": 0.26817804810616536,
      "grad_norm": 0.38632261753082275,
      "learning_rate": 2e-05,
      "loss": 0.5214,
      "step": 1940
    },
    {
      "epoch": 0.2695604091788775,
      "grad_norm": 0.4493684768676758,
      "learning_rate": 2e-05,
      "loss": 0.5082,
      "step": 1950
    },
    {
      "epoch": 0.2709427702515897,
      "grad_norm": 0.3479729890823364,
      "learning_rate": 2e-05,
      "loss": 0.5313,
      "step": 1960
    },
    {
      "epoch": 0.2723251313243019,
      "grad_norm": 0.43744829297065735,
      "learning_rate": 2e-05,
      "loss": 0.5076,
      "step": 1970
    },
    {
      "epoch": 0.2737074923970141,
      "grad_norm": 0.38938048481941223,
      "learning_rate": 2e-05,
      "loss": 0.5211,
      "step": 1980
    },
    {
      "epoch": 0.2750898534697263,
      "grad_norm": 0.364580363035202,
      "learning_rate": 2e-05,
      "loss": 0.5018,
      "step": 1990
    },
    {
      "epoch": 0.2764722145424385,
      "grad_norm": 0.3710174560546875,
      "learning_rate": 2e-05,
      "loss": 0.5339,
      "step": 2000
    },
    {
      "epoch": 0.2778545756151507,
      "grad_norm": 0.36626002192497253,
      "learning_rate": 2e-05,
      "loss": 0.5203,
      "step": 2010
    },
    {
      "epoch": 0.27923693668786287,
      "grad_norm": 0.3862822651863098,
      "learning_rate": 2e-05,
      "loss": 0.5405,
      "step": 2020
    },
    {
      "epoch": 0.28061929776057504,
      "grad_norm": 0.3611544370651245,
      "learning_rate": 2e-05,
      "loss": 0.5201,
      "step": 2030
    },
    {
      "epoch": 0.28200165883328726,
      "grad_norm": 0.3436617851257324,
      "learning_rate": 2e-05,
      "loss": 0.5321,
      "step": 2040
    },
    {
      "epoch": 0.28338401990599943,
      "grad_norm": 0.414676308631897,
      "learning_rate": 2e-05,
      "loss": 0.5134,
      "step": 2050
    },
    {
      "epoch": 0.28476638097871165,
      "grad_norm": 0.36941394209861755,
      "learning_rate": 2e-05,
      "loss": 0.5259,
      "step": 2060
    },
    {
      "epoch": 0.2861487420514238,
      "grad_norm": 0.39532580971717834,
      "learning_rate": 2e-05,
      "loss": 0.54,
      "step": 2070
    },
    {
      "epoch": 0.28753110312413604,
      "grad_norm": 0.3761785328388214,
      "learning_rate": 2e-05,
      "loss": 0.5329,
      "step": 2080
    },
    {
      "epoch": 0.2889134641968482,
      "grad_norm": 0.3895875811576843,
      "learning_rate": 2e-05,
      "loss": 0.5145,
      "step": 2090
    },
    {
      "epoch": 0.29029582526956044,
      "grad_norm": 0.38536500930786133,
      "learning_rate": 2e-05,
      "loss": 0.536,
      "step": 2100
    },
    {
      "epoch": 0.2916781863422726,
      "grad_norm": 0.4088023900985718,
      "learning_rate": 2e-05,
      "loss": 0.5186,
      "step": 2110
    },
    {
      "epoch": 0.2930605474149848,
      "grad_norm": 0.3533695936203003,
      "learning_rate": 2e-05,
      "loss": 0.5218,
      "step": 2120
    },
    {
      "epoch": 0.294442908487697,
      "grad_norm": 0.3686179518699646,
      "learning_rate": 2e-05,
      "loss": 0.5128,
      "step": 2130
    },
    {
      "epoch": 0.29582526956040917,
      "grad_norm": 0.36640217900276184,
      "learning_rate": 2e-05,
      "loss": 0.5428,
      "step": 2140
    },
    {
      "epoch": 0.2972076306331214,
      "grad_norm": 0.3724895417690277,
      "learning_rate": 2e-05,
      "loss": 0.5393,
      "step": 2150
    },
    {
      "epoch": 0.29858999170583356,
      "grad_norm": 0.385895699262619,
      "learning_rate": 2e-05,
      "loss": 0.5451,
      "step": 2160
    },
    {
      "epoch": 0.2999723527785458,
      "grad_norm": 0.4075011909008026,
      "learning_rate": 2e-05,
      "loss": 0.5297,
      "step": 2170
    },
    {
      "epoch": 0.30135471385125795,
      "grad_norm": 0.3829037547111511,
      "learning_rate": 2e-05,
      "loss": 0.5178,
      "step": 2180
    },
    {
      "epoch": 0.3027370749239701,
      "grad_norm": 0.37158817052841187,
      "learning_rate": 2e-05,
      "loss": 0.5165,
      "step": 2190
    },
    {
      "epoch": 0.30411943599668234,
      "grad_norm": 0.3635076880455017,
      "learning_rate": 2e-05,
      "loss": 0.5096,
      "step": 2200
    },
    {
      "epoch": 0.3055017970693945,
      "grad_norm": 0.5580878853797913,
      "learning_rate": 2e-05,
      "loss": 0.5335,
      "step": 2210
    },
    {
      "epoch": 0.30688415814210673,
      "grad_norm": 0.3738490045070648,
      "learning_rate": 2e-05,
      "loss": 0.5106,
      "step": 2220
    },
    {
      "epoch": 0.3082665192148189,
      "grad_norm": 0.3896075487136841,
      "learning_rate": 2e-05,
      "loss": 0.5069,
      "step": 2230
    },
    {
      "epoch": 0.3096488802875311,
      "grad_norm": 0.39652112126350403,
      "learning_rate": 2e-05,
      "loss": 0.5407,
      "step": 2240
    },
    {
      "epoch": 0.3110312413602433,
      "grad_norm": 0.35033077001571655,
      "learning_rate": 2e-05,
      "loss": 0.5154,
      "step": 2250
    },
    {
      "epoch": 0.31241360243295546,
      "grad_norm": 0.37589964270591736,
      "learning_rate": 2e-05,
      "loss": 0.5284,
      "step": 2260
    },
    {
      "epoch": 0.3137959635056677,
      "grad_norm": 0.3693106472492218,
      "learning_rate": 2e-05,
      "loss": 0.5072,
      "step": 2270
    },
    {
      "epoch": 0.31517832457837985,
      "grad_norm": 0.34492287039756775,
      "learning_rate": 2e-05,
      "loss": 0.5032,
      "step": 2280
    },
    {
      "epoch": 0.3165606856510921,
      "grad_norm": 0.36448386311531067,
      "learning_rate": 2e-05,
      "loss": 0.5296,
      "step": 2290
    },
    {
      "epoch": 0.31794304672380425,
      "grad_norm": 0.37113285064697266,
      "learning_rate": 2e-05,
      "loss": 0.5112,
      "step": 2300
    },
    {
      "epoch": 0.31932540779651647,
      "grad_norm": 0.39453989267349243,
      "learning_rate": 2e-05,
      "loss": 0.4949,
      "step": 2310
    },
    {
      "epoch": 0.32070776886922864,
      "grad_norm": 0.3524867594242096,
      "learning_rate": 2e-05,
      "loss": 0.5205,
      "step": 2320
    },
    {
      "epoch": 0.32209012994194086,
      "grad_norm": 0.36552318930625916,
      "learning_rate": 2e-05,
      "loss": 0.5137,
      "step": 2330
    },
    {
      "epoch": 0.32347249101465303,
      "grad_norm": 0.37899231910705566,
      "learning_rate": 2e-05,
      "loss": 0.5157,
      "step": 2340
    },
    {
      "epoch": 0.3248548520873652,
      "grad_norm": 0.351675420999527,
      "learning_rate": 2e-05,
      "loss": 0.5346,
      "step": 2350
    },
    {
      "epoch": 0.3262372131600774,
      "grad_norm": 0.3652631342411041,
      "learning_rate": 2e-05,
      "loss": 0.5191,
      "step": 2360
    },
    {
      "epoch": 0.3276195742327896,
      "grad_norm": 0.44226595759391785,
      "learning_rate": 2e-05,
      "loss": 0.5066,
      "step": 2370
    },
    {
      "epoch": 0.3290019353055018,
      "grad_norm": 0.38316550850868225,
      "learning_rate": 2e-05,
      "loss": 0.5232,
      "step": 2380
    },
    {
      "epoch": 0.330384296378214,
      "grad_norm": 0.3634336590766907,
      "learning_rate": 2e-05,
      "loss": 0.5202,
      "step": 2390
    },
    {
      "epoch": 0.3317666574509262,
      "grad_norm": 0.4114074409008026,
      "learning_rate": 2e-05,
      "loss": 0.5031,
      "step": 2400
    },
    {
      "epoch": 0.3331490185236384,
      "grad_norm": 0.3629259765148163,
      "learning_rate": 2e-05,
      "loss": 0.5163,
      "step": 2410
    },
    {
      "epoch": 0.33453137959635054,
      "grad_norm": 0.7599861025810242,
      "learning_rate": 2e-05,
      "loss": 0.5066,
      "step": 2420
    },
    {
      "epoch": 0.33591374066906277,
      "grad_norm": 0.3549429178237915,
      "learning_rate": 2e-05,
      "loss": 0.5138,
      "step": 2430
    },
    {
      "epoch": 0.33729610174177493,
      "grad_norm": 0.3634311556816101,
      "learning_rate": 2e-05,
      "loss": 0.5144,
      "step": 2440
    },
    {
      "epoch": 0.33867846281448716,
      "grad_norm": 0.39256978034973145,
      "learning_rate": 2e-05,
      "loss": 0.5328,
      "step": 2450
    },
    {
      "epoch": 0.3400608238871993,
      "grad_norm": 0.37388917803764343,
      "learning_rate": 2e-05,
      "loss": 0.5132,
      "step": 2460
    },
    {
      "epoch": 0.34144318495991155,
      "grad_norm": 0.3479449152946472,
      "learning_rate": 2e-05,
      "loss": 0.5047,
      "step": 2470
    },
    {
      "epoch": 0.3428255460326237,
      "grad_norm": 0.382307231426239,
      "learning_rate": 2e-05,
      "loss": 0.4952,
      "step": 2480
    },
    {
      "epoch": 0.3442079071053359,
      "grad_norm": 0.36051592230796814,
      "learning_rate": 2e-05,
      "loss": 0.517,
      "step": 2490
    },
    {
      "epoch": 0.3455902681780481,
      "grad_norm": 0.35527852177619934,
      "learning_rate": 2e-05,
      "loss": 0.5255,
      "step": 2500
    },
    {
      "epoch": 0.3469726292507603,
      "grad_norm": 0.37652215361595154,
      "learning_rate": 2e-05,
      "loss": 0.5153,
      "step": 2510
    },
    {
      "epoch": 0.3483549903234725,
      "grad_norm": 0.5380536913871765,
      "learning_rate": 2e-05,
      "loss": 0.5078,
      "step": 2520
    },
    {
      "epoch": 0.34973735139618467,
      "grad_norm": 0.3670841455459595,
      "learning_rate": 2e-05,
      "loss": 0.5074,
      "step": 2530
    },
    {
      "epoch": 0.3511197124688969,
      "grad_norm": 0.37626156210899353,
      "learning_rate": 2e-05,
      "loss": 0.5306,
      "step": 2540
    },
    {
      "epoch": 0.35250207354160906,
      "grad_norm": 0.3564271926879883,
      "learning_rate": 2e-05,
      "loss": 0.4997,
      "step": 2550
    },
    {
      "epoch": 0.3538844346143213,
      "grad_norm": 0.42474862933158875,
      "learning_rate": 2e-05,
      "loss": 0.5237,
      "step": 2560
    },
    {
      "epoch": 0.35526679568703345,
      "grad_norm": 0.37016427516937256,
      "learning_rate": 2e-05,
      "loss": 0.5135,
      "step": 2570
    },
    {
      "epoch": 0.3566491567597456,
      "grad_norm": 0.3544558584690094,
      "learning_rate": 2e-05,
      "loss": 0.5161,
      "step": 2580
    },
    {
      "epoch": 0.35803151783245785,
      "grad_norm": 0.5672499537467957,
      "learning_rate": 2e-05,
      "loss": 0.5186,
      "step": 2590
    },
    {
      "epoch": 0.35941387890517,
      "grad_norm": 0.35322484374046326,
      "learning_rate": 2e-05,
      "loss": 0.5049,
      "step": 2600
    },
    {
      "epoch": 0.36079623997788224,
      "grad_norm": 0.34540778398513794,
      "learning_rate": 2e-05,
      "loss": 0.5102,
      "step": 2610
    },
    {
      "epoch": 0.3621786010505944,
      "grad_norm": 0.384725958108902,
      "learning_rate": 2e-05,
      "loss": 0.53,
      "step": 2620
    },
    {
      "epoch": 0.36356096212330663,
      "grad_norm": 0.36552974581718445,
      "learning_rate": 2e-05,
      "loss": 0.5131,
      "step": 2630
    },
    {
      "epoch": 0.3649433231960188,
      "grad_norm": 0.36756059527397156,
      "learning_rate": 2e-05,
      "loss": 0.5194,
      "step": 2640
    },
    {
      "epoch": 0.36632568426873097,
      "grad_norm": 0.3563382625579834,
      "learning_rate": 2e-05,
      "loss": 0.5152,
      "step": 2650
    },
    {
      "epoch": 0.3677080453414432,
      "grad_norm": 0.3669905364513397,
      "learning_rate": 2e-05,
      "loss": 0.4927,
      "step": 2660
    },
    {
      "epoch": 0.36909040641415536,
      "grad_norm": 0.5866174697875977,
      "learning_rate": 2e-05,
      "loss": 0.5009,
      "step": 2670
    },
    {
      "epoch": 0.3704727674868676,
      "grad_norm": 0.3552144169807434,
      "learning_rate": 2e-05,
      "loss": 0.5045,
      "step": 2680
    },
    {
      "epoch": 0.37185512855957975,
      "grad_norm": 0.3726100027561188,
      "learning_rate": 2e-05,
      "loss": 0.5314,
      "step": 2690
    },
    {
      "epoch": 0.373237489632292,
      "grad_norm": 0.3901733160018921,
      "learning_rate": 2e-05,
      "loss": 0.5279,
      "step": 2700
    },
    {
      "epoch": 0.37461985070500414,
      "grad_norm": 0.3929930627346039,
      "learning_rate": 2e-05,
      "loss": 0.5098,
      "step": 2710
    },
    {
      "epoch": 0.37600221177771637,
      "grad_norm": 0.35530325770378113,
      "learning_rate": 2e-05,
      "loss": 0.4987,
      "step": 2720
    },
    {
      "epoch": 0.37738457285042853,
      "grad_norm": 0.4379098117351532,
      "learning_rate": 2e-05,
      "loss": 0.5014,
      "step": 2730
    },
    {
      "epoch": 0.3787669339231407,
      "grad_norm": 0.3846440315246582,
      "learning_rate": 2e-05,
      "loss": 0.5143,
      "step": 2740
    },
    {
      "epoch": 0.3801492949958529,
      "grad_norm": 0.35410434007644653,
      "learning_rate": 2e-05,
      "loss": 0.504,
      "step": 2750
    },
    {
      "epoch": 0.3815316560685651,
      "grad_norm": 0.363675981760025,
      "learning_rate": 2e-05,
      "loss": 0.5168,
      "step": 2760
    },
    {
      "epoch": 0.3829140171412773,
      "grad_norm": 0.38489505648612976,
      "learning_rate": 2e-05,
      "loss": 0.5152,
      "step": 2770
    },
    {
      "epoch": 0.3842963782139895,
      "grad_norm": 0.37016889452934265,
      "learning_rate": 2e-05,
      "loss": 0.5121,
      "step": 2780
    },
    {
      "epoch": 0.3856787392867017,
      "grad_norm": 0.3752250075340271,
      "learning_rate": 2e-05,
      "loss": 0.5101,
      "step": 2790
    },
    {
      "epoch": 0.3870611003594139,
      "grad_norm": 0.38032618165016174,
      "learning_rate": 2e-05,
      "loss": 0.5158,
      "step": 2800
    },
    {
      "epoch": 0.38844346143212605,
      "grad_norm": 0.401915967464447,
      "learning_rate": 2e-05,
      "loss": 0.5009,
      "step": 2810
    },
    {
      "epoch": 0.38982582250483827,
      "grad_norm": 0.3516676723957062,
      "learning_rate": 2e-05,
      "loss": 0.5002,
      "step": 2820
    },
    {
      "epoch": 0.39120818357755044,
      "grad_norm": 0.3500385284423828,
      "learning_rate": 2e-05,
      "loss": 0.5128,
      "step": 2830
    },
    {
      "epoch": 0.39259054465026266,
      "grad_norm": 0.35959821939468384,
      "learning_rate": 2e-05,
      "loss": 0.5179,
      "step": 2840
    },
    {
      "epoch": 0.39397290572297483,
      "grad_norm": 0.3963475227355957,
      "learning_rate": 2e-05,
      "loss": 0.5162,
      "step": 2850
    },
    {
      "epoch": 0.39535526679568705,
      "grad_norm": 0.37910184264183044,
      "learning_rate": 2e-05,
      "loss": 0.5012,
      "step": 2860
    },
    {
      "epoch": 0.3967376278683992,
      "grad_norm": 0.3836962878704071,
      "learning_rate": 2e-05,
      "loss": 0.5171,
      "step": 2870
    },
    {
      "epoch": 0.3981199889411114,
      "grad_norm": 0.35654759407043457,
      "learning_rate": 2e-05,
      "loss": 0.506,
      "step": 2880
    },
    {
      "epoch": 0.3995023500138236,
      "grad_norm": 0.3719756603240967,
      "learning_rate": 2e-05,
      "loss": 0.5138,
      "step": 2890
    },
    {
      "epoch": 0.4008847110865358,
      "grad_norm": 0.3636404871940613,
      "learning_rate": 2e-05,
      "loss": 0.5291,
      "step": 2900
    },
    {
      "epoch": 0.402267072159248,
      "grad_norm": 0.369764119386673,
      "learning_rate": 2e-05,
      "loss": 0.5031,
      "step": 2910
    },
    {
      "epoch": 0.4036494332319602,
      "grad_norm": 0.36248886585235596,
      "learning_rate": 2e-05,
      "loss": 0.5133,
      "step": 2920
    },
    {
      "epoch": 0.4050317943046724,
      "grad_norm": 0.3792344927787781,
      "learning_rate": 2e-05,
      "loss": 0.5044,
      "step": 2930
    },
    {
      "epoch": 0.40641415537738457,
      "grad_norm": 0.5317441821098328,
      "learning_rate": 2e-05,
      "loss": 0.5176,
      "step": 2940
    },
    {
      "epoch": 0.4077965164500968,
      "grad_norm": 0.3509214520454407,
      "learning_rate": 2e-05,
      "loss": 0.5221,
      "step": 2950
    },
    {
      "epoch": 0.40917887752280896,
      "grad_norm": 0.4713679254055023,
      "learning_rate": 2e-05,
      "loss": 0.513,
      "step": 2960
    },
    {
      "epoch": 0.4105612385955211,
      "grad_norm": 0.332958847284317,
      "learning_rate": 2e-05,
      "loss": 0.507,
      "step": 2970
    },
    {
      "epoch": 0.41194359966823335,
      "grad_norm": 0.377456933259964,
      "learning_rate": 2e-05,
      "loss": 0.5016,
      "step": 2980
    },
    {
      "epoch": 0.4133259607409455,
      "grad_norm": 0.3597404360771179,
      "learning_rate": 2e-05,
      "loss": 0.5079,
      "step": 2990
    },
    {
      "epoch": 0.41470832181365774,
      "grad_norm": 0.36518895626068115,
      "learning_rate": 2e-05,
      "loss": 0.5003,
      "step": 3000
    },
    {
      "epoch": 0.4160906828863699,
      "grad_norm": 0.4057992398738861,
      "learning_rate": 2e-05,
      "loss": 0.5241,
      "step": 3010
    },
    {
      "epoch": 0.41747304395908214,
      "grad_norm": 0.37498101592063904,
      "learning_rate": 2e-05,
      "loss": 0.5132,
      "step": 3020
    },
    {
      "epoch": 0.4188554050317943,
      "grad_norm": 0.37146303057670593,
      "learning_rate": 2e-05,
      "loss": 0.4971,
      "step": 3030
    },
    {
      "epoch": 0.42023776610450647,
      "grad_norm": 0.38399770855903625,
      "learning_rate": 2e-05,
      "loss": 0.5251,
      "step": 3040
    },
    {
      "epoch": 0.4216201271772187,
      "grad_norm": 0.40745237469673157,
      "learning_rate": 2e-05,
      "loss": 0.4912,
      "step": 3050
    },
    {
      "epoch": 0.42300248824993086,
      "grad_norm": 0.38469985127449036,
      "learning_rate": 2e-05,
      "loss": 0.5049,
      "step": 3060
    },
    {
      "epoch": 0.4243848493226431,
      "grad_norm": 0.3462386429309845,
      "learning_rate": 2e-05,
      "loss": 0.4991,
      "step": 3070
    },
    {
      "epoch": 0.42576721039535526,
      "grad_norm": 0.39158323407173157,
      "learning_rate": 2e-05,
      "loss": 0.5244,
      "step": 3080
    },
    {
      "epoch": 0.4271495714680675,
      "grad_norm": 0.38380932807922363,
      "learning_rate": 2e-05,
      "loss": 0.511,
      "step": 3090
    },
    {
      "epoch": 0.42853193254077965,
      "grad_norm": 0.3770333230495453,
      "learning_rate": 2e-05,
      "loss": 0.4997,
      "step": 3100
    },
    {
      "epoch": 0.42991429361349187,
      "grad_norm": 0.40797197818756104,
      "learning_rate": 2e-05,
      "loss": 0.5037,
      "step": 3110
    },
    {
      "epoch": 0.43129665468620404,
      "grad_norm": 0.35173916816711426,
      "learning_rate": 2e-05,
      "loss": 0.5086,
      "step": 3120
    },
    {
      "epoch": 0.4326790157589162,
      "grad_norm": 0.3891461193561554,
      "learning_rate": 2e-05,
      "loss": 0.507,
      "step": 3130
    },
    {
      "epoch": 0.43406137683162843,
      "grad_norm": 0.3675945997238159,
      "learning_rate": 2e-05,
      "loss": 0.4975,
      "step": 3140
    },
    {
      "epoch": 0.4354437379043406,
      "grad_norm": 0.40966328978538513,
      "learning_rate": 2e-05,
      "loss": 0.5092,
      "step": 3150
    },
    {
      "epoch": 0.4368260989770528,
      "grad_norm": 0.3550536632537842,
      "learning_rate": 2e-05,
      "loss": 0.5083,
      "step": 3160
    },
    {
      "epoch": 0.438208460049765,
      "grad_norm": 0.5163268446922302,
      "learning_rate": 2e-05,
      "loss": 0.5235,
      "step": 3170
    },
    {
      "epoch": 0.4395908211224772,
      "grad_norm": 0.3860738277435303,
      "learning_rate": 2e-05,
      "loss": 0.5277,
      "step": 3180
    },
    {
      "epoch": 0.4409731821951894,
      "grad_norm": 0.4370495080947876,
      "learning_rate": 2e-05,
      "loss": 0.4986,
      "step": 3190
    },
    {
      "epoch": 0.44235554326790155,
      "grad_norm": 0.6449421048164368,
      "learning_rate": 2e-05,
      "loss": 0.5026,
      "step": 3200
    },
    {
      "epoch": 0.4437379043406138,
      "grad_norm": 0.39056459069252014,
      "learning_rate": 2e-05,
      "loss": 0.5147,
      "step": 3210
    },
    {
      "epoch": 0.44512026541332594,
      "grad_norm": 0.3760967552661896,
      "learning_rate": 2e-05,
      "loss": 0.5118,
      "step": 3220
    },
    {
      "epoch": 0.44650262648603817,
      "grad_norm": 0.36903247237205505,
      "learning_rate": 2e-05,
      "loss": 0.5053,
      "step": 3230
    },
    {
      "epoch": 0.44788498755875034,
      "grad_norm": 0.3520955741405487,
      "learning_rate": 2e-05,
      "loss": 0.4909,
      "step": 3240
    },
    {
      "epoch": 0.44926734863146256,
      "grad_norm": 0.34639033675193787,
      "learning_rate": 2e-05,
      "loss": 0.5239,
      "step": 3250
    },
    {
      "epoch": 0.45064970970417473,
      "grad_norm": 0.3815230429172516,
      "learning_rate": 2e-05,
      "loss": 0.512,
      "step": 3260
    },
    {
      "epoch": 0.4520320707768869,
      "grad_norm": 0.3396725654602051,
      "learning_rate": 2e-05,
      "loss": 0.5013,
      "step": 3270
    },
    {
      "epoch": 0.4534144318495991,
      "grad_norm": 0.3416706919670105,
      "learning_rate": 2e-05,
      "loss": 0.5199,
      "step": 3280
    },
    {
      "epoch": 0.4547967929223113,
      "grad_norm": 0.39162614941596985,
      "learning_rate": 2e-05,
      "loss": 0.5016,
      "step": 3290
    },
    {
      "epoch": 0.4561791539950235,
      "grad_norm": 0.3534219264984131,
      "learning_rate": 2e-05,
      "loss": 0.4907,
      "step": 3300
    },
    {
      "epoch": 0.4575615150677357,
      "grad_norm": 0.37462279200553894,
      "learning_rate": 2e-05,
      "loss": 0.489,
      "step": 3310
    },
    {
      "epoch": 0.4589438761404479,
      "grad_norm": 0.44896674156188965,
      "learning_rate": 2e-05,
      "loss": 0.4918,
      "step": 3320
    },
    {
      "epoch": 0.46032623721316007,
      "grad_norm": 0.3766739070415497,
      "learning_rate": 2e-05,
      "loss": 0.5015,
      "step": 3330
    },
    {
      "epoch": 0.4617085982858723,
      "grad_norm": 0.3731243312358856,
      "learning_rate": 2e-05,
      "loss": 0.4834,
      "step": 3340
    },
    {
      "epoch": 0.46309095935858446,
      "grad_norm": 0.3779064416885376,
      "learning_rate": 2e-05,
      "loss": 0.5125,
      "step": 3350
    },
    {
      "epoch": 0.46447332043129663,
      "grad_norm": 0.3384222388267517,
      "learning_rate": 2e-05,
      "loss": 0.5121,
      "step": 3360
    },
    {
      "epoch": 0.46585568150400886,
      "grad_norm": 0.33813780546188354,
      "learning_rate": 2e-05,
      "loss": 0.5083,
      "step": 3370
    },
    {
      "epoch": 0.467238042576721,
      "grad_norm": 0.39807358384132385,
      "learning_rate": 2e-05,
      "loss": 0.5035,
      "step": 3380
    },
    {
      "epoch": 0.46862040364943325,
      "grad_norm": 0.37986233830451965,
      "learning_rate": 2e-05,
      "loss": 0.5095,
      "step": 3390
    },
    {
      "epoch": 0.4700027647221454,
      "grad_norm": 0.36730825901031494,
      "learning_rate": 2e-05,
      "loss": 0.5017,
      "step": 3400
    },
    {
      "epoch": 0.47138512579485764,
      "grad_norm": 0.36764654517173767,
      "learning_rate": 2e-05,
      "loss": 0.5006,
      "step": 3410
    },
    {
      "epoch": 0.4727674868675698,
      "grad_norm": 0.34106895327568054,
      "learning_rate": 2e-05,
      "loss": 0.5166,
      "step": 3420
    },
    {
      "epoch": 0.474149847940282,
      "grad_norm": 0.3782879114151001,
      "learning_rate": 2e-05,
      "loss": 0.5011,
      "step": 3430
    },
    {
      "epoch": 0.4755322090129942,
      "grad_norm": 0.3349928557872772,
      "learning_rate": 2e-05,
      "loss": 0.5027,
      "step": 3440
    },
    {
      "epoch": 0.47691457008570637,
      "grad_norm": 0.35555174946784973,
      "learning_rate": 2e-05,
      "loss": 0.4916,
      "step": 3450
    },
    {
      "epoch": 0.4782969311584186,
      "grad_norm": 0.3594384491443634,
      "learning_rate": 2e-05,
      "loss": 0.4863,
      "step": 3460
    },
    {
      "epoch": 0.47967929223113076,
      "grad_norm": 0.3686355948448181,
      "learning_rate": 2e-05,
      "loss": 0.5063,
      "step": 3470
    },
    {
      "epoch": 0.481061653303843,
      "grad_norm": 0.354671835899353,
      "learning_rate": 2e-05,
      "loss": 0.5211,
      "step": 3480
    },
    {
      "epoch": 0.48244401437655515,
      "grad_norm": 0.39377838373184204,
      "learning_rate": 2e-05,
      "loss": 0.488,
      "step": 3490
    },
    {
      "epoch": 0.4838263754492674,
      "grad_norm": 0.368236780166626,
      "learning_rate": 2e-05,
      "loss": 0.5039,
      "step": 3500
    },
    {
      "epoch": 0.48520873652197954,
      "grad_norm": 0.3925241231918335,
      "learning_rate": 2e-05,
      "loss": 0.5174,
      "step": 3510
    },
    {
      "epoch": 0.4865910975946917,
      "grad_norm": 0.3547084629535675,
      "learning_rate": 2e-05,
      "loss": 0.4848,
      "step": 3520
    },
    {
      "epoch": 0.48797345866740394,
      "grad_norm": 0.3589189946651459,
      "learning_rate": 2e-05,
      "loss": 0.5159,
      "step": 3530
    },
    {
      "epoch": 0.4893558197401161,
      "grad_norm": 0.36982518434524536,
      "learning_rate": 2e-05,
      "loss": 0.4872,
      "step": 3540
    },
    {
      "epoch": 0.49073818081282833,
      "grad_norm": 0.3496567904949188,
      "learning_rate": 2e-05,
      "loss": 0.4819,
      "step": 3550
    },
    {
      "epoch": 0.4921205418855405,
      "grad_norm": 0.3593372106552124,
      "learning_rate": 2e-05,
      "loss": 0.4988,
      "step": 3560
    },
    {
      "epoch": 0.4935029029582527,
      "grad_norm": 0.3650154173374176,
      "learning_rate": 2e-05,
      "loss": 0.5192,
      "step": 3570
    },
    {
      "epoch": 0.4948852640309649,
      "grad_norm": 0.36254391074180603,
      "learning_rate": 2e-05,
      "loss": 0.5112,
      "step": 3580
    },
    {
      "epoch": 0.49626762510367706,
      "grad_norm": 0.3586868941783905,
      "learning_rate": 2e-05,
      "loss": 0.4747,
      "step": 3590
    },
    {
      "epoch": 0.4976499861763893,
      "grad_norm": 0.46653470396995544,
      "learning_rate": 2e-05,
      "loss": 0.5191,
      "step": 3600
    },
    {
      "epoch": 0.49903234724910145,
      "grad_norm": 0.34913384914398193,
      "learning_rate": 2e-05,
      "loss": 0.5077,
      "step": 3610
    },
    {
      "epoch": 0.5004147083218137,
      "grad_norm": 0.35827919840812683,
      "learning_rate": 2e-05,
      "loss": 0.5069,
      "step": 3620
    },
    {
      "epoch": 0.5017970693945258,
      "grad_norm": 0.40602752566337585,
      "learning_rate": 2e-05,
      "loss": 0.4973,
      "step": 3630
    },
    {
      "epoch": 0.503179430467238,
      "grad_norm": 0.3897165060043335,
      "learning_rate": 2e-05,
      "loss": 0.5035,
      "step": 3640
    },
    {
      "epoch": 0.5045617915399503,
      "grad_norm": 0.3288012742996216,
      "learning_rate": 2e-05,
      "loss": 0.5113,
      "step": 3650
    },
    {
      "epoch": 0.5059441526126625,
      "grad_norm": 0.36094406247138977,
      "learning_rate": 2e-05,
      "loss": 0.4987,
      "step": 3660
    },
    {
      "epoch": 0.5073265136853746,
      "grad_norm": 0.3636637330055237,
      "learning_rate": 2e-05,
      "loss": 0.4888,
      "step": 3670
    },
    {
      "epoch": 0.5087088747580868,
      "grad_norm": 0.3737115263938904,
      "learning_rate": 2e-05,
      "loss": 0.4958,
      "step": 3680
    },
    {
      "epoch": 0.510091235830799,
      "grad_norm": 0.3680795729160309,
      "learning_rate": 2e-05,
      "loss": 0.5039,
      "step": 3690
    },
    {
      "epoch": 0.5114735969035112,
      "grad_norm": 0.3601042628288269,
      "learning_rate": 2e-05,
      "loss": 0.52,
      "step": 3700
    },
    {
      "epoch": 0.5128559579762234,
      "grad_norm": 0.40435078740119934,
      "learning_rate": 2e-05,
      "loss": 0.4984,
      "step": 3710
    },
    {
      "epoch": 0.5142383190489356,
      "grad_norm": 0.37081679701805115,
      "learning_rate": 2e-05,
      "loss": 0.5268,
      "step": 3720
    },
    {
      "epoch": 0.5156206801216477,
      "grad_norm": 0.3774346709251404,
      "learning_rate": 2e-05,
      "loss": 0.5093,
      "step": 3730
    },
    {
      "epoch": 0.5170030411943599,
      "grad_norm": 0.36081093549728394,
      "learning_rate": 2e-05,
      "loss": 0.4973,
      "step": 3740
    },
    {
      "epoch": 0.5183854022670722,
      "grad_norm": 0.3778867721557617,
      "learning_rate": 2e-05,
      "loss": 0.5098,
      "step": 3750
    },
    {
      "epoch": 0.5197677633397844,
      "grad_norm": 0.36873292922973633,
      "learning_rate": 2e-05,
      "loss": 0.4972,
      "step": 3760
    },
    {
      "epoch": 0.5211501244124965,
      "grad_norm": 0.36456185579299927,
      "learning_rate": 2e-05,
      "loss": 0.5108,
      "step": 3770
    },
    {
      "epoch": 0.5225324854852087,
      "grad_norm": 0.34830570220947266,
      "learning_rate": 2e-05,
      "loss": 0.5081,
      "step": 3780
    },
    {
      "epoch": 0.523914846557921,
      "grad_norm": 0.3849759101867676,
      "learning_rate": 2e-05,
      "loss": 0.5017,
      "step": 3790
    },
    {
      "epoch": 0.5252972076306331,
      "grad_norm": 0.3629772663116455,
      "learning_rate": 2e-05,
      "loss": 0.4929,
      "step": 3800
    },
    {
      "epoch": 0.5266795687033453,
      "grad_norm": 0.3344227075576782,
      "learning_rate": 2e-05,
      "loss": 0.5077,
      "step": 3810
    },
    {
      "epoch": 0.5280619297760575,
      "grad_norm": 0.37643933296203613,
      "learning_rate": 2e-05,
      "loss": 0.4838,
      "step": 3820
    },
    {
      "epoch": 0.5294442908487696,
      "grad_norm": 0.36004793643951416,
      "learning_rate": 2e-05,
      "loss": 0.482,
      "step": 3830
    },
    {
      "epoch": 0.5308266519214819,
      "grad_norm": 0.36973950266838074,
      "learning_rate": 2e-05,
      "loss": 0.4927,
      "step": 3840
    },
    {
      "epoch": 0.5322090129941941,
      "grad_norm": 0.36242756247520447,
      "learning_rate": 2e-05,
      "loss": 0.4914,
      "step": 3850
    },
    {
      "epoch": 0.5335913740669063,
      "grad_norm": 0.3760611414909363,
      "learning_rate": 2e-05,
      "loss": 0.4955,
      "step": 3860
    },
    {
      "epoch": 0.5349737351396184,
      "grad_norm": 0.3864375054836273,
      "learning_rate": 2e-05,
      "loss": 0.4927,
      "step": 3870
    },
    {
      "epoch": 0.5363560962123307,
      "grad_norm": 0.37925985455513,
      "learning_rate": 2e-05,
      "loss": 0.5067,
      "step": 3880
    },
    {
      "epoch": 0.5377384572850429,
      "grad_norm": 0.34099358320236206,
      "learning_rate": 2e-05,
      "loss": 0.4926,
      "step": 3890
    },
    {
      "epoch": 0.539120818357755,
      "grad_norm": 0.35715991258621216,
      "learning_rate": 2e-05,
      "loss": 0.4923,
      "step": 3900
    },
    {
      "epoch": 0.5405031794304672,
      "grad_norm": 0.3545444905757904,
      "learning_rate": 2e-05,
      "loss": 0.4858,
      "step": 3910
    },
    {
      "epoch": 0.5418855405031794,
      "grad_norm": 0.3889092803001404,
      "learning_rate": 2e-05,
      "loss": 0.4853,
      "step": 3920
    },
    {
      "epoch": 0.5432679015758917,
      "grad_norm": 0.3511466383934021,
      "learning_rate": 2e-05,
      "loss": 0.4842,
      "step": 3930
    },
    {
      "epoch": 0.5446502626486038,
      "grad_norm": 0.3911827504634857,
      "learning_rate": 2e-05,
      "loss": 0.4971,
      "step": 3940
    },
    {
      "epoch": 0.546032623721316,
      "grad_norm": 0.3663742244243622,
      "learning_rate": 2e-05,
      "loss": 0.4804,
      "step": 3950
    },
    {
      "epoch": 0.5474149847940282,
      "grad_norm": 0.38275814056396484,
      "learning_rate": 2e-05,
      "loss": 0.5039,
      "step": 3960
    },
    {
      "epoch": 0.5487973458667403,
      "grad_norm": 0.41375941038131714,
      "learning_rate": 2e-05,
      "loss": 0.484,
      "step": 3970
    },
    {
      "epoch": 0.5501797069394526,
      "grad_norm": 0.39458703994750977,
      "learning_rate": 2e-05,
      "loss": 0.4777,
      "step": 3980
    },
    {
      "epoch": 0.5515620680121648,
      "grad_norm": 0.36154913902282715,
      "learning_rate": 2e-05,
      "loss": 0.5077,
      "step": 3990
    },
    {
      "epoch": 0.552944429084877,
      "grad_norm": 0.35634762048721313,
      "learning_rate": 2e-05,
      "loss": 0.5032,
      "step": 4000
    },
    {
      "epoch": 0.5543267901575891,
      "grad_norm": 0.40761202573776245,
      "learning_rate": 2e-05,
      "loss": 0.4792,
      "step": 4010
    },
    {
      "epoch": 0.5557091512303014,
      "grad_norm": 0.39307764172554016,
      "learning_rate": 2e-05,
      "loss": 0.4788,
      "step": 4020
    },
    {
      "epoch": 0.5570915123030136,
      "grad_norm": 0.356644868850708,
      "learning_rate": 2e-05,
      "loss": 0.4875,
      "step": 4030
    },
    {
      "epoch": 0.5584738733757257,
      "grad_norm": 0.35265010595321655,
      "learning_rate": 2e-05,
      "loss": 0.51,
      "step": 4040
    },
    {
      "epoch": 0.5598562344484379,
      "grad_norm": 0.3889898657798767,
      "learning_rate": 2e-05,
      "loss": 0.5033,
      "step": 4050
    },
    {
      "epoch": 0.5612385955211501,
      "grad_norm": 0.3594723343849182,
      "learning_rate": 2e-05,
      "loss": 0.4873,
      "step": 4060
    },
    {
      "epoch": 0.5626209565938624,
      "grad_norm": 0.3885984420776367,
      "learning_rate": 2e-05,
      "loss": 0.505,
      "step": 4070
    },
    {
      "epoch": 0.5640033176665745,
      "grad_norm": 0.3806542754173279,
      "learning_rate": 2e-05,
      "loss": 0.5084,
      "step": 4080
    },
    {
      "epoch": 0.5653856787392867,
      "grad_norm": 0.360491007566452,
      "learning_rate": 2e-05,
      "loss": 0.4819,
      "step": 4090
    },
    {
      "epoch": 0.5667680398119989,
      "grad_norm": 0.5734226107597351,
      "learning_rate": 2e-05,
      "loss": 0.4883,
      "step": 4100
    },
    {
      "epoch": 0.5681504008847111,
      "grad_norm": 0.3655814826488495,
      "learning_rate": 2e-05,
      "loss": 0.4814,
      "step": 4110
    },
    {
      "epoch": 0.5695327619574233,
      "grad_norm": 0.32986176013946533,
      "learning_rate": 2e-05,
      "loss": 0.4839,
      "step": 4120
    },
    {
      "epoch": 0.5709151230301355,
      "grad_norm": 0.36655423045158386,
      "learning_rate": 2e-05,
      "loss": 0.4956,
      "step": 4130
    },
    {
      "epoch": 0.5722974841028476,
      "grad_norm": 0.3643620014190674,
      "learning_rate": 2e-05,
      "loss": 0.4654,
      "step": 4140
    },
    {
      "epoch": 0.5736798451755598,
      "grad_norm": 0.3567427694797516,
      "learning_rate": 2e-05,
      "loss": 0.4926,
      "step": 4150
    },
    {
      "epoch": 0.5750622062482721,
      "grad_norm": 0.36955100297927856,
      "learning_rate": 2e-05,
      "loss": 0.4841,
      "step": 4160
    },
    {
      "epoch": 0.5764445673209843,
      "grad_norm": 0.378263384103775,
      "learning_rate": 2e-05,
      "loss": 0.4894,
      "step": 4170
    },
    {
      "epoch": 0.5778269283936964,
      "grad_norm": 0.3758687376976013,
      "learning_rate": 2e-05,
      "loss": 0.4971,
      "step": 4180
    },
    {
      "epoch": 0.5792092894664086,
      "grad_norm": 0.3725149631500244,
      "learning_rate": 2e-05,
      "loss": 0.4948,
      "step": 4190
    },
    {
      "epoch": 0.5805916505391209,
      "grad_norm": 0.40945640206336975,
      "learning_rate": 2e-05,
      "loss": 0.4933,
      "step": 4200
    },
    {
      "epoch": 0.581974011611833,
      "grad_norm": 0.3702208697795868,
      "learning_rate": 2e-05,
      "loss": 0.4891,
      "step": 4210
    },
    {
      "epoch": 0.5833563726845452,
      "grad_norm": 0.3664851784706116,
      "learning_rate": 2e-05,
      "loss": 0.4939,
      "step": 4220
    },
    {
      "epoch": 0.5847387337572574,
      "grad_norm": 0.3586764931678772,
      "learning_rate": 2e-05,
      "loss": 0.4932,
      "step": 4230
    },
    {
      "epoch": 0.5861210948299695,
      "grad_norm": 0.38907358050346375,
      "learning_rate": 2e-05,
      "loss": 0.4742,
      "step": 4240
    },
    {
      "epoch": 0.5875034559026818,
      "grad_norm": 0.3497750759124756,
      "learning_rate": 2e-05,
      "loss": 0.4815,
      "step": 4250
    },
    {
      "epoch": 0.588885816975394,
      "grad_norm": 0.3587239980697632,
      "learning_rate": 2e-05,
      "loss": 0.4923,
      "step": 4260
    },
    {
      "epoch": 0.5902681780481062,
      "grad_norm": 0.38422808051109314,
      "learning_rate": 2e-05,
      "loss": 0.463,
      "step": 4270
    },
    {
      "epoch": 0.5916505391208183,
      "grad_norm": 0.3566284477710724,
      "learning_rate": 2e-05,
      "loss": 0.4955,
      "step": 4280
    },
    {
      "epoch": 0.5930329001935305,
      "grad_norm": 0.3858462870121002,
      "learning_rate": 2e-05,
      "loss": 0.4798,
      "step": 4290
    },
    {
      "epoch": 0.5944152612662428,
      "grad_norm": 0.34307661652565,
      "learning_rate": 2e-05,
      "loss": 0.4798,
      "step": 4300
    },
    {
      "epoch": 0.595797622338955,
      "grad_norm": 0.37319737672805786,
      "learning_rate": 2e-05,
      "loss": 0.4848,
      "step": 4310
    },
    {
      "epoch": 0.5971799834116671,
      "grad_norm": 0.4352557361125946,
      "learning_rate": 2e-05,
      "loss": 0.4938,
      "step": 4320
    },
    {
      "epoch": 0.5985623444843793,
      "grad_norm": 0.35341876745224,
      "learning_rate": 2e-05,
      "loss": 0.4912,
      "step": 4330
    },
    {
      "epoch": 0.5999447055570916,
      "grad_norm": 0.3408118188381195,
      "learning_rate": 2e-05,
      "loss": 0.4926,
      "step": 4340
    },
    {
      "epoch": 0.6013270666298037,
      "grad_norm": 0.3943605422973633,
      "learning_rate": 2e-05,
      "loss": 0.4714,
      "step": 4350
    },
    {
      "epoch": 0.6027094277025159,
      "grad_norm": 0.4143805503845215,
      "learning_rate": 2e-05,
      "loss": 0.4909,
      "step": 4360
    },
    {
      "epoch": 0.6040917887752281,
      "grad_norm": 0.3695594370365143,
      "learning_rate": 2e-05,
      "loss": 0.4858,
      "step": 4370
    },
    {
      "epoch": 0.6054741498479402,
      "grad_norm": 0.37762024998664856,
      "learning_rate": 2e-05,
      "loss": 0.4853,
      "step": 4380
    },
    {
      "epoch": 0.6068565109206525,
      "grad_norm": 0.38498401641845703,
      "learning_rate": 1.9984835874829866e-05,
      "loss": 0.4963,
      "step": 4390
    },
    {
      "epoch": 0.6082388719933647,
      "grad_norm": 0.3566693663597107,
      "learning_rate": 1.993248227822551e-05,
      "loss": 0.4887,
      "step": 4400
    },
    {
      "epoch": 0.6096212330660769,
      "grad_norm": 0.36355316638946533,
      "learning_rate": 1.984296973671751e-05,
      "loss": 0.4943,
      "step": 4410
    },
    {
      "epoch": 0.611003594138789,
      "grad_norm": 0.365328311920166,
      "learning_rate": 1.9716670620660985e-05,
      "loss": 0.4772,
      "step": 4420
    },
    {
      "epoch": 0.6123859552115013,
      "grad_norm": 0.3775816559791565,
      "learning_rate": 1.9554110331814234e-05,
      "loss": 0.4897,
      "step": 4430
    },
    {
      "epoch": 0.6137683162842135,
      "grad_norm": 0.35593676567077637,
      "learning_rate": 1.9355965117678084e-05,
      "loss": 0.4886,
      "step": 4440
    },
    {
      "epoch": 0.6151506773569256,
      "grad_norm": 0.3660953640937805,
      "learning_rate": 1.912305925832e-05,
      "loss": 0.4941,
      "step": 4450
    },
    {
      "epoch": 0.6165330384296378,
      "grad_norm": 0.36347121000289917,
      "learning_rate": 1.885636163738577e-05,
      "loss": 0.4982,
      "step": 4460
    },
    {
      "epoch": 0.61791539950235,
      "grad_norm": 0.3549681305885315,
      "learning_rate": 1.8556981711563194e-05,
      "loss": 0.4799,
      "step": 4470
    },
    {
      "epoch": 0.6192977605750623,
      "grad_norm": 0.3446556329727173,
      "learning_rate": 1.822616489526479e-05,
      "loss": 0.4952,
      "step": 4480
    },
    {
      "epoch": 0.6206801216477744,
      "grad_norm": 0.3710188865661621,
      "learning_rate": 1.7865287379729137e-05,
      "loss": 0.4945,
      "step": 4490
    },
    {
      "epoch": 0.6220624827204866,
      "grad_norm": 0.6089836955070496,
      "learning_rate": 1.7475850408093257e-05,
      "loss": 0.4886,
      "step": 4500
    },
    {
      "epoch": 0.6234448437931988,
      "grad_norm": 0.36110493540763855,
      "learning_rate": 1.705947403025165e-05,
      "loss": 0.4989,
      "step": 4510
    },
    {
      "epoch": 0.6248272048659109,
      "grad_norm": 0.3587700128555298,
      "learning_rate": 1.6617890363481623e-05,
      "loss": 0.4797,
      "step": 4520
    },
    {
      "epoch": 0.6262095659386232,
      "grad_norm": 0.365161657333374,
      "learning_rate": 1.6152936386870604e-05,
      "loss": 0.4972,
      "step": 4530
    },
    {
      "epoch": 0.6275919270113354,
      "grad_norm": 0.3511389195919037,
      "learning_rate": 1.5666546299520414e-05,
      "loss": 0.4979,
      "step": 4540
    },
    {
      "epoch": 0.6289742880840475,
      "grad_norm": 0.3604752719402313,
      "learning_rate": 1.5160743474318298e-05,
      "loss": 0.4872,
      "step": 4550
    },
    {
      "epoch": 0.6303566491567597,
      "grad_norm": 0.4664958417415619,
      "learning_rate": 1.463763204074678e-05,
      "loss": 0.4872,
      "step": 4560
    },
    {
      "epoch": 0.631739010229472,
      "grad_norm": 0.35334518551826477,
      "learning_rate": 1.4099388131747764e-05,
      "loss": 0.4855,
      "step": 4570
    },
    {
      "epoch": 0.6331213713021842,
      "grad_norm": 0.37638890743255615,
      "learning_rate": 1.3548250831053656e-05,
      "loss": 0.4993,
      "step": 4580
    },
    {
      "epoch": 0.6345037323748963,
      "grad_norm": 0.3355979919433594,
      "learning_rate": 1.2986512858644518e-05,
      "loss": 0.4725,
      "step": 4590
    },
    {
      "epoch": 0.6358860934476085,
      "grad_norm": 0.3452688455581665,
      "learning_rate": 1.2416511033079498e-05,
      "loss": 0.4969,
      "step": 4600
    },
    {
      "epoch": 0.6372684545203207,
      "grad_norm": 0.3958473205566406,
      "learning_rate": 1.1840616550379082e-05,
      "loss": 0.477,
      "step": 4610
    },
    {
      "epoch": 0.6386508155930329,
      "grad_norm": 0.423064261674881,
      "learning_rate": 1.1261225119897781e-05,
      "loss": 0.4751,
      "step": 4620
    },
    {
      "epoch": 0.6400331766657451,
      "grad_norm": 0.3663679361343384,
      "learning_rate": 1.068074699822187e-05,
      "loss": 0.4924,
      "step": 4630
    },
    {
      "epoch": 0.6414155377384573,
      "grad_norm": 0.3992937505245209,
      "learning_rate": 1.0101596962550937e-05,
      "loss": 0.4919,
      "step": 4640
    },
    {
      "epoch": 0.6427978988111694,
      "grad_norm": 0.3676553964614868,
      "learning_rate": 9.526184265273748e-06,
      "loss": 0.4872,
      "step": 4650
    },
    {
      "epoch": 0.6441802598838817,
      "grad_norm": 0.36466097831726074,
      "learning_rate": 8.956902611527237e-06,
      "loss": 0.4775,
      "step": 4660
    },
    {
      "epoch": 0.6455626209565939,
      "grad_norm": 0.3609284460544586,
      "learning_rate": 8.396120201431742e-06,
      "loss": 0.4788,
      "step": 4670
    },
    {
      "epoch": 0.6469449820293061,
      "grad_norm": 0.341800719499588,
      "learning_rate": 7.846169878426593e-06,
      "loss": 0.4913,
      "step": 4680
    },
    {
      "epoch": 0.6483273431020182,
      "grad_norm": 0.3842666447162628,
      "learning_rate": 7.3093394246887686e-06,
      "loss": 0.4739,
      "step": 4690
    },
    {
      "epoch": 0.6497097041747304,
      "grad_norm": 0.37957844138145447,
      "learning_rate": 6.787862044005491e-06,
      "loss": 0.4873,
      "step": 4700
    },
    {
      "epoch": 0.6510920652474427,
      "grad_norm": 0.3528563976287842,
      "learning_rate": 6.2839070716918235e-06,
      "loss": 0.4773,
      "step": 4710
    },
    {
      "epoch": 0.6524744263201548,
      "grad_norm": 0.3529062271118164,
      "learning_rate": 5.799570950199793e-06,
      "loss": 0.4879,
      "step": 4720
    },
    {
      "epoch": 0.653856787392867,
      "grad_norm": 0.3497636020183563,
      "learning_rate": 5.336868507960386e-06,
      "loss": 0.4789,
      "step": 4730
    },
    {
      "epoch": 0.6552391484655792,
      "grad_norm": 0.4196619391441345,
      "learning_rate": 4.897724577738157e-06,
      "loss": 0.4721,
      "step": 4740
    },
    {
      "epoch": 0.6566215095382913,
      "grad_norm": 0.38888242840766907,
      "learning_rate": 4.483965989366025e-06,
      "loss": 0.5121,
      "step": 4750
    },
    {
      "epoch": 0.6580038706110036,
      "grad_norm": 0.35125547647476196,
      "learning_rate": 4.097313970170263e-06,
      "loss": 0.4673,
      "step": 4760
    },
    {
      "epoch": 0.6593862316837158,
      "grad_norm": 0.3662374019622803,
      "learning_rate": 3.73937698469977e-06,
      "loss": 0.4719,
      "step": 4770
    },
    {
      "epoch": 0.660768592756428,
      "grad_norm": 0.34830427169799805,
      "learning_rate": 3.411644043546202e-06,
      "loss": 0.4725,
      "step": 4780
    },
    {
      "epoch": 0.6621509538291401,
      "grad_norm": 0.3560945391654968,
      "learning_rate": 3.1154785090901505e-06,
      "loss": 0.4858,
      "step": 4790
    },
    {
      "epoch": 0.6635333149018524,
      "grad_norm": 0.381537526845932,
      "learning_rate": 2.8521124239413213e-06,
      "loss": 0.4743,
      "step": 4800
    },
    {
      "epoch": 0.6649156759745646,
      "grad_norm": 0.3499518930912018,
      "learning_rate": 2.622641385666294e-06,
      "loss": 0.4698,
      "step": 4810
    },
    {
      "epoch": 0.6662980370472767,
      "grad_norm": 0.35871583223342896,
      "learning_rate": 2.428019989124892e-06,
      "loss": 0.474,
      "step": 4820
    },
    {
      "epoch": 0.6676803981199889,
      "grad_norm": 0.33109673857688904,
      "learning_rate": 2.269057855374944e-06,
      "loss": 0.4782,
      "step": 4830
    },
    {
      "epoch": 0.6690627591927011,
      "grad_norm": 0.3982751965522766,
      "learning_rate": 2.1464162636651277e-06,
      "loss": 0.4897,
      "step": 4840
    },
    {
      "epoch": 0.6704451202654134,
      "grad_norm": 0.3669286072254181,
      "learning_rate": 2.0606054005267244e-06,
      "loss": 0.4881,
      "step": 4850
    },
    {
      "epoch": 0.6718274813381255,
      "grad_norm": 0.36520859599113464,
      "learning_rate": 2.011982237408003e-06,
      "loss": 0.4514,
      "step": 4860
    }
  ],
  "logging_steps": 10,
  "max_steps": 4867,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2035,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.1726419832950227e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}