{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998234442495793,
  "eval_steps": 1000,
  "global_step": 5309,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0009416306689108865,
      "grad_norm": 13.400167465209961,
      "learning_rate": 1e-05,
      "loss": 4.0168,
      "step": 5
    },
    {
      "epoch": 0.001883261337821773,
      "grad_norm": 12.05469036102295,
      "learning_rate": 1e-05,
      "loss": 3.4972,
      "step": 10
    },
    {
      "epoch": 0.0028248920067326594,
      "grad_norm": 11.709904670715332,
      "learning_rate": 1e-05,
      "loss": 3.5447,
      "step": 15
    },
    {
      "epoch": 0.003766522675643546,
      "grad_norm": 10.864925384521484,
      "learning_rate": 1e-05,
      "loss": 3.4348,
      "step": 20
    },
    {
      "epoch": 0.004708153344554432,
      "grad_norm": 12.948342323303223,
      "learning_rate": 1e-05,
      "loss": 3.4288,
      "step": 25
    },
    {
      "epoch": 0.005649784013465319,
      "grad_norm": 10.90282917022705,
      "learning_rate": 1e-05,
      "loss": 3.4199,
      "step": 30
    },
    {
      "epoch": 0.006591414682376205,
      "grad_norm": 10.574660301208496,
      "learning_rate": 1e-05,
      "loss": 3.4564,
      "step": 35
    },
    {
      "epoch": 0.007533045351287092,
      "grad_norm": 12.737661361694336,
      "learning_rate": 1e-05,
      "loss": 3.4071,
      "step": 40
    },
    {
      "epoch": 0.008474676020197977,
      "grad_norm": 11.787413597106934,
      "learning_rate": 1e-05,
      "loss": 3.3169,
      "step": 45
    },
    {
      "epoch": 0.009416306689108865,
      "grad_norm": 11.487077713012695,
      "learning_rate": 1e-05,
      "loss": 3.4083,
      "step": 50
    },
    {
      "epoch": 0.01035793735801975,
      "grad_norm": 10.931989669799805,
      "learning_rate": 1e-05,
      "loss": 3.3384,
      "step": 55
    },
    {
      "epoch": 0.011299568026930638,
      "grad_norm": 11.073975563049316,
      "learning_rate": 1e-05,
      "loss": 3.2959,
      "step": 60
    },
    {
      "epoch": 0.012241198695841523,
      "grad_norm": 10.520795822143555,
      "learning_rate": 1e-05,
      "loss": 3.3513,
      "step": 65
    },
    {
      "epoch": 0.01318282936475241,
      "grad_norm": 10.708564758300781,
      "learning_rate": 1e-05,
      "loss": 3.1843,
      "step": 70
    },
    {
      "epoch": 0.014124460033663296,
      "grad_norm": 9.278353691101074,
      "learning_rate": 1e-05,
      "loss": 3.2907,
      "step": 75
    },
    {
      "epoch": 0.015066090702574184,
      "grad_norm": 9.356634140014648,
      "learning_rate": 1e-05,
      "loss": 3.3033,
      "step": 80
    },
    {
      "epoch": 0.01600772137148507,
      "grad_norm": 10.317462921142578,
      "learning_rate": 1e-05,
      "loss": 3.402,
      "step": 85
    },
    {
      "epoch": 0.016949352040395955,
      "grad_norm": 9.032744407653809,
      "learning_rate": 1e-05,
      "loss": 3.3137,
      "step": 90
    },
    {
      "epoch": 0.017890982709306842,
      "grad_norm": 9.985954284667969,
      "learning_rate": 1e-05,
      "loss": 3.3074,
      "step": 95
    },
    {
      "epoch": 0.01883261337821773,
      "grad_norm": 9.465397834777832,
      "learning_rate": 1e-05,
      "loss": 3.3329,
      "step": 100
    },
    {
      "epoch": 0.019774244047128613,
      "grad_norm": 9.615800857543945,
      "learning_rate": 1e-05,
      "loss": 3.3839,
      "step": 105
    },
    {
      "epoch": 0.0207158747160395,
      "grad_norm": 9.558786392211914,
      "learning_rate": 1e-05,
      "loss": 3.3383,
      "step": 110
    },
    {
      "epoch": 0.021657505384950388,
      "grad_norm": 10.330078125,
      "learning_rate": 1e-05,
      "loss": 3.2845,
      "step": 115
    },
    {
      "epoch": 0.022599136053861275,
      "grad_norm": 9.462120056152344,
      "learning_rate": 1e-05,
      "loss": 3.2918,
      "step": 120
    },
    {
      "epoch": 0.02354076672277216,
      "grad_norm": 10.415687561035156,
      "learning_rate": 1e-05,
      "loss": 3.3509,
      "step": 125
    },
    {
      "epoch": 0.024482397391683047,
      "grad_norm": 9.891234397888184,
      "learning_rate": 1e-05,
      "loss": 3.3196,
      "step": 130
    },
    {
      "epoch": 0.025424028060593934,
      "grad_norm": 10.6549072265625,
      "learning_rate": 1e-05,
      "loss": 3.2702,
      "step": 135
    },
    {
      "epoch": 0.02636565872950482,
      "grad_norm": 10.514317512512207,
      "learning_rate": 1e-05,
      "loss": 3.3639,
      "step": 140
    },
    {
      "epoch": 0.027307289398415705,
      "grad_norm": 10.233137130737305,
      "learning_rate": 1e-05,
      "loss": 3.3693,
      "step": 145
    },
    {
      "epoch": 0.028248920067326592,
      "grad_norm": 9.689544677734375,
      "learning_rate": 1e-05,
      "loss": 3.2754,
      "step": 150
    },
    {
      "epoch": 0.02919055073623748,
      "grad_norm": 10.0300931930542,
      "learning_rate": 1e-05,
      "loss": 3.2058,
      "step": 155
    },
    {
      "epoch": 0.030132181405148367,
      "grad_norm": 9.759441375732422,
      "learning_rate": 1e-05,
      "loss": 3.216,
      "step": 160
    },
    {
      "epoch": 0.03107381207405925,
      "grad_norm": 9.909936904907227,
      "learning_rate": 1e-05,
      "loss": 3.3169,
      "step": 165
    },
    {
      "epoch": 0.03201544274297014,
      "grad_norm": 10.085428237915039,
      "learning_rate": 1e-05,
      "loss": 3.3579,
      "step": 170
    },
    {
      "epoch": 0.032957073411881026,
      "grad_norm": 11.790485382080078,
      "learning_rate": 1e-05,
      "loss": 3.3376,
      "step": 175
    },
    {
      "epoch": 0.03389870408079191,
      "grad_norm": 9.588286399841309,
      "learning_rate": 1e-05,
      "loss": 3.2441,
      "step": 180
    },
    {
      "epoch": 0.0348403347497028,
      "grad_norm": 9.073866844177246,
      "learning_rate": 1e-05,
      "loss": 3.3118,
      "step": 185
    },
    {
      "epoch": 0.035781965418613684,
      "grad_norm": 11.346445083618164,
      "learning_rate": 1e-05,
      "loss": 3.2795,
      "step": 190
    },
    {
      "epoch": 0.03672359608752457,
      "grad_norm": 8.872209548950195,
      "learning_rate": 1e-05,
      "loss": 3.1868,
      "step": 195
    },
    {
      "epoch": 0.03766522675643546,
      "grad_norm": 10.55789566040039,
      "learning_rate": 1e-05,
      "loss": 3.2716,
      "step": 200
    },
    {
      "epoch": 0.03860685742534634,
      "grad_norm": 9.99712085723877,
      "learning_rate": 1e-05,
      "loss": 3.2368,
      "step": 205
    },
    {
      "epoch": 0.03954848809425723,
      "grad_norm": 10.883671760559082,
      "learning_rate": 1e-05,
      "loss": 3.237,
      "step": 210
    },
    {
      "epoch": 0.04049011876316812,
      "grad_norm": 10.810287475585938,
      "learning_rate": 1e-05,
      "loss": 3.2724,
      "step": 215
    },
    {
      "epoch": 0.041431749432079,
      "grad_norm": 8.746783256530762,
      "learning_rate": 1e-05,
      "loss": 3.2741,
      "step": 220
    },
    {
      "epoch": 0.04237338010098989,
      "grad_norm": 10.720795631408691,
      "learning_rate": 1e-05,
      "loss": 3.2725,
      "step": 225
    },
    {
      "epoch": 0.043315010769900776,
      "grad_norm": 10.715919494628906,
      "learning_rate": 1e-05,
      "loss": 3.2604,
      "step": 230
    },
    {
      "epoch": 0.04425664143881166,
      "grad_norm": 9.999095916748047,
      "learning_rate": 1e-05,
      "loss": 3.3685,
      "step": 235
    },
    {
      "epoch": 0.04519827210772255,
      "grad_norm": 9.06040096282959,
      "learning_rate": 1e-05,
      "loss": 3.3883,
      "step": 240
    },
    {
      "epoch": 0.046139902776633435,
      "grad_norm": 9.397573471069336,
      "learning_rate": 1e-05,
      "loss": 3.2377,
      "step": 245
    },
    {
      "epoch": 0.04708153344554432,
      "grad_norm": 11.011545181274414,
      "learning_rate": 1e-05,
      "loss": 3.1805,
      "step": 250
    },
    {
      "epoch": 0.04802316411445521,
      "grad_norm": 8.947606086730957,
      "learning_rate": 1e-05,
      "loss": 3.1641,
      "step": 255
    },
    {
      "epoch": 0.04896479478336609,
      "grad_norm": 9.644259452819824,
      "learning_rate": 1e-05,
      "loss": 3.2127,
      "step": 260
    },
    {
      "epoch": 0.049906425452276984,
      "grad_norm": 9.601791381835938,
      "learning_rate": 1e-05,
      "loss": 3.2503,
      "step": 265
    },
    {
      "epoch": 0.05084805612118787,
      "grad_norm": 9.706978797912598,
      "learning_rate": 1e-05,
      "loss": 3.2276,
      "step": 270
    },
    {
      "epoch": 0.05178968679009875,
      "grad_norm": 10.092705726623535,
      "learning_rate": 1e-05,
      "loss": 3.2853,
      "step": 275
    },
    {
      "epoch": 0.05273131745900964,
      "grad_norm": 10.306342124938965,
      "learning_rate": 1e-05,
      "loss": 3.3613,
      "step": 280
    },
    {
      "epoch": 0.053672948127920526,
      "grad_norm": 9.216752052307129,
      "learning_rate": 1e-05,
      "loss": 3.1688,
      "step": 285
    },
    {
      "epoch": 0.05461457879683141,
      "grad_norm": 9.452301979064941,
      "learning_rate": 1e-05,
      "loss": 3.2417,
      "step": 290
    },
    {
      "epoch": 0.0555562094657423,
      "grad_norm": 9.074745178222656,
      "learning_rate": 1e-05,
      "loss": 3.3112,
      "step": 295
    },
    {
      "epoch": 0.056497840134653185,
      "grad_norm": 8.492777824401855,
      "learning_rate": 1e-05,
      "loss": 3.2811,
      "step": 300
    },
    {
      "epoch": 0.05743947080356407,
      "grad_norm": 9.102340698242188,
      "learning_rate": 1e-05,
      "loss": 3.2543,
      "step": 305
    },
    {
      "epoch": 0.05838110147247496,
      "grad_norm": 9.913708686828613,
      "learning_rate": 1e-05,
      "loss": 3.3047,
      "step": 310
    },
    {
      "epoch": 0.059322732141385844,
      "grad_norm": 8.82358455657959,
      "learning_rate": 1e-05,
      "loss": 3.2744,
      "step": 315
    },
    {
      "epoch": 0.060264362810296734,
      "grad_norm": 10.01116943359375,
      "learning_rate": 1e-05,
      "loss": 3.3184,
      "step": 320
    },
    {
      "epoch": 0.06120599347920762,
      "grad_norm": 8.861343383789062,
      "learning_rate": 1e-05,
      "loss": 3.3169,
      "step": 325
    },
    {
      "epoch": 0.0621476241481185,
      "grad_norm": 8.552498817443848,
      "learning_rate": 1e-05,
      "loss": 3.3267,
      "step": 330
    },
    {
      "epoch": 0.06308925481702939,
      "grad_norm": 9.180069923400879,
      "learning_rate": 1e-05,
      "loss": 3.1593,
      "step": 335
    },
    {
      "epoch": 0.06403088548594028,
      "grad_norm": 8.194535255432129,
      "learning_rate": 1e-05,
      "loss": 3.252,
      "step": 340
    },
    {
      "epoch": 0.06497251615485117,
      "grad_norm": 9.234159469604492,
      "learning_rate": 1e-05,
      "loss": 3.3233,
      "step": 345
    },
    {
      "epoch": 0.06591414682376205,
      "grad_norm": 9.615317344665527,
      "learning_rate": 1e-05,
      "loss": 3.1951,
      "step": 350
    },
    {
      "epoch": 0.06685577749267294,
      "grad_norm": 9.555344581604004,
      "learning_rate": 1e-05,
      "loss": 3.3024,
      "step": 355
    },
    {
      "epoch": 0.06779740816158382,
      "grad_norm": 9.033773422241211,
      "learning_rate": 1e-05,
      "loss": 3.2833,
      "step": 360
    },
    {
      "epoch": 0.0687390388304947,
      "grad_norm": 10.110182762145996,
      "learning_rate": 1e-05,
      "loss": 3.3022,
      "step": 365
    },
    {
      "epoch": 0.0696806694994056,
      "grad_norm": 8.350312232971191,
      "learning_rate": 1e-05,
      "loss": 3.1566,
      "step": 370
    },
    {
      "epoch": 0.07062230016831648,
      "grad_norm": 9.642204284667969,
      "learning_rate": 1e-05,
      "loss": 3.2279,
      "step": 375
    },
    {
      "epoch": 0.07156393083722737,
      "grad_norm": 8.922101020812988,
      "learning_rate": 1e-05,
      "loss": 3.1227,
      "step": 380
    },
    {
      "epoch": 0.07250556150613825,
      "grad_norm": 8.6968412399292,
      "learning_rate": 1e-05,
      "loss": 3.2537,
      "step": 385
    },
    {
      "epoch": 0.07344719217504914,
      "grad_norm": 9.628043174743652,
      "learning_rate": 1e-05,
      "loss": 3.335,
      "step": 390
    },
    {
      "epoch": 0.07438882284396003,
      "grad_norm": 9.418974876403809,
      "learning_rate": 1e-05,
      "loss": 3.2609,
      "step": 395
    },
    {
      "epoch": 0.07533045351287092,
      "grad_norm": 9.786445617675781,
      "learning_rate": 1e-05,
      "loss": 3.1916,
      "step": 400
    },
    {
      "epoch": 0.0762720841817818,
      "grad_norm": 9.372411727905273,
      "learning_rate": 1e-05,
      "loss": 3.2404,
      "step": 405
    },
    {
      "epoch": 0.07721371485069269,
      "grad_norm": 9.255708694458008,
      "learning_rate": 1e-05,
      "loss": 3.1984,
      "step": 410
    },
    {
      "epoch": 0.07815534551960357,
      "grad_norm": 8.31712818145752,
      "learning_rate": 1e-05,
      "loss": 3.2435,
      "step": 415
    },
    {
      "epoch": 0.07909697618851445,
      "grad_norm": 9.45557975769043,
      "learning_rate": 1e-05,
      "loss": 3.271,
      "step": 420
    },
    {
      "epoch": 0.08003860685742535,
      "grad_norm": 9.2068510055542,
      "learning_rate": 1e-05,
      "loss": 3.2629,
      "step": 425
    },
    {
      "epoch": 0.08098023752633624,
      "grad_norm": 9.854654312133789,
      "learning_rate": 1e-05,
      "loss": 3.2466,
      "step": 430
    },
    {
      "epoch": 0.08192186819524712,
      "grad_norm": 9.899443626403809,
      "learning_rate": 1e-05,
      "loss": 3.303,
      "step": 435
    },
    {
      "epoch": 0.082863498864158,
      "grad_norm": 8.167075157165527,
      "learning_rate": 1e-05,
      "loss": 3.1332,
      "step": 440
    },
    {
      "epoch": 0.08380512953306889,
      "grad_norm": 9.357007026672363,
      "learning_rate": 1e-05,
      "loss": 3.2161,
      "step": 445
    },
    {
      "epoch": 0.08474676020197978,
      "grad_norm": 8.328939437866211,
      "learning_rate": 1e-05,
      "loss": 3.2558,
      "step": 450
    },
    {
      "epoch": 0.08568839087089067,
      "grad_norm": 10.344287872314453,
      "learning_rate": 1e-05,
      "loss": 3.1148,
      "step": 455
    },
    {
      "epoch": 0.08663002153980155,
      "grad_norm": 8.459121704101562,
      "learning_rate": 1e-05,
      "loss": 3.2366,
      "step": 460
    },
    {
      "epoch": 0.08757165220871244,
      "grad_norm": 9.339311599731445,
      "learning_rate": 1e-05,
      "loss": 3.3596,
      "step": 465
    },
    {
      "epoch": 0.08851328287762332,
      "grad_norm": 9.49207592010498,
      "learning_rate": 1e-05,
      "loss": 3.2823,
      "step": 470
    },
    {
      "epoch": 0.0894549135465342,
      "grad_norm": 9.24524974822998,
      "learning_rate": 1e-05,
      "loss": 3.2593,
      "step": 475
    },
    {
      "epoch": 0.0903965442154451,
      "grad_norm": 9.082175254821777,
      "learning_rate": 1e-05,
      "loss": 3.1761,
      "step": 480
    },
    {
      "epoch": 0.09133817488435599,
      "grad_norm": 8.596846580505371,
      "learning_rate": 1e-05,
      "loss": 3.2005,
      "step": 485
    },
    {
      "epoch": 0.09227980555326687,
      "grad_norm": 9.297995567321777,
      "learning_rate": 1e-05,
      "loss": 3.2414,
      "step": 490
    },
    {
      "epoch": 0.09322143622217775,
      "grad_norm": 9.254470825195312,
      "learning_rate": 1e-05,
      "loss": 3.2309,
      "step": 495
    },
    {
      "epoch": 0.09416306689108864,
      "grad_norm": 9.504804611206055,
      "learning_rate": 1e-05,
      "loss": 3.2978,
      "step": 500
    },
    {
      "epoch": 0.09510469755999953,
      "grad_norm": 9.160736083984375,
      "learning_rate": 1e-05,
      "loss": 3.2522,
      "step": 505
    },
    {
      "epoch": 0.09604632822891042,
      "grad_norm": 8.80601692199707,
      "learning_rate": 1e-05,
      "loss": 3.1128,
      "step": 510
    },
    {
      "epoch": 0.0969879588978213,
      "grad_norm": 9.246788024902344,
      "learning_rate": 1e-05,
      "loss": 3.2469,
      "step": 515
    },
    {
      "epoch": 0.09792958956673219,
      "grad_norm": 8.799399375915527,
      "learning_rate": 1e-05,
      "loss": 3.1663,
      "step": 520
    },
    {
      "epoch": 0.09887122023564307,
      "grad_norm": 8.722356796264648,
      "learning_rate": 1e-05,
      "loss": 3.2616,
      "step": 525
    },
    {
      "epoch": 0.09981285090455397,
      "grad_norm": 9.724103927612305,
      "learning_rate": 1e-05,
      "loss": 3.2367,
      "step": 530
    },
    {
      "epoch": 0.10075448157346485,
      "grad_norm": 9.14875602722168,
      "learning_rate": 1e-05,
      "loss": 3.1932,
      "step": 535
    },
    {
      "epoch": 0.10169611224237574,
      "grad_norm": 8.882226943969727,
      "learning_rate": 1e-05,
      "loss": 3.1863,
      "step": 540
    },
    {
      "epoch": 0.10263774291128662,
      "grad_norm": 8.96247673034668,
      "learning_rate": 1e-05,
      "loss": 3.3411,
      "step": 545
    },
    {
      "epoch": 0.1035793735801975,
      "grad_norm": 7.7259202003479,
      "learning_rate": 1e-05,
      "loss": 3.2209,
      "step": 550
    },
    {
      "epoch": 0.10452100424910839,
      "grad_norm": 8.8307466506958,
      "learning_rate": 1e-05,
      "loss": 3.1653,
      "step": 555
    },
    {
      "epoch": 0.10546263491801929,
      "grad_norm": 9.439279556274414,
      "learning_rate": 1e-05,
      "loss": 3.2451,
      "step": 560
    },
    {
      "epoch": 0.10640426558693017,
      "grad_norm": 8.951433181762695,
      "learning_rate": 1e-05,
      "loss": 3.2327,
      "step": 565
    },
    {
      "epoch": 0.10734589625584105,
      "grad_norm": 8.996528625488281,
      "learning_rate": 1e-05,
      "loss": 3.177,
      "step": 570
    },
    {
      "epoch": 0.10828752692475194,
      "grad_norm": 8.661666870117188,
      "learning_rate": 1e-05,
      "loss": 3.2389,
      "step": 575
    },
    {
      "epoch": 0.10922915759366282,
      "grad_norm": 7.827404499053955,
      "learning_rate": 1e-05,
      "loss": 3.2133,
      "step": 580
    },
    {
      "epoch": 0.11017078826257372,
      "grad_norm": 8.948150634765625,
      "learning_rate": 1e-05,
      "loss": 3.1059,
      "step": 585
    },
    {
      "epoch": 0.1111124189314846,
      "grad_norm": 9.571684837341309,
      "learning_rate": 1e-05,
      "loss": 3.2737,
      "step": 590
    },
    {
      "epoch": 0.11205404960039549,
      "grad_norm": 8.415332794189453,
      "learning_rate": 1e-05,
      "loss": 3.181,
      "step": 595
    },
    {
      "epoch": 0.11299568026930637,
      "grad_norm": 8.199889183044434,
      "learning_rate": 1e-05,
      "loss": 3.1989,
      "step": 600
    },
    {
      "epoch": 0.11393731093821725,
      "grad_norm": 8.935426712036133,
      "learning_rate": 1e-05,
      "loss": 3.165,
      "step": 605
    },
    {
      "epoch": 0.11487894160712814,
      "grad_norm": 9.254772186279297,
      "learning_rate": 1e-05,
      "loss": 3.2167,
      "step": 610
    },
    {
      "epoch": 0.11582057227603904,
      "grad_norm": 8.660184860229492,
      "learning_rate": 1e-05,
      "loss": 3.1573,
      "step": 615
    },
    {
      "epoch": 0.11676220294494992,
      "grad_norm": 9.153916358947754,
      "learning_rate": 1e-05,
      "loss": 3.2069,
      "step": 620
    },
    {
      "epoch": 0.1177038336138608,
      "grad_norm": 10.136688232421875,
      "learning_rate": 1e-05,
      "loss": 3.184,
      "step": 625
    },
    {
      "epoch": 0.11864546428277169,
      "grad_norm": 8.210442543029785,
      "learning_rate": 1e-05,
      "loss": 3.2377,
      "step": 630
    },
    {
      "epoch": 0.11958709495168257,
      "grad_norm": 10.24669361114502,
      "learning_rate": 1e-05,
      "loss": 3.2338,
      "step": 635
    },
    {
      "epoch": 0.12052872562059347,
      "grad_norm": 8.631328582763672,
      "learning_rate": 1e-05,
      "loss": 3.3123,
      "step": 640
    },
    {
      "epoch": 0.12147035628950435,
      "grad_norm": 8.293212890625,
      "learning_rate": 1e-05,
      "loss": 3.1506,
      "step": 645
    },
    {
      "epoch": 0.12241198695841524,
      "grad_norm": 9.084228515625,
      "learning_rate": 1e-05,
      "loss": 3.1423,
      "step": 650
    },
    {
      "epoch": 0.12335361762732612,
      "grad_norm": 7.838293552398682,
      "learning_rate": 1e-05,
      "loss": 3.0779,
      "step": 655
    },
    {
      "epoch": 0.124295248296237,
      "grad_norm": 10.947461128234863,
      "learning_rate": 1e-05,
      "loss": 3.1623,
      "step": 660
    },
    {
      "epoch": 0.1252368789651479,
      "grad_norm": 9.293968200683594,
      "learning_rate": 1e-05,
      "loss": 3.2047,
      "step": 665
    },
    {
      "epoch": 0.12617850963405877,
      "grad_norm": 8.150403022766113,
      "learning_rate": 1e-05,
      "loss": 3.1688,
      "step": 670
    },
    {
      "epoch": 0.12712014030296967,
      "grad_norm": 8.54336929321289,
      "learning_rate": 1e-05,
      "loss": 3.2182,
      "step": 675
    },
    {
      "epoch": 0.12806177097188057,
      "grad_norm": 7.81638240814209,
      "learning_rate": 1e-05,
      "loss": 3.2553,
      "step": 680
    },
    {
      "epoch": 0.12900340164079144,
      "grad_norm": 8.324007034301758,
      "learning_rate": 1e-05,
      "loss": 3.1702,
      "step": 685
    },
    {
      "epoch": 0.12994503230970234,
      "grad_norm": 9.187409400939941,
      "learning_rate": 1e-05,
      "loss": 3.1752,
      "step": 690
    },
    {
      "epoch": 0.1308866629786132,
      "grad_norm": 8.79953670501709,
      "learning_rate": 1e-05,
      "loss": 3.2211,
      "step": 695
    },
    {
      "epoch": 0.1318282936475241,
      "grad_norm": 9.9981689453125,
      "learning_rate": 1e-05,
      "loss": 3.1846,
      "step": 700
    },
    {
      "epoch": 0.13276992431643497,
      "grad_norm": 8.74207592010498,
      "learning_rate": 1e-05,
      "loss": 3.2351,
      "step": 705
    },
    {
      "epoch": 0.13371155498534587,
      "grad_norm": 8.988236427307129,
      "learning_rate": 1e-05,
      "loss": 3.2312,
      "step": 710
    },
    {
      "epoch": 0.13465318565425677,
      "grad_norm": 8.005229949951172,
      "learning_rate": 1e-05,
      "loss": 3.218,
      "step": 715
    },
    {
      "epoch": 0.13559481632316764,
      "grad_norm": 9.276175498962402,
      "learning_rate": 1e-05,
      "loss": 3.279,
      "step": 720
    },
    {
      "epoch": 0.13653644699207854,
      "grad_norm": 8.09757137298584,
      "learning_rate": 1e-05,
      "loss": 3.2876,
      "step": 725
    },
    {
      "epoch": 0.1374780776609894,
      "grad_norm": 8.40072250366211,
      "learning_rate": 1e-05,
      "loss": 3.2547,
      "step": 730
    },
    {
      "epoch": 0.1384197083299003,
      "grad_norm": 8.614055633544922,
      "learning_rate": 1e-05,
      "loss": 3.2795,
      "step": 735
    },
    {
      "epoch": 0.1393613389988112,
      "grad_norm": 9.18989086151123,
      "learning_rate": 1e-05,
      "loss": 3.0988,
      "step": 740
    },
    {
      "epoch": 0.14030296966772207,
      "grad_norm": 8.401784896850586,
      "learning_rate": 1e-05,
      "loss": 3.1612,
      "step": 745
    },
    {
      "epoch": 0.14124460033663297,
      "grad_norm": 8.722879409790039,
      "learning_rate": 1e-05,
      "loss": 3.1921,
      "step": 750
    },
    {
      "epoch": 0.14218623100554384,
      "grad_norm": 8.14240550994873,
      "learning_rate": 1e-05,
      "loss": 3.2371,
      "step": 755
    },
    {
      "epoch": 0.14312786167445474,
      "grad_norm": 10.812965393066406,
      "learning_rate": 1e-05,
      "loss": 3.2239,
      "step": 760
    },
    {
      "epoch": 0.14406949234336563,
      "grad_norm": 8.65410327911377,
      "learning_rate": 1e-05,
      "loss": 3.267,
      "step": 765
    },
    {
      "epoch": 0.1450111230122765,
      "grad_norm": 7.997138977050781,
      "learning_rate": 1e-05,
      "loss": 3.2387,
      "step": 770
    },
    {
      "epoch": 0.1459527536811874,
      "grad_norm": 8.726889610290527,
      "learning_rate": 1e-05,
      "loss": 3.1542,
      "step": 775
    },
    {
      "epoch": 0.14689438435009827,
      "grad_norm": 8.375578880310059,
      "learning_rate": 1e-05,
      "loss": 3.1742,
      "step": 780
    },
    {
      "epoch": 0.14783601501900917,
      "grad_norm": 8.577352523803711,
      "learning_rate": 1e-05,
      "loss": 3.2441,
      "step": 785
    },
    {
      "epoch": 0.14877764568792007,
      "grad_norm": 8.30477523803711,
      "learning_rate": 1e-05,
      "loss": 3.133,
      "step": 790
    },
    {
      "epoch": 0.14971927635683094,
      "grad_norm": 8.190656661987305,
      "learning_rate": 1e-05,
      "loss": 3.1822,
      "step": 795
    },
    {
      "epoch": 0.15066090702574184,
      "grad_norm": 8.631675720214844,
      "learning_rate": 1e-05,
      "loss": 3.2068,
      "step": 800
    },
    {
      "epoch": 0.1516025376946527,
      "grad_norm": 9.275361061096191,
      "learning_rate": 1e-05,
      "loss": 3.2194,
      "step": 805
    },
    {
      "epoch": 0.1525441683635636,
      "grad_norm": 8.464102745056152,
      "learning_rate": 1e-05,
      "loss": 3.1456,
      "step": 810
    },
    {
      "epoch": 0.1534857990324745,
      "grad_norm": 8.44454288482666,
      "learning_rate": 1e-05,
      "loss": 3.1675,
      "step": 815
    },
    {
      "epoch": 0.15442742970138537,
      "grad_norm": 8.282485961914062,
      "learning_rate": 1e-05,
      "loss": 3.1728,
      "step": 820
    },
    {
      "epoch": 0.15536906037029627,
      "grad_norm": 8.142829895019531,
      "learning_rate": 1e-05,
      "loss": 3.2012,
      "step": 825
    },
    {
      "epoch": 0.15631069103920714,
      "grad_norm": 8.29863452911377,
      "learning_rate": 1e-05,
      "loss": 3.1008,
      "step": 830
    },
    {
      "epoch": 0.15725232170811804,
      "grad_norm": 8.145419120788574,
      "learning_rate": 1e-05,
      "loss": 3.0711,
      "step": 835
    },
    {
      "epoch": 0.1581939523770289,
      "grad_norm": 8.715362548828125,
      "learning_rate": 1e-05,
      "loss": 3.1625,
      "step": 840
    },
    {
      "epoch": 0.1591355830459398,
      "grad_norm": 8.283182144165039,
      "learning_rate": 1e-05,
      "loss": 3.066,
      "step": 845
    },
    {
      "epoch": 0.1600772137148507,
      "grad_norm": 9.210651397705078,
      "learning_rate": 1e-05,
      "loss": 3.2498,
      "step": 850
    },
    {
      "epoch": 0.16101884438376157,
      "grad_norm": 8.161669731140137,
      "learning_rate": 1e-05,
      "loss": 3.2414,
      "step": 855
    },
    {
      "epoch": 0.16196047505267247,
      "grad_norm": 8.109793663024902,
      "learning_rate": 1e-05,
      "loss": 3.1307,
      "step": 860
    },
    {
      "epoch": 0.16290210572158334,
      "grad_norm": 8.473872184753418,
      "learning_rate": 1e-05,
      "loss": 3.1473,
      "step": 865
    },
    {
      "epoch": 0.16384373639049424,
      "grad_norm": 8.496044158935547,
      "learning_rate": 1e-05,
      "loss": 3.1971,
      "step": 870
    },
    {
      "epoch": 0.16478536705940514,
      "grad_norm": 7.473423004150391,
      "learning_rate": 1e-05,
      "loss": 3.1617,
      "step": 875
    },
    {
      "epoch": 0.165726997728316,
      "grad_norm": 7.770011901855469,
      "learning_rate": 1e-05,
      "loss": 3.1806,
      "step": 880
    },
    {
      "epoch": 0.1666686283972269,
      "grad_norm": 8.22207260131836,
      "learning_rate": 1e-05,
      "loss": 3.2673,
      "step": 885
    },
    {
      "epoch": 0.16761025906613777,
      "grad_norm": 9.465033531188965,
      "learning_rate": 1e-05,
      "loss": 3.273,
      "step": 890
    },
    {
      "epoch": 0.16855188973504867,
      "grad_norm": 8.1371431350708,
      "learning_rate": 1e-05,
      "loss": 3.1663,
      "step": 895
    },
    {
      "epoch": 0.16949352040395957,
      "grad_norm": 8.64779281616211,
      "learning_rate": 1e-05,
      "loss": 3.1979,
      "step": 900
    },
    {
      "epoch": 0.17043515107287044,
      "grad_norm": 8.860164642333984,
      "learning_rate": 1e-05,
      "loss": 3.2013,
      "step": 905
    },
    {
      "epoch": 0.17137678174178134,
      "grad_norm": 8.785137176513672,
      "learning_rate": 1e-05,
      "loss": 3.1721,
      "step": 910
    },
    {
      "epoch": 0.1723184124106922,
      "grad_norm": 8.557008743286133,
      "learning_rate": 1e-05,
      "loss": 3.1774,
      "step": 915
    },
    {
      "epoch": 0.1732600430796031,
      "grad_norm": 8.913068771362305,
      "learning_rate": 1e-05,
      "loss": 3.095,
      "step": 920
    },
    {
      "epoch": 0.174201673748514,
      "grad_norm": 7.609038352966309,
      "learning_rate": 1e-05,
      "loss": 3.1806,
      "step": 925
    },
    {
      "epoch": 0.17514330441742487,
      "grad_norm": 7.761782169342041,
      "learning_rate": 1e-05,
      "loss": 3.2352,
      "step": 930
    },
    {
      "epoch": 0.17608493508633577,
      "grad_norm": 8.180496215820312,
      "learning_rate": 1e-05,
      "loss": 3.2167,
      "step": 935
    },
    {
      "epoch": 0.17702656575524664,
      "grad_norm": 8.675491333007812,
      "learning_rate": 1e-05,
      "loss": 3.12,
      "step": 940
    },
    {
      "epoch": 0.17796819642415754,
      "grad_norm": 8.088050842285156,
      "learning_rate": 1e-05,
      "loss": 3.1027,
      "step": 945
    },
    {
      "epoch": 0.1789098270930684,
      "grad_norm": 8.452052116394043,
      "learning_rate": 1e-05,
      "loss": 2.9981,
      "step": 950
    },
    {
      "epoch": 0.1798514577619793,
      "grad_norm": 8.1648588180542,
      "learning_rate": 1e-05,
      "loss": 3.0539,
      "step": 955
    },
    {
      "epoch": 0.1807930884308902,
      "grad_norm": 8.245767593383789,
      "learning_rate": 1e-05,
      "loss": 3.1362,
      "step": 960
    },
    {
      "epoch": 0.18173471909980107,
      "grad_norm": 8.551701545715332,
      "learning_rate": 1e-05,
      "loss": 3.1821,
      "step": 965
    },
    {
      "epoch": 0.18267634976871197,
      "grad_norm": 8.178792953491211,
      "learning_rate": 1e-05,
      "loss": 3.1523,
      "step": 970
    },
    {
      "epoch": 0.18361798043762284,
      "grad_norm": 8.187594413757324,
      "learning_rate": 1e-05,
      "loss": 3.2531,
      "step": 975
    },
    {
      "epoch": 0.18455961110653374,
      "grad_norm": 7.719356060028076,
      "learning_rate": 1e-05,
      "loss": 3.1006,
      "step": 980
    },
    {
      "epoch": 0.18550124177544464,
      "grad_norm": 8.289334297180176,
      "learning_rate": 1e-05,
      "loss": 3.1259,
      "step": 985
    },
    {
      "epoch": 0.1864428724443555,
      "grad_norm": 8.719657897949219,
      "learning_rate": 1e-05,
      "loss": 3.1813,
      "step": 990
    },
    {
      "epoch": 0.1873845031132664,
      "grad_norm": 8.192325592041016,
      "learning_rate": 1e-05,
      "loss": 3.1389,
      "step": 995
    },
    {
      "epoch": 0.18832613378217727,
      "grad_norm": 7.759548664093018,
      "learning_rate": 1e-05,
      "loss": 3.1203,
      "step": 1000
    },
    {
      "epoch": 0.18832613378217727,
      "eval_accuracy": 0.3959961907900079,
      "eval_loss": 3.179180860519409,
      "eval_runtime": 1109.8083,
      "eval_samples_per_second": 34.024,
      "eval_steps_per_second": 8.506,
      "step": 1000
    },
    {
      "epoch": 0.18926776445108817,
      "grad_norm": 8.558948516845703,
      "learning_rate": 1e-05,
      "loss": 3.2909,
      "step": 1005
    },
    {
      "epoch": 0.19020939511999907,
      "grad_norm": 7.582333087921143,
      "learning_rate": 1e-05,
      "loss": 3.1388,
      "step": 1010
    },
    {
      "epoch": 0.19115102578890994,
      "grad_norm": 8.941636085510254,
      "learning_rate": 1e-05,
      "loss": 3.1176,
      "step": 1015
    },
    {
      "epoch": 0.19209265645782084,
      "grad_norm": 8.161066055297852,
      "learning_rate": 1e-05,
      "loss": 3.1302,
      "step": 1020
    },
    {
      "epoch": 0.1930342871267317,
      "grad_norm": 8.573451042175293,
      "learning_rate": 1e-05,
      "loss": 3.1978,
      "step": 1025
    },
    {
      "epoch": 0.1939759177956426,
      "grad_norm": 9.676179885864258,
      "learning_rate": 1e-05,
      "loss": 3.2441,
      "step": 1030
    },
    {
      "epoch": 0.1949175484645535,
      "grad_norm": 9.518370628356934,
      "learning_rate": 1e-05,
      "loss": 3.209,
      "step": 1035
    },
    {
      "epoch": 0.19585917913346437,
      "grad_norm": 8.120454788208008,
      "learning_rate": 1e-05,
      "loss": 3.0915,
      "step": 1040
    },
    {
      "epoch": 0.19680080980237527,
      "grad_norm": 8.362229347229004,
      "learning_rate": 1e-05,
      "loss": 3.2262,
      "step": 1045
    },
    {
      "epoch": 0.19774244047128614,
      "grad_norm": 8.197416305541992,
      "learning_rate": 1e-05,
      "loss": 3.1164,
      "step": 1050
    },
    {
      "epoch": 0.19868407114019704,
      "grad_norm": 8.58997631072998,
      "learning_rate": 1e-05,
      "loss": 3.2323,
      "step": 1055
    },
    {
      "epoch": 0.19962570180910794,
      "grad_norm": 7.581511497497559,
      "learning_rate": 1e-05,
      "loss": 3.1852,
      "step": 1060
    },
    {
      "epoch": 0.2005673324780188,
      "grad_norm": 7.385310649871826,
      "learning_rate": 1e-05,
      "loss": 3.2516,
      "step": 1065
    },
    {
      "epoch": 0.2015089631469297,
      "grad_norm": 7.678237438201904,
      "learning_rate": 1e-05,
      "loss": 3.1669,
      "step": 1070
    },
    {
      "epoch": 0.20245059381584057,
      "grad_norm": 8.25271224975586,
      "learning_rate": 1e-05,
      "loss": 3.1575,
      "step": 1075
    },
    {
      "epoch": 0.20339222448475147,
      "grad_norm": 7.587473392486572,
      "learning_rate": 1e-05,
      "loss": 3.182,
      "step": 1080
    },
    {
      "epoch": 0.20433385515366234,
      "grad_norm": 7.0264153480529785,
      "learning_rate": 1e-05,
      "loss": 3.1049,
      "step": 1085
    },
    {
      "epoch": 0.20527548582257324,
      "grad_norm": 7.00242280960083,
      "learning_rate": 1e-05,
      "loss": 3.0769,
      "step": 1090
    },
    {
      "epoch": 0.20621711649148414,
      "grad_norm": 8.321081161499023,
      "learning_rate": 1e-05,
      "loss": 3.0977,
      "step": 1095
    },
    {
      "epoch": 0.207158747160395,
      "grad_norm": 8.19550609588623,
      "learning_rate": 1e-05,
      "loss": 3.1317,
      "step": 1100
    },
    {
      "epoch": 0.2081003778293059,
      "grad_norm": 7.681639671325684,
      "learning_rate": 1e-05,
      "loss": 3.2471,
      "step": 1105
    },
    {
      "epoch": 0.20904200849821677,
      "grad_norm": 8.00836181640625,
      "learning_rate": 1e-05,
      "loss": 3.183,
      "step": 1110
    },
    {
      "epoch": 0.20998363916712767,
      "grad_norm": 7.7261576652526855,
      "learning_rate": 1e-05,
      "loss": 3.2104,
      "step": 1115
    },
    {
      "epoch": 0.21092526983603857,
      "grad_norm": 7.512089729309082,
      "learning_rate": 1e-05,
      "loss": 3.0842,
      "step": 1120
    },
    {
      "epoch": 0.21186690050494944,
      "grad_norm": 8.911776542663574,
      "learning_rate": 1e-05,
      "loss": 3.2527,
      "step": 1125
    },
    {
      "epoch": 0.21280853117386034,
      "grad_norm": 7.831639766693115,
      "learning_rate": 1e-05,
      "loss": 3.2335,
      "step": 1130
    },
    {
      "epoch": 0.2137501618427712,
      "grad_norm": 7.919592380523682,
      "learning_rate": 1e-05,
      "loss": 3.1974,
      "step": 1135
    },
    {
      "epoch": 0.2146917925116821,
      "grad_norm": 7.262505531311035,
      "learning_rate": 1e-05,
      "loss": 3.0964,
      "step": 1140
    },
    {
      "epoch": 0.215633423180593,
      "grad_norm": 7.787773132324219,
      "learning_rate": 1e-05,
      "loss": 3.1731,
      "step": 1145
    },
    {
      "epoch": 0.21657505384950387,
      "grad_norm": 8.574644088745117,
      "learning_rate": 1e-05,
      "loss": 3.203,
      "step": 1150
    },
    {
      "epoch": 0.21751668451841477,
      "grad_norm": 8.63223648071289,
      "learning_rate": 1e-05,
      "loss": 3.0797,
      "step": 1155
    },
    {
      "epoch": 0.21845831518732564,
      "grad_norm": 7.836942195892334,
      "learning_rate": 1e-05,
      "loss": 3.1964,
      "step": 1160
    },
    {
      "epoch": 0.21939994585623654,
      "grad_norm": 8.048019409179688,
      "learning_rate": 1e-05,
      "loss": 3.0157,
      "step": 1165
    },
    {
      "epoch": 0.22034157652514744,
      "grad_norm": 9.863574981689453,
      "learning_rate": 1e-05,
      "loss": 3.1937,
      "step": 1170
    },
    {
      "epoch": 0.2212832071940583,
      "grad_norm": 7.458190441131592,
      "learning_rate": 1e-05,
      "loss": 3.1185,
      "step": 1175
    },
    {
      "epoch": 0.2222248378629692,
      "grad_norm": 7.304734230041504,
      "learning_rate": 1e-05,
      "loss": 3.0673,
      "step": 1180
    },
    {
      "epoch": 0.22316646853188007,
      "grad_norm": 8.21141529083252,
      "learning_rate": 1e-05,
      "loss": 3.1101,
      "step": 1185
    },
    {
      "epoch": 0.22410809920079097,
      "grad_norm": 8.499639511108398,
      "learning_rate": 1e-05,
      "loss": 3.2017,
      "step": 1190
    },
    {
      "epoch": 0.22504972986970187,
      "grad_norm": 8.153023719787598,
      "learning_rate": 1e-05,
      "loss": 3.1853,
      "step": 1195
    },
    {
      "epoch": 0.22599136053861274,
      "grad_norm": 8.166133880615234,
      "learning_rate": 1e-05,
      "loss": 3.0598,
      "step": 1200
    },
    {
      "epoch": 0.22693299120752364,
      "grad_norm": 8.11247730255127,
      "learning_rate": 1e-05,
      "loss": 3.2231,
      "step": 1205
    },
    {
      "epoch": 0.2278746218764345,
      "grad_norm": 9.046285629272461,
      "learning_rate": 1e-05,
      "loss": 3.0795,
      "step": 1210
    },
    {
      "epoch": 0.2288162525453454,
      "grad_norm": 8.127588272094727,
      "learning_rate": 1e-05,
      "loss": 3.1155,
      "step": 1215
    },
    {
      "epoch": 0.22975788321425628,
      "grad_norm": 7.537458419799805,
      "learning_rate": 1e-05,
      "loss": 3.1393,
      "step": 1220
    },
    {
      "epoch": 0.23069951388316717,
      "grad_norm": 7.949858665466309,
      "learning_rate": 1e-05,
      "loss": 3.1817,
      "step": 1225
    },
    {
      "epoch": 0.23164114455207807,
      "grad_norm": 8.178318977355957,
      "learning_rate": 1e-05,
      "loss": 3.1628,
      "step": 1230
    },
    {
      "epoch": 0.23258277522098894,
      "grad_norm": 7.652297019958496,
      "learning_rate": 1e-05,
      "loss": 3.1317,
      "step": 1235
    },
    {
      "epoch": 0.23352440588989984,
      "grad_norm": 7.234852313995361,
      "learning_rate": 1e-05,
      "loss": 3.1783,
      "step": 1240
    },
    {
      "epoch": 0.2344660365588107,
      "grad_norm": 7.958459854125977,
      "learning_rate": 1e-05,
      "loss": 3.1752,
      "step": 1245
    },
    {
      "epoch": 0.2354076672277216,
      "grad_norm": 7.870058536529541,
      "learning_rate": 1e-05,
      "loss": 3.2413,
      "step": 1250
    },
    {
      "epoch": 0.2363492978966325,
      "grad_norm": 8.846611976623535,
      "learning_rate": 1e-05,
      "loss": 3.1678,
      "step": 1255
    },
    {
      "epoch": 0.23729092856554337,
      "grad_norm": 8.129776000976562,
      "learning_rate": 1e-05,
      "loss": 3.1669,
      "step": 1260
    },
    {
      "epoch": 0.23823255923445427,
      "grad_norm": 6.927892208099365,
      "learning_rate": 1e-05,
      "loss": 3.1201,
      "step": 1265
    },
    {
      "epoch": 0.23917418990336514,
      "grad_norm": 9.028277397155762,
      "learning_rate": 1e-05,
      "loss": 3.1909,
      "step": 1270
    },
    {
      "epoch": 0.24011582057227604,
      "grad_norm": 8.353940963745117,
      "learning_rate": 1e-05,
      "loss": 3.2572,
      "step": 1275
    },
    {
      "epoch": 0.24105745124118694,
      "grad_norm": 7.9163737297058105,
      "learning_rate": 1e-05,
      "loss": 3.1206,
      "step": 1280
    },
    {
      "epoch": 0.2419990819100978,
      "grad_norm": 8.831631660461426,
      "learning_rate": 1e-05,
      "loss": 3.2485,
      "step": 1285
    },
    {
      "epoch": 0.2429407125790087,
      "grad_norm": 7.537317752838135,
      "learning_rate": 1e-05,
      "loss": 3.1398,
      "step": 1290
    },
    {
      "epoch": 0.24388234324791958,
      "grad_norm": 7.471547603607178,
      "learning_rate": 1e-05,
      "loss": 3.0652,
      "step": 1295
    },
    {
      "epoch": 0.24482397391683047,
      "grad_norm": 7.851377487182617,
      "learning_rate": 1e-05,
      "loss": 3.0686,
      "step": 1300
    },
    {
      "epoch": 0.24576560458574137,
      "grad_norm": 8.015359878540039,
      "learning_rate": 1e-05,
      "loss": 3.2359,
      "step": 1305
    },
    {
      "epoch": 0.24670723525465224,
      "grad_norm": 8.371912002563477,
      "learning_rate": 1e-05,
      "loss": 3.0857,
      "step": 1310
    },
    {
      "epoch": 0.24764886592356314,
      "grad_norm": 7.510787487030029,
      "learning_rate": 1e-05,
      "loss": 3.244,
      "step": 1315
    },
    {
      "epoch": 0.248590496592474,
      "grad_norm": 7.756171703338623,
      "learning_rate": 1e-05,
      "loss": 3.1444,
      "step": 1320
    },
    {
      "epoch": 0.2495321272613849,
      "grad_norm": 8.038382530212402,
      "learning_rate": 1e-05,
      "loss": 3.2096,
      "step": 1325
    },
    {
      "epoch": 0.2504737579302958,
      "grad_norm": 7.765430927276611,
      "learning_rate": 1e-05,
      "loss": 3.139,
      "step": 1330
    },
    {
      "epoch": 0.2514153885992067,
      "grad_norm": 8.474689483642578,
      "learning_rate": 1e-05,
      "loss": 3.1436,
      "step": 1335
    },
    {
      "epoch": 0.25235701926811754,
      "grad_norm": 9.304710388183594,
      "learning_rate": 1e-05,
      "loss": 3.1479,
      "step": 1340
    },
    {
      "epoch": 0.25329864993702844,
      "grad_norm": 10.669320106506348,
      "learning_rate": 1e-05,
      "loss": 3.0753,
      "step": 1345
    },
    {
      "epoch": 0.25424028060593934,
      "grad_norm": 6.909900665283203,
      "learning_rate": 1e-05,
      "loss": 3.0703,
      "step": 1350
    },
    {
      "epoch": 0.25518191127485024,
      "grad_norm": 8.000332832336426,
      "learning_rate": 1e-05,
      "loss": 3.0837,
      "step": 1355
    },
    {
      "epoch": 0.25612354194376113,
      "grad_norm": 7.857721328735352,
      "learning_rate": 1e-05,
      "loss": 3.121,
      "step": 1360
    },
    {
      "epoch": 0.257065172612672,
      "grad_norm": 7.735369682312012,
      "learning_rate": 1e-05,
      "loss": 3.122,
      "step": 1365
    },
    {
      "epoch": 0.2580068032815829,
      "grad_norm": 7.965452194213867,
      "learning_rate": 1e-05,
      "loss": 3.1381,
      "step": 1370
    },
    {
      "epoch": 0.2589484339504938,
      "grad_norm": 8.671842575073242,
      "learning_rate": 1e-05,
      "loss": 3.1744,
      "step": 1375
    },
    {
      "epoch": 0.25989006461940467,
      "grad_norm": 7.975509166717529,
      "learning_rate": 1e-05,
      "loss": 3.1222,
      "step": 1380
    },
    {
      "epoch": 0.26083169528831557,
      "grad_norm": 7.30696964263916,
      "learning_rate": 1e-05,
      "loss": 3.1122,
      "step": 1385
    },
    {
      "epoch": 0.2617733259572264,
      "grad_norm": 8.272529602050781,
      "learning_rate": 1e-05,
      "loss": 3.0969,
      "step": 1390
    },
    {
      "epoch": 0.2627149566261373,
      "grad_norm": 8.145869255065918,
      "learning_rate": 1e-05,
      "loss": 3.0817,
      "step": 1395
    },
    {
      "epoch": 0.2636565872950482,
      "grad_norm": 7.628718852996826,
      "learning_rate": 1e-05,
      "loss": 3.105,
      "step": 1400
    },
    {
      "epoch": 0.2645982179639591,
      "grad_norm": 6.900229454040527,
      "learning_rate": 1e-05,
      "loss": 3.2028,
      "step": 1405
    },
    {
      "epoch": 0.26553984863286995,
      "grad_norm": 7.9082207679748535,
      "learning_rate": 1e-05,
      "loss": 3.014,
      "step": 1410
    },
    {
      "epoch": 0.26648147930178084,
      "grad_norm": 8.532185554504395,
      "learning_rate": 1e-05,
      "loss": 3.143,
      "step": 1415
    },
    {
      "epoch": 0.26742310997069174,
      "grad_norm": 8.344785690307617,
      "learning_rate": 1e-05,
      "loss": 3.1865,
      "step": 1420
    },
    {
      "epoch": 0.26836474063960264,
      "grad_norm": 7.369927883148193,
      "learning_rate": 1e-05,
      "loss": 3.0858,
      "step": 1425
    },
    {
      "epoch": 0.26930637130851354,
      "grad_norm": 7.539013385772705,
      "learning_rate": 1e-05,
      "loss": 2.9854,
      "step": 1430
    },
    {
      "epoch": 0.2702480019774244,
      "grad_norm": 8.182465553283691,
      "learning_rate": 1e-05,
      "loss": 3.0504,
      "step": 1435
    },
    {
      "epoch": 0.2711896326463353,
      "grad_norm": 8.410788536071777,
      "learning_rate": 1e-05,
      "loss": 3.0354,
      "step": 1440
    },
    {
      "epoch": 0.2721312633152462,
      "grad_norm": 7.914114475250244,
      "learning_rate": 1e-05,
      "loss": 3.067,
      "step": 1445
    },
    {
      "epoch": 0.2730728939841571,
      "grad_norm": 8.113722801208496,
      "learning_rate": 1e-05,
      "loss": 3.1172,
      "step": 1450
    },
    {
      "epoch": 0.27401452465306797,
      "grad_norm": 7.887296676635742,
      "learning_rate": 1e-05,
      "loss": 3.139,
      "step": 1455
    },
    {
      "epoch": 0.2749561553219788,
      "grad_norm": 9.083703994750977,
      "learning_rate": 1e-05,
      "loss": 3.1805,
      "step": 1460
    },
    {
      "epoch": 0.2758977859908897,
      "grad_norm": 7.937193393707275,
      "learning_rate": 1e-05,
      "loss": 3.1389,
      "step": 1465
    },
    {
      "epoch": 0.2768394166598006,
      "grad_norm": 8.576871871948242,
      "learning_rate": 1e-05,
      "loss": 3.0739,
      "step": 1470
    },
    {
      "epoch": 0.2777810473287115,
      "grad_norm": 8.345330238342285,
      "learning_rate": 1e-05,
      "loss": 3.043,
      "step": 1475
    },
    {
      "epoch": 0.2787226779976224,
      "grad_norm": 8.3008394241333,
      "learning_rate": 1e-05,
      "loss": 3.2218,
      "step": 1480
    },
    {
      "epoch": 0.27966430866653325,
      "grad_norm": 7.56035852432251,
      "learning_rate": 1e-05,
      "loss": 3.1249,
      "step": 1485
    },
    {
      "epoch": 0.28060593933544414,
      "grad_norm": 7.499727249145508,
      "learning_rate": 1e-05,
      "loss": 3.0944,
      "step": 1490
    },
    {
      "epoch": 0.28154757000435504,
      "grad_norm": 8.389445304870605,
      "learning_rate": 1e-05,
      "loss": 3.0844,
      "step": 1495
    },
    {
      "epoch": 0.28248920067326594,
      "grad_norm": 9.202953338623047,
      "learning_rate": 1e-05,
      "loss": 3.2153,
      "step": 1500
    },
    {
      "epoch": 0.28343083134217684,
      "grad_norm": 7.818172454833984,
      "learning_rate": 1e-05,
      "loss": 3.0676,
      "step": 1505
    },
    {
      "epoch": 0.2843724620110877,
      "grad_norm": 7.743332386016846,
      "learning_rate": 1e-05,
      "loss": 3.0716,
      "step": 1510
    },
    {
      "epoch": 0.2853140926799986,
      "grad_norm": 8.619895935058594,
      "learning_rate": 1e-05,
      "loss": 3.142,
      "step": 1515
    },
    {
      "epoch": 0.2862557233489095,
      "grad_norm": 7.999989986419678,
      "learning_rate": 1e-05,
      "loss": 3.2461,
      "step": 1520
    },
    {
      "epoch": 0.28719735401782037,
      "grad_norm": 7.5509772300720215,
      "learning_rate": 1e-05,
      "loss": 3.1051,
      "step": 1525
    },
    {
      "epoch": 0.28813898468673127,
      "grad_norm": 7.679853916168213,
      "learning_rate": 1e-05,
      "loss": 3.1131,
      "step": 1530
    },
    {
      "epoch": 0.2890806153556421,
      "grad_norm": 7.623820781707764,
      "learning_rate": 1e-05,
      "loss": 3.1323,
      "step": 1535
    },
    {
      "epoch": 0.290022246024553,
      "grad_norm": 7.860290050506592,
      "learning_rate": 1e-05,
      "loss": 3.1595,
      "step": 1540
    },
    {
      "epoch": 0.2909638766934639,
      "grad_norm": 7.90553092956543,
      "learning_rate": 1e-05,
      "loss": 3.0943,
      "step": 1545
    },
    {
      "epoch": 0.2919055073623748,
      "grad_norm": 8.377348899841309,
      "learning_rate": 1e-05,
      "loss": 3.194,
      "step": 1550
    },
    {
      "epoch": 0.2928471380312857,
      "grad_norm": 7.801243305206299,
      "learning_rate": 1e-05,
      "loss": 3.0033,
      "step": 1555
    },
    {
      "epoch": 0.29378876870019655,
      "grad_norm": 8.063822746276855,
      "learning_rate": 1e-05,
      "loss": 3.1661,
      "step": 1560
    },
    {
      "epoch": 0.29473039936910744,
      "grad_norm": 7.1917314529418945,
      "learning_rate": 1e-05,
      "loss": 3.1411,
      "step": 1565
    },
    {
      "epoch": 0.29567203003801834,
      "grad_norm": 7.78692102432251,
      "learning_rate": 1e-05,
      "loss": 3.0471,
      "step": 1570
    },
    {
      "epoch": 0.29661366070692924,
      "grad_norm": 8.212677001953125,
      "learning_rate": 1e-05,
      "loss": 3.1417,
      "step": 1575
    },
    {
      "epoch": 0.29755529137584014,
      "grad_norm": 6.906081676483154,
      "learning_rate": 1e-05,
      "loss": 3.13,
      "step": 1580
    },
    {
      "epoch": 0.298496922044751,
      "grad_norm": 7.585642337799072,
      "learning_rate": 1e-05,
      "loss": 3.1489,
      "step": 1585
    },
    {
      "epoch": 0.2994385527136619,
      "grad_norm": 7.452507495880127,
      "learning_rate": 1e-05,
      "loss": 3.1203,
      "step": 1590
    },
    {
      "epoch": 0.3003801833825728,
      "grad_norm": 9.468456268310547,
      "learning_rate": 1e-05,
      "loss": 3.1018,
      "step": 1595
    },
    {
      "epoch": 0.30132181405148367,
      "grad_norm": 8.016668319702148,
      "learning_rate": 1e-05,
      "loss": 3.209,
      "step": 1600
    },
    {
      "epoch": 0.30226344472039457,
      "grad_norm": 7.062180995941162,
      "learning_rate": 1e-05,
      "loss": 3.2921,
      "step": 1605
    },
    {
      "epoch": 0.3032050753893054,
      "grad_norm": 7.844501495361328,
      "learning_rate": 1e-05,
      "loss": 3.1812,
      "step": 1610
    },
    {
      "epoch": 0.3041467060582163,
      "grad_norm": 6.8861308097839355,
      "learning_rate": 1e-05,
      "loss": 3.0092,
      "step": 1615
    },
    {
      "epoch": 0.3050883367271272,
      "grad_norm": 7.58292818069458,
      "learning_rate": 1e-05,
      "loss": 3.1616,
      "step": 1620
    },
    {
      "epoch": 0.3060299673960381,
      "grad_norm": 7.547516822814941,
      "learning_rate": 1e-05,
      "loss": 3.1064,
      "step": 1625
    },
    {
      "epoch": 0.306971598064949,
      "grad_norm": 7.924961090087891,
      "learning_rate": 1e-05,
      "loss": 3.1486,
      "step": 1630
    },
    {
      "epoch": 0.30791322873385985,
      "grad_norm": 9.10555362701416,
      "learning_rate": 1e-05,
      "loss": 2.995,
      "step": 1635
    },
    {
      "epoch": 0.30885485940277074,
      "grad_norm": 7.4836931228637695,
      "learning_rate": 1e-05,
      "loss": 3.1971,
      "step": 1640
    },
    {
      "epoch": 0.30979649007168164,
      "grad_norm": 9.865035057067871,
      "learning_rate": 1e-05,
      "loss": 3.1604,
      "step": 1645
    },
    {
      "epoch": 0.31073812074059254,
      "grad_norm": 8.200763702392578,
      "learning_rate": 1e-05,
      "loss": 3.0831,
      "step": 1650
    },
    {
      "epoch": 0.3116797514095034,
      "grad_norm": 8.044021606445312,
      "learning_rate": 1e-05,
      "loss": 3.1665,
      "step": 1655
    },
    {
      "epoch": 0.3126213820784143,
      "grad_norm": 7.945106029510498,
      "learning_rate": 1e-05,
      "loss": 3.0876,
      "step": 1660
    },
    {
      "epoch": 0.3135630127473252,
      "grad_norm": 8.18039608001709,
      "learning_rate": 1e-05,
      "loss": 3.1172,
      "step": 1665
    },
    {
      "epoch": 0.3145046434162361,
      "grad_norm": 7.830636024475098,
      "learning_rate": 1e-05,
      "loss": 3.1627,
      "step": 1670
    },
    {
      "epoch": 0.31544627408514697,
      "grad_norm": 8.153444290161133,
      "learning_rate": 1e-05,
      "loss": 3.0749,
      "step": 1675
    },
    {
      "epoch": 0.3163879047540578,
      "grad_norm": 8.141214370727539,
      "learning_rate": 1e-05,
      "loss": 3.0414,
      "step": 1680
    },
    {
      "epoch": 0.3173295354229687,
      "grad_norm": 7.207879543304443,
      "learning_rate": 1e-05,
      "loss": 3.1829,
      "step": 1685
    },
    {
      "epoch": 0.3182711660918796,
      "grad_norm": 6.900830268859863,
      "learning_rate": 1e-05,
      "loss": 3.0745,
      "step": 1690
    },
    {
      "epoch": 0.3192127967607905,
      "grad_norm": 7.9972243309021,
      "learning_rate": 1e-05,
      "loss": 3.1691,
      "step": 1695
    },
    {
      "epoch": 0.3201544274297014,
      "grad_norm": 8.405791282653809,
      "learning_rate": 1e-05,
      "loss": 3.0937,
      "step": 1700
    },
    {
      "epoch": 0.32109605809861225,
      "grad_norm": 7.26245641708374,
      "learning_rate": 1e-05,
      "loss": 3.0187,
      "step": 1705
    },
    {
      "epoch": 0.32203768876752314,
      "grad_norm": 8.00014591217041,
      "learning_rate": 1e-05,
      "loss": 3.0608,
      "step": 1710
    },
    {
      "epoch": 0.32297931943643404,
      "grad_norm": 7.747166156768799,
      "learning_rate": 1e-05,
      "loss": 3.1721,
      "step": 1715
    },
    {
      "epoch": 0.32392095010534494,
      "grad_norm": 8.84945011138916,
      "learning_rate": 1e-05,
      "loss": 3.0493,
      "step": 1720
    },
    {
      "epoch": 0.32486258077425584,
      "grad_norm": 8.65839958190918,
      "learning_rate": 1e-05,
      "loss": 3.0396,
      "step": 1725
    },
    {
      "epoch": 0.3258042114431667,
      "grad_norm": 8.27173900604248,
      "learning_rate": 1e-05,
      "loss": 3.0956,
      "step": 1730
    },
    {
      "epoch": 0.3267458421120776,
      "grad_norm": 7.819243907928467,
      "learning_rate": 1e-05,
      "loss": 3.1334,
      "step": 1735
    },
    {
      "epoch": 0.3276874727809885,
      "grad_norm": 8.155767440795898,
      "learning_rate": 1e-05,
      "loss": 3.1312,
      "step": 1740
    },
    {
      "epoch": 0.3286291034498994,
      "grad_norm": 7.6382155418396,
      "learning_rate": 1e-05,
      "loss": 2.9995,
      "step": 1745
    },
    {
      "epoch": 0.32957073411881027,
      "grad_norm": 7.503223896026611,
      "learning_rate": 1e-05,
      "loss": 3.1386,
      "step": 1750
    },
    {
      "epoch": 0.3305123647877211,
      "grad_norm": 8.2285737991333,
      "learning_rate": 1e-05,
      "loss": 3.0442,
      "step": 1755
    },
    {
      "epoch": 0.331453995456632,
      "grad_norm": 7.820169925689697,
      "learning_rate": 1e-05,
      "loss": 3.0377,
      "step": 1760
    },
    {
      "epoch": 0.3323956261255429,
      "grad_norm": 7.6100969314575195,
      "learning_rate": 1e-05,
      "loss": 3.1311,
      "step": 1765
    },
    {
      "epoch": 0.3333372567944538,
      "grad_norm": 7.763628005981445,
      "learning_rate": 1e-05,
      "loss": 3.0599,
      "step": 1770
    },
    {
      "epoch": 0.3342788874633647,
      "grad_norm": 8.461652755737305,
      "learning_rate": 1e-05,
      "loss": 3.222,
      "step": 1775
    },
    {
      "epoch": 0.33522051813227555,
      "grad_norm": 7.414519309997559,
      "learning_rate": 1e-05,
      "loss": 3.0587,
      "step": 1780
    },
    {
      "epoch": 0.33616214880118644,
      "grad_norm": 7.964784145355225,
      "learning_rate": 1e-05,
      "loss": 3.2038,
      "step": 1785
    },
    {
      "epoch": 0.33710377947009734,
      "grad_norm": 7.656503200531006,
      "learning_rate": 1e-05,
      "loss": 3.1324,
      "step": 1790
    },
    {
      "epoch": 0.33804541013900824,
      "grad_norm": 8.035988807678223,
      "learning_rate": 1e-05,
      "loss": 3.1101,
      "step": 1795
    },
    {
      "epoch": 0.33898704080791914,
      "grad_norm": 7.202066898345947,
      "learning_rate": 1e-05,
      "loss": 3.0915,
      "step": 1800
    },
    {
      "epoch": 0.33992867147683,
      "grad_norm": 8.242351531982422,
      "learning_rate": 1e-05,
      "loss": 3.2059,
      "step": 1805
    },
    {
      "epoch": 0.3408703021457409,
      "grad_norm": 7.794888019561768,
      "learning_rate": 1e-05,
      "loss": 3.1364,
      "step": 1810
    },
    {
      "epoch": 0.3418119328146518,
      "grad_norm": 7.34774923324585,
      "learning_rate": 1e-05,
      "loss": 3.0507,
      "step": 1815
    },
    {
      "epoch": 0.3427535634835627,
      "grad_norm": 7.673720359802246,
      "learning_rate": 1e-05,
      "loss": 3.0383,
      "step": 1820
    },
    {
      "epoch": 0.34369519415247357,
      "grad_norm": 7.948644638061523,
      "learning_rate": 1e-05,
      "loss": 3.1888,
      "step": 1825
    },
    {
      "epoch": 0.3446368248213844,
      "grad_norm": 7.747712135314941,
      "learning_rate": 1e-05,
      "loss": 3.089,
      "step": 1830
    },
    {
      "epoch": 0.3455784554902953,
      "grad_norm": 7.607177257537842,
      "learning_rate": 1e-05,
      "loss": 3.1111,
      "step": 1835
    },
    {
      "epoch": 0.3465200861592062,
      "grad_norm": 6.7083916664123535,
      "learning_rate": 1e-05,
      "loss": 3.0944,
      "step": 1840
    },
    {
      "epoch": 0.3474617168281171,
      "grad_norm": 8.02619743347168,
      "learning_rate": 1e-05,
      "loss": 3.1129,
      "step": 1845
    },
    {
      "epoch": 0.348403347497028,
      "grad_norm": 8.047721862792969,
      "learning_rate": 1e-05,
      "loss": 3.1843,
      "step": 1850
    },
    {
      "epoch": 0.34934497816593885,
      "grad_norm": 7.408081531524658,
      "learning_rate": 1e-05,
      "loss": 3.0514,
      "step": 1855
    },
    {
      "epoch": 0.35028660883484974,
      "grad_norm": 7.935153484344482,
      "learning_rate": 1e-05,
      "loss": 3.1919,
      "step": 1860
    },
    {
      "epoch": 0.35122823950376064,
      "grad_norm": 7.98247766494751,
      "learning_rate": 1e-05,
      "loss": 3.1084,
      "step": 1865
    },
    {
      "epoch": 0.35216987017267154,
      "grad_norm": 6.453451156616211,
      "learning_rate": 1e-05,
      "loss": 3.0292,
      "step": 1870
    },
    {
      "epoch": 0.35311150084158244,
      "grad_norm": 7.293237686157227,
      "learning_rate": 1e-05,
      "loss": 3.0266,
      "step": 1875
    },
    {
      "epoch": 0.3540531315104933,
      "grad_norm": 7.254806041717529,
      "learning_rate": 1e-05,
      "loss": 3.1074,
      "step": 1880
    },
    {
      "epoch": 0.3549947621794042,
      "grad_norm": 7.4231743812561035,
      "learning_rate": 1e-05,
      "loss": 3.1157,
      "step": 1885
    },
    {
      "epoch": 0.3559363928483151,
      "grad_norm": 7.126735210418701,
      "learning_rate": 1e-05,
      "loss": 3.0207,
      "step": 1890
    },
    {
      "epoch": 0.356878023517226,
      "grad_norm": 7.252379417419434,
      "learning_rate": 1e-05,
      "loss": 3.1247,
      "step": 1895
    },
    {
      "epoch": 0.3578196541861368,
      "grad_norm": 8.155769348144531,
      "learning_rate": 1e-05,
      "loss": 3.1562,
      "step": 1900
    },
    {
      "epoch": 0.3587612848550477,
      "grad_norm": 7.091341972351074,
      "learning_rate": 1e-05,
      "loss": 3.0676,
      "step": 1905
    },
    {
      "epoch": 0.3597029155239586,
      "grad_norm": 7.706187725067139,
      "learning_rate": 1e-05,
      "loss": 3.1094,
      "step": 1910
    },
    {
      "epoch": 0.3606445461928695,
      "grad_norm": 7.110264301300049,
      "learning_rate": 1e-05,
      "loss": 3.12,
      "step": 1915
    },
    {
      "epoch": 0.3615861768617804,
      "grad_norm": 6.751636505126953,
      "learning_rate": 1e-05,
      "loss": 3.189,
      "step": 1920
    },
    {
      "epoch": 0.36252780753069125,
      "grad_norm": 7.525967597961426,
      "learning_rate": 1e-05,
      "loss": 3.1649,
      "step": 1925
    },
    {
      "epoch": 0.36346943819960215,
      "grad_norm": 7.6558356285095215,
      "learning_rate": 1e-05,
      "loss": 3.103,
      "step": 1930
    },
    {
      "epoch": 0.36441106886851304,
      "grad_norm": 7.773464679718018,
      "learning_rate": 1e-05,
      "loss": 3.1464,
      "step": 1935
    },
    {
      "epoch": 0.36535269953742394,
      "grad_norm": 8.413071632385254,
      "learning_rate": 1e-05,
      "loss": 3.0434,
      "step": 1940
    },
    {
      "epoch": 0.36629433020633484,
      "grad_norm": 9.84329605102539,
      "learning_rate": 1e-05,
      "loss": 3.077,
      "step": 1945
    },
    {
      "epoch": 0.3672359608752457,
      "grad_norm": 7.708522319793701,
      "learning_rate": 1e-05,
      "loss": 3.0714,
      "step": 1950
    },
    {
      "epoch": 0.3681775915441566,
      "grad_norm": 7.770474910736084,
      "learning_rate": 1e-05,
      "loss": 3.1796,
      "step": 1955
    },
    {
      "epoch": 0.3691192222130675,
      "grad_norm": 6.997617244720459,
      "learning_rate": 1e-05,
      "loss": 3.1673,
      "step": 1960
    },
    {
      "epoch": 0.3700608528819784,
      "grad_norm": 7.298404216766357,
      "learning_rate": 1e-05,
      "loss": 2.955,
      "step": 1965
    },
    {
      "epoch": 0.3710024835508893,
      "grad_norm": 7.35360050201416,
      "learning_rate": 1e-05,
      "loss": 2.9934,
      "step": 1970
    },
    {
      "epoch": 0.3719441142198001,
      "grad_norm": 7.945688247680664,
      "learning_rate": 1e-05,
      "loss": 3.0656,
      "step": 1975
    },
    {
      "epoch": 0.372885744888711,
      "grad_norm": 8.263771057128906,
      "learning_rate": 1e-05,
      "loss": 3.2,
      "step": 1980
    },
    {
      "epoch": 0.3738273755576219,
      "grad_norm": 7.973668575286865,
      "learning_rate": 1e-05,
      "loss": 3.129,
      "step": 1985
    },
    {
      "epoch": 0.3747690062265328,
      "grad_norm": 7.491722583770752,
      "learning_rate": 1e-05,
      "loss": 3.1392,
      "step": 1990
    },
    {
      "epoch": 0.3757106368954437,
      "grad_norm": 7.867580413818359,
      "learning_rate": 1e-05,
      "loss": 3.1261,
      "step": 1995
    },
    {
      "epoch": 0.37665226756435455,
      "grad_norm": 7.192398548126221,
      "learning_rate": 1e-05,
      "loss": 3.1165,
      "step": 2000
    },
    {
      "epoch": 0.37665226756435455,
      "eval_accuracy": 0.40667947257976045,
      "eval_loss": 3.105938673019409,
      "eval_runtime": 1037.0565,
      "eval_samples_per_second": 36.411,
      "eval_steps_per_second": 9.103,
      "step": 2000
    },
    {
      "epoch": 0.37759389823326545,
      "grad_norm": 7.978450775146484,
      "learning_rate": 1e-05,
      "loss": 3.0121,
      "step": 2005
    },
    {
      "epoch": 0.37853552890217634,
      "grad_norm": 7.5898847579956055,
      "learning_rate": 1e-05,
      "loss": 3.0836,
      "step": 2010
    },
    {
      "epoch": 0.37947715957108724,
      "grad_norm": 7.317259311676025,
      "learning_rate": 1e-05,
      "loss": 3.0859,
      "step": 2015
    },
    {
      "epoch": 0.38041879023999814,
      "grad_norm": 7.094460964202881,
      "learning_rate": 1e-05,
      "loss": 3.0992,
      "step": 2020
    },
    {
      "epoch": 0.381360420908909,
      "grad_norm": 7.557990074157715,
      "learning_rate": 1e-05,
      "loss": 3.0912,
      "step": 2025
    },
    {
      "epoch": 0.3823020515778199,
      "grad_norm": 6.940727710723877,
      "learning_rate": 1e-05,
      "loss": 3.0669,
      "step": 2030
    },
    {
      "epoch": 0.3832436822467308,
      "grad_norm": 8.001688957214355,
      "learning_rate": 1e-05,
      "loss": 3.1704,
      "step": 2035
    },
    {
      "epoch": 0.3841853129156417,
      "grad_norm": 7.38444185256958,
      "learning_rate": 1e-05,
      "loss": 3.0961,
      "step": 2040
    },
    {
      "epoch": 0.3851269435845526,
      "grad_norm": 6.947928428649902,
      "learning_rate": 1e-05,
      "loss": 3.1942,
      "step": 2045
    },
    {
      "epoch": 0.3860685742534634,
      "grad_norm": 7.699880123138428,
      "learning_rate": 1e-05,
      "loss": 3.0334,
      "step": 2050
    },
    {
      "epoch": 0.3870102049223743,
      "grad_norm": 7.490096092224121,
      "learning_rate": 1e-05,
      "loss": 3.1301,
      "step": 2055
    },
    {
      "epoch": 0.3879518355912852,
      "grad_norm": 7.5343451499938965,
      "learning_rate": 1e-05,
      "loss": 3.0588,
      "step": 2060
    },
    {
      "epoch": 0.3888934662601961,
      "grad_norm": 9.956011772155762,
      "learning_rate": 1e-05,
      "loss": 3.0914,
      "step": 2065
    },
    {
      "epoch": 0.389835096929107,
      "grad_norm": 7.387354373931885,
      "learning_rate": 1e-05,
      "loss": 3.1279,
      "step": 2070
    },
    {
      "epoch": 0.39077672759801785,
      "grad_norm": 7.3654351234436035,
      "learning_rate": 1e-05,
      "loss": 3.0307,
      "step": 2075
    },
    {
      "epoch": 0.39171835826692875,
      "grad_norm": 7.671773910522461,
      "learning_rate": 1e-05,
      "loss": 3.0223,
      "step": 2080
    },
    {
      "epoch": 0.39265998893583964,
      "grad_norm": 8.229500770568848,
      "learning_rate": 1e-05,
      "loss": 3.107,
      "step": 2085
    },
    {
      "epoch": 0.39360161960475054,
      "grad_norm": 7.7551798820495605,
      "learning_rate": 1e-05,
      "loss": 3.1048,
      "step": 2090
    },
    {
      "epoch": 0.39454325027366144,
      "grad_norm": 6.759220600128174,
      "learning_rate": 1e-05,
      "loss": 3.0824,
      "step": 2095
    },
    {
      "epoch": 0.3954848809425723,
      "grad_norm": 8.242779731750488,
      "learning_rate": 1e-05,
      "loss": 3.0604,
      "step": 2100
    },
    {
      "epoch": 0.3964265116114832,
      "grad_norm": 7.663089752197266,
      "learning_rate": 1e-05,
      "loss": 3.1625,
      "step": 2105
    },
    {
      "epoch": 0.3973681422803941,
      "grad_norm": 8.07699203491211,
      "learning_rate": 1e-05,
      "loss": 3.0494,
      "step": 2110
    },
    {
      "epoch": 0.398309772949305,
      "grad_norm": 7.712186336517334,
      "learning_rate": 1e-05,
      "loss": 3.1758,
      "step": 2115
    },
    {
      "epoch": 0.39925140361821587,
      "grad_norm": 9.279346466064453,
      "learning_rate": 1e-05,
      "loss": 3.1234,
      "step": 2120
    },
    {
      "epoch": 0.4001930342871267,
      "grad_norm": 6.985004901885986,
      "learning_rate": 1e-05,
      "loss": 3.013,
      "step": 2125
    },
    {
      "epoch": 0.4011346649560376,
      "grad_norm": 7.505834102630615,
      "learning_rate": 1e-05,
      "loss": 3.1199,
      "step": 2130
    },
    {
      "epoch": 0.4020762956249485,
      "grad_norm": 7.714130401611328,
      "learning_rate": 1e-05,
      "loss": 3.0762,
      "step": 2135
    },
    {
      "epoch": 0.4030179262938594,
      "grad_norm": 7.358199119567871,
      "learning_rate": 1e-05,
      "loss": 3.053,
      "step": 2140
    },
    {
      "epoch": 0.40395955696277025,
      "grad_norm": 7.569924354553223,
      "learning_rate": 1e-05,
      "loss": 3.1715,
      "step": 2145
    },
    {
      "epoch": 0.40490118763168115,
      "grad_norm": 7.3484697341918945,
      "learning_rate": 1e-05,
      "loss": 3.093,
      "step": 2150
    },
    {
      "epoch": 0.40584281830059205,
      "grad_norm": 7.914135456085205,
      "learning_rate": 1e-05,
      "loss": 3.0454,
      "step": 2155
    },
    {
      "epoch": 0.40678444896950294,
      "grad_norm": 7.99644660949707,
      "learning_rate": 1e-05,
      "loss": 3.1014,
      "step": 2160
    },
    {
      "epoch": 0.40772607963841384,
      "grad_norm": 8.003348350524902,
      "learning_rate": 1e-05,
      "loss": 3.1274,
      "step": 2165
    },
    {
      "epoch": 0.4086677103073247,
      "grad_norm": 7.18101692199707,
      "learning_rate": 1e-05,
      "loss": 3.0354,
      "step": 2170
    },
    {
      "epoch": 0.4096093409762356,
      "grad_norm": 7.236114501953125,
      "learning_rate": 1e-05,
      "loss": 3.0585,
      "step": 2175
    },
    {
      "epoch": 0.4105509716451465,
      "grad_norm": 7.164278984069824,
      "learning_rate": 1e-05,
      "loss": 3.0792,
      "step": 2180
    },
    {
      "epoch": 0.4114926023140574,
      "grad_norm": 7.827737808227539,
      "learning_rate": 1e-05,
      "loss": 3.1141,
      "step": 2185
    },
    {
      "epoch": 0.4124342329829683,
      "grad_norm": 10.366262435913086,
      "learning_rate": 1e-05,
      "loss": 3.0372,
      "step": 2190
    },
    {
      "epoch": 0.4133758636518791,
      "grad_norm": 8.009645462036133,
      "learning_rate": 1e-05,
      "loss": 3.0653,
      "step": 2195
    },
    {
      "epoch": 0.41431749432079,
      "grad_norm": 8.994948387145996,
      "learning_rate": 1e-05,
      "loss": 3.1723,
      "step": 2200
    },
    {
      "epoch": 0.4152591249897009,
      "grad_norm": 6.850546360015869,
      "learning_rate": 1e-05,
      "loss": 3.1088,
      "step": 2205
    },
    {
      "epoch": 0.4162007556586118,
      "grad_norm": 7.697965145111084,
      "learning_rate": 1e-05,
      "loss": 3.1864,
      "step": 2210
    },
    {
      "epoch": 0.4171423863275227,
      "grad_norm": 7.226153373718262,
      "learning_rate": 1e-05,
      "loss": 3.1957,
      "step": 2215
    },
    {
      "epoch": 0.41808401699643355,
      "grad_norm": 7.277571678161621,
      "learning_rate": 1e-05,
      "loss": 3.1946,
      "step": 2220
    },
    {
      "epoch": 0.41902564766534445,
      "grad_norm": 7.740057945251465,
      "learning_rate": 1e-05,
      "loss": 3.0261,
      "step": 2225
    },
    {
      "epoch": 0.41996727833425535,
      "grad_norm": 7.6060028076171875,
      "learning_rate": 1e-05,
      "loss": 3.0848,
      "step": 2230
    },
    {
      "epoch": 0.42090890900316624,
      "grad_norm": 8.158476829528809,
      "learning_rate": 1e-05,
      "loss": 3.0151,
      "step": 2235
    },
    {
      "epoch": 0.42185053967207714,
      "grad_norm": 7.340221405029297,
      "learning_rate": 1e-05,
      "loss": 2.9477,
      "step": 2240
    },
    {
      "epoch": 0.422792170340988,
      "grad_norm": 6.894491195678711,
      "learning_rate": 1e-05,
      "loss": 3.0536,
      "step": 2245
    },
    {
      "epoch": 0.4237338010098989,
      "grad_norm": 7.707582473754883,
      "learning_rate": 1e-05,
      "loss": 3.0879,
      "step": 2250
    },
    {
      "epoch": 0.4246754316788098,
      "grad_norm": 7.966876029968262,
      "learning_rate": 1e-05,
      "loss": 3.0493,
      "step": 2255
    },
    {
      "epoch": 0.4256170623477207,
      "grad_norm": 7.2440185546875,
      "learning_rate": 1e-05,
      "loss": 2.9738,
      "step": 2260
    },
    {
      "epoch": 0.4265586930166316,
      "grad_norm": 7.6277337074279785,
      "learning_rate": 1e-05,
      "loss": 3.2125,
      "step": 2265
    },
    {
      "epoch": 0.4275003236855424,
      "grad_norm": 7.30360746383667,
      "learning_rate": 1e-05,
      "loss": 3.1131,
      "step": 2270
    },
    {
      "epoch": 0.4284419543544533,
      "grad_norm": 7.502569675445557,
      "learning_rate": 1e-05,
      "loss": 3.0646,
      "step": 2275
    },
    {
      "epoch": 0.4293835850233642,
      "grad_norm": 7.918311595916748,
      "learning_rate": 1e-05,
      "loss": 3.1952,
      "step": 2280
    },
    {
      "epoch": 0.4303252156922751,
      "grad_norm": 7.016907691955566,
      "learning_rate": 1e-05,
      "loss": 3.1168,
      "step": 2285
    },
    {
      "epoch": 0.431266846361186,
      "grad_norm": 7.167459964752197,
      "learning_rate": 1e-05,
      "loss": 3.126,
      "step": 2290
    },
    {
      "epoch": 0.43220847703009685,
      "grad_norm": 7.548356533050537,
      "learning_rate": 1e-05,
      "loss": 3.0791,
      "step": 2295
    },
    {
      "epoch": 0.43315010769900775,
      "grad_norm": 6.640120506286621,
      "learning_rate": 1e-05,
      "loss": 3.1876,
      "step": 2300
    },
    {
      "epoch": 0.43409173836791864,
      "grad_norm": 7.116699695587158,
      "learning_rate": 1e-05,
      "loss": 3.0427,
      "step": 2305
    },
    {
      "epoch": 0.43503336903682954,
      "grad_norm": 7.0854668617248535,
      "learning_rate": 1e-05,
      "loss": 3.1455,
      "step": 2310
    },
    {
      "epoch": 0.43597499970574044,
      "grad_norm": 7.1804423332214355,
      "learning_rate": 1e-05,
      "loss": 3.0863,
      "step": 2315
    },
    {
      "epoch": 0.4369166303746513,
      "grad_norm": 7.1510210037231445,
      "learning_rate": 1e-05,
      "loss": 3.1586,
      "step": 2320
    },
    {
      "epoch": 0.4378582610435622,
      "grad_norm": 7.871683120727539,
      "learning_rate": 1e-05,
      "loss": 3.1336,
      "step": 2325
    },
    {
      "epoch": 0.4387998917124731,
      "grad_norm": 7.599919319152832,
      "learning_rate": 1e-05,
      "loss": 3.05,
      "step": 2330
    },
    {
      "epoch": 0.439741522381384,
      "grad_norm": 8.056229591369629,
      "learning_rate": 1e-05,
      "loss": 3.0782,
      "step": 2335
    },
    {
      "epoch": 0.4406831530502949,
      "grad_norm": 7.520589351654053,
      "learning_rate": 1e-05,
      "loss": 3.113,
      "step": 2340
    },
    {
      "epoch": 0.4416247837192057,
      "grad_norm": 6.8444952964782715,
      "learning_rate": 1e-05,
      "loss": 3.0913,
      "step": 2345
    },
    {
      "epoch": 0.4425664143881166,
      "grad_norm": 7.817233562469482,
      "learning_rate": 1e-05,
      "loss": 3.1279,
      "step": 2350
    },
    {
      "epoch": 0.4435080450570275,
      "grad_norm": 6.604560375213623,
      "learning_rate": 1e-05,
      "loss": 3.1026,
      "step": 2355
    },
    {
      "epoch": 0.4444496757259384,
      "grad_norm": 7.087409496307373,
      "learning_rate": 1e-05,
      "loss": 3.1523,
      "step": 2360
    },
    {
      "epoch": 0.4453913063948493,
      "grad_norm": 7.123996257781982,
      "learning_rate": 1e-05,
      "loss": 3.0798,
      "step": 2365
    },
    {
      "epoch": 0.44633293706376015,
      "grad_norm": 7.211167812347412,
      "learning_rate": 1e-05,
      "loss": 3.1813,
      "step": 2370
    },
    {
      "epoch": 0.44727456773267105,
      "grad_norm": 7.160671710968018,
      "learning_rate": 1e-05,
      "loss": 3.0597,
      "step": 2375
    },
    {
      "epoch": 0.44821619840158194,
      "grad_norm": 7.699546813964844,
      "learning_rate": 1e-05,
      "loss": 3.0837,
      "step": 2380
    },
    {
      "epoch": 0.44915782907049284,
      "grad_norm": 7.012664794921875,
      "learning_rate": 1e-05,
      "loss": 2.9755,
      "step": 2385
    },
    {
      "epoch": 0.45009945973940374,
      "grad_norm": 9.324268341064453,
      "learning_rate": 1e-05,
      "loss": 3.0906,
      "step": 2390
    },
    {
      "epoch": 0.4510410904083146,
      "grad_norm": 7.166225433349609,
      "learning_rate": 1e-05,
      "loss": 3.0489,
      "step": 2395
    },
    {
      "epoch": 0.4519827210772255,
      "grad_norm": 7.982870578765869,
      "learning_rate": 1e-05,
      "loss": 3.1248,
      "step": 2400
    },
    {
      "epoch": 0.4529243517461364,
      "grad_norm": 7.513890743255615,
      "learning_rate": 1e-05,
      "loss": 3.1008,
      "step": 2405
    },
    {
      "epoch": 0.4538659824150473,
      "grad_norm": 6.794632911682129,
      "learning_rate": 1e-05,
      "loss": 3.1117,
      "step": 2410
    },
    {
      "epoch": 0.4548076130839581,
      "grad_norm": 8.143996238708496,
      "learning_rate": 1e-05,
      "loss": 3.1321,
      "step": 2415
    },
    {
      "epoch": 0.455749243752869,
      "grad_norm": 7.631880283355713,
      "learning_rate": 1e-05,
      "loss": 3.1185,
      "step": 2420
    },
    {
      "epoch": 0.4566908744217799,
      "grad_norm": 7.6734466552734375,
      "learning_rate": 1e-05,
      "loss": 3.0592,
      "step": 2425
    },
    {
      "epoch": 0.4576325050906908,
      "grad_norm": 6.934384346008301,
      "learning_rate": 1e-05,
      "loss": 3.1555,
      "step": 2430
    },
    {
      "epoch": 0.4585741357596017,
      "grad_norm": 7.925302505493164,
      "learning_rate": 1e-05,
      "loss": 3.0891,
      "step": 2435
    },
    {
      "epoch": 0.45951576642851255,
      "grad_norm": 7.213337421417236,
      "learning_rate": 1e-05,
      "loss": 2.9779,
      "step": 2440
    },
    {
      "epoch": 0.46045739709742345,
      "grad_norm": 7.504617691040039,
      "learning_rate": 1e-05,
      "loss": 3.0613,
      "step": 2445
    },
    {
      "epoch": 0.46139902776633435,
      "grad_norm": 7.625884056091309,
      "learning_rate": 1e-05,
      "loss": 2.9949,
      "step": 2450
    },
    {
      "epoch": 0.46234065843524524,
      "grad_norm": 7.4358696937561035,
      "learning_rate": 1e-05,
      "loss": 3.1089,
      "step": 2455
    },
    {
      "epoch": 0.46328228910415614,
      "grad_norm": 7.1684489250183105,
      "learning_rate": 1e-05,
      "loss": 2.9965,
      "step": 2460
    },
    {
      "epoch": 0.464223919773067,
      "grad_norm": 7.672368049621582,
      "learning_rate": 1e-05,
      "loss": 3.0439,
      "step": 2465
    },
    {
      "epoch": 0.4651655504419779,
      "grad_norm": 7.329806327819824,
      "learning_rate": 1e-05,
      "loss": 3.0025,
      "step": 2470
    },
    {
      "epoch": 0.4661071811108888,
      "grad_norm": 7.104738235473633,
      "learning_rate": 1e-05,
      "loss": 3.0738,
      "step": 2475
    },
    {
      "epoch": 0.4670488117797997,
      "grad_norm": 6.561854839324951,
      "learning_rate": 1e-05,
      "loss": 2.9749,
      "step": 2480
    },
    {
      "epoch": 0.4679904424487106,
      "grad_norm": 7.665432453155518,
      "learning_rate": 1e-05,
      "loss": 3.0647,
      "step": 2485
    },
    {
      "epoch": 0.4689320731176214,
      "grad_norm": 7.229370594024658,
      "learning_rate": 1e-05,
      "loss": 3.0231,
      "step": 2490
    },
    {
      "epoch": 0.4698737037865323,
      "grad_norm": 7.576310157775879,
      "learning_rate": 1e-05,
      "loss": 3.0132,
      "step": 2495
    },
    {
      "epoch": 0.4708153344554432,
      "grad_norm": 6.800790786743164,
      "learning_rate": 1e-05,
      "loss": 3.0343,
      "step": 2500
    },
    {
      "epoch": 0.4717569651243541,
      "grad_norm": 7.0642266273498535,
      "learning_rate": 1e-05,
      "loss": 3.0681,
      "step": 2505
    },
    {
      "epoch": 0.472698595793265,
      "grad_norm": 6.809929370880127,
      "learning_rate": 1e-05,
      "loss": 3.1242,
      "step": 2510
    },
    {
      "epoch": 0.47364022646217585,
      "grad_norm": 7.387331008911133,
      "learning_rate": 1e-05,
      "loss": 3.0638,
      "step": 2515
    },
    {
      "epoch": 0.47458185713108675,
      "grad_norm": 7.186639308929443,
      "learning_rate": 1e-05,
      "loss": 3.0472,
      "step": 2520
    },
    {
      "epoch": 0.47552348779999765,
      "grad_norm": 7.330045223236084,
      "learning_rate": 1e-05,
      "loss": 3.1256,
      "step": 2525
    },
    {
      "epoch": 0.47646511846890854,
      "grad_norm": 7.658766746520996,
      "learning_rate": 1e-05,
      "loss": 3.0334,
      "step": 2530
    },
    {
      "epoch": 0.47740674913781944,
      "grad_norm": 7.821640968322754,
      "learning_rate": 1e-05,
      "loss": 3.1666,
      "step": 2535
    },
    {
      "epoch": 0.4783483798067303,
      "grad_norm": 8.806328773498535,
      "learning_rate": 1e-05,
      "loss": 3.1129,
      "step": 2540
    },
    {
      "epoch": 0.4792900104756412,
      "grad_norm": 7.119672775268555,
      "learning_rate": 1e-05,
      "loss": 3.0084,
      "step": 2545
    },
    {
      "epoch": 0.4802316411445521,
      "grad_norm": 7.20904541015625,
      "learning_rate": 1e-05,
      "loss": 3.0855,
      "step": 2550
    },
    {
      "epoch": 0.481173271813463,
      "grad_norm": 7.136516571044922,
      "learning_rate": 1e-05,
      "loss": 2.9989,
      "step": 2555
    },
    {
      "epoch": 0.4821149024823739,
      "grad_norm": 6.163409233093262,
      "learning_rate": 1e-05,
      "loss": 2.9992,
      "step": 2560
    },
    {
      "epoch": 0.4830565331512847,
      "grad_norm": 7.667194843292236,
      "learning_rate": 1e-05,
      "loss": 2.996,
      "step": 2565
    },
    {
      "epoch": 0.4839981638201956,
      "grad_norm": 7.100743293762207,
      "learning_rate": 1e-05,
      "loss": 3.0643,
      "step": 2570
    },
    {
      "epoch": 0.4849397944891065,
      "grad_norm": 7.50702428817749,
      "learning_rate": 1e-05,
      "loss": 3.1647,
      "step": 2575
    },
    {
      "epoch": 0.4858814251580174,
      "grad_norm": 7.247680187225342,
      "learning_rate": 1e-05,
      "loss": 3.0211,
      "step": 2580
    },
    {
      "epoch": 0.4868230558269283,
      "grad_norm": 7.403180122375488,
      "learning_rate": 1e-05,
      "loss": 3.0319,
      "step": 2585
    },
    {
      "epoch": 0.48776468649583915,
      "grad_norm": 7.5679473876953125,
      "learning_rate": 1e-05,
      "loss": 3.0748,
      "step": 2590
    },
    {
      "epoch": 0.48870631716475005,
      "grad_norm": 7.389653205871582,
      "learning_rate": 1e-05,
      "loss": 3.0973,
      "step": 2595
    },
    {
      "epoch": 0.48964794783366095,
      "grad_norm": 12.070070266723633,
      "learning_rate": 1e-05,
      "loss": 3.0398,
      "step": 2600
    },
    {
      "epoch": 0.49058957850257184,
      "grad_norm": 7.190162658691406,
      "learning_rate": 1e-05,
      "loss": 3.16,
      "step": 2605
    },
    {
      "epoch": 0.49153120917148274,
      "grad_norm": 6.753963470458984,
      "learning_rate": 1e-05,
      "loss": 3.0305,
      "step": 2610
    },
    {
      "epoch": 0.4924728398403936,
      "grad_norm": 7.3352789878845215,
      "learning_rate": 1e-05,
      "loss": 3.0226,
      "step": 2615
    },
    {
      "epoch": 0.4934144705093045,
      "grad_norm": 6.455969333648682,
      "learning_rate": 1e-05,
      "loss": 3.0626,
      "step": 2620
    },
    {
      "epoch": 0.4943561011782154,
      "grad_norm": 6.789318561553955,
      "learning_rate": 1e-05,
      "loss": 3.1405,
      "step": 2625
    },
    {
      "epoch": 0.4952977318471263,
      "grad_norm": 7.1296186447143555,
      "learning_rate": 1e-05,
      "loss": 2.9996,
      "step": 2630
    },
    {
      "epoch": 0.4962393625160372,
      "grad_norm": 7.188891887664795,
      "learning_rate": 1e-05,
      "loss": 3.0108,
      "step": 2635
    },
    {
      "epoch": 0.497180993184948,
      "grad_norm": 8.604632377624512,
      "learning_rate": 1e-05,
      "loss": 3.1121,
      "step": 2640
    },
    {
      "epoch": 0.4981226238538589,
      "grad_norm": 8.063715934753418,
      "learning_rate": 1e-05,
      "loss": 3.1481,
      "step": 2645
    },
    {
      "epoch": 0.4990642545227698,
      "grad_norm": 7.208702087402344,
      "learning_rate": 1e-05,
      "loss": 3.1499,
      "step": 2650
    },
    {
      "epoch": 0.5000058851916807,
      "grad_norm": 6.898314952850342,
      "learning_rate": 1e-05,
      "loss": 3.0624,
      "step": 2655
    },
    {
      "epoch": 0.5009475158605916,
      "grad_norm": 7.062436103820801,
      "learning_rate": 1e-05,
      "loss": 2.9495,
      "step": 2660
    },
    {
      "epoch": 0.5018891465295025,
      "grad_norm": 7.108069896697998,
      "learning_rate": 1e-05,
      "loss": 2.9833,
      "step": 2665
    },
    {
      "epoch": 0.5028307771984134,
      "grad_norm": 8.10451602935791,
      "learning_rate": 1e-05,
      "loss": 3.1446,
      "step": 2670
    },
    {
      "epoch": 0.5037724078673242,
      "grad_norm": 7.409706115722656,
      "learning_rate": 1e-05,
      "loss": 3.1812,
      "step": 2675
    },
    {
      "epoch": 0.5047140385362351,
      "grad_norm": 7.078268527984619,
      "learning_rate": 1e-05,
      "loss": 3.0981,
      "step": 2680
    },
    {
      "epoch": 0.505655669205146,
      "grad_norm": 6.625565052032471,
      "learning_rate": 1e-05,
      "loss": 3.0404,
      "step": 2685
    },
    {
      "epoch": 0.5065972998740569,
      "grad_norm": 7.314438819885254,
      "learning_rate": 1e-05,
      "loss": 3.0807,
      "step": 2690
    },
    {
      "epoch": 0.5075389305429678,
      "grad_norm": 6.884581565856934,
      "learning_rate": 1e-05,
      "loss": 3.0895,
      "step": 2695
    },
    {
      "epoch": 0.5084805612118787,
      "grad_norm": 7.702692031860352,
      "learning_rate": 1e-05,
      "loss": 3.0731,
      "step": 2700
    },
    {
      "epoch": 0.5094221918807895,
      "grad_norm": 7.680057525634766,
      "learning_rate": 1e-05,
      "loss": 3.0837,
      "step": 2705
    },
    {
      "epoch": 0.5103638225497005,
      "grad_norm": 7.11198616027832,
      "learning_rate": 1e-05,
      "loss": 2.9995,
      "step": 2710
    },
    {
      "epoch": 0.5113054532186113,
      "grad_norm": 7.569001197814941,
      "learning_rate": 1e-05,
      "loss": 3.1655,
      "step": 2715
    },
    {
      "epoch": 0.5122470838875223,
      "grad_norm": 7.120611667633057,
      "learning_rate": 1e-05,
      "loss": 3.0245,
      "step": 2720
    },
    {
      "epoch": 0.5131887145564331,
      "grad_norm": 7.674874305725098,
      "learning_rate": 1e-05,
      "loss": 3.0993,
      "step": 2725
    },
    {
      "epoch": 0.514130345225344,
      "grad_norm": 7.820282459259033,
      "learning_rate": 1e-05,
      "loss": 3.0897,
      "step": 2730
    },
    {
      "epoch": 0.5150719758942549,
      "grad_norm": 7.010729789733887,
      "learning_rate": 1e-05,
      "loss": 2.9622,
      "step": 2735
    },
    {
      "epoch": 0.5160136065631657,
      "grad_norm": 7.443545818328857,
      "learning_rate": 1e-05,
      "loss": 3.1474,
      "step": 2740
    },
    {
      "epoch": 0.5169552372320767,
      "grad_norm": 8.485078811645508,
      "learning_rate": 1e-05,
      "loss": 3.095,
      "step": 2745
    },
    {
      "epoch": 0.5178968679009875,
      "grad_norm": 7.487083435058594,
      "learning_rate": 1e-05,
      "loss": 3.1509,
      "step": 2750
    },
    {
      "epoch": 0.5188384985698984,
      "grad_norm": 7.581494331359863,
      "learning_rate": 1e-05,
      "loss": 3.0771,
      "step": 2755
    },
    {
      "epoch": 0.5197801292388093,
      "grad_norm": 7.041471481323242,
      "learning_rate": 1e-05,
      "loss": 2.9578,
      "step": 2760
    },
    {
      "epoch": 0.5207217599077202,
      "grad_norm": 6.6093034744262695,
      "learning_rate": 1e-05,
      "loss": 3.0864,
      "step": 2765
    },
    {
      "epoch": 0.5216633905766311,
      "grad_norm": 6.595880031585693,
      "learning_rate": 1e-05,
      "loss": 2.9654,
      "step": 2770
    },
    {
      "epoch": 0.522605021245542,
      "grad_norm": 7.0556511878967285,
      "learning_rate": 1e-05,
      "loss": 2.9616,
      "step": 2775
    },
    {
      "epoch": 0.5235466519144528,
      "grad_norm": 7.266999244689941,
      "learning_rate": 1e-05,
      "loss": 3.1041,
      "step": 2780
    },
    {
      "epoch": 0.5244882825833638,
      "grad_norm": 7.564467430114746,
      "learning_rate": 1e-05,
      "loss": 3.1307,
      "step": 2785
    },
    {
      "epoch": 0.5254299132522746,
      "grad_norm": 6.620129108428955,
      "learning_rate": 1e-05,
      "loss": 3.0366,
      "step": 2790
    },
    {
      "epoch": 0.5263715439211855,
      "grad_norm": 6.7112321853637695,
      "learning_rate": 1e-05,
      "loss": 3.0846,
      "step": 2795
    },
    {
      "epoch": 0.5273131745900964,
      "grad_norm": 8.041632652282715,
      "learning_rate": 1e-05,
      "loss": 3.161,
      "step": 2800
    },
    {
      "epoch": 0.5282548052590073,
      "grad_norm": 7.796072006225586,
      "learning_rate": 1e-05,
      "loss": 3.0004,
      "step": 2805
    },
    {
      "epoch": 0.5291964359279182,
      "grad_norm": 7.121657371520996,
      "learning_rate": 1e-05,
      "loss": 3.0108,
      "step": 2810
    },
    {
      "epoch": 0.530138066596829,
      "grad_norm": 7.073033809661865,
      "learning_rate": 1e-05,
      "loss": 3.0985,
      "step": 2815
    },
    {
      "epoch": 0.5310796972657399,
      "grad_norm": 6.580442905426025,
      "learning_rate": 1e-05,
      "loss": 3.081,
      "step": 2820
    },
    {
      "epoch": 0.5320213279346508,
      "grad_norm": 7.59156608581543,
      "learning_rate": 1e-05,
      "loss": 3.1838,
      "step": 2825
    },
    {
      "epoch": 0.5329629586035617,
      "grad_norm": 6.721426963806152,
      "learning_rate": 1e-05,
      "loss": 3.12,
      "step": 2830
    },
    {
      "epoch": 0.5339045892724726,
      "grad_norm": 6.488240718841553,
      "learning_rate": 1e-05,
      "loss": 3.0373,
      "step": 2835
    },
    {
      "epoch": 0.5348462199413835,
      "grad_norm": 7.797455787658691,
      "learning_rate": 1e-05,
      "loss": 3.0716,
      "step": 2840
    },
    {
      "epoch": 0.5357878506102943,
      "grad_norm": 7.475494861602783,
      "learning_rate": 1e-05,
      "loss": 2.9321,
      "step": 2845
    },
    {
      "epoch": 0.5367294812792053,
      "grad_norm": 7.374056339263916,
      "learning_rate": 1e-05,
      "loss": 3.1087,
      "step": 2850
    },
    {
      "epoch": 0.5376711119481161,
      "grad_norm": 6.967336654663086,
      "learning_rate": 1e-05,
      "loss": 3.0788,
      "step": 2855
    },
    {
      "epoch": 0.5386127426170271,
      "grad_norm": 7.319347858428955,
      "learning_rate": 1e-05,
      "loss": 3.1399,
      "step": 2860
    },
    {
      "epoch": 0.5395543732859379,
      "grad_norm": 7.5937957763671875,
      "learning_rate": 1e-05,
      "loss": 3.0926,
      "step": 2865
    },
    {
      "epoch": 0.5404960039548488,
      "grad_norm": 7.879610538482666,
      "learning_rate": 1e-05,
      "loss": 3.0853,
      "step": 2870
    },
    {
      "epoch": 0.5414376346237597,
      "grad_norm": 7.16407585144043,
      "learning_rate": 1e-05,
      "loss": 2.8899,
      "step": 2875
    },
    {
      "epoch": 0.5423792652926706,
      "grad_norm": 7.709092617034912,
      "learning_rate": 1e-05,
      "loss": 3.1303,
      "step": 2880
    },
    {
      "epoch": 0.5433208959615815,
      "grad_norm": 7.1035356521606445,
      "learning_rate": 1e-05,
      "loss": 3.1267,
      "step": 2885
    },
    {
      "epoch": 0.5442625266304923,
      "grad_norm": 6.998748302459717,
      "learning_rate": 1e-05,
      "loss": 3.1023,
      "step": 2890
    },
    {
      "epoch": 0.5452041572994032,
      "grad_norm": 7.0011186599731445,
      "learning_rate": 1e-05,
      "loss": 3.1474,
      "step": 2895
    },
    {
      "epoch": 0.5461457879683141,
      "grad_norm": 6.540005207061768,
      "learning_rate": 1e-05,
      "loss": 2.9699,
      "step": 2900
    },
    {
      "epoch": 0.547087418637225,
      "grad_norm": 7.625404357910156,
      "learning_rate": 1e-05,
      "loss": 3.2087,
      "step": 2905
    },
    {
      "epoch": 0.5480290493061359,
      "grad_norm": 8.491080284118652,
      "learning_rate": 1e-05,
      "loss": 3.0275,
      "step": 2910
    },
    {
      "epoch": 0.5489706799750468,
      "grad_norm": 7.326952934265137,
      "learning_rate": 1e-05,
      "loss": 3.119,
      "step": 2915
    },
    {
      "epoch": 0.5499123106439576,
      "grad_norm": 6.557469367980957,
      "learning_rate": 1e-05,
      "loss": 3.0491,
      "step": 2920
    },
    {
      "epoch": 0.5508539413128686,
      "grad_norm": 8.412656784057617,
      "learning_rate": 1e-05,
      "loss": 2.9752,
      "step": 2925
    },
    {
      "epoch": 0.5517955719817794,
      "grad_norm": 7.805161476135254,
      "learning_rate": 1e-05,
      "loss": 3.188,
      "step": 2930
    },
    {
      "epoch": 0.5527372026506904,
      "grad_norm": 7.706869602203369,
      "learning_rate": 1e-05,
      "loss": 3.01,
      "step": 2935
    },
    {
      "epoch": 0.5536788333196012,
      "grad_norm": 6.797297477722168,
      "learning_rate": 1e-05,
      "loss": 2.9473,
      "step": 2940
    },
    {
      "epoch": 0.5546204639885121,
      "grad_norm": 7.4034624099731445,
      "learning_rate": 1e-05,
      "loss": 3.0252,
      "step": 2945
    },
    {
      "epoch": 0.555562094657423,
      "grad_norm": 8.384190559387207,
      "learning_rate": 1e-05,
      "loss": 3.0013,
      "step": 2950
    },
    {
      "epoch": 0.5565037253263339,
      "grad_norm": 7.6911187171936035,
      "learning_rate": 1e-05,
      "loss": 3.1842,
      "step": 2955
    },
    {
      "epoch": 0.5574453559952448,
      "grad_norm": 6.948124408721924,
      "learning_rate": 1e-05,
      "loss": 2.9562,
      "step": 2960
    },
    {
      "epoch": 0.5583869866641556,
      "grad_norm": 7.233328819274902,
      "learning_rate": 1e-05,
      "loss": 3.0687,
      "step": 2965
    },
    {
      "epoch": 0.5593286173330665,
      "grad_norm": 7.2016425132751465,
      "learning_rate": 1e-05,
      "loss": 3.0819,
      "step": 2970
    },
    {
      "epoch": 0.5602702480019774,
      "grad_norm": 6.626002311706543,
      "learning_rate": 1e-05,
      "loss": 3.078,
      "step": 2975
    },
    {
      "epoch": 0.5612118786708883,
      "grad_norm": 7.238675117492676,
      "learning_rate": 1e-05,
      "loss": 3.1172,
      "step": 2980
    },
    {
      "epoch": 0.5621535093397992,
      "grad_norm": 6.863603115081787,
      "learning_rate": 1e-05,
      "loss": 3.0753,
      "step": 2985
    },
    {
      "epoch": 0.5630951400087101,
      "grad_norm": 7.731668949127197,
      "learning_rate": 1e-05,
      "loss": 2.9837,
      "step": 2990
    },
    {
      "epoch": 0.5640367706776209,
      "grad_norm": 6.4692702293396,
      "learning_rate": 1e-05,
      "loss": 3.1187,
      "step": 2995
    },
    {
      "epoch": 0.5649784013465319,
      "grad_norm": 7.253861427307129,
      "learning_rate": 1e-05,
      "loss": 2.9771,
      "step": 3000
    },
    {
      "epoch": 0.5649784013465319,
      "eval_accuracy": 0.4138860273852878,
      "eval_loss": 3.0637450218200684,
      "eval_runtime": 1049.1644,
      "eval_samples_per_second": 35.991,
      "eval_steps_per_second": 8.998,
      "step": 3000
    },
    {
      "epoch": 0.5659200320154427,
      "grad_norm": 7.105886459350586,
      "learning_rate": 1e-05,
      "loss": 3.0877,
      "step": 3005
    },
    {
      "epoch": 0.5668616626843537,
      "grad_norm": 6.724785327911377,
      "learning_rate": 1e-05,
      "loss": 3.0216,
      "step": 3010
    },
    {
      "epoch": 0.5678032933532645,
      "grad_norm": 6.603315353393555,
      "learning_rate": 1e-05,
      "loss": 3.1276,
      "step": 3015
    },
    {
      "epoch": 0.5687449240221754,
      "grad_norm": 6.701154708862305,
      "learning_rate": 1e-05,
      "loss": 2.9892,
      "step": 3020
    },
    {
      "epoch": 0.5696865546910863,
      "grad_norm": 6.806232929229736,
      "learning_rate": 1e-05,
      "loss": 3.0323,
      "step": 3025
    },
    {
      "epoch": 0.5706281853599972,
      "grad_norm": 7.004578113555908,
      "learning_rate": 1e-05,
      "loss": 3.0609,
      "step": 3030
    },
    {
      "epoch": 0.5715698160289081,
      "grad_norm": 6.511326789855957,
      "learning_rate": 1e-05,
      "loss": 3.1092,
      "step": 3035
    },
    {
      "epoch": 0.572511446697819,
      "grad_norm": 7.045777320861816,
      "learning_rate": 1e-05,
      "loss": 3.1217,
      "step": 3040
    },
    {
      "epoch": 0.5734530773667298,
      "grad_norm": 6.780973434448242,
      "learning_rate": 1e-05,
      "loss": 3.0726,
      "step": 3045
    },
    {
      "epoch": 0.5743947080356407,
      "grad_norm": 7.139321804046631,
      "learning_rate": 1e-05,
      "loss": 3.065,
      "step": 3050
    },
    {
      "epoch": 0.5753363387045516,
      "grad_norm": 7.129504203796387,
      "learning_rate": 1e-05,
      "loss": 3.1518,
      "step": 3055
    },
    {
      "epoch": 0.5762779693734625,
      "grad_norm": 8.042350769042969,
      "learning_rate": 1e-05,
      "loss": 3.0006,
      "step": 3060
    },
    {
      "epoch": 0.5772196000423734,
      "grad_norm": 6.995969772338867,
      "learning_rate": 1e-05,
      "loss": 3.0849,
      "step": 3065
    },
    {
      "epoch": 0.5781612307112842,
      "grad_norm": 6.369194030761719,
      "learning_rate": 1e-05,
      "loss": 2.9541,
      "step": 3070
    },
    {
      "epoch": 0.5791028613801952,
      "grad_norm": 6.521533012390137,
      "learning_rate": 1e-05,
      "loss": 3.0632,
      "step": 3075
    },
    {
      "epoch": 0.580044492049106,
      "grad_norm": 7.185681343078613,
      "learning_rate": 1e-05,
      "loss": 3.1407,
      "step": 3080
    },
    {
      "epoch": 0.580986122718017,
      "grad_norm": 6.984232425689697,
      "learning_rate": 1e-05,
      "loss": 3.1047,
      "step": 3085
    },
    {
      "epoch": 0.5819277533869278,
      "grad_norm": 7.491443634033203,
      "learning_rate": 1e-05,
      "loss": 3.0165,
      "step": 3090
    },
    {
      "epoch": 0.5828693840558387,
      "grad_norm": 6.652406215667725,
      "learning_rate": 1e-05,
      "loss": 3.1079,
      "step": 3095
    },
    {
      "epoch": 0.5838110147247496,
      "grad_norm": 7.801830291748047,
      "learning_rate": 1e-05,
      "loss": 3.1699,
      "step": 3100
    },
    {
      "epoch": 0.5847526453936605,
      "grad_norm": 7.229469299316406,
      "learning_rate": 1e-05,
      "loss": 3.0419,
      "step": 3105
    },
    {
      "epoch": 0.5856942760625714,
      "grad_norm": 6.888019561767578,
      "learning_rate": 1e-05,
      "loss": 2.9814,
      "step": 3110
    },
    {
      "epoch": 0.5866359067314822,
      "grad_norm": 7.353707790374756,
      "learning_rate": 1e-05,
      "loss": 3.0864,
      "step": 3115
    },
    {
      "epoch": 0.5875775374003931,
      "grad_norm": 7.126093864440918,
      "learning_rate": 1e-05,
      "loss": 3.1298,
      "step": 3120
    },
    {
      "epoch": 0.588519168069304,
      "grad_norm": 7.8923211097717285,
      "learning_rate": 1e-05,
      "loss": 3.0013,
      "step": 3125
    },
    {
      "epoch": 0.5894607987382149,
      "grad_norm": 6.3733649253845215,
      "learning_rate": 1e-05,
      "loss": 2.9838,
      "step": 3130
    },
    {
      "epoch": 0.5904024294071258,
      "grad_norm": 7.295039176940918,
      "learning_rate": 1e-05,
      "loss": 3.1155,
      "step": 3135
    },
    {
      "epoch": 0.5913440600760367,
      "grad_norm": 7.227429389953613,
      "learning_rate": 1e-05,
      "loss": 3.1303,
      "step": 3140
    },
    {
      "epoch": 0.5922856907449475,
      "grad_norm": 7.039018630981445,
      "learning_rate": 1e-05,
      "loss": 2.9929,
      "step": 3145
    },
    {
      "epoch": 0.5932273214138585,
      "grad_norm": 8.034615516662598,
      "learning_rate": 1e-05,
      "loss": 3.0979,
      "step": 3150
    },
    {
      "epoch": 0.5941689520827693,
      "grad_norm": 7.214796543121338,
      "learning_rate": 1e-05,
      "loss": 3.0098,
      "step": 3155
    },
    {
      "epoch": 0.5951105827516803,
      "grad_norm": 7.179065227508545,
      "learning_rate": 1e-05,
      "loss": 3.0168,
      "step": 3160
    },
    {
      "epoch": 0.5960522134205911,
      "grad_norm": 7.219829559326172,
      "learning_rate": 1e-05,
      "loss": 2.9791,
      "step": 3165
    },
    {
      "epoch": 0.596993844089502,
      "grad_norm": 7.285015106201172,
      "learning_rate": 1e-05,
      "loss": 2.9036,
      "step": 3170
    },
    {
      "epoch": 0.5979354747584129,
      "grad_norm": 8.418420791625977,
      "learning_rate": 1e-05,
      "loss": 2.9316,
      "step": 3175
    },
    {
      "epoch": 0.5988771054273238,
      "grad_norm": 7.653911590576172,
      "learning_rate": 1e-05,
      "loss": 3.0915,
      "step": 3180
    },
    {
      "epoch": 0.5998187360962347,
      "grad_norm": 6.934592247009277,
      "learning_rate": 1e-05,
      "loss": 2.9547,
      "step": 3185
    },
    {
      "epoch": 0.6007603667651455,
      "grad_norm": 6.826767444610596,
      "learning_rate": 1e-05,
      "loss": 2.9792,
      "step": 3190
    },
    {
      "epoch": 0.6017019974340564,
      "grad_norm": 7.79368782043457,
      "learning_rate": 1e-05,
      "loss": 3.0121,
      "step": 3195
    },
    {
      "epoch": 0.6026436281029673,
      "grad_norm": 7.679154872894287,
      "learning_rate": 1e-05,
      "loss": 3.1288,
      "step": 3200
    },
    {
      "epoch": 0.6035852587718782,
      "grad_norm": 7.1232781410217285,
      "learning_rate": 1e-05,
      "loss": 3.0423,
      "step": 3205
    },
    {
      "epoch": 0.6045268894407891,
      "grad_norm": 7.5240983963012695,
      "learning_rate": 1e-05,
      "loss": 3.056,
      "step": 3210
    },
    {
      "epoch": 0.6054685201097,
      "grad_norm": 7.4613237380981445,
      "learning_rate": 1e-05,
      "loss": 3.0777,
      "step": 3215
    },
    {
      "epoch": 0.6064101507786108,
      "grad_norm": 6.664481163024902,
      "learning_rate": 1e-05,
      "loss": 3.0872,
      "step": 3220
    },
    {
      "epoch": 0.6073517814475218,
      "grad_norm": 6.648809432983398,
      "learning_rate": 1e-05,
      "loss": 3.1206,
      "step": 3225
    },
    {
      "epoch": 0.6082934121164326,
      "grad_norm": 6.968840599060059,
      "learning_rate": 1e-05,
      "loss": 3.052,
      "step": 3230
    },
    {
      "epoch": 0.6092350427853436,
      "grad_norm": 7.363081932067871,
      "learning_rate": 1e-05,
      "loss": 3.1386,
      "step": 3235
    },
    {
      "epoch": 0.6101766734542544,
      "grad_norm": 6.684266567230225,
      "learning_rate": 1e-05,
      "loss": 2.9426,
      "step": 3240
    },
    {
      "epoch": 0.6111183041231653,
      "grad_norm": 7.6497650146484375,
      "learning_rate": 1e-05,
      "loss": 2.9756,
      "step": 3245
    },
    {
      "epoch": 0.6120599347920762,
      "grad_norm": 7.052670955657959,
      "learning_rate": 1e-05,
      "loss": 3.0383,
      "step": 3250
    },
    {
      "epoch": 0.613001565460987,
      "grad_norm": 6.5792083740234375,
      "learning_rate": 1e-05,
      "loss": 2.8681,
      "step": 3255
    },
    {
      "epoch": 0.613943196129898,
      "grad_norm": 7.075092315673828,
      "learning_rate": 1e-05,
      "loss": 3.0287,
      "step": 3260
    },
    {
      "epoch": 0.6148848267988088,
      "grad_norm": 7.760746002197266,
      "learning_rate": 1e-05,
      "loss": 3.0866,
      "step": 3265
    },
    {
      "epoch": 0.6158264574677197,
      "grad_norm": 7.239394664764404,
      "learning_rate": 1e-05,
      "loss": 2.9918,
      "step": 3270
    },
    {
      "epoch": 0.6167680881366306,
      "grad_norm": 7.3641276359558105,
      "learning_rate": 1e-05,
      "loss": 2.9678,
      "step": 3275
    },
    {
      "epoch": 0.6177097188055415,
      "grad_norm": 10.26201057434082,
      "learning_rate": 1e-05,
      "loss": 3.1028,
      "step": 3280
    },
    {
      "epoch": 0.6186513494744523,
      "grad_norm": 6.868218421936035,
      "learning_rate": 1e-05,
      "loss": 3.0633,
      "step": 3285
    },
    {
      "epoch": 0.6195929801433633,
      "grad_norm": 6.771929740905762,
      "learning_rate": 1e-05,
      "loss": 3.0126,
      "step": 3290
    },
    {
      "epoch": 0.6205346108122741,
      "grad_norm": 7.173276901245117,
      "learning_rate": 1e-05,
      "loss": 3.0549,
      "step": 3295
    },
    {
      "epoch": 0.6214762414811851,
      "grad_norm": 6.571986198425293,
      "learning_rate": 1e-05,
      "loss": 3.0469,
      "step": 3300
    },
    {
      "epoch": 0.6224178721500959,
      "grad_norm": 8.061853408813477,
      "learning_rate": 1e-05,
      "loss": 3.0797,
      "step": 3305
    },
    {
      "epoch": 0.6233595028190068,
      "grad_norm": 6.856041431427002,
      "learning_rate": 1e-05,
      "loss": 3.0835,
      "step": 3310
    },
    {
      "epoch": 0.6243011334879177,
      "grad_norm": 6.356860637664795,
      "learning_rate": 1e-05,
      "loss": 3.0321,
      "step": 3315
    },
    {
      "epoch": 0.6252427641568286,
      "grad_norm": 7.281395435333252,
      "learning_rate": 1e-05,
      "loss": 2.9796,
      "step": 3320
    },
    {
      "epoch": 0.6261843948257395,
      "grad_norm": 7.561255931854248,
      "learning_rate": 1e-05,
      "loss": 3.0791,
      "step": 3325
    },
    {
      "epoch": 0.6271260254946504,
      "grad_norm": 5.9859185218811035,
      "learning_rate": 1e-05,
      "loss": 3.0435,
      "step": 3330
    },
    {
      "epoch": 0.6280676561635612,
      "grad_norm": 7.344315528869629,
      "learning_rate": 1e-05,
      "loss": 3.0616,
      "step": 3335
    },
    {
      "epoch": 0.6290092868324721,
      "grad_norm": 7.146785736083984,
      "learning_rate": 1e-05,
      "loss": 3.0322,
      "step": 3340
    },
    {
      "epoch": 0.629950917501383,
      "grad_norm": 7.084141731262207,
      "learning_rate": 1e-05,
      "loss": 3.0367,
      "step": 3345
    },
    {
      "epoch": 0.6308925481702939,
      "grad_norm": 7.505505084991455,
      "learning_rate": 1e-05,
      "loss": 3.052,
      "step": 3350
    },
    {
      "epoch": 0.6318341788392048,
      "grad_norm": 7.202043533325195,
      "learning_rate": 1e-05,
      "loss": 3.0902,
      "step": 3355
    },
    {
      "epoch": 0.6327758095081156,
      "grad_norm": 6.772159099578857,
      "learning_rate": 1e-05,
      "loss": 3.015,
      "step": 3360
    },
    {
      "epoch": 0.6337174401770266,
      "grad_norm": 6.495903968811035,
      "learning_rate": 1e-05,
      "loss": 3.0254,
      "step": 3365
    },
    {
      "epoch": 0.6346590708459374,
      "grad_norm": 7.2963457107543945,
      "learning_rate": 1e-05,
      "loss": 3.0161,
      "step": 3370
    },
    {
      "epoch": 0.6356007015148484,
      "grad_norm": 7.240870475769043,
      "learning_rate": 1e-05,
      "loss": 3.0329,
      "step": 3375
    },
    {
      "epoch": 0.6365423321837592,
      "grad_norm": 6.670111656188965,
      "learning_rate": 1e-05,
      "loss": 2.9956,
      "step": 3380
    },
    {
      "epoch": 0.6374839628526701,
      "grad_norm": 7.213159561157227,
      "learning_rate": 1e-05,
      "loss": 2.9224,
      "step": 3385
    },
    {
      "epoch": 0.638425593521581,
      "grad_norm": 7.563082218170166,
      "learning_rate": 1e-05,
      "loss": 3.14,
      "step": 3390
    },
    {
      "epoch": 0.6393672241904919,
      "grad_norm": 7.692990779876709,
      "learning_rate": 1e-05,
      "loss": 2.9937,
      "step": 3395
    },
    {
      "epoch": 0.6403088548594028,
      "grad_norm": 7.73374605178833,
      "learning_rate": 1e-05,
      "loss": 3.0108,
      "step": 3400
    },
    {
      "epoch": 0.6412504855283137,
      "grad_norm": 6.7092976570129395,
      "learning_rate": 1e-05,
      "loss": 3.0887,
      "step": 3405
    },
    {
      "epoch": 0.6421921161972245,
      "grad_norm": 7.405087471008301,
      "learning_rate": 1e-05,
      "loss": 2.9877,
      "step": 3410
    },
    {
      "epoch": 0.6431337468661354,
      "grad_norm": 7.017783164978027,
      "learning_rate": 1e-05,
      "loss": 3.0624,
      "step": 3415
    },
    {
      "epoch": 0.6440753775350463,
      "grad_norm": 6.923324108123779,
      "learning_rate": 1e-05,
      "loss": 2.9223,
      "step": 3420
    },
    {
      "epoch": 0.6450170082039572,
      "grad_norm": 7.183239936828613,
      "learning_rate": 1e-05,
      "loss": 3.1286,
      "step": 3425
    },
    {
      "epoch": 0.6459586388728681,
      "grad_norm": 7.303028106689453,
      "learning_rate": 1e-05,
      "loss": 3.0108,
      "step": 3430
    },
    {
      "epoch": 0.6469002695417789,
      "grad_norm": 6.967830181121826,
      "learning_rate": 1e-05,
      "loss": 3.0991,
      "step": 3435
    },
    {
      "epoch": 0.6478419002106899,
      "grad_norm": 7.331003189086914,
      "learning_rate": 1e-05,
      "loss": 3.1454,
      "step": 3440
    },
    {
      "epoch": 0.6487835308796007,
      "grad_norm": 7.24896764755249,
      "learning_rate": 1e-05,
      "loss": 2.9953,
      "step": 3445
    },
    {
      "epoch": 0.6497251615485117,
      "grad_norm": 8.16205883026123,
      "learning_rate": 1e-05,
      "loss": 2.9901,
      "step": 3450
    },
    {
      "epoch": 0.6506667922174225,
      "grad_norm": 7.7972412109375,
      "learning_rate": 1e-05,
      "loss": 3.1011,
      "step": 3455
    },
    {
      "epoch": 0.6516084228863334,
      "grad_norm": 6.8451714515686035,
      "learning_rate": 1e-05,
      "loss": 2.9883,
      "step": 3460
    },
    {
      "epoch": 0.6525500535552443,
      "grad_norm": 6.7832465171813965,
      "learning_rate": 1e-05,
      "loss": 2.9701,
      "step": 3465
    },
    {
      "epoch": 0.6534916842241552,
      "grad_norm": 7.421736717224121,
      "learning_rate": 1e-05,
      "loss": 3.0205,
      "step": 3470
    },
    {
      "epoch": 0.6544333148930661,
      "grad_norm": 7.19790506362915,
      "learning_rate": 1e-05,
      "loss": 2.9935,
      "step": 3475
    },
    {
      "epoch": 0.655374945561977,
      "grad_norm": 7.348209857940674,
      "learning_rate": 1e-05,
      "loss": 2.8969,
      "step": 3480
    },
    {
      "epoch": 0.6563165762308878,
      "grad_norm": 7.604790687561035,
      "learning_rate": 1e-05,
      "loss": 3.0523,
      "step": 3485
    },
    {
      "epoch": 0.6572582068997987,
      "grad_norm": 6.750668525695801,
      "learning_rate": 1e-05,
      "loss": 3.0571,
      "step": 3490
    },
    {
      "epoch": 0.6581998375687096,
      "grad_norm": 6.694459915161133,
      "learning_rate": 1e-05,
      "loss": 3.0252,
      "step": 3495
    },
    {
      "epoch": 0.6591414682376205,
      "grad_norm": 7.250037670135498,
      "learning_rate": 1e-05,
      "loss": 3.0203,
      "step": 3500
    },
    {
      "epoch": 0.6600830989065314,
      "grad_norm": 6.481886863708496,
      "learning_rate": 1e-05,
      "loss": 3.1012,
      "step": 3505
    },
    {
      "epoch": 0.6610247295754422,
      "grad_norm": 7.44242525100708,
      "learning_rate": 1e-05,
      "loss": 3.031,
      "step": 3510
    },
    {
      "epoch": 0.6619663602443532,
      "grad_norm": 6.313452243804932,
      "learning_rate": 1e-05,
      "loss": 3.0551,
      "step": 3515
    },
    {
      "epoch": 0.662907990913264,
      "grad_norm": 7.033621788024902,
      "learning_rate": 1e-05,
      "loss": 3.0459,
      "step": 3520
    },
    {
      "epoch": 0.663849621582175,
      "grad_norm": 6.128735065460205,
      "learning_rate": 1e-05,
      "loss": 2.9017,
      "step": 3525
    },
    {
      "epoch": 0.6647912522510858,
      "grad_norm": 6.252425670623779,
      "learning_rate": 1e-05,
      "loss": 3.0382,
      "step": 3530
    },
    {
      "epoch": 0.6657328829199967,
      "grad_norm": 6.670257091522217,
      "learning_rate": 1e-05,
      "loss": 3.1805,
      "step": 3535
    },
    {
      "epoch": 0.6666745135889076,
      "grad_norm": 6.33909797668457,
      "learning_rate": 1e-05,
      "loss": 3.0568,
      "step": 3540
    },
    {
      "epoch": 0.6676161442578185,
      "grad_norm": 7.0296630859375,
      "learning_rate": 1e-05,
      "loss": 3.0804,
      "step": 3545
    },
    {
      "epoch": 0.6685577749267294,
      "grad_norm": 6.852359294891357,
      "learning_rate": 1e-05,
      "loss": 3.003,
      "step": 3550
    },
    {
      "epoch": 0.6694994055956403,
      "grad_norm": 7.319458961486816,
      "learning_rate": 1e-05,
      "loss": 2.9733,
      "step": 3555
    },
    {
      "epoch": 0.6704410362645511,
      "grad_norm": 7.213205337524414,
      "learning_rate": 1e-05,
      "loss": 3.0515,
      "step": 3560
    },
    {
      "epoch": 0.671382666933462,
      "grad_norm": 6.743834972381592,
      "learning_rate": 1e-05,
      "loss": 2.8611,
      "step": 3565
    },
    {
      "epoch": 0.6723242976023729,
      "grad_norm": 7.25566291809082,
      "learning_rate": 1e-05,
      "loss": 3.0633,
      "step": 3570
    },
    {
      "epoch": 0.6732659282712838,
      "grad_norm": 6.3295578956604,
      "learning_rate": 1e-05,
      "loss": 3.0196,
      "step": 3575
    },
    {
      "epoch": 0.6742075589401947,
      "grad_norm": 7.30181884765625,
      "learning_rate": 1e-05,
      "loss": 3.052,
      "step": 3580
    },
    {
      "epoch": 0.6751491896091055,
      "grad_norm": 6.492465496063232,
      "learning_rate": 1e-05,
      "loss": 2.9969,
      "step": 3585
    },
    {
      "epoch": 0.6760908202780165,
      "grad_norm": 7.620621204376221,
      "learning_rate": 1e-05,
      "loss": 3.0297,
      "step": 3590
    },
    {
      "epoch": 0.6770324509469273,
      "grad_norm": 6.799111843109131,
      "learning_rate": 1e-05,
      "loss": 2.9475,
      "step": 3595
    },
    {
      "epoch": 0.6779740816158383,
      "grad_norm": 7.955085277557373,
      "learning_rate": 1e-05,
      "loss": 2.9165,
      "step": 3600
    },
    {
      "epoch": 0.6789157122847491,
      "grad_norm": 6.649621486663818,
      "learning_rate": 1e-05,
      "loss": 3.0665,
      "step": 3605
    },
    {
      "epoch": 0.67985734295366,
      "grad_norm": 7.301611423492432,
      "learning_rate": 1e-05,
      "loss": 3.0468,
      "step": 3610
    },
    {
      "epoch": 0.6807989736225709,
      "grad_norm": 6.720338821411133,
      "learning_rate": 1e-05,
      "loss": 2.9307,
      "step": 3615
    },
    {
      "epoch": 0.6817406042914818,
      "grad_norm": 6.150376796722412,
      "learning_rate": 1e-05,
      "loss": 3.0759,
      "step": 3620
    },
    {
      "epoch": 0.6826822349603927,
      "grad_norm": 6.534663200378418,
      "learning_rate": 1e-05,
      "loss": 3.0509,
      "step": 3625
    },
    {
      "epoch": 0.6836238656293036,
      "grad_norm": 6.672440052032471,
      "learning_rate": 1e-05,
      "loss": 2.9175,
      "step": 3630
    },
    {
      "epoch": 0.6845654962982144,
      "grad_norm": 6.849086761474609,
      "learning_rate": 1e-05,
      "loss": 3.1881,
      "step": 3635
    },
    {
      "epoch": 0.6855071269671253,
      "grad_norm": 7.532481670379639,
      "learning_rate": 1e-05,
      "loss": 3.0881,
      "step": 3640
    },
    {
      "epoch": 0.6864487576360362,
      "grad_norm": 7.62110710144043,
      "learning_rate": 1e-05,
      "loss": 3.0303,
      "step": 3645
    },
    {
      "epoch": 0.6873903883049471,
      "grad_norm": 6.990483283996582,
      "learning_rate": 1e-05,
      "loss": 3.1092,
      "step": 3650
    },
    {
      "epoch": 0.688332018973858,
      "grad_norm": 7.549264907836914,
      "learning_rate": 1e-05,
      "loss": 3.0515,
      "step": 3655
    },
    {
      "epoch": 0.6892736496427688,
      "grad_norm": 7.345057010650635,
      "learning_rate": 1e-05,
      "loss": 3.0485,
      "step": 3660
    },
    {
      "epoch": 0.6902152803116798,
      "grad_norm": 7.437522888183594,
      "learning_rate": 1e-05,
      "loss": 3.0337,
      "step": 3665
    },
    {
      "epoch": 0.6911569109805906,
      "grad_norm": 6.142207145690918,
      "learning_rate": 1e-05,
      "loss": 2.9561,
      "step": 3670
    },
    {
      "epoch": 0.6920985416495016,
      "grad_norm": 6.626426696777344,
      "learning_rate": 1e-05,
      "loss": 3.1157,
      "step": 3675
    },
    {
      "epoch": 0.6930401723184124,
      "grad_norm": 6.62837553024292,
      "learning_rate": 1e-05,
      "loss": 3.0824,
      "step": 3680
    },
    {
      "epoch": 0.6939818029873233,
      "grad_norm": 7.575382232666016,
      "learning_rate": 1e-05,
      "loss": 3.0039,
      "step": 3685
    },
    {
      "epoch": 0.6949234336562342,
      "grad_norm": 7.333104133605957,
      "learning_rate": 1e-05,
      "loss": 3.0786,
      "step": 3690
    },
    {
      "epoch": 0.695865064325145,
      "grad_norm": 7.4377923011779785,
      "learning_rate": 1e-05,
      "loss": 2.9884,
      "step": 3695
    },
    {
      "epoch": 0.696806694994056,
      "grad_norm": 6.842844486236572,
      "learning_rate": 1e-05,
      "loss": 2.9574,
      "step": 3700
    },
    {
      "epoch": 0.6977483256629669,
      "grad_norm": 6.345764636993408,
      "learning_rate": 1e-05,
      "loss": 3.0264,
      "step": 3705
    },
    {
      "epoch": 0.6986899563318777,
      "grad_norm": 7.15539026260376,
      "learning_rate": 1e-05,
      "loss": 3.059,
      "step": 3710
    },
    {
      "epoch": 0.6996315870007886,
      "grad_norm": 7.269350528717041,
      "learning_rate": 1e-05,
      "loss": 3.0736,
      "step": 3715
    },
    {
      "epoch": 0.7005732176696995,
      "grad_norm": 6.788912296295166,
      "learning_rate": 1e-05,
      "loss": 2.9954,
      "step": 3720
    },
    {
      "epoch": 0.7015148483386104,
      "grad_norm": 6.4122185707092285,
      "learning_rate": 1e-05,
      "loss": 2.9234,
      "step": 3725
    },
    {
      "epoch": 0.7024564790075213,
      "grad_norm": 7.263458251953125,
      "learning_rate": 1e-05,
      "loss": 2.9974,
      "step": 3730
    },
    {
      "epoch": 0.7033981096764321,
      "grad_norm": 6.88037633895874,
      "learning_rate": 1e-05,
      "loss": 3.0908,
      "step": 3735
    },
    {
      "epoch": 0.7043397403453431,
      "grad_norm": 7.234920978546143,
      "learning_rate": 1e-05,
      "loss": 3.0355,
      "step": 3740
    },
    {
      "epoch": 0.7052813710142539,
      "grad_norm": 7.346100807189941,
      "learning_rate": 1e-05,
      "loss": 2.938,
      "step": 3745
    },
    {
      "epoch": 0.7062230016831649,
      "grad_norm": 6.825830936431885,
      "learning_rate": 1e-05,
      "loss": 3.1014,
      "step": 3750
    },
    {
      "epoch": 0.7071646323520757,
      "grad_norm": 7.167881965637207,
      "learning_rate": 1e-05,
      "loss": 3.064,
      "step": 3755
    },
    {
      "epoch": 0.7081062630209866,
      "grad_norm": 6.6397294998168945,
      "learning_rate": 1e-05,
      "loss": 3.0352,
      "step": 3760
    },
    {
      "epoch": 0.7090478936898975,
      "grad_norm": 7.553011894226074,
      "learning_rate": 1e-05,
      "loss": 3.0674,
      "step": 3765
    },
    {
      "epoch": 0.7099895243588084,
      "grad_norm": 7.612958908081055,
      "learning_rate": 1e-05,
      "loss": 3.0081,
      "step": 3770
    },
    {
      "epoch": 0.7109311550277193,
      "grad_norm": 6.634482383728027,
      "learning_rate": 1e-05,
      "loss": 3.0408,
      "step": 3775
    },
    {
      "epoch": 0.7118727856966302,
      "grad_norm": 6.5582356452941895,
      "learning_rate": 1e-05,
      "loss": 3.0377,
      "step": 3780
    },
    {
      "epoch": 0.712814416365541,
      "grad_norm": 6.804215431213379,
      "learning_rate": 1e-05,
      "loss": 3.001,
      "step": 3785
    },
    {
      "epoch": 0.713756047034452,
      "grad_norm": 7.458028316497803,
      "learning_rate": 1e-05,
      "loss": 2.9582,
      "step": 3790
    },
    {
      "epoch": 0.7146976777033628,
      "grad_norm": 6.639705181121826,
      "learning_rate": 1e-05,
      "loss": 3.0887,
      "step": 3795
    },
    {
      "epoch": 0.7156393083722736,
      "grad_norm": 6.617265701293945,
      "learning_rate": 1e-05,
      "loss": 3.0967,
      "step": 3800
    },
    {
      "epoch": 0.7165809390411846,
      "grad_norm": 8.8301362991333,
      "learning_rate": 1e-05,
      "loss": 3.0354,
      "step": 3805
    },
    {
      "epoch": 0.7175225697100954,
      "grad_norm": 6.7238264083862305,
      "learning_rate": 1e-05,
      "loss": 3.0144,
      "step": 3810
    },
    {
      "epoch": 0.7184642003790064,
      "grad_norm": 6.8456926345825195,
      "learning_rate": 1e-05,
      "loss": 2.9888,
      "step": 3815
    },
    {
      "epoch": 0.7194058310479172,
      "grad_norm": 8.741981506347656,
      "learning_rate": 1e-05,
      "loss": 3.1182,
      "step": 3820
    },
    {
      "epoch": 0.7203474617168281,
      "grad_norm": 6.671243667602539,
      "learning_rate": 1e-05,
      "loss": 3.0422,
      "step": 3825
    },
    {
      "epoch": 0.721289092385739,
      "grad_norm": 7.100765228271484,
      "learning_rate": 1e-05,
      "loss": 3.0205,
      "step": 3830
    },
    {
      "epoch": 0.7222307230546499,
      "grad_norm": 7.116576671600342,
      "learning_rate": 1e-05,
      "loss": 3.0333,
      "step": 3835
    },
    {
      "epoch": 0.7231723537235608,
      "grad_norm": 6.72273588180542,
      "learning_rate": 1e-05,
      "loss": 3.081,
      "step": 3840
    },
    {
      "epoch": 0.7241139843924717,
      "grad_norm": 6.83878231048584,
      "learning_rate": 1e-05,
      "loss": 2.9635,
      "step": 3845
    },
    {
      "epoch": 0.7250556150613825,
      "grad_norm": 7.474968433380127,
      "learning_rate": 1e-05,
      "loss": 3.0792,
      "step": 3850
    },
    {
      "epoch": 0.7259972457302935,
      "grad_norm": 7.929852485656738,
      "learning_rate": 1e-05,
      "loss": 2.9793,
      "step": 3855
    },
    {
      "epoch": 0.7269388763992043,
      "grad_norm": 6.638747215270996,
      "learning_rate": 1e-05,
      "loss": 3.0311,
      "step": 3860
    },
    {
      "epoch": 0.7278805070681152,
      "grad_norm": 6.270215034484863,
      "learning_rate": 1e-05,
      "loss": 2.9846,
      "step": 3865
    },
    {
      "epoch": 0.7288221377370261,
      "grad_norm": 7.199500560760498,
      "learning_rate": 1e-05,
      "loss": 3.0651,
      "step": 3870
    },
    {
      "epoch": 0.7297637684059369,
      "grad_norm": 6.566064834594727,
      "learning_rate": 1e-05,
      "loss": 2.9806,
      "step": 3875
    },
    {
      "epoch": 0.7307053990748479,
      "grad_norm": 7.1128830909729,
      "learning_rate": 1e-05,
      "loss": 3.0375,
      "step": 3880
    },
    {
      "epoch": 0.7316470297437587,
      "grad_norm": 6.685820579528809,
      "learning_rate": 1e-05,
      "loss": 3.058,
      "step": 3885
    },
    {
      "epoch": 0.7325886604126697,
      "grad_norm": 6.155799388885498,
      "learning_rate": 1e-05,
      "loss": 3.0514,
      "step": 3890
    },
    {
      "epoch": 0.7335302910815805,
      "grad_norm": 6.671477794647217,
      "learning_rate": 1e-05,
      "loss": 2.9818,
      "step": 3895
    },
    {
      "epoch": 0.7344719217504914,
      "grad_norm": 7.42489767074585,
      "learning_rate": 1e-05,
      "loss": 3.1053,
      "step": 3900
    },
    {
      "epoch": 0.7354135524194023,
      "grad_norm": 6.796725749969482,
      "learning_rate": 1e-05,
      "loss": 2.9821,
      "step": 3905
    },
    {
      "epoch": 0.7363551830883132,
      "grad_norm": 7.848191261291504,
      "learning_rate": 1e-05,
      "loss": 2.9765,
      "step": 3910
    },
    {
      "epoch": 0.7372968137572241,
      "grad_norm": 6.922936916351318,
      "learning_rate": 1e-05,
      "loss": 3.1376,
      "step": 3915
    },
    {
      "epoch": 0.738238444426135,
      "grad_norm": 6.367747783660889,
      "learning_rate": 1e-05,
      "loss": 3.1063,
      "step": 3920
    },
    {
      "epoch": 0.7391800750950458,
      "grad_norm": 6.697858810424805,
      "learning_rate": 1e-05,
      "loss": 2.9815,
      "step": 3925
    },
    {
      "epoch": 0.7401217057639567,
      "grad_norm": 6.511905193328857,
      "learning_rate": 1e-05,
      "loss": 2.9989,
      "step": 3930
    },
    {
      "epoch": 0.7410633364328676,
      "grad_norm": 7.045902729034424,
      "learning_rate": 1e-05,
      "loss": 3.0436,
      "step": 3935
    },
    {
      "epoch": 0.7420049671017785,
      "grad_norm": 6.697847843170166,
      "learning_rate": 1e-05,
      "loss": 2.9933,
      "step": 3940
    },
    {
      "epoch": 0.7429465977706894,
      "grad_norm": 8.465171813964844,
      "learning_rate": 1e-05,
      "loss": 3.1475,
      "step": 3945
    },
    {
      "epoch": 0.7438882284396002,
      "grad_norm": 7.368391990661621,
      "learning_rate": 1e-05,
      "loss": 2.9639,
      "step": 3950
    },
    {
      "epoch": 0.7448298591085112,
      "grad_norm": 9.309432983398438,
      "learning_rate": 1e-05,
      "loss": 3.0413,
      "step": 3955
    },
    {
      "epoch": 0.745771489777422,
      "grad_norm": 7.960738658905029,
      "learning_rate": 1e-05,
      "loss": 2.9888,
      "step": 3960
    },
    {
      "epoch": 0.746713120446333,
      "grad_norm": 7.475559711456299,
      "learning_rate": 1e-05,
      "loss": 3.0662,
      "step": 3965
    },
    {
      "epoch": 0.7476547511152438,
      "grad_norm": 7.966568946838379,
      "learning_rate": 1e-05,
      "loss": 3.0452,
      "step": 3970
    },
    {
      "epoch": 0.7485963817841547,
      "grad_norm": 7.25874137878418,
      "learning_rate": 1e-05,
      "loss": 2.9464,
      "step": 3975
    },
    {
      "epoch": 0.7495380124530656,
      "grad_norm": 7.146042346954346,
      "learning_rate": 1e-05,
      "loss": 3.0599,
      "step": 3980
    },
    {
      "epoch": 0.7504796431219765,
      "grad_norm": 7.571315765380859,
      "learning_rate": 1e-05,
      "loss": 3.0561,
      "step": 3985
    },
    {
      "epoch": 0.7514212737908874,
      "grad_norm": 7.106105327606201,
      "learning_rate": 1e-05,
      "loss": 3.0022,
      "step": 3990
    },
    {
      "epoch": 0.7523629044597983,
      "grad_norm": 7.547785758972168,
      "learning_rate": 1e-05,
      "loss": 3.1197,
      "step": 3995
    },
    {
      "epoch": 0.7533045351287091,
      "grad_norm": 6.921874046325684,
      "learning_rate": 1e-05,
      "loss": 2.946,
      "step": 4000
    },
    {
      "epoch": 0.7533045351287091,
      "eval_accuracy": 0.4179943871149811,
      "eval_loss": 3.0350847244262695,
      "eval_runtime": 1060.3761,
      "eval_samples_per_second": 35.61,
      "eval_steps_per_second": 8.903,
      "step": 4000
    },
    {
      "epoch": 0.75424616579762,
      "grad_norm": 7.115399360656738,
      "learning_rate": 1e-05,
      "loss": 2.9728,
      "step": 4005
    },
    {
      "epoch": 0.7551877964665309,
      "grad_norm": 6.1880269050598145,
      "learning_rate": 1e-05,
      "loss": 3.0188,
      "step": 4010
    },
    {
      "epoch": 0.7561294271354418,
      "grad_norm": 6.350123405456543,
      "learning_rate": 1e-05,
      "loss": 2.9826,
      "step": 4015
    },
    {
      "epoch": 0.7570710578043527,
      "grad_norm": 6.536036968231201,
      "learning_rate": 1e-05,
      "loss": 3.0136,
      "step": 4020
    },
    {
      "epoch": 0.7580126884732635,
      "grad_norm": 6.593567848205566,
      "learning_rate": 1e-05,
      "loss": 2.9847,
      "step": 4025
    },
    {
      "epoch": 0.7589543191421745,
      "grad_norm": 6.684474468231201,
      "learning_rate": 1e-05,
      "loss": 3.1012,
      "step": 4030
    },
    {
      "epoch": 0.7598959498110853,
      "grad_norm": 6.77077579498291,
      "learning_rate": 1e-05,
      "loss": 3.0014,
      "step": 4035
    },
    {
      "epoch": 0.7608375804799963,
      "grad_norm": 7.389578819274902,
      "learning_rate": 1e-05,
      "loss": 3.014,
      "step": 4040
    },
    {
      "epoch": 0.7617792111489071,
      "grad_norm": 6.702242374420166,
      "learning_rate": 1e-05,
      "loss": 2.9964,
      "step": 4045
    },
    {
      "epoch": 0.762720841817818,
      "grad_norm": 7.013911247253418,
      "learning_rate": 1e-05,
      "loss": 3.0237,
      "step": 4050
    },
    {
      "epoch": 0.7636624724867289,
      "grad_norm": 6.621683597564697,
      "learning_rate": 1e-05,
      "loss": 2.9814,
      "step": 4055
    },
    {
      "epoch": 0.7646041031556398,
      "grad_norm": 6.537116050720215,
      "learning_rate": 1e-05,
      "loss": 3.0512,
      "step": 4060
    },
    {
      "epoch": 0.7655457338245507,
      "grad_norm": 7.781442165374756,
      "learning_rate": 1e-05,
      "loss": 3.0855,
      "step": 4065
    },
    {
      "epoch": 0.7664873644934616,
      "grad_norm": 7.431787967681885,
      "learning_rate": 1e-05,
      "loss": 2.9926,
      "step": 4070
    },
    {
      "epoch": 0.7674289951623724,
      "grad_norm": 7.550678730010986,
      "learning_rate": 1e-05,
      "loss": 2.9967,
      "step": 4075
    },
    {
      "epoch": 0.7683706258312833,
      "grad_norm": 7.120694637298584,
      "learning_rate": 1e-05,
      "loss": 3.0271,
      "step": 4080
    },
    {
      "epoch": 0.7693122565001942,
      "grad_norm": 6.2324748039245605,
      "learning_rate": 1e-05,
      "loss": 2.979,
      "step": 4085
    },
    {
      "epoch": 0.7702538871691051,
      "grad_norm": 7.045401573181152,
      "learning_rate": 1e-05,
      "loss": 3.0638,
      "step": 4090
    },
    {
      "epoch": 0.771195517838016,
      "grad_norm": 7.148534774780273,
      "learning_rate": 1e-05,
      "loss": 2.9982,
      "step": 4095
    },
    {
      "epoch": 0.7721371485069268,
      "grad_norm": 6.389626979827881,
      "learning_rate": 1e-05,
      "loss": 2.9228,
      "step": 4100
    },
    {
      "epoch": 0.7730787791758378,
      "grad_norm": 7.732367992401123,
      "learning_rate": 1e-05,
      "loss": 3.0264,
      "step": 4105
    },
    {
      "epoch": 0.7740204098447486,
      "grad_norm": 6.456219673156738,
      "learning_rate": 1e-05,
      "loss": 3.1302,
      "step": 4110
    },
    {
      "epoch": 0.7749620405136596,
      "grad_norm": 6.384737014770508,
      "learning_rate": 1e-05,
      "loss": 2.9592,
      "step": 4115
    },
    {
      "epoch": 0.7759036711825704,
      "grad_norm": 7.74468469619751,
      "learning_rate": 1e-05,
      "loss": 3.0449,
      "step": 4120
    },
    {
      "epoch": 0.7768453018514813,
      "grad_norm": 6.250748634338379,
      "learning_rate": 1e-05,
      "loss": 3.0414,
      "step": 4125
    },
    {
      "epoch": 0.7777869325203922,
      "grad_norm": 6.97851037979126,
      "learning_rate": 1e-05,
      "loss": 3.0961,
      "step": 4130
    },
    {
      "epoch": 0.7787285631893031,
      "grad_norm": 7.248284816741943,
      "learning_rate": 1e-05,
      "loss": 2.9958,
      "step": 4135
    },
    {
      "epoch": 0.779670193858214,
      "grad_norm": 6.732598304748535,
      "learning_rate": 1e-05,
      "loss": 2.9861,
      "step": 4140
    },
    {
      "epoch": 0.7806118245271249,
      "grad_norm": 6.4773173332214355,
      "learning_rate": 1e-05,
      "loss": 2.9421,
      "step": 4145
    },
    {
      "epoch": 0.7815534551960357,
      "grad_norm": 6.1240620613098145,
      "learning_rate": 1e-05,
      "loss": 2.9885,
      "step": 4150
    },
    {
      "epoch": 0.7824950858649466,
      "grad_norm": 8.18076229095459,
      "learning_rate": 1e-05,
      "loss": 2.8914,
      "step": 4155
    },
    {
      "epoch": 0.7834367165338575,
      "grad_norm": 6.479029178619385,
      "learning_rate": 1e-05,
      "loss": 2.9848,
      "step": 4160
    },
    {
      "epoch": 0.7843783472027684,
      "grad_norm": 7.001804828643799,
      "learning_rate": 1e-05,
      "loss": 2.9447,
      "step": 4165
    },
    {
      "epoch": 0.7853199778716793,
      "grad_norm": 7.894457817077637,
      "learning_rate": 1e-05,
      "loss": 2.9969,
      "step": 4170
    },
    {
      "epoch": 0.7862616085405901,
      "grad_norm": 6.224025726318359,
      "learning_rate": 1e-05,
      "loss": 3.1038,
      "step": 4175
    },
    {
      "epoch": 0.7872032392095011,
      "grad_norm": 7.672967910766602,
      "learning_rate": 1e-05,
      "loss": 3.1315,
      "step": 4180
    },
    {
      "epoch": 0.7881448698784119,
      "grad_norm": 6.595861434936523,
      "learning_rate": 1e-05,
      "loss": 3.0766,
      "step": 4185
    },
    {
      "epoch": 0.7890865005473229,
      "grad_norm": 6.691225051879883,
      "learning_rate": 1e-05,
      "loss": 3.0528,
      "step": 4190
    },
    {
      "epoch": 0.7900281312162337,
      "grad_norm": 6.80654764175415,
      "learning_rate": 1e-05,
      "loss": 3.0533,
      "step": 4195
    },
    {
      "epoch": 0.7909697618851446,
      "grad_norm": 7.244470119476318,
      "learning_rate": 1e-05,
      "loss": 3.0414,
      "step": 4200
    },
    {
      "epoch": 0.7919113925540555,
      "grad_norm": 6.781643390655518,
      "learning_rate": 1e-05,
      "loss": 3.0211,
      "step": 4205
    },
    {
      "epoch": 0.7928530232229664,
      "grad_norm": 6.594189167022705,
      "learning_rate": 1e-05,
      "loss": 3.0595,
      "step": 4210
    },
    {
      "epoch": 0.7937946538918773,
      "grad_norm": 6.970313549041748,
      "learning_rate": 1e-05,
      "loss": 3.0505,
      "step": 4215
    },
    {
      "epoch": 0.7947362845607882,
      "grad_norm": 7.164793491363525,
      "learning_rate": 1e-05,
      "loss": 2.9477,
      "step": 4220
    },
    {
      "epoch": 0.795677915229699,
      "grad_norm": 6.909482955932617,
      "learning_rate": 1e-05,
      "loss": 2.9963,
      "step": 4225
    },
    {
      "epoch": 0.79661954589861,
      "grad_norm": 6.645529747009277,
      "learning_rate": 1e-05,
      "loss": 3.0773,
      "step": 4230
    },
    {
      "epoch": 0.7975611765675208,
      "grad_norm": 6.645266532897949,
      "learning_rate": 1e-05,
      "loss": 3.0217,
      "step": 4235
    },
    {
      "epoch": 0.7985028072364317,
      "grad_norm": 6.193962574005127,
      "learning_rate": 1e-05,
      "loss": 2.9698,
      "step": 4240
    },
    {
      "epoch": 0.7994444379053426,
      "grad_norm": 6.405771255493164,
      "learning_rate": 1e-05,
      "loss": 3.012,
      "step": 4245
    },
    {
      "epoch": 0.8003860685742534,
      "grad_norm": 6.351288795471191,
      "learning_rate": 1e-05,
      "loss": 2.9623,
      "step": 4250
    },
    {
      "epoch": 0.8013276992431644,
      "grad_norm": 6.759338855743408,
      "learning_rate": 1e-05,
      "loss": 3.004,
      "step": 4255
    },
    {
      "epoch": 0.8022693299120752,
      "grad_norm": 6.525925636291504,
      "learning_rate": 1e-05,
      "loss": 3.0707,
      "step": 4260
    },
    {
      "epoch": 0.8032109605809862,
      "grad_norm": 7.245208263397217,
      "learning_rate": 1e-05,
      "loss": 3.033,
      "step": 4265
    },
    {
      "epoch": 0.804152591249897,
      "grad_norm": 7.056334972381592,
      "learning_rate": 1e-05,
      "loss": 2.9285,
      "step": 4270
    },
    {
      "epoch": 0.8050942219188079,
      "grad_norm": 7.097318172454834,
      "learning_rate": 1e-05,
      "loss": 2.9806,
      "step": 4275
    },
    {
      "epoch": 0.8060358525877188,
      "grad_norm": 6.930456638336182,
      "learning_rate": 1e-05,
      "loss": 3.0609,
      "step": 4280
    },
    {
      "epoch": 0.8069774832566297,
      "grad_norm": 6.811648368835449,
      "learning_rate": 1e-05,
      "loss": 3.1415,
      "step": 4285
    },
    {
      "epoch": 0.8079191139255405,
      "grad_norm": 6.777252197265625,
      "learning_rate": 1e-05,
      "loss": 3.0246,
      "step": 4290
    },
    {
      "epoch": 0.8088607445944515,
      "grad_norm": 6.139156818389893,
      "learning_rate": 1e-05,
      "loss": 2.9366,
      "step": 4295
    },
    {
      "epoch": 0.8098023752633623,
      "grad_norm": 6.63328742980957,
      "learning_rate": 1e-05,
      "loss": 3.0745,
      "step": 4300
    },
    {
      "epoch": 0.8107440059322732,
      "grad_norm": 7.7045183181762695,
      "learning_rate": 1e-05,
      "loss": 3.0592,
      "step": 4305
    },
    {
      "epoch": 0.8116856366011841,
      "grad_norm": 6.513607978820801,
      "learning_rate": 1e-05,
      "loss": 2.9483,
      "step": 4310
    },
    {
      "epoch": 0.8126272672700949,
      "grad_norm": 6.670443534851074,
      "learning_rate": 1e-05,
      "loss": 3.0345,
      "step": 4315
    },
    {
      "epoch": 0.8135688979390059,
      "grad_norm": 6.660811901092529,
      "learning_rate": 1e-05,
      "loss": 2.9852,
      "step": 4320
    },
    {
      "epoch": 0.8145105286079167,
      "grad_norm": 7.2338361740112305,
      "learning_rate": 1e-05,
      "loss": 3.0945,
      "step": 4325
    },
    {
      "epoch": 0.8154521592768277,
      "grad_norm": 6.750393390655518,
      "learning_rate": 1e-05,
      "loss": 2.9639,
      "step": 4330
    },
    {
      "epoch": 0.8163937899457385,
      "grad_norm": 7.605077743530273,
      "learning_rate": 1e-05,
      "loss": 3.0707,
      "step": 4335
    },
    {
      "epoch": 0.8173354206146494,
      "grad_norm": 7.224457263946533,
      "learning_rate": 1e-05,
      "loss": 3.0978,
      "step": 4340
    },
    {
      "epoch": 0.8182770512835603,
      "grad_norm": 6.255330562591553,
      "learning_rate": 1e-05,
      "loss": 3.0075,
      "step": 4345
    },
    {
      "epoch": 0.8192186819524712,
      "grad_norm": 6.431050777435303,
      "learning_rate": 1e-05,
      "loss": 3.1404,
      "step": 4350
    },
    {
      "epoch": 0.8201603126213821,
      "grad_norm": 7.618997097015381,
      "learning_rate": 1e-05,
      "loss": 2.8578,
      "step": 4355
    },
    {
      "epoch": 0.821101943290293,
      "grad_norm": 6.874431133270264,
      "learning_rate": 1e-05,
      "loss": 2.9544,
      "step": 4360
    },
    {
      "epoch": 0.8220435739592038,
      "grad_norm": 6.51139497756958,
      "learning_rate": 1e-05,
      "loss": 3.0135,
      "step": 4365
    },
    {
      "epoch": 0.8229852046281148,
      "grad_norm": 6.711349964141846,
      "learning_rate": 1e-05,
      "loss": 3.0015,
      "step": 4370
    },
    {
      "epoch": 0.8239268352970256,
      "grad_norm": 7.047435283660889,
      "learning_rate": 1e-05,
      "loss": 3.0321,
      "step": 4375
    },
    {
      "epoch": 0.8248684659659365,
      "grad_norm": 7.979480266571045,
      "learning_rate": 1e-05,
      "loss": 3.1258,
      "step": 4380
    },
    {
      "epoch": 0.8258100966348474,
      "grad_norm": 6.74650239944458,
      "learning_rate": 1e-05,
      "loss": 3.1225,
      "step": 4385
    },
    {
      "epoch": 0.8267517273037582,
      "grad_norm": 5.72318696975708,
      "learning_rate": 1e-05,
      "loss": 2.9543,
      "step": 4390
    },
    {
      "epoch": 0.8276933579726692,
      "grad_norm": 6.581451416015625,
      "learning_rate": 1e-05,
      "loss": 2.9904,
      "step": 4395
    },
    {
      "epoch": 0.82863498864158,
      "grad_norm": 6.8295769691467285,
      "learning_rate": 1e-05,
      "loss": 2.994,
      "step": 4400
    },
    {
      "epoch": 0.829576619310491,
      "grad_norm": 7.150579929351807,
      "learning_rate": 1e-05,
      "loss": 3.024,
      "step": 4405
    },
    {
      "epoch": 0.8305182499794018,
      "grad_norm": 6.949995994567871,
      "learning_rate": 1e-05,
      "loss": 3.0905,
      "step": 4410
    },
    {
      "epoch": 0.8314598806483127,
      "grad_norm": 7.133937835693359,
      "learning_rate": 1e-05,
      "loss": 3.0603,
      "step": 4415
    },
    {
      "epoch": 0.8324015113172236,
      "grad_norm": 6.495321750640869,
      "learning_rate": 1e-05,
      "loss": 3.0027,
      "step": 4420
    },
    {
      "epoch": 0.8333431419861345,
      "grad_norm": 6.963568210601807,
      "learning_rate": 1e-05,
      "loss": 2.9662,
      "step": 4425
    },
    {
      "epoch": 0.8342847726550454,
      "grad_norm": 7.362428188323975,
      "learning_rate": 1e-05,
      "loss": 2.9869,
      "step": 4430
    },
    {
      "epoch": 0.8352264033239563,
      "grad_norm": 6.788746356964111,
      "learning_rate": 1e-05,
      "loss": 3.0145,
      "step": 4435
    },
    {
      "epoch": 0.8361680339928671,
      "grad_norm": 7.730389595031738,
      "learning_rate": 1e-05,
      "loss": 3.1651,
      "step": 4440
    },
    {
      "epoch": 0.837109664661778,
      "grad_norm": 7.5514116287231445,
      "learning_rate": 1e-05,
      "loss": 3.0095,
      "step": 4445
    },
    {
      "epoch": 0.8380512953306889,
      "grad_norm": 6.9045257568359375,
      "learning_rate": 1e-05,
      "loss": 3.0569,
      "step": 4450
    },
    {
      "epoch": 0.8389929259995998,
      "grad_norm": 7.573888778686523,
      "learning_rate": 1e-05,
      "loss": 2.9869,
      "step": 4455
    },
    {
      "epoch": 0.8399345566685107,
      "grad_norm": 7.188941478729248,
      "learning_rate": 1e-05,
      "loss": 3.0701,
      "step": 4460
    },
    {
      "epoch": 0.8408761873374215,
      "grad_norm": 6.4194817543029785,
      "learning_rate": 1e-05,
      "loss": 3.1,
      "step": 4465
    },
    {
      "epoch": 0.8418178180063325,
      "grad_norm": 6.900866508483887,
      "learning_rate": 1e-05,
      "loss": 2.9665,
      "step": 4470
    },
    {
      "epoch": 0.8427594486752433,
      "grad_norm": 7.416345596313477,
      "learning_rate": 1e-05,
      "loss": 3.0122,
      "step": 4475
    },
    {
      "epoch": 0.8437010793441543,
      "grad_norm": 7.08629846572876,
      "learning_rate": 1e-05,
      "loss": 3.1027,
      "step": 4480
    },
    {
      "epoch": 0.8446427100130651,
      "grad_norm": 6.6842803955078125,
      "learning_rate": 1e-05,
      "loss": 2.9872,
      "step": 4485
    },
    {
      "epoch": 0.845584340681976,
      "grad_norm": 7.190927982330322,
      "learning_rate": 1e-05,
      "loss": 2.9336,
      "step": 4490
    },
    {
      "epoch": 0.8465259713508869,
      "grad_norm": 7.118557929992676,
      "learning_rate": 1e-05,
      "loss": 2.9181,
      "step": 4495
    },
    {
      "epoch": 0.8474676020197978,
      "grad_norm": 6.6827006340026855,
      "learning_rate": 1e-05,
      "loss": 3.0798,
      "step": 4500
    },
    {
      "epoch": 0.8484092326887087,
      "grad_norm": 6.979405403137207,
      "learning_rate": 1e-05,
      "loss": 2.9763,
      "step": 4505
    },
    {
      "epoch": 0.8493508633576196,
      "grad_norm": 6.2492780685424805,
      "learning_rate": 1e-05,
      "loss": 3.021,
      "step": 4510
    },
    {
      "epoch": 0.8502924940265304,
      "grad_norm": 7.063942909240723,
      "learning_rate": 1e-05,
      "loss": 2.9505,
      "step": 4515
    },
    {
      "epoch": 0.8512341246954414,
      "grad_norm": 6.5366716384887695,
      "learning_rate": 1e-05,
      "loss": 2.8787,
      "step": 4520
    },
    {
      "epoch": 0.8521757553643522,
      "grad_norm": 6.5581769943237305,
      "learning_rate": 1e-05,
      "loss": 3.1491,
      "step": 4525
    },
    {
      "epoch": 0.8531173860332631,
      "grad_norm": 6.441840171813965,
      "learning_rate": 1e-05,
      "loss": 3.0369,
      "step": 4530
    },
    {
      "epoch": 0.854059016702174,
      "grad_norm": 7.320792198181152,
      "learning_rate": 1e-05,
      "loss": 3.0956,
      "step": 4535
    },
    {
      "epoch": 0.8550006473710848,
      "grad_norm": 7.050429821014404,
      "learning_rate": 1e-05,
      "loss": 3.0892,
      "step": 4540
    },
    {
      "epoch": 0.8559422780399958,
      "grad_norm": 6.61350679397583,
      "learning_rate": 1e-05,
      "loss": 2.9785,
      "step": 4545
    },
    {
      "epoch": 0.8568839087089066,
      "grad_norm": 6.520270347595215,
      "learning_rate": 1e-05,
      "loss": 3.0025,
      "step": 4550
    },
    {
      "epoch": 0.8578255393778176,
      "grad_norm": 6.5053791999816895,
      "learning_rate": 1e-05,
      "loss": 2.8551,
      "step": 4555
    },
    {
      "epoch": 0.8587671700467284,
      "grad_norm": 6.498419761657715,
      "learning_rate": 1e-05,
      "loss": 2.9634,
      "step": 4560
    },
    {
      "epoch": 0.8597088007156393,
      "grad_norm": 7.01683235168457,
      "learning_rate": 1e-05,
      "loss": 3.1556,
      "step": 4565
    },
    {
      "epoch": 0.8606504313845502,
      "grad_norm": 6.802071571350098,
      "learning_rate": 1e-05,
      "loss": 3.0464,
      "step": 4570
    },
    {
      "epoch": 0.8615920620534611,
      "grad_norm": 6.664157390594482,
      "learning_rate": 1e-05,
      "loss": 3.012,
      "step": 4575
    },
    {
      "epoch": 0.862533692722372,
      "grad_norm": 6.3640456199646,
      "learning_rate": 1e-05,
      "loss": 2.934,
      "step": 4580
    },
    {
      "epoch": 0.8634753233912829,
      "grad_norm": 6.966064929962158,
      "learning_rate": 1e-05,
      "loss": 2.9922,
      "step": 4585
    },
    {
      "epoch": 0.8644169540601937,
      "grad_norm": 6.394281387329102,
      "learning_rate": 1e-05,
      "loss": 2.904,
      "step": 4590
    },
    {
      "epoch": 0.8653585847291047,
      "grad_norm": 8.291240692138672,
      "learning_rate": 1e-05,
      "loss": 3.0824,
      "step": 4595
    },
    {
      "epoch": 0.8663002153980155,
      "grad_norm": 6.770839214324951,
      "learning_rate": 1e-05,
      "loss": 3.0608,
      "step": 4600
    },
    {
      "epoch": 0.8672418460669264,
      "grad_norm": 7.17354679107666,
      "learning_rate": 1e-05,
      "loss": 3.007,
      "step": 4605
    },
    {
      "epoch": 0.8681834767358373,
      "grad_norm": 6.194829940795898,
      "learning_rate": 1e-05,
      "loss": 2.9562,
      "step": 4610
    },
    {
      "epoch": 0.8691251074047481,
      "grad_norm": 6.281735420227051,
      "learning_rate": 1e-05,
      "loss": 2.9716,
      "step": 4615
    },
    {
      "epoch": 0.8700667380736591,
      "grad_norm": 7.197927951812744,
      "learning_rate": 1e-05,
      "loss": 3.008,
      "step": 4620
    },
    {
      "epoch": 0.8710083687425699,
      "grad_norm": 6.667356967926025,
      "learning_rate": 1e-05,
      "loss": 3.1845,
      "step": 4625
    },
    {
      "epoch": 0.8719499994114809,
      "grad_norm": 7.937153339385986,
      "learning_rate": 1e-05,
      "loss": 3.0967,
      "step": 4630
    },
    {
      "epoch": 0.8728916300803917,
      "grad_norm": 6.7486252784729,
      "learning_rate": 1e-05,
      "loss": 3.0172,
      "step": 4635
    },
    {
      "epoch": 0.8738332607493026,
      "grad_norm": 7.023075103759766,
      "learning_rate": 1e-05,
      "loss": 3.0151,
      "step": 4640
    },
    {
      "epoch": 0.8747748914182135,
      "grad_norm": 6.361885070800781,
      "learning_rate": 1e-05,
      "loss": 3.0446,
      "step": 4645
    },
    {
      "epoch": 0.8757165220871244,
      "grad_norm": 6.498104572296143,
      "learning_rate": 1e-05,
      "loss": 3.0111,
      "step": 4650
    },
    {
      "epoch": 0.8766581527560353,
      "grad_norm": 6.258391380310059,
      "learning_rate": 1e-05,
      "loss": 3.006,
      "step": 4655
    },
    {
      "epoch": 0.8775997834249462,
      "grad_norm": 7.305341720581055,
      "learning_rate": 1e-05,
      "loss": 2.9864,
      "step": 4660
    },
    {
      "epoch": 0.878541414093857,
      "grad_norm": 6.587819576263428,
      "learning_rate": 1e-05,
      "loss": 2.9275,
      "step": 4665
    },
    {
      "epoch": 0.879483044762768,
      "grad_norm": 6.967113018035889,
      "learning_rate": 1e-05,
      "loss": 3.0478,
      "step": 4670
    },
    {
      "epoch": 0.8804246754316788,
      "grad_norm": 6.257725715637207,
      "learning_rate": 1e-05,
      "loss": 2.9357,
      "step": 4675
    },
    {
      "epoch": 0.8813663061005897,
      "grad_norm": 7.08759069442749,
      "learning_rate": 1e-05,
      "loss": 2.9292,
      "step": 4680
    },
    {
      "epoch": 0.8823079367695006,
      "grad_norm": 6.993216514587402,
      "learning_rate": 1e-05,
      "loss": 2.9556,
      "step": 4685
    },
    {
      "epoch": 0.8832495674384114,
      "grad_norm": 6.2222771644592285,
      "learning_rate": 1e-05,
      "loss": 2.95,
      "step": 4690
    },
    {
      "epoch": 0.8841911981073224,
      "grad_norm": 5.418234825134277,
      "learning_rate": 1e-05,
      "loss": 2.9276,
      "step": 4695
    },
    {
      "epoch": 0.8851328287762332,
      "grad_norm": 7.821156024932861,
      "learning_rate": 1e-05,
      "loss": 2.957,
      "step": 4700
    },
    {
      "epoch": 0.8860744594451442,
      "grad_norm": 7.50763463973999,
      "learning_rate": 1e-05,
      "loss": 3.0653,
      "step": 4705
    },
    {
      "epoch": 0.887016090114055,
      "grad_norm": 5.917377471923828,
      "learning_rate": 1e-05,
      "loss": 3.0135,
      "step": 4710
    },
    {
      "epoch": 0.8879577207829659,
      "grad_norm": 6.086848258972168,
      "learning_rate": 1e-05,
      "loss": 2.9784,
      "step": 4715
    },
    {
      "epoch": 0.8888993514518768,
      "grad_norm": 7.318466663360596,
      "learning_rate": 1e-05,
      "loss": 2.9473,
      "step": 4720
    },
    {
      "epoch": 0.8898409821207877,
      "grad_norm": 6.821482181549072,
      "learning_rate": 1e-05,
      "loss": 3.0772,
      "step": 4725
    },
    {
      "epoch": 0.8907826127896986,
      "grad_norm": 6.979645252227783,
      "learning_rate": 1e-05,
      "loss": 3.0302,
      "step": 4730
    },
    {
      "epoch": 0.8917242434586095,
      "grad_norm": 7.538966655731201,
      "learning_rate": 1e-05,
      "loss": 3.0771,
      "step": 4735
    },
    {
      "epoch": 0.8926658741275203,
      "grad_norm": 6.447031497955322,
      "learning_rate": 1e-05,
      "loss": 3.0053,
      "step": 4740
    },
    {
      "epoch": 0.8936075047964313,
      "grad_norm": 6.903289794921875,
      "learning_rate": 1e-05,
      "loss": 3.0288,
      "step": 4745
    },
    {
      "epoch": 0.8945491354653421,
      "grad_norm": 6.24967098236084,
      "learning_rate": 1e-05,
      "loss": 3.0103,
      "step": 4750
    },
    {
      "epoch": 0.895490766134253,
      "grad_norm": 6.205604553222656,
      "learning_rate": 1e-05,
      "loss": 3.0107,
      "step": 4755
    },
    {
      "epoch": 0.8964323968031639,
      "grad_norm": 6.532459735870361,
      "learning_rate": 1e-05,
      "loss": 3.0275,
      "step": 4760
    },
    {
      "epoch": 0.8973740274720747,
      "grad_norm": 6.63011360168457,
      "learning_rate": 1e-05,
      "loss": 2.9595,
      "step": 4765
    },
    {
      "epoch": 0.8983156581409857,
      "grad_norm": 6.74818229675293,
      "learning_rate": 1e-05,
      "loss": 3.0929,
      "step": 4770
    },
    {
      "epoch": 0.8992572888098965,
      "grad_norm": 6.410697937011719,
      "learning_rate": 1e-05,
      "loss": 3.1027,
      "step": 4775
    },
    {
      "epoch": 0.9001989194788075,
      "grad_norm": 6.183323860168457,
      "learning_rate": 1e-05,
      "loss": 2.9818,
      "step": 4780
    },
    {
      "epoch": 0.9011405501477183,
      "grad_norm": 6.527472496032715,
      "learning_rate": 1e-05,
      "loss": 2.996,
      "step": 4785
    },
    {
      "epoch": 0.9020821808166292,
      "grad_norm": 6.853177547454834,
      "learning_rate": 1e-05,
      "loss": 3.0178,
      "step": 4790
    },
    {
      "epoch": 0.9030238114855401,
      "grad_norm": 7.206171035766602,
      "learning_rate": 1e-05,
      "loss": 3.0277,
      "step": 4795
    },
    {
      "epoch": 0.903965442154451,
      "grad_norm": 7.369021892547607,
      "learning_rate": 1e-05,
      "loss": 2.9376,
      "step": 4800
    },
    {
      "epoch": 0.9049070728233618,
      "grad_norm": 7.516411781311035,
      "learning_rate": 1e-05,
      "loss": 3.0268,
      "step": 4805
    },
    {
      "epoch": 0.9058487034922728,
      "grad_norm": 6.495131015777588,
      "learning_rate": 1e-05,
      "loss": 3.0405,
      "step": 4810
    },
    {
      "epoch": 0.9067903341611836,
      "grad_norm": 7.050095558166504,
      "learning_rate": 1e-05,
      "loss": 3.1039,
      "step": 4815
    },
    {
      "epoch": 0.9077319648300946,
      "grad_norm": 6.463887691497803,
      "learning_rate": 1e-05,
      "loss": 3.0336,
      "step": 4820
    },
    {
      "epoch": 0.9086735954990054,
      "grad_norm": 6.821866512298584,
      "learning_rate": 1e-05,
      "loss": 3.1019,
      "step": 4825
    },
    {
      "epoch": 0.9096152261679162,
      "grad_norm": 9.358695983886719,
      "learning_rate": 1e-05,
      "loss": 2.9516,
      "step": 4830
    },
    {
      "epoch": 0.9105568568368272,
      "grad_norm": 7.6679863929748535,
      "learning_rate": 1e-05,
      "loss": 2.979,
      "step": 4835
    },
    {
      "epoch": 0.911498487505738,
      "grad_norm": 6.348465919494629,
      "learning_rate": 1e-05,
      "loss": 3.0011,
      "step": 4840
    },
    {
      "epoch": 0.912440118174649,
      "grad_norm": 6.158372402191162,
      "learning_rate": 1e-05,
      "loss": 2.8976,
      "step": 4845
    },
    {
      "epoch": 0.9133817488435598,
      "grad_norm": 6.827818870544434,
      "learning_rate": 1e-05,
      "loss": 3.0961,
      "step": 4850
    },
    {
      "epoch": 0.9143233795124707,
      "grad_norm": 6.508996486663818,
      "learning_rate": 1e-05,
      "loss": 2.9493,
      "step": 4855
    },
    {
      "epoch": 0.9152650101813816,
      "grad_norm": 6.057750701904297,
      "learning_rate": 1e-05,
      "loss": 2.9642,
      "step": 4860
    },
    {
      "epoch": 0.9162066408502925,
      "grad_norm": 6.342018127441406,
      "learning_rate": 1e-05,
      "loss": 2.9965,
      "step": 4865
    },
    {
      "epoch": 0.9171482715192034,
      "grad_norm": 7.247106552124023,
      "learning_rate": 1e-05,
      "loss": 3.0096,
      "step": 4870
    },
    {
      "epoch": 0.9180899021881143,
      "grad_norm": 7.869892120361328,
      "learning_rate": 1e-05,
      "loss": 3.0186,
      "step": 4875
    },
    {
      "epoch": 0.9190315328570251,
      "grad_norm": 6.655325412750244,
      "learning_rate": 1e-05,
      "loss": 2.9532,
      "step": 4880
    },
    {
      "epoch": 0.919973163525936,
      "grad_norm": 7.238323211669922,
      "learning_rate": 1e-05,
      "loss": 2.964,
      "step": 4885
    },
    {
      "epoch": 0.9209147941948469,
      "grad_norm": 6.327097415924072,
      "learning_rate": 1e-05,
      "loss": 2.9265,
      "step": 4890
    },
    {
      "epoch": 0.9218564248637579,
      "grad_norm": 6.87644624710083,
      "learning_rate": 1e-05,
      "loss": 2.9748,
      "step": 4895
    },
    {
      "epoch": 0.9227980555326687,
      "grad_norm": 7.045034408569336,
      "learning_rate": 1e-05,
      "loss": 2.9084,
      "step": 4900
    },
    {
      "epoch": 0.9237396862015795,
      "grad_norm": 7.708536624908447,
      "learning_rate": 1e-05,
      "loss": 3.0383,
      "step": 4905
    },
    {
      "epoch": 0.9246813168704905,
      "grad_norm": 6.771285057067871,
      "learning_rate": 1e-05,
      "loss": 2.9808,
      "step": 4910
    },
    {
      "epoch": 0.9256229475394013,
      "grad_norm": 6.201882362365723,
      "learning_rate": 1e-05,
      "loss": 3.0511,
      "step": 4915
    },
    {
      "epoch": 0.9265645782083123,
      "grad_norm": 6.08769416809082,
      "learning_rate": 1e-05,
      "loss": 3.0086,
      "step": 4920
    },
    {
      "epoch": 0.9275062088772231,
      "grad_norm": 6.0023393630981445,
      "learning_rate": 1e-05,
      "loss": 3.0951,
      "step": 4925
    },
    {
      "epoch": 0.928447839546134,
      "grad_norm": 6.94964075088501,
      "learning_rate": 1e-05,
      "loss": 3.0002,
      "step": 4930
    },
    {
      "epoch": 0.9293894702150449,
      "grad_norm": 6.350240707397461,
      "learning_rate": 1e-05,
      "loss": 2.9763,
      "step": 4935
    },
    {
      "epoch": 0.9303311008839558,
      "grad_norm": 7.173480033874512,
      "learning_rate": 1e-05,
      "loss": 2.9946,
      "step": 4940
    },
    {
      "epoch": 0.9312727315528667,
      "grad_norm": 6.645884990692139,
      "learning_rate": 1e-05,
      "loss": 3.0732,
      "step": 4945
    },
    {
      "epoch": 0.9322143622217776,
      "grad_norm": 6.9946370124816895,
      "learning_rate": 1e-05,
      "loss": 3.0506,
      "step": 4950
    },
    {
      "epoch": 0.9331559928906884,
      "grad_norm": 6.4223456382751465,
      "learning_rate": 1e-05,
      "loss": 2.9862,
      "step": 4955
    },
    {
      "epoch": 0.9340976235595994,
      "grad_norm": 6.558778762817383,
      "learning_rate": 1e-05,
      "loss": 2.9428,
      "step": 4960
    },
    {
      "epoch": 0.9350392542285102,
      "grad_norm": 7.190298080444336,
      "learning_rate": 1e-05,
      "loss": 3.0667,
      "step": 4965
    },
    {
      "epoch": 0.9359808848974212,
      "grad_norm": 6.2120184898376465,
      "learning_rate": 1e-05,
      "loss": 2.9411,
      "step": 4970
    },
    {
      "epoch": 0.936922515566332,
      "grad_norm": 7.766112804412842,
      "learning_rate": 1e-05,
      "loss": 3.0971,
      "step": 4975
    },
    {
      "epoch": 0.9378641462352428,
      "grad_norm": 7.066897392272949,
      "learning_rate": 1e-05,
      "loss": 3.1249,
      "step": 4980
    },
    {
      "epoch": 0.9388057769041538,
      "grad_norm": 6.099008560180664,
      "learning_rate": 1e-05,
      "loss": 3.0034,
      "step": 4985
    },
    {
      "epoch": 0.9397474075730646,
      "grad_norm": 6.696260452270508,
      "learning_rate": 1e-05,
      "loss": 3.0705,
      "step": 4990
    },
    {
      "epoch": 0.9406890382419756,
      "grad_norm": 6.140739440917969,
      "learning_rate": 1e-05,
      "loss": 2.9683,
      "step": 4995
    },
    {
      "epoch": 0.9416306689108864,
      "grad_norm": 7.1208600997924805,
      "learning_rate": 1e-05,
      "loss": 2.9827,
      "step": 5000
    },
    {
      "epoch": 0.9416306689108864,
      "eval_accuracy": 0.4198751663662041,
      "eval_loss": 3.008730888366699,
      "eval_runtime": 1038.4567,
      "eval_samples_per_second": 36.362,
      "eval_steps_per_second": 9.09,
      "step": 5000
    },
    {
      "epoch": 0.9425722995797973,
      "grad_norm": 6.645845890045166,
      "learning_rate": 1e-05,
      "loss": 2.9817,
      "step": 5005
    },
    {
      "epoch": 0.9435139302487082,
      "grad_norm": 6.95515251159668,
      "learning_rate": 1e-05,
      "loss": 3.0721,
      "step": 5010
    },
    {
      "epoch": 0.9444555609176191,
      "grad_norm": 6.197011947631836,
      "learning_rate": 1e-05,
      "loss": 3.0179,
      "step": 5015
    },
    {
      "epoch": 0.94539719158653,
      "grad_norm": 6.877534866333008,
      "learning_rate": 1e-05,
      "loss": 3.0148,
      "step": 5020
    },
    {
      "epoch": 0.9463388222554409,
      "grad_norm": 6.991530418395996,
      "learning_rate": 1e-05,
      "loss": 3.0165,
      "step": 5025
    },
    {
      "epoch": 0.9472804529243517,
      "grad_norm": 7.398330211639404,
      "learning_rate": 1e-05,
      "loss": 2.9934,
      "step": 5030
    },
    {
      "epoch": 0.9482220835932627,
      "grad_norm": 7.08344030380249,
      "learning_rate": 1e-05,
      "loss": 3.0171,
      "step": 5035
    },
    {
      "epoch": 0.9491637142621735,
      "grad_norm": 6.782553195953369,
      "learning_rate": 1e-05,
      "loss": 2.932,
      "step": 5040
    },
    {
      "epoch": 0.9501053449310845,
      "grad_norm": 6.938860893249512,
      "learning_rate": 1e-05,
      "loss": 2.9003,
      "step": 5045
    },
    {
      "epoch": 0.9510469755999953,
      "grad_norm": 6.443044185638428,
      "learning_rate": 1e-05,
      "loss": 3.066,
      "step": 5050
    },
    {
      "epoch": 0.9519886062689061,
      "grad_norm": 7.133784770965576,
      "learning_rate": 1e-05,
      "loss": 3.0384,
      "step": 5055
    },
    {
      "epoch": 0.9529302369378171,
      "grad_norm": 7.460980415344238,
      "learning_rate": 1e-05,
      "loss": 3.1658,
      "step": 5060
    },
    {
      "epoch": 0.9538718676067279,
      "grad_norm": 6.758791923522949,
      "learning_rate": 1e-05,
      "loss": 3.0877,
      "step": 5065
    },
    {
      "epoch": 0.9548134982756389,
      "grad_norm": 6.621381759643555,
      "learning_rate": 1e-05,
      "loss": 2.9102,
      "step": 5070
    },
    {
      "epoch": 0.9557551289445497,
      "grad_norm": 7.349249362945557,
      "learning_rate": 1e-05,
      "loss": 2.9883,
      "step": 5075
    },
    {
      "epoch": 0.9566967596134606,
      "grad_norm": 7.027257919311523,
      "learning_rate": 1e-05,
      "loss": 3.0714,
      "step": 5080
    },
    {
      "epoch": 0.9576383902823715,
      "grad_norm": 7.285141468048096,
      "learning_rate": 1e-05,
      "loss": 2.9703,
      "step": 5085
    },
    {
      "epoch": 0.9585800209512824,
      "grad_norm": 6.913536548614502,
      "learning_rate": 1e-05,
      "loss": 3.0111,
      "step": 5090
    },
    {
      "epoch": 0.9595216516201933,
      "grad_norm": 6.543398857116699,
      "learning_rate": 1e-05,
      "loss": 3.0237,
      "step": 5095
    },
    {
      "epoch": 0.9604632822891042,
      "grad_norm": 6.569692611694336,
      "learning_rate": 1e-05,
      "loss": 2.9803,
      "step": 5100
    },
    {
      "epoch": 0.961404912958015,
      "grad_norm": 6.915118217468262,
      "learning_rate": 1e-05,
      "loss": 2.941,
      "step": 5105
    },
    {
      "epoch": 0.962346543626926,
      "grad_norm": 7.568630695343018,
      "learning_rate": 1e-05,
      "loss": 3.1096,
      "step": 5110
    },
    {
      "epoch": 0.9632881742958368,
      "grad_norm": 6.865966320037842,
      "learning_rate": 1e-05,
      "loss": 3.0179,
      "step": 5115
    },
    {
      "epoch": 0.9642298049647477,
      "grad_norm": 6.584255695343018,
      "learning_rate": 1e-05,
      "loss": 3.1118,
      "step": 5120
    },
    {
      "epoch": 0.9651714356336586,
      "grad_norm": 6.9417314529418945,
      "learning_rate": 1e-05,
      "loss": 2.9616,
      "step": 5125
    },
    {
      "epoch": 0.9661130663025694,
      "grad_norm": 7.483455181121826,
      "learning_rate": 1e-05,
      "loss": 2.8741,
      "step": 5130
    },
    {
      "epoch": 0.9670546969714804,
      "grad_norm": 6.249847412109375,
      "learning_rate": 1e-05,
      "loss": 3.0775,
      "step": 5135
    },
    {
      "epoch": 0.9679963276403912,
      "grad_norm": 7.1288018226623535,
      "learning_rate": 1e-05,
      "loss": 3.0476,
      "step": 5140
    },
    {
      "epoch": 0.9689379583093022,
      "grad_norm": 6.932041168212891,
      "learning_rate": 1e-05,
      "loss": 3.0245,
      "step": 5145
    },
    {
      "epoch": 0.969879588978213,
      "grad_norm": 7.062252044677734,
      "learning_rate": 1e-05,
      "loss": 2.9943,
      "step": 5150
    },
    {
      "epoch": 0.9708212196471239,
      "grad_norm": 7.1132121086120605,
      "learning_rate": 1e-05,
      "loss": 3.0385,
      "step": 5155
    },
    {
      "epoch": 0.9717628503160348,
      "grad_norm": 6.818399429321289,
      "learning_rate": 1e-05,
      "loss": 3.0848,
      "step": 5160
    },
    {
      "epoch": 0.9727044809849457,
      "grad_norm": 6.7855610847473145,
      "learning_rate": 1e-05,
      "loss": 3.0152,
      "step": 5165
    },
    {
      "epoch": 0.9736461116538566,
      "grad_norm": 6.324885368347168,
      "learning_rate": 1e-05,
      "loss": 2.91,
      "step": 5170
    },
    {
      "epoch": 0.9745877423227675,
      "grad_norm": 6.335012435913086,
      "learning_rate": 1e-05,
      "loss": 3.0045,
      "step": 5175
    },
    {
      "epoch": 0.9755293729916783,
      "grad_norm": 6.50994873046875,
      "learning_rate": 1e-05,
      "loss": 2.9847,
      "step": 5180
    },
    {
      "epoch": 0.9764710036605893,
      "grad_norm": 6.5178680419921875,
      "learning_rate": 1e-05,
      "loss": 2.9757,
      "step": 5185
    },
    {
      "epoch": 0.9774126343295001,
      "grad_norm": 7.039647102355957,
      "learning_rate": 1e-05,
      "loss": 2.9804,
      "step": 5190
    },
    {
      "epoch": 0.978354264998411,
      "grad_norm": 7.136976718902588,
      "learning_rate": 1e-05,
      "loss": 2.9841,
      "step": 5195
    },
    {
      "epoch": 0.9792958956673219,
      "grad_norm": 6.711559772491455,
      "learning_rate": 1e-05,
      "loss": 3.1176,
      "step": 5200
    },
    {
      "epoch": 0.9802375263362327,
      "grad_norm": 6.82883882522583,
      "learning_rate": 1e-05,
      "loss": 2.9664,
      "step": 5205
    },
    {
      "epoch": 0.9811791570051437,
      "grad_norm": 7.072768688201904,
      "learning_rate": 1e-05,
      "loss": 2.9684,
      "step": 5210
    },
    {
      "epoch": 0.9821207876740545,
      "grad_norm": 6.364959716796875,
      "learning_rate": 1e-05,
      "loss": 2.8905,
      "step": 5215
    },
    {
      "epoch": 0.9830624183429655,
      "grad_norm": 6.766102313995361,
      "learning_rate": 1e-05,
      "loss": 3.0164,
      "step": 5220
    },
    {
      "epoch": 0.9840040490118763,
      "grad_norm": 8.253533363342285,
      "learning_rate": 1e-05,
      "loss": 2.9722,
      "step": 5225
    },
    {
      "epoch": 0.9849456796807872,
      "grad_norm": 7.092304229736328,
      "learning_rate": 1e-05,
      "loss": 2.9922,
      "step": 5230
    },
    {
      "epoch": 0.9858873103496981,
      "grad_norm": 6.579424858093262,
      "learning_rate": 1e-05,
      "loss": 2.9465,
      "step": 5235
    },
    {
      "epoch": 0.986828941018609,
      "grad_norm": 7.015688896179199,
      "learning_rate": 1e-05,
      "loss": 2.9417,
      "step": 5240
    },
    {
      "epoch": 0.9877705716875199,
      "grad_norm": 7.938333511352539,
      "learning_rate": 1e-05,
      "loss": 3.0678,
      "step": 5245
    },
    {
      "epoch": 0.9887122023564308,
      "grad_norm": 6.95517635345459,
      "learning_rate": 1e-05,
      "loss": 2.9374,
      "step": 5250
    },
    {
      "epoch": 0.9896538330253416,
      "grad_norm": 6.90593957901001,
      "learning_rate": 1e-05,
      "loss": 2.9633,
      "step": 5255
    },
    {
      "epoch": 0.9905954636942526,
      "grad_norm": 7.352856636047363,
      "learning_rate": 1e-05,
      "loss": 3.0061,
      "step": 5260
    },
    {
      "epoch": 0.9915370943631634,
      "grad_norm": 7.105273723602295,
      "learning_rate": 1e-05,
      "loss": 2.9133,
      "step": 5265
    },
    {
      "epoch": 0.9924787250320743,
      "grad_norm": 6.221724033355713,
      "learning_rate": 1e-05,
      "loss": 2.9518,
      "step": 5270
    },
    {
      "epoch": 0.9934203557009852,
      "grad_norm": 6.9957380294799805,
      "learning_rate": 1e-05,
      "loss": 3.0659,
      "step": 5275
    },
    {
      "epoch": 0.994361986369896,
      "grad_norm": 6.70432186126709,
      "learning_rate": 1e-05,
      "loss": 3.0431,
      "step": 5280
    },
    {
      "epoch": 0.995303617038807,
      "grad_norm": 6.815949440002441,
      "learning_rate": 1e-05,
      "loss": 2.9397,
      "step": 5285
    },
    {
      "epoch": 0.9962452477077178,
      "grad_norm": 6.253696918487549,
      "learning_rate": 1e-05,
      "loss": 2.9773,
      "step": 5290
    },
    {
      "epoch": 0.9971868783766287,
      "grad_norm": 6.235177040100098,
      "learning_rate": 1e-05,
      "loss": 2.9691,
      "step": 5295
    },
    {
      "epoch": 0.9981285090455396,
      "grad_norm": 8.551080703735352,
      "learning_rate": 1e-05,
      "loss": 3.0107,
      "step": 5300
    },
    {
      "epoch": 0.9990701397144505,
      "grad_norm": 6.636978626251221,
      "learning_rate": 1e-05,
      "loss": 2.9,
      "step": 5305
    },
    {
      "epoch": 0.9998234442495793,
      "step": 5309,
      "total_flos": 7.961680948018545e+18,
      "train_loss": 3.093570432952174,
      "train_runtime": 64923.0121,
      "train_samples_per_second": 5.234,
      "train_steps_per_second": 0.082
    }
  ],
  "logging_steps": 5,
  "max_steps": 5309,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.961680948018545e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}