{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2757479663587481,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013787398317937406,
      "grad_norm": 1.0734721422195435,
      "learning_rate": 0.0002,
      "loss": 2.8003,
      "step": 10
    },
    {
      "epoch": 0.002757479663587481,
      "grad_norm": 0.19643057882785797,
      "learning_rate": 0.0002,
      "loss": 0.3727,
      "step": 20
    },
    {
      "epoch": 0.004136219495381221,
      "grad_norm": 0.8137874007225037,
      "learning_rate": 0.0002,
      "loss": 0.3519,
      "step": 30
    },
    {
      "epoch": 0.005514959327174962,
      "grad_norm": 0.4064357578754425,
      "learning_rate": 0.0002,
      "loss": 0.357,
      "step": 40
    },
    {
      "epoch": 0.0068936991589687024,
      "grad_norm": 1.0673978328704834,
      "learning_rate": 0.0002,
      "loss": 0.3515,
      "step": 50
    },
    {
      "epoch": 0.008272438990762443,
      "grad_norm": 2.2639453411102295,
      "learning_rate": 0.0002,
      "loss": 0.4068,
      "step": 60
    },
    {
      "epoch": 0.009651178822556184,
      "grad_norm": 1.2247616052627563,
      "learning_rate": 0.0002,
      "loss": 0.5108,
      "step": 70
    },
    {
      "epoch": 0.011029918654349925,
      "grad_norm": 0.07231716066598892,
      "learning_rate": 0.0002,
      "loss": 0.3644,
      "step": 80
    },
    {
      "epoch": 0.012408658486143665,
      "grad_norm": 1.4329577684402466,
      "learning_rate": 0.0002,
      "loss": 0.3571,
      "step": 90
    },
    {
      "epoch": 0.013787398317937405,
      "grad_norm": 0.8798255920410156,
      "learning_rate": 0.0002,
      "loss": 0.3521,
      "step": 100
    },
    {
      "epoch": 0.015166138149731145,
      "grad_norm": 0.06499180942773819,
      "learning_rate": 0.0002,
      "loss": 0.3597,
      "step": 110
    },
    {
      "epoch": 0.016544877981524885,
      "grad_norm": 0.284839928150177,
      "learning_rate": 0.0002,
      "loss": 0.4099,
      "step": 120
    },
    {
      "epoch": 0.017923617813318627,
      "grad_norm": 0.12186373770236969,
      "learning_rate": 0.0002,
      "loss": 0.3572,
      "step": 130
    },
    {
      "epoch": 0.01930235764511237,
      "grad_norm": 0.21747085452079773,
      "learning_rate": 0.0002,
      "loss": 0.3497,
      "step": 140
    },
    {
      "epoch": 0.020681097476906107,
      "grad_norm": 0.22812429070472717,
      "learning_rate": 0.0002,
      "loss": 0.3778,
      "step": 150
    },
    {
      "epoch": 0.02205983730869985,
      "grad_norm": 1.1110987663269043,
      "learning_rate": 0.0002,
      "loss": 0.3607,
      "step": 160
    },
    {
      "epoch": 0.023438577140493588,
      "grad_norm": 0.8457236886024475,
      "learning_rate": 0.0002,
      "loss": 0.9183,
      "step": 170
    },
    {
      "epoch": 0.02481731697228733,
      "grad_norm": 0.46427470445632935,
      "learning_rate": 0.0002,
      "loss": 0.3503,
      "step": 180
    },
    {
      "epoch": 0.02619605680408107,
      "grad_norm": 2.4730632305145264,
      "learning_rate": 0.0002,
      "loss": 0.3824,
      "step": 190
    },
    {
      "epoch": 0.02757479663587481,
      "grad_norm": 0.3589227497577667,
      "learning_rate": 0.0002,
      "loss": 1.0105,
      "step": 200
    },
    {
      "epoch": 0.02895353646766855,
      "grad_norm": 3.5308260917663574,
      "learning_rate": 0.0002,
      "loss": 0.3605,
      "step": 210
    },
    {
      "epoch": 0.03033227629946229,
      "grad_norm": 0.5237946510314941,
      "learning_rate": 0.0002,
      "loss": 0.3983,
      "step": 220
    },
    {
      "epoch": 0.031711016131256035,
      "grad_norm": 0.5702632665634155,
      "learning_rate": 0.0002,
      "loss": 0.3521,
      "step": 230
    },
    {
      "epoch": 0.03308975596304977,
      "grad_norm": 1.1318608522415161,
      "learning_rate": 0.0002,
      "loss": 0.3617,
      "step": 240
    },
    {
      "epoch": 0.03446849579484351,
      "grad_norm": 0.536571741104126,
      "learning_rate": 0.0002,
      "loss": 0.3555,
      "step": 250
    },
    {
      "epoch": 0.035847235626637254,
      "grad_norm": 0.12125778943300247,
      "learning_rate": 0.0002,
      "loss": 0.3733,
      "step": 260
    },
    {
      "epoch": 0.037225975458430996,
      "grad_norm": 0.045536063611507416,
      "learning_rate": 0.0002,
      "loss": 0.3514,
      "step": 270
    },
    {
      "epoch": 0.03860471529022474,
      "grad_norm": 0.31765612959861755,
      "learning_rate": 0.0002,
      "loss": 0.3536,
      "step": 280
    },
    {
      "epoch": 0.03998345512201847,
      "grad_norm": 0.27900660037994385,
      "learning_rate": 0.0002,
      "loss": 0.3495,
      "step": 290
    },
    {
      "epoch": 0.041362194953812215,
      "grad_norm": 0.15112566947937012,
      "learning_rate": 0.0002,
      "loss": 0.355,
      "step": 300
    },
    {
      "epoch": 0.04274093478560596,
      "grad_norm": 0.2682303786277771,
      "learning_rate": 0.0002,
      "loss": 0.3609,
      "step": 310
    },
    {
      "epoch": 0.0441196746173997,
      "grad_norm": 1.0106860399246216,
      "learning_rate": 0.0002,
      "loss": 0.3697,
      "step": 320
    },
    {
      "epoch": 0.04549841444919344,
      "grad_norm": 1.0782426595687866,
      "learning_rate": 0.0002,
      "loss": 0.3594,
      "step": 330
    },
    {
      "epoch": 0.046877154280987175,
      "grad_norm": 2.294581651687622,
      "learning_rate": 0.0002,
      "loss": 0.3676,
      "step": 340
    },
    {
      "epoch": 0.04825589411278092,
      "grad_norm": 0.6223801970481873,
      "learning_rate": 0.0002,
      "loss": 0.3741,
      "step": 350
    },
    {
      "epoch": 0.04963463394457466,
      "grad_norm": 0.2735952138900757,
      "learning_rate": 0.0002,
      "loss": 0.3628,
      "step": 360
    },
    {
      "epoch": 0.0510133737763684,
      "grad_norm": 0.7569056153297424,
      "learning_rate": 0.0002,
      "loss": 1.1827,
      "step": 370
    },
    {
      "epoch": 0.05239211360816214,
      "grad_norm": 0.6536706686019897,
      "learning_rate": 0.0002,
      "loss": 0.3543,
      "step": 380
    },
    {
      "epoch": 0.05377085343995588,
      "grad_norm": 0.3573110103607178,
      "learning_rate": 0.0002,
      "loss": 0.3529,
      "step": 390
    },
    {
      "epoch": 0.05514959327174962,
      "grad_norm": 0.8121228218078613,
      "learning_rate": 0.0002,
      "loss": 0.3566,
      "step": 400
    },
    {
      "epoch": 0.05652833310354336,
      "grad_norm": 0.7444269061088562,
      "learning_rate": 0.0002,
      "loss": 0.7803,
      "step": 410
    },
    {
      "epoch": 0.0579070729353371,
      "grad_norm": 0.5015038847923279,
      "learning_rate": 0.0002,
      "loss": 0.6134,
      "step": 420
    },
    {
      "epoch": 0.059285812767130845,
      "grad_norm": 0.08748292177915573,
      "learning_rate": 0.0002,
      "loss": 0.3501,
      "step": 430
    },
    {
      "epoch": 0.06066455259892458,
      "grad_norm": 0.3987080156803131,
      "learning_rate": 0.0002,
      "loss": 0.3484,
      "step": 440
    },
    {
      "epoch": 0.06204329243071832,
      "grad_norm": 347.7005920410156,
      "learning_rate": 0.0002,
      "loss": 0.8413,
      "step": 450
    },
    {
      "epoch": 0.06342203226251207,
      "grad_norm": 88.2750473022461,
      "learning_rate": 0.0002,
      "loss": 2.8013,
      "step": 460
    },
    {
      "epoch": 0.0648007720943058,
      "grad_norm": 0.8716701865196228,
      "learning_rate": 0.0002,
      "loss": 0.8356,
      "step": 470
    },
    {
      "epoch": 0.06617951192609954,
      "grad_norm": 0.8243119120597839,
      "learning_rate": 0.0002,
      "loss": 0.3616,
      "step": 480
    },
    {
      "epoch": 0.06755825175789329,
      "grad_norm": 1.1744294166564941,
      "learning_rate": 0.0002,
      "loss": 0.3998,
      "step": 490
    },
    {
      "epoch": 0.06893699158968702,
      "grad_norm": 0.03163053095340729,
      "learning_rate": 0.0002,
      "loss": 0.3549,
      "step": 500
    },
    {
      "epoch": 0.07031573142148077,
      "grad_norm": 3.4403915405273438,
      "learning_rate": 0.0002,
      "loss": 0.3722,
      "step": 510
    },
    {
      "epoch": 0.07169447125327451,
      "grad_norm": 1.0608879327774048,
      "learning_rate": 0.0002,
      "loss": 0.3701,
      "step": 520
    },
    {
      "epoch": 0.07307321108506824,
      "grad_norm": 1.2809940576553345,
      "learning_rate": 0.0002,
      "loss": 0.3602,
      "step": 530
    },
    {
      "epoch": 0.07445195091686199,
      "grad_norm": 0.40460118651390076,
      "learning_rate": 0.0002,
      "loss": 0.3532,
      "step": 540
    },
    {
      "epoch": 0.07583069074865573,
      "grad_norm": 0.6290703415870667,
      "learning_rate": 0.0002,
      "loss": 0.3477,
      "step": 550
    },
    {
      "epoch": 0.07720943058044948,
      "grad_norm": 0.2159261256456375,
      "learning_rate": 0.0002,
      "loss": 0.405,
      "step": 560
    },
    {
      "epoch": 0.07858817041224321,
      "grad_norm": 0.37101301550865173,
      "learning_rate": 0.0002,
      "loss": 0.3752,
      "step": 570
    },
    {
      "epoch": 0.07996691024403695,
      "grad_norm": 1.3007190227508545,
      "learning_rate": 0.0002,
      "loss": 0.3503,
      "step": 580
    },
    {
      "epoch": 0.0813456500758307,
      "grad_norm": 0.4508918225765228,
      "learning_rate": 0.0002,
      "loss": 0.3531,
      "step": 590
    },
    {
      "epoch": 0.08272438990762443,
      "grad_norm": 0.46898791193962097,
      "learning_rate": 0.0002,
      "loss": 0.347,
      "step": 600
    },
    {
      "epoch": 0.08410312973941818,
      "grad_norm": 0.8449831604957581,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 610
    },
    {
      "epoch": 0.08548186957121191,
      "grad_norm": 0.7988163232803345,
      "learning_rate": 0.0002,
      "loss": 0.3505,
      "step": 620
    },
    {
      "epoch": 0.08686060940300565,
      "grad_norm": 0.4426226317882538,
      "learning_rate": 0.0002,
      "loss": 0.3649,
      "step": 630
    },
    {
      "epoch": 0.0882393492347994,
      "grad_norm": 0.2260913848876953,
      "learning_rate": 0.0002,
      "loss": 0.3499,
      "step": 640
    },
    {
      "epoch": 0.08961808906659313,
      "grad_norm": 1.476747751235962,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 650
    },
    {
      "epoch": 0.09099682889838688,
      "grad_norm": 0.7640777230262756,
      "learning_rate": 0.0002,
      "loss": 0.3568,
      "step": 660
    },
    {
      "epoch": 0.09237556873018062,
      "grad_norm": 0.8559088706970215,
      "learning_rate": 0.0002,
      "loss": 0.3507,
      "step": 670
    },
    {
      "epoch": 0.09375430856197435,
      "grad_norm": 0.20833595097064972,
      "learning_rate": 0.0002,
      "loss": 0.3556,
      "step": 680
    },
    {
      "epoch": 0.0951330483937681,
      "grad_norm": 1.1485021114349365,
      "learning_rate": 0.0002,
      "loss": 0.3516,
      "step": 690
    },
    {
      "epoch": 0.09651178822556183,
      "grad_norm": 1.0206815004348755,
      "learning_rate": 0.0002,
      "loss": 0.3591,
      "step": 700
    },
    {
      "epoch": 0.09789052805735558,
      "grad_norm": 0.9966775178909302,
      "learning_rate": 0.0002,
      "loss": 0.359,
      "step": 710
    },
    {
      "epoch": 0.09926926788914932,
      "grad_norm": 0.8833585977554321,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 720
    },
    {
      "epoch": 0.10064800772094305,
      "grad_norm": 1.0842584371566772,
      "learning_rate": 0.0002,
      "loss": 0.3556,
      "step": 730
    },
    {
      "epoch": 0.1020267475527368,
      "grad_norm": 0.3791511058807373,
      "learning_rate": 0.0002,
      "loss": 0.3571,
      "step": 740
    },
    {
      "epoch": 0.10340548738453054,
      "grad_norm": 0.24666732549667358,
      "learning_rate": 0.0002,
      "loss": 0.3502,
      "step": 750
    },
    {
      "epoch": 0.10478422721632429,
      "grad_norm": 0.21794968843460083,
      "learning_rate": 0.0002,
      "loss": 0.3483,
      "step": 760
    },
    {
      "epoch": 0.10616296704811802,
      "grad_norm": 0.47017499804496765,
      "learning_rate": 0.0002,
      "loss": 0.3499,
      "step": 770
    },
    {
      "epoch": 0.10754170687991176,
      "grad_norm": 0.2813131809234619,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 780
    },
    {
      "epoch": 0.1089204467117055,
      "grad_norm": 1.2175363302230835,
      "learning_rate": 0.0002,
      "loss": 0.3524,
      "step": 790
    },
    {
      "epoch": 0.11029918654349924,
      "grad_norm": 0.2712210416793823,
      "learning_rate": 0.0002,
      "loss": 0.3526,
      "step": 800
    },
    {
      "epoch": 0.11167792637529299,
      "grad_norm": 0.1428445726633072,
      "learning_rate": 0.0002,
      "loss": 0.3518,
      "step": 810
    },
    {
      "epoch": 0.11305666620708672,
      "grad_norm": 0.23716595768928528,
      "learning_rate": 0.0002,
      "loss": 0.3501,
      "step": 820
    },
    {
      "epoch": 0.11443540603888046,
      "grad_norm": 0.07993923872709274,
      "learning_rate": 0.0002,
      "loss": 0.349,
      "step": 830
    },
    {
      "epoch": 0.1158141458706742,
      "grad_norm": 1.2958595752716064,
      "learning_rate": 0.0002,
      "loss": 0.352,
      "step": 840
    },
    {
      "epoch": 0.11719288570246794,
      "grad_norm": 1.6257132291793823,
      "learning_rate": 0.0002,
      "loss": 0.3589,
      "step": 850
    },
    {
      "epoch": 0.11857162553426169,
      "grad_norm": 0.20367591083049774,
      "learning_rate": 0.0002,
      "loss": 0.3602,
      "step": 860
    },
    {
      "epoch": 0.11995036536605543,
      "grad_norm": 1.147210955619812,
      "learning_rate": 0.0002,
      "loss": 0.3619,
      "step": 870
    },
    {
      "epoch": 0.12132910519784916,
      "grad_norm": 0.19706425070762634,
      "learning_rate": 0.0002,
      "loss": 0.3536,
      "step": 880
    },
    {
      "epoch": 0.12270784502964291,
      "grad_norm": 0.17990930378437042,
      "learning_rate": 0.0002,
      "loss": 0.3501,
      "step": 890
    },
    {
      "epoch": 0.12408658486143664,
      "grad_norm": 0.5770463943481445,
      "learning_rate": 0.0002,
      "loss": 0.3509,
      "step": 900
    },
    {
      "epoch": 0.12546532469323038,
      "grad_norm": 0.24645955860614777,
      "learning_rate": 0.0002,
      "loss": 0.3526,
      "step": 910
    },
    {
      "epoch": 0.12684406452502414,
      "grad_norm": 0.15745119750499725,
      "learning_rate": 0.0002,
      "loss": 0.3503,
      "step": 920
    },
    {
      "epoch": 0.12822280435681788,
      "grad_norm": 0.054484590888023376,
      "learning_rate": 0.0002,
      "loss": 0.3508,
      "step": 930
    },
    {
      "epoch": 0.1296015441886116,
      "grad_norm": 0.30564025044441223,
      "learning_rate": 0.0002,
      "loss": 0.3489,
      "step": 940
    },
    {
      "epoch": 0.13098028402040535,
      "grad_norm": 0.3614678382873535,
      "learning_rate": 0.0002,
      "loss": 0.3477,
      "step": 950
    },
    {
      "epoch": 0.13235902385219908,
      "grad_norm": 0.703029990196228,
      "learning_rate": 0.0002,
      "loss": 0.3552,
      "step": 960
    },
    {
      "epoch": 0.13373776368399284,
      "grad_norm": 1.1954560279846191,
      "learning_rate": 0.0002,
      "loss": 0.3528,
      "step": 970
    },
    {
      "epoch": 0.13511650351578658,
      "grad_norm": 0.8106504678726196,
      "learning_rate": 0.0002,
      "loss": 0.3586,
      "step": 980
    },
    {
      "epoch": 0.13649524334758031,
      "grad_norm": 0.40758854150772095,
      "learning_rate": 0.0002,
      "loss": 0.3908,
      "step": 990
    },
    {
      "epoch": 0.13787398317937405,
      "grad_norm": 0.613096296787262,
      "learning_rate": 0.0002,
      "loss": 0.3519,
      "step": 1000
    },
    {
      "epoch": 0.13925272301116778,
      "grad_norm": 0.38185614347457886,
      "learning_rate": 0.0002,
      "loss": 0.3506,
      "step": 1010
    },
    {
      "epoch": 0.14063146284296155,
      "grad_norm": 0.07220327854156494,
      "learning_rate": 0.0002,
      "loss": 0.3472,
      "step": 1020
    },
    {
      "epoch": 0.14201020267475528,
      "grad_norm": 0.1451689898967743,
      "learning_rate": 0.0002,
      "loss": 0.3534,
      "step": 1030
    },
    {
      "epoch": 0.14338894250654902,
      "grad_norm": 0.08052591234445572,
      "learning_rate": 0.0002,
      "loss": 0.3476,
      "step": 1040
    },
    {
      "epoch": 0.14476768233834275,
      "grad_norm": 1.0108163356781006,
      "learning_rate": 0.0002,
      "loss": 0.3508,
      "step": 1050
    },
    {
      "epoch": 0.1461464221701365,
      "grad_norm": 0.5895722508430481,
      "learning_rate": 0.0002,
      "loss": 0.3541,
      "step": 1060
    },
    {
      "epoch": 0.14752516200193025,
      "grad_norm": 0.6988415718078613,
      "learning_rate": 0.0002,
      "loss": 0.3512,
      "step": 1070
    },
    {
      "epoch": 0.14890390183372398,
      "grad_norm": 0.54078608751297,
      "learning_rate": 0.0002,
      "loss": 0.3479,
      "step": 1080
    },
    {
      "epoch": 0.15028264166551772,
      "grad_norm": 0.19162333011627197,
      "learning_rate": 0.0002,
      "loss": 0.3518,
      "step": 1090
    },
    {
      "epoch": 0.15166138149731145,
      "grad_norm": 0.36928215622901917,
      "learning_rate": 0.0002,
      "loss": 0.3505,
      "step": 1100
    },
    {
      "epoch": 0.1530401213291052,
      "grad_norm": 0.572607696056366,
      "learning_rate": 0.0002,
      "loss": 0.355,
      "step": 1110
    },
    {
      "epoch": 0.15441886116089895,
      "grad_norm": 0.20841191709041595,
      "learning_rate": 0.0002,
      "loss": 0.348,
      "step": 1120
    },
    {
      "epoch": 0.1557976009926927,
      "grad_norm": 0.04682110995054245,
      "learning_rate": 0.0002,
      "loss": 0.3502,
      "step": 1130
    },
    {
      "epoch": 0.15717634082448642,
      "grad_norm": 0.867899477481842,
      "learning_rate": 0.0002,
      "loss": 0.3476,
      "step": 1140
    },
    {
      "epoch": 0.15855508065628016,
      "grad_norm": 0.2828502655029297,
      "learning_rate": 0.0002,
      "loss": 0.3525,
      "step": 1150
    },
    {
      "epoch": 0.1599338204880739,
      "grad_norm": 0.44510889053344727,
      "learning_rate": 0.0002,
      "loss": 0.35,
      "step": 1160
    },
    {
      "epoch": 0.16131256031986765,
      "grad_norm": 0.1896822154521942,
      "learning_rate": 0.0002,
      "loss": 0.3493,
      "step": 1170
    },
    {
      "epoch": 0.1626913001516614,
      "grad_norm": 0.15781590342521667,
      "learning_rate": 0.0002,
      "loss": 0.3477,
      "step": 1180
    },
    {
      "epoch": 0.16407003998345512,
      "grad_norm": 0.2315225899219513,
      "learning_rate": 0.0002,
      "loss": 0.3498,
      "step": 1190
    },
    {
      "epoch": 0.16544877981524886,
      "grad_norm": 0.2198018729686737,
      "learning_rate": 0.0002,
      "loss": 0.3484,
      "step": 1200
    },
    {
      "epoch": 0.1668275196470426,
      "grad_norm": 0.2039571851491928,
      "learning_rate": 0.0002,
      "loss": 0.348,
      "step": 1210
    },
    {
      "epoch": 0.16820625947883636,
      "grad_norm": 0.009352603927254677,
      "learning_rate": 0.0002,
      "loss": 0.3481,
      "step": 1220
    },
    {
      "epoch": 0.1695849993106301,
      "grad_norm": 0.2558707892894745,
      "learning_rate": 0.0002,
      "loss": 0.3475,
      "step": 1230
    },
    {
      "epoch": 0.17096373914242383,
      "grad_norm": 0.07278712838888168,
      "learning_rate": 0.0002,
      "loss": 0.3471,
      "step": 1240
    },
    {
      "epoch": 0.17234247897421756,
      "grad_norm": 0.4133436381816864,
      "learning_rate": 0.0002,
      "loss": 0.3453,
      "step": 1250
    },
    {
      "epoch": 0.1737212188060113,
      "grad_norm": 0.16729828715324402,
      "learning_rate": 0.0002,
      "loss": 0.3527,
      "step": 1260
    },
    {
      "epoch": 0.17509995863780506,
      "grad_norm": 0.33326980471611023,
      "learning_rate": 0.0002,
      "loss": 0.3463,
      "step": 1270
    },
    {
      "epoch": 0.1764786984695988,
      "grad_norm": 0.7140666246414185,
      "learning_rate": 0.0002,
      "loss": 0.3627,
      "step": 1280
    },
    {
      "epoch": 0.17785743830139253,
      "grad_norm": 0.17751634120941162,
      "learning_rate": 0.0002,
      "loss": 0.3506,
      "step": 1290
    },
    {
      "epoch": 0.17923617813318626,
      "grad_norm": 0.44009125232696533,
      "learning_rate": 0.0002,
      "loss": 0.3516,
      "step": 1300
    },
    {
      "epoch": 0.18061491796498,
      "grad_norm": 0.07371579110622406,
      "learning_rate": 0.0002,
      "loss": 0.3479,
      "step": 1310
    },
    {
      "epoch": 0.18199365779677376,
      "grad_norm": 0.6804266571998596,
      "learning_rate": 0.0002,
      "loss": 0.3476,
      "step": 1320
    },
    {
      "epoch": 0.1833723976285675,
      "grad_norm": 0.19634029269218445,
      "learning_rate": 0.0002,
      "loss": 0.354,
      "step": 1330
    },
    {
      "epoch": 0.18475113746036123,
      "grad_norm": 0.34020882844924927,
      "learning_rate": 0.0002,
      "loss": 0.3481,
      "step": 1340
    },
    {
      "epoch": 0.18612987729215497,
      "grad_norm": 0.38502731919288635,
      "learning_rate": 0.0002,
      "loss": 0.3501,
      "step": 1350
    },
    {
      "epoch": 0.1875086171239487,
      "grad_norm": 0.0810522586107254,
      "learning_rate": 0.0002,
      "loss": 0.3473,
      "step": 1360
    },
    {
      "epoch": 0.18888735695574246,
      "grad_norm": 0.4057389497756958,
      "learning_rate": 0.0002,
      "loss": 0.3489,
      "step": 1370
    },
    {
      "epoch": 0.1902660967875362,
      "grad_norm": 0.17514599859714508,
      "learning_rate": 0.0002,
      "loss": 0.3472,
      "step": 1380
    },
    {
      "epoch": 0.19164483661932993,
      "grad_norm": 0.10964088141918182,
      "learning_rate": 0.0002,
      "loss": 0.3479,
      "step": 1390
    },
    {
      "epoch": 0.19302357645112367,
      "grad_norm": 0.20920871198177338,
      "learning_rate": 0.0002,
      "loss": 0.3488,
      "step": 1400
    },
    {
      "epoch": 0.1944023162829174,
      "grad_norm": 1.149121880531311,
      "learning_rate": 0.0002,
      "loss": 0.3548,
      "step": 1410
    },
    {
      "epoch": 0.19578105611471117,
      "grad_norm": 1.3394649028778076,
      "learning_rate": 0.0002,
      "loss": 0.3495,
      "step": 1420
    },
    {
      "epoch": 0.1971597959465049,
      "grad_norm": 1.2763960361480713,
      "learning_rate": 0.0002,
      "loss": 0.3679,
      "step": 1430
    },
    {
      "epoch": 0.19853853577829864,
      "grad_norm": 0.5421571731567383,
      "learning_rate": 0.0002,
      "loss": 0.3538,
      "step": 1440
    },
    {
      "epoch": 0.19991727561009237,
      "grad_norm": 0.22273503243923187,
      "learning_rate": 0.0002,
      "loss": 0.3518,
      "step": 1450
    },
    {
      "epoch": 0.2012960154418861,
      "grad_norm": 0.6335702538490295,
      "learning_rate": 0.0002,
      "loss": 0.3481,
      "step": 1460
    },
    {
      "epoch": 0.20267475527367987,
      "grad_norm": 0.7090324759483337,
      "learning_rate": 0.0002,
      "loss": 0.3486,
      "step": 1470
    },
    {
      "epoch": 0.2040534951054736,
      "grad_norm": 0.011333847418427467,
      "learning_rate": 0.0002,
      "loss": 0.3476,
      "step": 1480
    },
    {
      "epoch": 0.20543223493726734,
      "grad_norm": 0.24088676273822784,
      "learning_rate": 0.0002,
      "loss": 0.3569,
      "step": 1490
    },
    {
      "epoch": 0.20681097476906107,
      "grad_norm": 0.8654371500015259,
      "learning_rate": 0.0002,
      "loss": 0.3528,
      "step": 1500
    },
    {
      "epoch": 0.2081897146008548,
      "grad_norm": 0.06135034188628197,
      "learning_rate": 0.0002,
      "loss": 0.3509,
      "step": 1510
    },
    {
      "epoch": 0.20956845443264857,
      "grad_norm": 0.38141730427742004,
      "learning_rate": 0.0002,
      "loss": 0.3474,
      "step": 1520
    },
    {
      "epoch": 0.2109471942644423,
      "grad_norm": 1.1622456312179565,
      "learning_rate": 0.0002,
      "loss": 0.353,
      "step": 1530
    },
    {
      "epoch": 0.21232593409623604,
      "grad_norm": 0.5747712254524231,
      "learning_rate": 0.0002,
      "loss": 0.3513,
      "step": 1540
    },
    {
      "epoch": 0.21370467392802978,
      "grad_norm": 0.09723293781280518,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 1550
    },
    {
      "epoch": 0.2150834137598235,
      "grad_norm": 0.18574804067611694,
      "learning_rate": 0.0002,
      "loss": 0.3512,
      "step": 1560
    },
    {
      "epoch": 0.21646215359161727,
      "grad_norm": 0.33651217818260193,
      "learning_rate": 0.0002,
      "loss": 0.349,
      "step": 1570
    },
    {
      "epoch": 0.217840893423411,
      "grad_norm": 0.07309216260910034,
      "learning_rate": 0.0002,
      "loss": 0.3493,
      "step": 1580
    },
    {
      "epoch": 0.21921963325520474,
      "grad_norm": 0.19346486032009125,
      "learning_rate": 0.0002,
      "loss": 0.3478,
      "step": 1590
    },
    {
      "epoch": 0.22059837308699848,
      "grad_norm": 0.3398933708667755,
      "learning_rate": 0.0002,
      "loss": 0.3496,
      "step": 1600
    },
    {
      "epoch": 0.2219771129187922,
      "grad_norm": 0.34032130241394043,
      "learning_rate": 0.0002,
      "loss": 0.3488,
      "step": 1610
    },
    {
      "epoch": 0.22335585275058598,
      "grad_norm": 0.901030421257019,
      "learning_rate": 0.0002,
      "loss": 0.3531,
      "step": 1620
    },
    {
      "epoch": 0.2247345925823797,
      "grad_norm": 0.500088632106781,
      "learning_rate": 0.0002,
      "loss": 0.3516,
      "step": 1630
    },
    {
      "epoch": 0.22611333241417345,
      "grad_norm": 0.3230324387550354,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 1640
    },
    {
      "epoch": 0.22749207224596718,
      "grad_norm": 1.2476601600646973,
      "learning_rate": 0.0002,
      "loss": 0.3512,
      "step": 1650
    },
    {
      "epoch": 0.22887081207776092,
      "grad_norm": 0.23318485915660858,
      "learning_rate": 0.0002,
      "loss": 0.3456,
      "step": 1660
    },
    {
      "epoch": 0.23024955190955468,
      "grad_norm": 0.472400963306427,
      "learning_rate": 0.0002,
      "loss": 0.3551,
      "step": 1670
    },
    {
      "epoch": 0.2316282917413484,
      "grad_norm": 0.04836912825703621,
      "learning_rate": 0.0002,
      "loss": 0.3505,
      "step": 1680
    },
    {
      "epoch": 0.23300703157314215,
      "grad_norm": 0.34590113162994385,
      "learning_rate": 0.0002,
      "loss": 0.3507,
      "step": 1690
    },
    {
      "epoch": 0.23438577140493588,
      "grad_norm": 0.23341989517211914,
      "learning_rate": 0.0002,
      "loss": 0.3478,
      "step": 1700
    },
    {
      "epoch": 0.23576451123672962,
      "grad_norm": 0.001271920627914369,
      "learning_rate": 0.0002,
      "loss": 0.3562,
      "step": 1710
    },
    {
      "epoch": 0.23714325106852338,
      "grad_norm": 0.20549911260604858,
      "learning_rate": 0.0002,
      "loss": 0.3479,
      "step": 1720
    },
    {
      "epoch": 0.23852199090031712,
      "grad_norm": 0.3825775384902954,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 1730
    },
    {
      "epoch": 0.23990073073211085,
      "grad_norm": 0.028804048895835876,
      "learning_rate": 0.0002,
      "loss": 0.3481,
      "step": 1740
    },
    {
      "epoch": 0.24127947056390459,
      "grad_norm": 0.04462611302733421,
      "learning_rate": 0.0002,
      "loss": 0.3477,
      "step": 1750
    },
    {
      "epoch": 0.24265821039569832,
      "grad_norm": 0.6634818315505981,
      "learning_rate": 0.0002,
      "loss": 0.3501,
      "step": 1760
    },
    {
      "epoch": 0.24403695022749208,
      "grad_norm": 1.3807406425476074,
      "learning_rate": 0.0002,
      "loss": 0.3547,
      "step": 1770
    },
    {
      "epoch": 0.24541569005928582,
      "grad_norm": 0.24347831308841705,
      "learning_rate": 0.0002,
      "loss": 0.3473,
      "step": 1780
    },
    {
      "epoch": 0.24679442989107955,
      "grad_norm": 0.61258465051651,
      "learning_rate": 0.0002,
      "loss": 0.3492,
      "step": 1790
    },
    {
      "epoch": 0.2481731697228733,
      "grad_norm": 0.011543272994458675,
      "learning_rate": 0.0002,
      "loss": 0.347,
      "step": 1800
    },
    {
      "epoch": 0.24955190955466702,
      "grad_norm": 0.09996844828128815,
      "learning_rate": 0.0002,
      "loss": 0.3474,
      "step": 1810
    },
    {
      "epoch": 0.25093064938646076,
      "grad_norm": 0.17044603824615479,
      "learning_rate": 0.0002,
      "loss": 0.3471,
      "step": 1820
    },
    {
      "epoch": 0.2523093892182545,
      "grad_norm": 0.17940489947795868,
      "learning_rate": 0.0002,
      "loss": 0.3478,
      "step": 1830
    },
    {
      "epoch": 0.2536881290500483,
      "grad_norm": 0.21834205090999603,
      "learning_rate": 0.0002,
      "loss": 0.3461,
      "step": 1840
    },
    {
      "epoch": 0.255066868881842,
      "grad_norm": 0.2272634655237198,
      "learning_rate": 0.0002,
      "loss": 0.347,
      "step": 1850
    },
    {
      "epoch": 0.25644560871363575,
      "grad_norm": 0.18734070658683777,
      "learning_rate": 0.0002,
      "loss": 0.3497,
      "step": 1860
    },
    {
      "epoch": 0.25782434854542946,
      "grad_norm": 0.04078834876418114,
      "learning_rate": 0.0002,
      "loss": 0.3493,
      "step": 1870
    },
    {
      "epoch": 0.2592030883772232,
      "grad_norm": 0.3463903069496155,
      "learning_rate": 0.0002,
      "loss": 0.3494,
      "step": 1880
    },
    {
      "epoch": 0.260581828209017,
      "grad_norm": 0.3256634771823883,
      "learning_rate": 0.0002,
      "loss": 0.3488,
      "step": 1890
    },
    {
      "epoch": 0.2619605680408107,
      "grad_norm": 0.04404434189200401,
      "learning_rate": 0.0002,
      "loss": 0.3509,
      "step": 1900
    },
    {
      "epoch": 0.26333930787260446,
      "grad_norm": 0.20446011424064636,
      "learning_rate": 0.0002,
      "loss": 0.3526,
      "step": 1910
    },
    {
      "epoch": 0.26471804770439816,
      "grad_norm": 0.06089179962873459,
      "learning_rate": 0.0002,
      "loss": 0.3491,
      "step": 1920
    },
    {
      "epoch": 0.2660967875361919,
      "grad_norm": 0.32661405205726624,
      "learning_rate": 0.0002,
      "loss": 0.35,
      "step": 1930
    },
    {
      "epoch": 0.2674755273679857,
      "grad_norm": 0.09823151677846909,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 1940
    },
    {
      "epoch": 0.2688542671997794,
      "grad_norm": 0.11397412419319153,
      "learning_rate": 0.0002,
      "loss": 0.347,
      "step": 1950
    },
    {
      "epoch": 0.27023300703157316,
      "grad_norm": 0.2632172703742981,
      "learning_rate": 0.0002,
      "loss": 0.352,
      "step": 1960
    },
    {
      "epoch": 0.27161174686336687,
      "grad_norm": 0.27215296030044556,
      "learning_rate": 0.0002,
      "loss": 0.3482,
      "step": 1970
    },
    {
      "epoch": 0.27299048669516063,
      "grad_norm": 0.20016005635261536,
      "learning_rate": 0.0002,
      "loss": 0.3489,
      "step": 1980
    },
    {
      "epoch": 0.2743692265269544,
      "grad_norm": 0.3071637749671936,
      "learning_rate": 0.0002,
      "loss": 0.354,
      "step": 1990
    },
    {
      "epoch": 0.2757479663587481,
      "grad_norm": 1.0373337268829346,
      "learning_rate": 0.0002,
      "loss": 0.3481,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 14,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.09149951218339e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}