{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.3076804738279297,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0015384023691396484,
      "grad_norm": 15216.6796875,
      "learning_rate": 1.9999999999999996e-07,
      "loss": 7.628,
      "step": 10
    },
    {
      "epoch": 0.003076804738279297,
      "grad_norm": 539.3572387695312,
      "learning_rate": 1.2e-06,
      "loss": 7.3644,
      "step": 20
    },
    {
      "epoch": 0.004615207107418946,
      "grad_norm": 496.3940734863281,
      "learning_rate": 2.1999999999999997e-06,
      "loss": 7.0137,
      "step": 30
    },
    {
      "epoch": 0.006153609476558594,
      "grad_norm": 110.5931396484375,
      "learning_rate": 3.1999999999999994e-06,
      "loss": 6.9299,
      "step": 40
    },
    {
      "epoch": 0.007692011845698243,
      "grad_norm": 442.7750549316406,
      "learning_rate": 4.2e-06,
      "loss": 6.88,
      "step": 50
    },
    {
      "epoch": 0.009230414214837892,
      "grad_norm": 285.75390625,
      "learning_rate": 5.199999999999999e-06,
      "loss": 6.7252,
      "step": 60
    },
    {
      "epoch": 0.010768816583977539,
      "grad_norm": 238.201416015625,
      "learning_rate": 6.199999999999999e-06,
      "loss": 6.7641,
      "step": 70
    },
    {
      "epoch": 0.012307218953117188,
      "grad_norm": 203.7633514404297,
      "learning_rate": 7.2e-06,
      "loss": 6.6961,
      "step": 80
    },
    {
      "epoch": 0.013845621322256836,
      "grad_norm": 537.8616943359375,
      "learning_rate": 8.2e-06,
      "loss": 6.5996,
      "step": 90
    },
    {
      "epoch": 0.015384023691396485,
      "grad_norm": 643.6299438476562,
      "learning_rate": 9.199999999999998e-06,
      "loss": 6.5856,
      "step": 100
    },
    {
      "epoch": 0.016922426060536132,
      "grad_norm": 679.996337890625,
      "learning_rate": 1.02e-05,
      "loss": 6.599,
      "step": 110
    },
    {
      "epoch": 0.018460828429675783,
      "grad_norm": 452.3795471191406,
      "learning_rate": 1.12e-05,
      "loss": 6.6354,
      "step": 120
    },
    {
      "epoch": 0.01999923079881543,
      "grad_norm": 117.24376678466797,
      "learning_rate": 1.2199999999999998e-05,
      "loss": 6.5363,
      "step": 130
    },
    {
      "epoch": 0.021537633167955077,
      "grad_norm": 145.65036010742188,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 6.5564,
      "step": 140
    },
    {
      "epoch": 0.023076035537094728,
      "grad_norm": 178.4263458251953,
      "learning_rate": 1.4199999999999998e-05,
      "loss": 6.5828,
      "step": 150
    },
    {
      "epoch": 0.024614437906234375,
      "grad_norm": 174.7183837890625,
      "learning_rate": 1.5199999999999998e-05,
      "loss": 6.4821,
      "step": 160
    },
    {
      "epoch": 0.026152840275374026,
      "grad_norm": 145.78704833984375,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 6.4711,
      "step": 170
    },
    {
      "epoch": 0.027691242644513673,
      "grad_norm": 217.81866455078125,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 6.4947,
      "step": 180
    },
    {
      "epoch": 0.02922964501365332,
      "grad_norm": 105.79635620117188,
      "learning_rate": 1.82e-05,
      "loss": 6.4997,
      "step": 190
    },
    {
      "epoch": 0.03076804738279297,
      "grad_norm": 86.61837768554688,
      "learning_rate": 1.92e-05,
      "loss": 6.4604,
      "step": 200
    },
    {
      "epoch": 0.03230644975193262,
      "grad_norm": 102.03221893310547,
      "learning_rate": 2.0199999999999996e-05,
      "loss": 6.4451,
      "step": 210
    },
    {
      "epoch": 0.033844852121072265,
      "grad_norm": 93.15425109863281,
      "learning_rate": 2.12e-05,
      "loss": 6.4653,
      "step": 220
    },
    {
      "epoch": 0.035383254490211916,
      "grad_norm": 120.66065979003906,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 6.3898,
      "step": 230
    },
    {
      "epoch": 0.036921656859351566,
      "grad_norm": 108.00463104248047,
      "learning_rate": 2.3199999999999998e-05,
      "loss": 6.3907,
      "step": 240
    },
    {
      "epoch": 0.03846005922849121,
      "grad_norm": 129.6912841796875,
      "learning_rate": 2.42e-05,
      "loss": 6.3551,
      "step": 250
    },
    {
      "epoch": 0.03999846159763086,
      "grad_norm": 66.96916961669922,
      "learning_rate": 2.52e-05,
      "loss": 6.3443,
      "step": 260
    },
    {
      "epoch": 0.04153686396677051,
      "grad_norm": 44.31541061401367,
      "learning_rate": 2.6199999999999996e-05,
      "loss": 6.3372,
      "step": 270
    },
    {
      "epoch": 0.043075266335910155,
      "grad_norm": 21.828279495239258,
      "learning_rate": 2.72e-05,
      "loss": 6.3368,
      "step": 280
    },
    {
      "epoch": 0.044613668705049805,
      "grad_norm": 38.62431716918945,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 6.2708,
      "step": 290
    },
    {
      "epoch": 0.046152071074189456,
      "grad_norm": 28.7634334564209,
      "learning_rate": 2.9199999999999995e-05,
      "loss": 6.3324,
      "step": 300
    },
    {
      "epoch": 0.0476904734433291,
      "grad_norm": 29.64866065979004,
      "learning_rate": 3.02e-05,
      "loss": 6.329,
      "step": 310
    },
    {
      "epoch": 0.04922887581246875,
      "grad_norm": 27.20194435119629,
      "learning_rate": 3.119999999999999e-05,
      "loss": 6.2864,
      "step": 320
    },
    {
      "epoch": 0.0507672781816084,
      "grad_norm": 24.59855079650879,
      "learning_rate": 3.22e-05,
      "loss": 6.2371,
      "step": 330
    },
    {
      "epoch": 0.05230568055074805,
      "grad_norm": 32.6953239440918,
      "learning_rate": 3.3199999999999994e-05,
      "loss": 6.2824,
      "step": 340
    },
    {
      "epoch": 0.053844082919887695,
      "grad_norm": 36.308067321777344,
      "learning_rate": 3.42e-05,
      "loss": 6.2977,
      "step": 350
    },
    {
      "epoch": 0.055382485289027346,
      "grad_norm": 25.23465347290039,
      "learning_rate": 3.5199999999999995e-05,
      "loss": 6.3222,
      "step": 360
    },
    {
      "epoch": 0.056920887658166996,
      "grad_norm": 47.594947814941406,
      "learning_rate": 3.62e-05,
      "loss": 6.2513,
      "step": 370
    },
    {
      "epoch": 0.05845929002730664,
      "grad_norm": 44.8812370300293,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 6.2244,
      "step": 380
    },
    {
      "epoch": 0.05999769239644629,
      "grad_norm": 22.950286865234375,
      "learning_rate": 3.8199999999999993e-05,
      "loss": 6.1613,
      "step": 390
    },
    {
      "epoch": 0.06153609476558594,
      "grad_norm": 23.65997314453125,
      "learning_rate": 3.919999999999999e-05,
      "loss": 6.1839,
      "step": 400
    },
    {
      "epoch": 0.06307449713472559,
      "grad_norm": 37.60873031616211,
      "learning_rate": 4.02e-05,
      "loss": 6.1783,
      "step": 410
    },
    {
      "epoch": 0.06461289950386524,
      "grad_norm": 30.03960418701172,
      "learning_rate": 4.12e-05,
      "loss": 6.1673,
      "step": 420
    },
    {
      "epoch": 0.06615130187300489,
      "grad_norm": 38.63550567626953,
      "learning_rate": 4.2199999999999996e-05,
      "loss": 6.1151,
      "step": 430
    },
    {
      "epoch": 0.06768970424214453,
      "grad_norm": 24.747684478759766,
      "learning_rate": 4.319999999999999e-05,
      "loss": 6.0666,
      "step": 440
    },
    {
      "epoch": 0.06922810661128419,
      "grad_norm": 22.29100799560547,
      "learning_rate": 4.42e-05,
      "loss": 6.1266,
      "step": 450
    },
    {
      "epoch": 0.07076650898042383,
      "grad_norm": 24.165313720703125,
      "learning_rate": 4.5199999999999994e-05,
      "loss": 6.1199,
      "step": 460
    },
    {
      "epoch": 0.07230491134956347,
      "grad_norm": 19.26260757446289,
      "learning_rate": 4.62e-05,
      "loss": 6.1002,
      "step": 470
    },
    {
      "epoch": 0.07384331371870313,
      "grad_norm": 21.792062759399414,
      "learning_rate": 4.7199999999999995e-05,
      "loss": 6.0919,
      "step": 480
    },
    {
      "epoch": 0.07538171608784278,
      "grad_norm": 25.019775390625,
      "learning_rate": 4.82e-05,
      "loss": 6.1171,
      "step": 490
    },
    {
      "epoch": 0.07692011845698242,
      "grad_norm": 21.131744384765625,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 6.0676,
      "step": 500
    },
    {
      "epoch": 0.07845852082612208,
      "grad_norm": 24.013105392456055,
      "learning_rate": 5.0199999999999994e-05,
      "loss": 6.0463,
      "step": 510
    },
    {
      "epoch": 0.07999692319526172,
      "grad_norm": 20.643056869506836,
      "learning_rate": 5.119999999999999e-05,
      "loss": 6.0317,
      "step": 520
    },
    {
      "epoch": 0.08153532556440136,
      "grad_norm": 27.55217170715332,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 5.9955,
      "step": 530
    },
    {
      "epoch": 0.08307372793354102,
      "grad_norm": 24.888734817504883,
      "learning_rate": 5.32e-05,
      "loss": 6.0481,
      "step": 540
    },
    {
      "epoch": 0.08461213030268067,
      "grad_norm": 19.217220306396484,
      "learning_rate": 5.4199999999999996e-05,
      "loss": 5.9471,
      "step": 550
    },
    {
      "epoch": 0.08615053267182031,
      "grad_norm": 19.650312423706055,
      "learning_rate": 5.519999999999999e-05,
      "loss": 5.9048,
      "step": 560
    },
    {
      "epoch": 0.08768893504095997,
      "grad_norm": 21.437850952148438,
      "learning_rate": 5.619999999999999e-05,
      "loss": 6.0098,
      "step": 570
    },
    {
      "epoch": 0.08922733741009961,
      "grad_norm": 23.66594696044922,
      "learning_rate": 5.72e-05,
      "loss": 5.9389,
      "step": 580
    },
    {
      "epoch": 0.09076573977923925,
      "grad_norm": 33.00392532348633,
      "learning_rate": 5.82e-05,
      "loss": 5.8897,
      "step": 590
    },
    {
      "epoch": 0.09230414214837891,
      "grad_norm": 20.737876892089844,
      "learning_rate": 5.9199999999999996e-05,
      "loss": 5.8598,
      "step": 600
    },
    {
      "epoch": 0.09384254451751856,
      "grad_norm": 23.151851654052734,
      "learning_rate": 6.019999999999999e-05,
      "loss": 5.8652,
      "step": 610
    },
    {
      "epoch": 0.0953809468866582,
      "grad_norm": 23.21453094482422,
      "learning_rate": 6.12e-05,
      "loss": 5.833,
      "step": 620
    },
    {
      "epoch": 0.09691934925579786,
      "grad_norm": 23.856670379638672,
      "learning_rate": 6.22e-05,
      "loss": 5.8686,
      "step": 630
    },
    {
      "epoch": 0.0984577516249375,
      "grad_norm": 18.7842960357666,
      "learning_rate": 6.319999999999999e-05,
      "loss": 5.8517,
      "step": 640
    },
    {
      "epoch": 0.09999615399407714,
      "grad_norm": 16.733356475830078,
      "learning_rate": 6.419999999999999e-05,
      "loss": 5.8614,
      "step": 650
    },
    {
      "epoch": 0.1015345563632168,
      "grad_norm": 18.41499137878418,
      "learning_rate": 6.519999999999999e-05,
      "loss": 5.8441,
      "step": 660
    },
    {
      "epoch": 0.10307295873235645,
      "grad_norm": 20.63882827758789,
      "learning_rate": 6.62e-05,
      "loss": 5.7316,
      "step": 670
    },
    {
      "epoch": 0.1046113611014961,
      "grad_norm": 17.48029899597168,
      "learning_rate": 6.72e-05,
      "loss": 5.6974,
      "step": 680
    },
    {
      "epoch": 0.10614976347063575,
      "grad_norm": 22.546825408935547,
      "learning_rate": 6.819999999999999e-05,
      "loss": 5.7741,
      "step": 690
    },
    {
      "epoch": 0.10768816583977539,
      "grad_norm": 16.246665954589844,
      "learning_rate": 6.919999999999999e-05,
      "loss": 5.6912,
      "step": 700
    },
    {
      "epoch": 0.10922656820891505,
      "grad_norm": 16.516803741455078,
      "learning_rate": 7.02e-05,
      "loss": 5.7397,
      "step": 710
    },
    {
      "epoch": 0.11076497057805469,
      "grad_norm": 19.068937301635742,
      "learning_rate": 7.12e-05,
      "loss": 5.7386,
      "step": 720
    },
    {
      "epoch": 0.11230337294719434,
      "grad_norm": 14.85578727722168,
      "learning_rate": 7.219999999999999e-05,
      "loss": 5.6994,
      "step": 730
    },
    {
      "epoch": 0.11384177531633399,
      "grad_norm": 18.3763427734375,
      "learning_rate": 7.319999999999999e-05,
      "loss": 5.6505,
      "step": 740
    },
    {
      "epoch": 0.11538017768547364,
      "grad_norm": 16.224000930786133,
      "learning_rate": 7.419999999999999e-05,
      "loss": 5.7087,
      "step": 750
    },
    {
      "epoch": 0.11691858005461328,
      "grad_norm": 15.365422248840332,
      "learning_rate": 7.519999999999998e-05,
      "loss": 5.6072,
      "step": 760
    },
    {
      "epoch": 0.11845698242375294,
      "grad_norm": 18.633525848388672,
      "learning_rate": 7.62e-05,
      "loss": 5.5342,
      "step": 770
    },
    {
      "epoch": 0.11999538479289258,
      "grad_norm": 19.027240753173828,
      "learning_rate": 7.72e-05,
      "loss": 5.5611,
      "step": 780
    },
    {
      "epoch": 0.12153378716203223,
      "grad_norm": 16.857473373413086,
      "learning_rate": 7.819999999999999e-05,
      "loss": 5.5638,
      "step": 790
    },
    {
      "epoch": 0.12307218953117188,
      "grad_norm": 15.174002647399902,
      "learning_rate": 7.92e-05,
      "loss": 5.5405,
      "step": 800
    },
    {
      "epoch": 0.12461059190031153,
      "grad_norm": 16.1444034576416,
      "learning_rate": 8.019999999999998e-05,
      "loss": 5.5565,
      "step": 810
    },
    {
      "epoch": 0.12614899426945117,
      "grad_norm": 18.24864959716797,
      "learning_rate": 8.12e-05,
      "loss": 5.4931,
      "step": 820
    },
    {
      "epoch": 0.12768739663859083,
      "grad_norm": 18.213571548461914,
      "learning_rate": 8.22e-05,
      "loss": 5.5312,
      "step": 830
    },
    {
      "epoch": 0.12922579900773049,
      "grad_norm": 17.174373626708984,
      "learning_rate": 8.319999999999999e-05,
      "loss": 5.4435,
      "step": 840
    },
    {
      "epoch": 0.13076420137687012,
      "grad_norm": 15.482538223266602,
      "learning_rate": 8.42e-05,
      "loss": 5.4082,
      "step": 850
    },
    {
      "epoch": 0.13230260374600977,
      "grad_norm": 17.8658504486084,
      "learning_rate": 8.519999999999998e-05,
      "loss": 5.473,
      "step": 860
    },
    {
      "epoch": 0.13384100611514943,
      "grad_norm": 15.455946922302246,
      "learning_rate": 8.62e-05,
      "loss": 5.3541,
      "step": 870
    },
    {
      "epoch": 0.13537940848428906,
      "grad_norm": 15.936956405639648,
      "learning_rate": 8.719999999999999e-05,
      "loss": 5.3488,
      "step": 880
    },
    {
      "epoch": 0.13691781085342872,
      "grad_norm": 19.47191619873047,
      "learning_rate": 8.819999999999999e-05,
      "loss": 5.4054,
      "step": 890
    },
    {
      "epoch": 0.13845621322256838,
      "grad_norm": 12.858948707580566,
      "learning_rate": 8.92e-05,
      "loss": 5.3355,
      "step": 900
    },
    {
      "epoch": 0.139994615591708,
      "grad_norm": 15.170941352844238,
      "learning_rate": 9.02e-05,
      "loss": 5.3799,
      "step": 910
    },
    {
      "epoch": 0.14153301796084766,
      "grad_norm": 15.187207221984863,
      "learning_rate": 9.12e-05,
      "loss": 5.2994,
      "step": 920
    },
    {
      "epoch": 0.14307142032998732,
      "grad_norm": 13.390646934509277,
      "learning_rate": 9.219999999999999e-05,
      "loss": 5.2335,
      "step": 930
    },
    {
      "epoch": 0.14460982269912695,
      "grad_norm": 18.35064697265625,
      "learning_rate": 9.319999999999999e-05,
      "loss": 5.3427,
      "step": 940
    },
    {
      "epoch": 0.1461482250682666,
      "grad_norm": 19.055871963500977,
      "learning_rate": 9.419999999999999e-05,
      "loss": 5.2078,
      "step": 950
    },
    {
      "epoch": 0.14768662743740626,
      "grad_norm": 16.91037940979004,
      "learning_rate": 9.52e-05,
      "loss": 5.2242,
      "step": 960
    },
    {
      "epoch": 0.1492250298065459,
      "grad_norm": 15.454617500305176,
      "learning_rate": 9.62e-05,
      "loss": 5.2252,
      "step": 970
    },
    {
      "epoch": 0.15076343217568555,
      "grad_norm": 13.046433448791504,
      "learning_rate": 9.719999999999999e-05,
      "loss": 5.2289,
      "step": 980
    },
    {
      "epoch": 0.1523018345448252,
      "grad_norm": 15.343282699584961,
      "learning_rate": 9.819999999999999e-05,
      "loss": 5.1872,
      "step": 990
    },
    {
      "epoch": 0.15384023691396484,
      "grad_norm": 13.18584156036377,
      "learning_rate": 9.919999999999999e-05,
      "loss": 5.1641,
      "step": 1000
    },
    {
      "epoch": 0.1553786392831045,
      "grad_norm": 16.466337203979492,
      "learning_rate": 0.0001002,
      "loss": 5.0973,
      "step": 1010
    },
    {
      "epoch": 0.15691704165224415,
      "grad_norm": 16.390913009643555,
      "learning_rate": 0.0001012,
      "loss": 5.1066,
      "step": 1020
    },
    {
      "epoch": 0.15845544402138378,
      "grad_norm": 13.89096736907959,
      "learning_rate": 0.00010219999999999999,
      "loss": 5.1375,
      "step": 1030
    },
    {
      "epoch": 0.15999384639052344,
      "grad_norm": 14.639664649963379,
      "learning_rate": 0.00010319999999999999,
      "loss": 5.0562,
      "step": 1040
    },
    {
      "epoch": 0.1615322487596631,
      "grad_norm": 14.390929222106934,
      "learning_rate": 0.00010419999999999998,
      "loss": 5.1309,
      "step": 1050
    },
    {
      "epoch": 0.16307065112880273,
      "grad_norm": 15.142280578613281,
      "learning_rate": 0.0001052,
      "loss": 5.0287,
      "step": 1060
    },
    {
      "epoch": 0.1646090534979424,
      "grad_norm": 13.173295021057129,
      "learning_rate": 0.00010619999999999998,
      "loss": 4.9568,
      "step": 1070
    },
    {
      "epoch": 0.16614745586708204,
      "grad_norm": 15.059986114501953,
      "learning_rate": 0.00010719999999999999,
      "loss": 5.01,
      "step": 1080
    },
    {
      "epoch": 0.16768585823622167,
      "grad_norm": 13.298657417297363,
      "learning_rate": 0.0001082,
      "loss": 4.9931,
      "step": 1090
    },
    {
      "epoch": 0.16922426060536133,
      "grad_norm": 14.215367317199707,
      "learning_rate": 0.00010919999999999998,
      "loss": 4.9235,
      "step": 1100
    },
    {
      "epoch": 0.170762662974501,
      "grad_norm": 14.457551002502441,
      "learning_rate": 0.0001102,
      "loss": 4.9374,
      "step": 1110
    },
    {
      "epoch": 0.17230106534364062,
      "grad_norm": 14.105072975158691,
      "learning_rate": 0.00011119999999999998,
      "loss": 4.8857,
      "step": 1120
    },
    {
      "epoch": 0.17383946771278028,
      "grad_norm": 16.460472106933594,
      "learning_rate": 0.00011219999999999999,
      "loss": 4.8934,
      "step": 1130
    },
    {
      "epoch": 0.17537787008191993,
      "grad_norm": 15.866796493530273,
      "learning_rate": 0.0001132,
      "loss": 4.8335,
      "step": 1140
    },
    {
      "epoch": 0.17691627245105956,
      "grad_norm": 14.246673583984375,
      "learning_rate": 0.00011419999999999998,
      "loss": 4.8547,
      "step": 1150
    },
    {
      "epoch": 0.17845467482019922,
      "grad_norm": 14.18166732788086,
      "learning_rate": 0.0001152,
      "loss": 4.7489,
      "step": 1160
    },
    {
      "epoch": 0.17999307718933888,
      "grad_norm": 14.299484252929688,
      "learning_rate": 0.00011619999999999998,
      "loss": 4.788,
      "step": 1170
    },
    {
      "epoch": 0.1815314795584785,
      "grad_norm": 13.284591674804688,
      "learning_rate": 0.00011719999999999999,
      "loss": 4.7395,
      "step": 1180
    },
    {
      "epoch": 0.18306988192761817,
      "grad_norm": 19.244670867919922,
      "learning_rate": 0.0001182,
      "loss": 4.7696,
      "step": 1190
    },
    {
      "epoch": 0.18460828429675782,
      "grad_norm": 12.124654769897461,
      "learning_rate": 0.00011919999999999998,
      "loss": 4.7223,
      "step": 1200
    },
    {
      "epoch": 0.18614668666589745,
      "grad_norm": 13.926753044128418,
      "learning_rate": 0.0001202,
      "loss": 4.7244,
      "step": 1210
    },
    {
      "epoch": 0.1876850890350371,
      "grad_norm": 14.33655071258545,
      "learning_rate": 0.00012119999999999999,
      "loss": 4.6641,
      "step": 1220
    },
    {
      "epoch": 0.18922349140417677,
      "grad_norm": 15.00214672088623,
      "learning_rate": 0.0001222,
      "loss": 4.6653,
      "step": 1230
    },
    {
      "epoch": 0.1907618937733164,
      "grad_norm": 15.199170112609863,
      "learning_rate": 0.00012319999999999999,
      "loss": 4.6486,
      "step": 1240
    },
    {
      "epoch": 0.19230029614245606,
      "grad_norm": 15.2691068649292,
      "learning_rate": 0.00012419999999999998,
      "loss": 4.545,
      "step": 1250
    },
    {
      "epoch": 0.19383869851159571,
      "grad_norm": 13.688949584960938,
      "learning_rate": 0.00012519999999999998,
      "loss": 4.5674,
      "step": 1260
    },
    {
      "epoch": 0.19537710088073534,
      "grad_norm": 14.25082015991211,
      "learning_rate": 0.0001262,
      "loss": 4.5894,
      "step": 1270
    },
    {
      "epoch": 0.196915503249875,
      "grad_norm": 14.78211498260498,
      "learning_rate": 0.00012719999999999997,
      "loss": 4.5822,
      "step": 1280
    },
    {
      "epoch": 0.19845390561901466,
      "grad_norm": 15.209827423095703,
      "learning_rate": 0.0001282,
      "loss": 4.5339,
      "step": 1290
    },
    {
      "epoch": 0.1999923079881543,
      "grad_norm": 13.816431999206543,
      "learning_rate": 0.00012919999999999997,
      "loss": 4.5117,
      "step": 1300
    },
    {
      "epoch": 0.20153071035729395,
      "grad_norm": 11.83894157409668,
      "learning_rate": 0.0001302,
      "loss": 4.4655,
      "step": 1310
    },
    {
      "epoch": 0.2030691127264336,
      "grad_norm": 12.231562614440918,
      "learning_rate": 0.0001312,
      "loss": 4.4517,
      "step": 1320
    },
    {
      "epoch": 0.20460751509557323,
      "grad_norm": 14.597062110900879,
      "learning_rate": 0.0001322,
      "loss": 4.4247,
      "step": 1330
    },
    {
      "epoch": 0.2061459174647129,
      "grad_norm": 14.041385650634766,
      "learning_rate": 0.00013319999999999999,
      "loss": 4.4515,
      "step": 1340
    },
    {
      "epoch": 0.20768431983385255,
      "grad_norm": 14.399479866027832,
      "learning_rate": 0.0001342,
      "loss": 4.4116,
      "step": 1350
    },
    {
      "epoch": 0.2092227222029922,
      "grad_norm": 11.779489517211914,
      "learning_rate": 0.00013519999999999998,
      "loss": 4.3499,
      "step": 1360
    },
    {
      "epoch": 0.21076112457213184,
      "grad_norm": 16.141752243041992,
      "learning_rate": 0.0001362,
      "loss": 4.263,
      "step": 1370
    },
    {
      "epoch": 0.2122995269412715,
      "grad_norm": 12.214835166931152,
      "learning_rate": 0.00013719999999999997,
      "loss": 4.3635,
      "step": 1380
    },
    {
      "epoch": 0.21383792931041115,
      "grad_norm": 11.679731369018555,
      "learning_rate": 0.0001382,
      "loss": 4.3867,
      "step": 1390
    },
    {
      "epoch": 0.21537633167955078,
      "grad_norm": 13.870141983032227,
      "learning_rate": 0.0001392,
      "loss": 4.2572,
      "step": 1400
    },
    {
      "epoch": 0.21691473404869044,
      "grad_norm": 15.715397834777832,
      "learning_rate": 0.0001402,
      "loss": 4.2449,
      "step": 1410
    },
    {
      "epoch": 0.2184531364178301,
      "grad_norm": 13.933563232421875,
      "learning_rate": 0.0001412,
      "loss": 4.3044,
      "step": 1420
    },
    {
      "epoch": 0.21999153878696973,
      "grad_norm": 11.775773048400879,
      "learning_rate": 0.0001422,
      "loss": 4.1688,
      "step": 1430
    },
    {
      "epoch": 0.22152994115610938,
      "grad_norm": 14.641937255859375,
      "learning_rate": 0.00014319999999999998,
      "loss": 4.2013,
      "step": 1440
    },
    {
      "epoch": 0.22306834352524904,
      "grad_norm": 12.434427261352539,
      "learning_rate": 0.00014419999999999998,
      "loss": 4.138,
      "step": 1450
    },
    {
      "epoch": 0.22460674589438867,
      "grad_norm": 13.412193298339844,
      "learning_rate": 0.00014519999999999998,
      "loss": 4.2305,
      "step": 1460
    },
    {
      "epoch": 0.22614514826352833,
      "grad_norm": 12.081714630126953,
      "learning_rate": 0.0001462,
      "loss": 4.1844,
      "step": 1470
    },
    {
      "epoch": 0.22768355063266799,
      "grad_norm": 14.079381942749023,
      "learning_rate": 0.00014719999999999997,
      "loss": 4.1119,
      "step": 1480
    },
    {
      "epoch": 0.22922195300180762,
      "grad_norm": 12.457756996154785,
      "learning_rate": 0.0001482,
      "loss": 4.1138,
      "step": 1490
    },
    {
      "epoch": 0.23076035537094727,
      "grad_norm": 12.655537605285645,
      "learning_rate": 0.0001492,
      "loss": 4.0457,
      "step": 1500
    },
    {
      "epoch": 0.23229875774008693,
      "grad_norm": 14.801631927490234,
      "learning_rate": 0.0001502,
      "loss": 3.972,
      "step": 1510
    },
    {
      "epoch": 0.23383716010922656,
      "grad_norm": 12.558164596557617,
      "learning_rate": 0.0001512,
      "loss": 4.0401,
      "step": 1520
    },
    {
      "epoch": 0.23537556247836622,
      "grad_norm": 12.809494972229004,
      "learning_rate": 0.00015219999999999999,
      "loss": 3.9904,
      "step": 1530
    },
    {
      "epoch": 0.23691396484750588,
      "grad_norm": 11.842789649963379,
      "learning_rate": 0.0001532,
      "loss": 4.0351,
      "step": 1540
    },
    {
      "epoch": 0.2384523672166455,
      "grad_norm": 10.249466896057129,
      "learning_rate": 0.00015419999999999998,
      "loss": 3.9697,
      "step": 1550
    },
    {
      "epoch": 0.23999076958578516,
      "grad_norm": 11.96493148803711,
      "learning_rate": 0.00015519999999999998,
      "loss": 3.891,
      "step": 1560
    },
    {
      "epoch": 0.24152917195492482,
      "grad_norm": 13.920326232910156,
      "learning_rate": 0.0001562,
      "loss": 3.9165,
      "step": 1570
    },
    {
      "epoch": 0.24306757432406445,
      "grad_norm": 10.648415565490723,
      "learning_rate": 0.0001572,
      "loss": 3.905,
      "step": 1580
    },
    {
      "epoch": 0.2446059766932041,
      "grad_norm": 12.895554542541504,
      "learning_rate": 0.00015819999999999997,
      "loss": 3.8309,
      "step": 1590
    },
    {
      "epoch": 0.24614437906234377,
      "grad_norm": 12.320569038391113,
      "learning_rate": 0.00015919999999999997,
      "loss": 3.8677,
      "step": 1600
    },
    {
      "epoch": 0.2476827814314834,
      "grad_norm": 12.256221771240234,
      "learning_rate": 0.0001602,
      "loss": 3.809,
      "step": 1610
    },
    {
      "epoch": 0.24922118380062305,
      "grad_norm": 12.378941535949707,
      "learning_rate": 0.0001612,
      "loss": 3.8226,
      "step": 1620
    },
    {
      "epoch": 0.2507595861697627,
      "grad_norm": 11.537256240844727,
      "learning_rate": 0.00016219999999999999,
      "loss": 3.7577,
      "step": 1630
    },
    {
      "epoch": 0.25229798853890234,
      "grad_norm": 11.635543823242188,
      "learning_rate": 0.0001632,
      "loss": 3.692,
      "step": 1640
    },
    {
      "epoch": 0.253836390908042,
      "grad_norm": 11.149394035339355,
      "learning_rate": 0.00016419999999999998,
      "loss": 3.7158,
      "step": 1650
    },
    {
      "epoch": 0.25537479327718166,
      "grad_norm": 12.258176803588867,
      "learning_rate": 0.00016519999999999998,
      "loss": 3.72,
      "step": 1660
    },
    {
      "epoch": 0.2569131956463213,
      "grad_norm": 11.55327033996582,
      "learning_rate": 0.0001662,
      "loss": 3.6468,
      "step": 1670
    },
    {
      "epoch": 0.25845159801546097,
      "grad_norm": 11.125475883483887,
      "learning_rate": 0.0001672,
      "loss": 3.6283,
      "step": 1680
    },
    {
      "epoch": 0.2599900003846006,
      "grad_norm": 11.03936767578125,
      "learning_rate": 0.00016819999999999997,
      "loss": 3.6279,
      "step": 1690
    },
    {
      "epoch": 0.26152840275374023,
      "grad_norm": 11.331442832946777,
      "learning_rate": 0.00016919999999999997,
      "loss": 3.5886,
      "step": 1700
    },
    {
      "epoch": 0.2630668051228799,
      "grad_norm": 10.849939346313477,
      "learning_rate": 0.0001702,
      "loss": 3.6896,
      "step": 1710
    },
    {
      "epoch": 0.26460520749201955,
      "grad_norm": 9.935115814208984,
      "learning_rate": 0.0001712,
      "loss": 3.5194,
      "step": 1720
    },
    {
      "epoch": 0.2661436098611592,
      "grad_norm": 11.35429859161377,
      "learning_rate": 0.00017219999999999998,
      "loss": 3.5311,
      "step": 1730
    },
    {
      "epoch": 0.26768201223029886,
      "grad_norm": 11.712488174438477,
      "learning_rate": 0.00017319999999999998,
      "loss": 3.6083,
      "step": 1740
    },
    {
      "epoch": 0.26922041459943846,
      "grad_norm": 13.349652290344238,
      "learning_rate": 0.00017419999999999998,
      "loss": 3.6181,
      "step": 1750
    },
    {
      "epoch": 0.2707588169685781,
      "grad_norm": 10.547552108764648,
      "learning_rate": 0.00017519999999999998,
      "loss": 3.6202,
      "step": 1760
    },
    {
      "epoch": 0.2722972193377178,
      "grad_norm": 11.436111450195312,
      "learning_rate": 0.0001762,
      "loss": 3.5857,
      "step": 1770
    },
    {
      "epoch": 0.27383562170685743,
      "grad_norm": 10.192728996276855,
      "learning_rate": 0.0001772,
      "loss": 3.3816,
      "step": 1780
    },
    {
      "epoch": 0.2753740240759971,
      "grad_norm": 9.88837718963623,
      "learning_rate": 0.00017819999999999997,
      "loss": 3.4138,
      "step": 1790
    },
    {
      "epoch": 0.27691242644513675,
      "grad_norm": 9.523514747619629,
      "learning_rate": 0.0001792,
      "loss": 3.4505,
      "step": 1800
    },
    {
      "epoch": 0.27845082881427635,
      "grad_norm": 9.86983585357666,
      "learning_rate": 0.0001802,
      "loss": 3.4563,
      "step": 1810
    },
    {
      "epoch": 0.279989231183416,
      "grad_norm": 10.267000198364258,
      "learning_rate": 0.00018119999999999999,
      "loss": 3.2901,
      "step": 1820
    },
    {
      "epoch": 0.28152763355255567,
      "grad_norm": 9.708111763000488,
      "learning_rate": 0.00018219999999999996,
      "loss": 3.4291,
      "step": 1830
    },
    {
      "epoch": 0.2830660359216953,
      "grad_norm": 16.580785751342773,
      "learning_rate": 0.00018319999999999998,
      "loss": 3.3824,
      "step": 1840
    },
    {
      "epoch": 0.284604438290835,
      "grad_norm": 10.267104148864746,
      "learning_rate": 0.00018419999999999998,
      "loss": 3.3334,
      "step": 1850
    },
    {
      "epoch": 0.28614284065997464,
      "grad_norm": 9.529206275939941,
      "learning_rate": 0.00018519999999999998,
      "loss": 3.3357,
      "step": 1860
    },
    {
      "epoch": 0.28768124302911424,
      "grad_norm": 9.670571327209473,
      "learning_rate": 0.0001862,
      "loss": 3.2536,
      "step": 1870
    },
    {
      "epoch": 0.2892196453982539,
      "grad_norm": 9.460128784179688,
      "learning_rate": 0.0001872,
      "loss": 3.3416,
      "step": 1880
    },
    {
      "epoch": 0.29075804776739356,
      "grad_norm": 10.193305015563965,
      "learning_rate": 0.00018819999999999997,
      "loss": 3.2714,
      "step": 1890
    },
    {
      "epoch": 0.2922964501365332,
      "grad_norm": 9.581829071044922,
      "learning_rate": 0.0001892,
      "loss": 3.2382,
      "step": 1900
    },
    {
      "epoch": 0.29383485250567287,
      "grad_norm": 10.401164054870605,
      "learning_rate": 0.0001902,
      "loss": 3.2288,
      "step": 1910
    },
    {
      "epoch": 0.29537325487481253,
      "grad_norm": 10.252750396728516,
      "learning_rate": 0.00019119999999999999,
      "loss": 3.2718,
      "step": 1920
    },
    {
      "epoch": 0.29691165724395213,
      "grad_norm": 9.973576545715332,
      "learning_rate": 0.0001922,
      "loss": 3.2742,
      "step": 1930
    },
    {
      "epoch": 0.2984500596130918,
      "grad_norm": 9.254643440246582,
      "learning_rate": 0.00019319999999999998,
      "loss": 3.2849,
      "step": 1940
    },
    {
      "epoch": 0.29998846198223145,
      "grad_norm": 9.65290641784668,
      "learning_rate": 0.00019419999999999998,
      "loss": 3.2829,
      "step": 1950
    },
    {
      "epoch": 0.3015268643513711,
      "grad_norm": 10.02481746673584,
      "learning_rate": 0.00019519999999999997,
      "loss": 3.168,
      "step": 1960
    },
    {
      "epoch": 0.30306526672051076,
      "grad_norm": 9.108298301696777,
      "learning_rate": 0.0001962,
      "loss": 3.1571,
      "step": 1970
    },
    {
      "epoch": 0.3046036690896504,
      "grad_norm": 9.827427864074707,
      "learning_rate": 0.00019719999999999997,
      "loss": 3.0885,
      "step": 1980
    },
    {
      "epoch": 0.30614207145879,
      "grad_norm": 10.132247924804688,
      "learning_rate": 0.00019819999999999997,
      "loss": 3.116,
      "step": 1990
    },
    {
      "epoch": 0.3076804738279297,
      "grad_norm": 8.130690574645996,
      "learning_rate": 0.0001992,
      "loss": 3.11,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 16,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.83071981314048e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}