{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.04566455356183518,
  "eval_steps": 500,
  "global_step": 1700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.6861502095197165e-05,
      "grad_norm": 0.9079805016517639,
      "learning_rate": 2e-05,
      "loss": 5.1449,
      "step": 1
    },
    {
      "epoch": 5.372300419039433e-05,
      "grad_norm": 1.1043980121612549,
      "learning_rate": 4e-05,
      "loss": 4.9052,
      "step": 2
    },
    {
      "epoch": 8.05845062855915e-05,
      "grad_norm": 1.1937360763549805,
      "learning_rate": 6e-05,
      "loss": 5.1465,
      "step": 3
    },
    {
      "epoch": 0.00010744600838078866,
      "grad_norm": 3.707120180130005,
      "learning_rate": 8e-05,
      "loss": 5.3884,
      "step": 4
    },
    {
      "epoch": 0.00013430751047598582,
      "grad_norm": 1.2312233448028564,
      "learning_rate": 0.0001,
      "loss": 5.7386,
      "step": 5
    },
    {
      "epoch": 0.000161169012571183,
      "grad_norm": 1.418683648109436,
      "learning_rate": 9.999999999999938e-05,
      "loss": 5.7863,
      "step": 6
    },
    {
      "epoch": 0.00018803051466638015,
      "grad_norm": 1.7924962043762207,
      "learning_rate": 9.999999999999754e-05,
      "loss": 5.23,
      "step": 7
    },
    {
      "epoch": 0.00021489201676157732,
      "grad_norm": 1.8357185125350952,
      "learning_rate": 9.999999999999445e-05,
      "loss": 4.9708,
      "step": 8
    },
    {
      "epoch": 0.00024175351885677448,
      "grad_norm": 1.7411339282989502,
      "learning_rate": 9.999999999999014e-05,
      "loss": 4.9414,
      "step": 9
    },
    {
      "epoch": 0.00026861502095197165,
      "grad_norm": 1.7914339303970337,
      "learning_rate": 9.999999999998458e-05,
      "loss": 4.6556,
      "step": 10
    },
    {
      "epoch": 0.0002954765230471688,
      "grad_norm": 1.8011003732681274,
      "learning_rate": 9.99999999999778e-05,
      "loss": 4.2169,
      "step": 11
    },
    {
      "epoch": 0.000322338025142366,
      "grad_norm": 1.3325257301330566,
      "learning_rate": 9.999999999996978e-05,
      "loss": 4.1794,
      "step": 12
    },
    {
      "epoch": 0.00034919952723756314,
      "grad_norm": 1.4623206853866577,
      "learning_rate": 9.999999999996053e-05,
      "loss": 4.4874,
      "step": 13
    },
    {
      "epoch": 0.0003760610293327603,
      "grad_norm": 1.2885961532592773,
      "learning_rate": 9.999999999995005e-05,
      "loss": 4.0216,
      "step": 14
    },
    {
      "epoch": 0.00040292253142795747,
      "grad_norm": 1.2243579626083374,
      "learning_rate": 9.999999999993833e-05,
      "loss": 3.9258,
      "step": 15
    },
    {
      "epoch": 0.00042978403352315463,
      "grad_norm": 1.3720817565917969,
      "learning_rate": 9.999999999992537e-05,
      "loss": 4.0791,
      "step": 16
    },
    {
      "epoch": 0.0004566455356183518,
      "grad_norm": 1.672309398651123,
      "learning_rate": 9.999999999991118e-05,
      "loss": 4.1378,
      "step": 17
    },
    {
      "epoch": 0.00048350703771354896,
      "grad_norm": 1.2076022624969482,
      "learning_rate": 9.999999999989576e-05,
      "loss": 3.5104,
      "step": 18
    },
    {
      "epoch": 0.0005103685398087461,
      "grad_norm": 1.3783760070800781,
      "learning_rate": 9.99999999998791e-05,
      "loss": 3.9612,
      "step": 19
    },
    {
      "epoch": 0.0005372300419039433,
      "grad_norm": 1.3485159873962402,
      "learning_rate": 9.999999999986121e-05,
      "loss": 3.8829,
      "step": 20
    },
    {
      "epoch": 0.0005640915439991404,
      "grad_norm": 1.6064331531524658,
      "learning_rate": 9.999999999984209e-05,
      "loss": 4.1337,
      "step": 21
    },
    {
      "epoch": 0.0005909530460943376,
      "grad_norm": 1.2591063976287842,
      "learning_rate": 9.999999999982173e-05,
      "loss": 3.4452,
      "step": 22
    },
    {
      "epoch": 0.0006178145481895347,
      "grad_norm": 1.456554651260376,
      "learning_rate": 9.999999999980015e-05,
      "loss": 3.49,
      "step": 23
    },
    {
      "epoch": 0.000644676050284732,
      "grad_norm": 1.4870679378509521,
      "learning_rate": 9.999999999977731e-05,
      "loss": 3.829,
      "step": 24
    },
    {
      "epoch": 0.0006715375523799291,
      "grad_norm": 1.3454641103744507,
      "learning_rate": 9.999999999975327e-05,
      "loss": 3.7703,
      "step": 25
    },
    {
      "epoch": 0.0006983990544751263,
      "grad_norm": 1.5091837644577026,
      "learning_rate": 9.999999999972798e-05,
      "loss": 3.7927,
      "step": 26
    },
    {
      "epoch": 0.0007252605565703234,
      "grad_norm": 1.527295708656311,
      "learning_rate": 9.999999999970144e-05,
      "loss": 3.6802,
      "step": 27
    },
    {
      "epoch": 0.0007521220586655206,
      "grad_norm": 1.5174264907836914,
      "learning_rate": 9.999999999967369e-05,
      "loss": 3.684,
      "step": 28
    },
    {
      "epoch": 0.0007789835607607177,
      "grad_norm": 1.5124770402908325,
      "learning_rate": 9.99999999996447e-05,
      "loss": 3.8185,
      "step": 29
    },
    {
      "epoch": 0.0008058450628559149,
      "grad_norm": 1.4548962116241455,
      "learning_rate": 9.999999999961446e-05,
      "loss": 3.7584,
      "step": 30
    },
    {
      "epoch": 0.000832706564951112,
      "grad_norm": 1.470106840133667,
      "learning_rate": 9.999999999958302e-05,
      "loss": 3.5163,
      "step": 31
    },
    {
      "epoch": 0.0008595680670463093,
      "grad_norm": 1.4147369861602783,
      "learning_rate": 9.999999999955032e-05,
      "loss": 3.6908,
      "step": 32
    },
    {
      "epoch": 0.0008864295691415064,
      "grad_norm": 1.5790003538131714,
      "learning_rate": 9.99999999995164e-05,
      "loss": 3.589,
      "step": 33
    },
    {
      "epoch": 0.0009132910712367036,
      "grad_norm": 1.5375813245773315,
      "learning_rate": 9.999999999948123e-05,
      "loss": 3.629,
      "step": 34
    },
    {
      "epoch": 0.0009401525733319007,
      "grad_norm": 1.5038788318634033,
      "learning_rate": 9.999999999944484e-05,
      "loss": 3.691,
      "step": 35
    },
    {
      "epoch": 0.0009670140754270979,
      "grad_norm": 1.4857861995697021,
      "learning_rate": 9.999999999940722e-05,
      "loss": 3.7313,
      "step": 36
    },
    {
      "epoch": 0.0009938755775222951,
      "grad_norm": 1.533119559288025,
      "learning_rate": 9.999999999936835e-05,
      "loss": 3.925,
      "step": 37
    },
    {
      "epoch": 0.0010207370796174921,
      "grad_norm": 1.562211036682129,
      "learning_rate": 9.999999999932825e-05,
      "loss": 3.8762,
      "step": 38
    },
    {
      "epoch": 0.0010475985817126894,
      "grad_norm": 1.3962745666503906,
      "learning_rate": 9.999999999928693e-05,
      "loss": 3.3351,
      "step": 39
    },
    {
      "epoch": 0.0010744600838078866,
      "grad_norm": 1.3966686725616455,
      "learning_rate": 9.999999999924437e-05,
      "loss": 3.7117,
      "step": 40
    },
    {
      "epoch": 0.0011013215859030838,
      "grad_norm": 1.4492931365966797,
      "learning_rate": 9.999999999920057e-05,
      "loss": 3.715,
      "step": 41
    },
    {
      "epoch": 0.0011281830879982808,
      "grad_norm": 1.4874073266983032,
      "learning_rate": 9.999999999915553e-05,
      "loss": 3.6364,
      "step": 42
    },
    {
      "epoch": 0.001155044590093478,
      "grad_norm": 1.5005569458007812,
      "learning_rate": 9.999999999910928e-05,
      "loss": 3.712,
      "step": 43
    },
    {
      "epoch": 0.0011819060921886752,
      "grad_norm": 1.3945828676223755,
      "learning_rate": 9.999999999906178e-05,
      "loss": 3.6228,
      "step": 44
    },
    {
      "epoch": 0.0012087675942838725,
      "grad_norm": 1.4671238660812378,
      "learning_rate": 9.999999999901304e-05,
      "loss": 3.5543,
      "step": 45
    },
    {
      "epoch": 0.0012356290963790695,
      "grad_norm": 1.5679062604904175,
      "learning_rate": 9.999999999896307e-05,
      "loss": 3.4235,
      "step": 46
    },
    {
      "epoch": 0.0012624905984742667,
      "grad_norm": 1.389269471168518,
      "learning_rate": 9.999999999891187e-05,
      "loss": 3.5772,
      "step": 47
    },
    {
      "epoch": 0.001289352100569464,
      "grad_norm": 1.628616452217102,
      "learning_rate": 9.999999999885945e-05,
      "loss": 3.9661,
      "step": 48
    },
    {
      "epoch": 0.001316213602664661,
      "grad_norm": 1.5256599187850952,
      "learning_rate": 9.999999999880578e-05,
      "loss": 3.6918,
      "step": 49
    },
    {
      "epoch": 0.0013430751047598581,
      "grad_norm": 1.6094127893447876,
      "learning_rate": 9.999999999875088e-05,
      "loss": 3.5672,
      "step": 50
    },
    {
      "epoch": 0.0013699366068550553,
      "grad_norm": 1.349199891090393,
      "learning_rate": 9.999999999869475e-05,
      "loss": 3.6309,
      "step": 51
    },
    {
      "epoch": 0.0013967981089502526,
      "grad_norm": 1.3306453227996826,
      "learning_rate": 9.999999999863738e-05,
      "loss": 3.3533,
      "step": 52
    },
    {
      "epoch": 0.0014236596110454496,
      "grad_norm": 1.3546539545059204,
      "learning_rate": 9.999999999857879e-05,
      "loss": 3.2053,
      "step": 53
    },
    {
      "epoch": 0.0014505211131406468,
      "grad_norm": 1.1287821531295776,
      "learning_rate": 9.999999999851894e-05,
      "loss": 3.3292,
      "step": 54
    },
    {
      "epoch": 0.001477382615235844,
      "grad_norm": 1.1314492225646973,
      "learning_rate": 9.999999999845788e-05,
      "loss": 2.9421,
      "step": 55
    },
    {
      "epoch": 0.0015042441173310412,
      "grad_norm": 1.3284823894500732,
      "learning_rate": 9.999999999839558e-05,
      "loss": 3.5217,
      "step": 56
    },
    {
      "epoch": 0.0015311056194262382,
      "grad_norm": 1.223236322402954,
      "learning_rate": 9.999999999833204e-05,
      "loss": 3.376,
      "step": 57
    },
    {
      "epoch": 0.0015579671215214354,
      "grad_norm": 1.043579339981079,
      "learning_rate": 9.999999999826727e-05,
      "loss": 3.1108,
      "step": 58
    },
    {
      "epoch": 0.0015848286236166327,
      "grad_norm": 1.1521614789962769,
      "learning_rate": 9.999999999820126e-05,
      "loss": 3.5073,
      "step": 59
    },
    {
      "epoch": 0.0016116901257118299,
      "grad_norm": 1.1598647832870483,
      "learning_rate": 9.999999999813403e-05,
      "loss": 3.0585,
      "step": 60
    },
    {
      "epoch": 0.0016385516278070269,
      "grad_norm": 1.1426210403442383,
      "learning_rate": 9.999999999806556e-05,
      "loss": 3.5695,
      "step": 61
    },
    {
      "epoch": 0.001665413129902224,
      "grad_norm": 1.1063036918640137,
      "learning_rate": 9.999999999799586e-05,
      "loss": 3.5192,
      "step": 62
    },
    {
      "epoch": 0.0016922746319974213,
      "grad_norm": 1.1488440036773682,
      "learning_rate": 9.999999999792492e-05,
      "loss": 3.3374,
      "step": 63
    },
    {
      "epoch": 0.0017191361340926185,
      "grad_norm": 1.1304515600204468,
      "learning_rate": 9.999999999785274e-05,
      "loss": 3.3304,
      "step": 64
    },
    {
      "epoch": 0.0017459976361878155,
      "grad_norm": 1.1385020017623901,
      "learning_rate": 9.999999999777934e-05,
      "loss": 3.2574,
      "step": 65
    },
    {
      "epoch": 0.0017728591382830128,
      "grad_norm": 1.1962132453918457,
      "learning_rate": 9.999999999770471e-05,
      "loss": 3.5773,
      "step": 66
    },
    {
      "epoch": 0.00179972064037821,
      "grad_norm": 1.2136801481246948,
      "learning_rate": 9.999999999762884e-05,
      "loss": 3.3551,
      "step": 67
    },
    {
      "epoch": 0.0018265821424734072,
      "grad_norm": 1.208842158317566,
      "learning_rate": 9.999999999755171e-05,
      "loss": 3.126,
      "step": 68
    },
    {
      "epoch": 0.0018534436445686042,
      "grad_norm": 1.1841886043548584,
      "learning_rate": 9.999999999747338e-05,
      "loss": 3.3473,
      "step": 69
    },
    {
      "epoch": 0.0018803051466638014,
      "grad_norm": 1.213433861732483,
      "learning_rate": 9.999999999739381e-05,
      "loss": 3.1907,
      "step": 70
    },
    {
      "epoch": 0.0019071666487589986,
      "grad_norm": 1.2414772510528564,
      "learning_rate": 9.9999999997313e-05,
      "loss": 3.5086,
      "step": 71
    },
    {
      "epoch": 0.0019340281508541959,
      "grad_norm": 1.267667293548584,
      "learning_rate": 9.999999999723097e-05,
      "loss": 3.2744,
      "step": 72
    },
    {
      "epoch": 0.001960889652949393,
      "grad_norm": 1.3424843549728394,
      "learning_rate": 9.999999999714769e-05,
      "loss": 3.5996,
      "step": 73
    },
    {
      "epoch": 0.0019877511550445903,
      "grad_norm": 1.2784979343414307,
      "learning_rate": 9.999999999706317e-05,
      "loss": 3.4,
      "step": 74
    },
    {
      "epoch": 0.0020146126571397873,
      "grad_norm": 1.4537830352783203,
      "learning_rate": 9.999999999697744e-05,
      "loss": 3.3768,
      "step": 75
    },
    {
      "epoch": 0.0020414741592349843,
      "grad_norm": 1.3794150352478027,
      "learning_rate": 9.999999999689046e-05,
      "loss": 3.1867,
      "step": 76
    },
    {
      "epoch": 0.0020683356613301817,
      "grad_norm": 1.411429762840271,
      "learning_rate": 9.999999999680225e-05,
      "loss": 3.4702,
      "step": 77
    },
    {
      "epoch": 0.0020951971634253787,
      "grad_norm": 1.3160576820373535,
      "learning_rate": 9.999999999671281e-05,
      "loss": 3.4361,
      "step": 78
    },
    {
      "epoch": 0.0021220586655205757,
      "grad_norm": 1.2753205299377441,
      "learning_rate": 9.999999999662213e-05,
      "loss": 3.3278,
      "step": 79
    },
    {
      "epoch": 0.002148920167615773,
      "grad_norm": 1.4154719114303589,
      "learning_rate": 9.999999999653022e-05,
      "loss": 3.7775,
      "step": 80
    },
    {
      "epoch": 0.00217578166971097,
      "grad_norm": 1.354750156402588,
      "learning_rate": 9.999999999643707e-05,
      "loss": 3.5165,
      "step": 81
    },
    {
      "epoch": 0.0022026431718061676,
      "grad_norm": 1.316260814666748,
      "learning_rate": 9.999999999634269e-05,
      "loss": 3.0653,
      "step": 82
    },
    {
      "epoch": 0.0022295046739013646,
      "grad_norm": 1.3837639093399048,
      "learning_rate": 9.999999999624708e-05,
      "loss": 3.3219,
      "step": 83
    },
    {
      "epoch": 0.0022563661759965616,
      "grad_norm": 1.3703302145004272,
      "learning_rate": 9.999999999615023e-05,
      "loss": 3.2097,
      "step": 84
    },
    {
      "epoch": 0.002283227678091759,
      "grad_norm": 1.4257659912109375,
      "learning_rate": 9.999999999605217e-05,
      "loss": 3.669,
      "step": 85
    },
    {
      "epoch": 0.002310089180186956,
      "grad_norm": 1.4506181478500366,
      "learning_rate": 9.999999999595284e-05,
      "loss": 3.6931,
      "step": 86
    },
    {
      "epoch": 0.002336950682282153,
      "grad_norm": 1.5057146549224854,
      "learning_rate": 9.99999999958523e-05,
      "loss": 3.3992,
      "step": 87
    },
    {
      "epoch": 0.0023638121843773505,
      "grad_norm": 1.5274763107299805,
      "learning_rate": 9.999999999575052e-05,
      "loss": 3.476,
      "step": 88
    },
    {
      "epoch": 0.0023906736864725475,
      "grad_norm": 1.5464972257614136,
      "learning_rate": 9.99999999956475e-05,
      "loss": 3.5653,
      "step": 89
    },
    {
      "epoch": 0.002417535188567745,
      "grad_norm": 1.5150189399719238,
      "learning_rate": 9.999999999554326e-05,
      "loss": 3.5639,
      "step": 90
    },
    {
      "epoch": 0.002444396690662942,
      "grad_norm": 1.5306520462036133,
      "learning_rate": 9.999999999543778e-05,
      "loss": 3.7718,
      "step": 91
    },
    {
      "epoch": 0.002471258192758139,
      "grad_norm": 1.562620997428894,
      "learning_rate": 9.999999999533106e-05,
      "loss": 3.5158,
      "step": 92
    },
    {
      "epoch": 0.0024981196948533364,
      "grad_norm": 1.4864206314086914,
      "learning_rate": 9.999999999522312e-05,
      "loss": 3.6478,
      "step": 93
    },
    {
      "epoch": 0.0025249811969485334,
      "grad_norm": 1.6090619564056396,
      "learning_rate": 9.999999999511394e-05,
      "loss": 3.5618,
      "step": 94
    },
    {
      "epoch": 0.0025518426990437304,
      "grad_norm": 1.538235068321228,
      "learning_rate": 9.999999999500351e-05,
      "loss": 3.3743,
      "step": 95
    },
    {
      "epoch": 0.002578704201138928,
      "grad_norm": 1.647749900817871,
      "learning_rate": 9.999999999489187e-05,
      "loss": 3.77,
      "step": 96
    },
    {
      "epoch": 0.002605565703234125,
      "grad_norm": 1.4851956367492676,
      "learning_rate": 9.999999999477898e-05,
      "loss": 3.3571,
      "step": 97
    },
    {
      "epoch": 0.002632427205329322,
      "grad_norm": 1.6766194105148315,
      "learning_rate": 9.999999999466486e-05,
      "loss": 3.7478,
      "step": 98
    },
    {
      "epoch": 0.0026592887074245192,
      "grad_norm": 1.7860498428344727,
      "learning_rate": 9.999999999454952e-05,
      "loss": 3.7822,
      "step": 99
    },
    {
      "epoch": 0.0026861502095197162,
      "grad_norm": 1.7103192806243896,
      "learning_rate": 9.999999999443292e-05,
      "loss": 3.4418,
      "step": 100
    },
    {
      "epoch": 0.0027130117116149137,
      "grad_norm": 1.218453288078308,
      "learning_rate": 9.999999999431511e-05,
      "loss": 3.5775,
      "step": 101
    },
    {
      "epoch": 0.0027398732137101107,
      "grad_norm": 1.3814231157302856,
      "learning_rate": 9.999999999419605e-05,
      "loss": 3.5856,
      "step": 102
    },
    {
      "epoch": 0.0027667347158053077,
      "grad_norm": 1.2434844970703125,
      "learning_rate": 9.999999999407576e-05,
      "loss": 3.0429,
      "step": 103
    },
    {
      "epoch": 0.002793596217900505,
      "grad_norm": 1.2515223026275635,
      "learning_rate": 9.999999999395425e-05,
      "loss": 3.2445,
      "step": 104
    },
    {
      "epoch": 0.002820457719995702,
      "grad_norm": 1.14759361743927,
      "learning_rate": 9.999999999383149e-05,
      "loss": 3.1756,
      "step": 105
    },
    {
      "epoch": 0.002847319222090899,
      "grad_norm": 1.1015360355377197,
      "learning_rate": 9.99999999937075e-05,
      "loss": 2.8039,
      "step": 106
    },
    {
      "epoch": 0.0028741807241860966,
      "grad_norm": 1.5735284090042114,
      "learning_rate": 9.99999999935823e-05,
      "loss": 3.6006,
      "step": 107
    },
    {
      "epoch": 0.0029010422262812936,
      "grad_norm": 1.27744460105896,
      "learning_rate": 9.999999999345584e-05,
      "loss": 3.1294,
      "step": 108
    },
    {
      "epoch": 0.002927903728376491,
      "grad_norm": 1.2897201776504517,
      "learning_rate": 9.999999999332814e-05,
      "loss": 3.217,
      "step": 109
    },
    {
      "epoch": 0.002954765230471688,
      "grad_norm": 1.176347255706787,
      "learning_rate": 9.999999999319922e-05,
      "loss": 3.1679,
      "step": 110
    },
    {
      "epoch": 0.002981626732566885,
      "grad_norm": 1.3510855436325073,
      "learning_rate": 9.999999999306907e-05,
      "loss": 3.3798,
      "step": 111
    },
    {
      "epoch": 0.0030084882346620824,
      "grad_norm": 1.310192346572876,
      "learning_rate": 9.999999999293767e-05,
      "loss": 3.3365,
      "step": 112
    },
    {
      "epoch": 0.0030353497367572794,
      "grad_norm": 1.339492678642273,
      "learning_rate": 9.999999999280506e-05,
      "loss": 3.1089,
      "step": 113
    },
    {
      "epoch": 0.0030622112388524764,
      "grad_norm": 1.184141755104065,
      "learning_rate": 9.99999999926712e-05,
      "loss": 3.2258,
      "step": 114
    },
    {
      "epoch": 0.003089072740947674,
      "grad_norm": 1.149617314338684,
      "learning_rate": 9.999999999253611e-05,
      "loss": 2.82,
      "step": 115
    },
    {
      "epoch": 0.003115934243042871,
      "grad_norm": 1.402675747871399,
      "learning_rate": 9.99999999923998e-05,
      "loss": 3.486,
      "step": 116
    },
    {
      "epoch": 0.0031427957451380683,
      "grad_norm": 1.5151649713516235,
      "learning_rate": 9.999999999226224e-05,
      "loss": 3.4411,
      "step": 117
    },
    {
      "epoch": 0.0031696572472332653,
      "grad_norm": 1.33055579662323,
      "learning_rate": 9.999999999212344e-05,
      "loss": 3.2941,
      "step": 118
    },
    {
      "epoch": 0.0031965187493284623,
      "grad_norm": 1.2644603252410889,
      "learning_rate": 9.999999999198341e-05,
      "loss": 2.8132,
      "step": 119
    },
    {
      "epoch": 0.0032233802514236598,
      "grad_norm": 1.3672884702682495,
      "learning_rate": 9.999999999184216e-05,
      "loss": 3.1897,
      "step": 120
    },
    {
      "epoch": 0.0032502417535188568,
      "grad_norm": 1.3743053674697876,
      "learning_rate": 9.999999999169966e-05,
      "loss": 3.4031,
      "step": 121
    },
    {
      "epoch": 0.0032771032556140538,
      "grad_norm": 1.4384634494781494,
      "learning_rate": 9.999999999155594e-05,
      "loss": 3.4043,
      "step": 122
    },
    {
      "epoch": 0.003303964757709251,
      "grad_norm": 1.497501015663147,
      "learning_rate": 9.999999999141097e-05,
      "loss": 3.1186,
      "step": 123
    },
    {
      "epoch": 0.003330826259804448,
      "grad_norm": 1.39578378200531,
      "learning_rate": 9.999999999126478e-05,
      "loss": 3.4345,
      "step": 124
    },
    {
      "epoch": 0.0033576877618996456,
      "grad_norm": 1.3490184545516968,
      "learning_rate": 9.999999999111735e-05,
      "loss": 3.3325,
      "step": 125
    },
    {
      "epoch": 0.0033845492639948426,
      "grad_norm": 1.410489797592163,
      "learning_rate": 9.99999999909687e-05,
      "loss": 3.3977,
      "step": 126
    },
    {
      "epoch": 0.0034114107660900396,
      "grad_norm": 1.3900330066680908,
      "learning_rate": 9.999999999081879e-05,
      "loss": 3.4452,
      "step": 127
    },
    {
      "epoch": 0.003438272268185237,
      "grad_norm": 1.3278859853744507,
      "learning_rate": 9.999999999066767e-05,
      "loss": 3.15,
      "step": 128
    },
    {
      "epoch": 0.003465133770280434,
      "grad_norm": 1.5235861539840698,
      "learning_rate": 9.999999999051531e-05,
      "loss": 3.5005,
      "step": 129
    },
    {
      "epoch": 0.003491995272375631,
      "grad_norm": 1.5267044305801392,
      "learning_rate": 9.999999999036172e-05,
      "loss": 3.5311,
      "step": 130
    },
    {
      "epoch": 0.0035188567744708285,
      "grad_norm": 1.4808764457702637,
      "learning_rate": 9.999999999020688e-05,
      "loss": 3.5163,
      "step": 131
    },
    {
      "epoch": 0.0035457182765660255,
      "grad_norm": 1.438101887702942,
      "learning_rate": 9.999999999005082e-05,
      "loss": 3.2459,
      "step": 132
    },
    {
      "epoch": 0.003572579778661223,
      "grad_norm": 1.4126555919647217,
      "learning_rate": 9.999999998989352e-05,
      "loss": 3.3198,
      "step": 133
    },
    {
      "epoch": 0.00359944128075642,
      "grad_norm": 1.53532874584198,
      "learning_rate": 9.9999999989735e-05,
      "loss": 3.3744,
      "step": 134
    },
    {
      "epoch": 0.003626302782851617,
      "grad_norm": 1.4298460483551025,
      "learning_rate": 9.999999998957524e-05,
      "loss": 3.5602,
      "step": 135
    },
    {
      "epoch": 0.0036531642849468144,
      "grad_norm": 1.5221728086471558,
      "learning_rate": 9.999999998941423e-05,
      "loss": 3.4989,
      "step": 136
    },
    {
      "epoch": 0.0036800257870420114,
      "grad_norm": 1.5173331499099731,
      "learning_rate": 9.9999999989252e-05,
      "loss": 3.2325,
      "step": 137
    },
    {
      "epoch": 0.0037068872891372084,
      "grad_norm": 1.512844204902649,
      "learning_rate": 9.999999998908853e-05,
      "loss": 3.5864,
      "step": 138
    },
    {
      "epoch": 0.003733748791232406,
      "grad_norm": 1.5138494968414307,
      "learning_rate": 9.999999998892384e-05,
      "loss": 3.2989,
      "step": 139
    },
    {
      "epoch": 0.003760610293327603,
      "grad_norm": 1.484354019165039,
      "learning_rate": 9.999999998875791e-05,
      "loss": 3.2201,
      "step": 140
    },
    {
      "epoch": 0.0037874717954228,
      "grad_norm": 1.5164704322814941,
      "learning_rate": 9.999999998859073e-05,
      "loss": 3.5061,
      "step": 141
    },
    {
      "epoch": 0.0038143332975179973,
      "grad_norm": 1.4222763776779175,
      "learning_rate": 9.999999998842234e-05,
      "loss": 3.3215,
      "step": 142
    },
    {
      "epoch": 0.0038411947996131943,
      "grad_norm": 1.4848490953445435,
      "learning_rate": 9.99999999882527e-05,
      "loss": 3.2796,
      "step": 143
    },
    {
      "epoch": 0.0038680563017083917,
      "grad_norm": 1.4712228775024414,
      "learning_rate": 9.999999998808183e-05,
      "loss": 3.4786,
      "step": 144
    },
    {
      "epoch": 0.0038949178038035887,
      "grad_norm": 1.654551386833191,
      "learning_rate": 9.999999998790973e-05,
      "loss": 3.4385,
      "step": 145
    },
    {
      "epoch": 0.003921779305898786,
      "grad_norm": 1.503765344619751,
      "learning_rate": 9.999999998773639e-05,
      "loss": 3.3224,
      "step": 146
    },
    {
      "epoch": 0.003948640807993983,
      "grad_norm": 1.7049174308776855,
      "learning_rate": 9.999999998756183e-05,
      "loss": 3.589,
      "step": 147
    },
    {
      "epoch": 0.003975502310089181,
      "grad_norm": 1.5706239938735962,
      "learning_rate": 9.999999998738603e-05,
      "loss": 3.5896,
      "step": 148
    },
    {
      "epoch": 0.004002363812184377,
      "grad_norm": 1.7528107166290283,
      "learning_rate": 9.9999999987209e-05,
      "loss": 3.8007,
      "step": 149
    },
    {
      "epoch": 0.004029225314279575,
      "grad_norm": 1.8006619215011597,
      "learning_rate": 9.999999998703072e-05,
      "loss": 3.7126,
      "step": 150
    },
    {
      "epoch": 0.004056086816374772,
      "grad_norm": 1.501328706741333,
      "learning_rate": 9.999999998685121e-05,
      "loss": 3.5138,
      "step": 151
    },
    {
      "epoch": 0.004082948318469969,
      "grad_norm": 1.5186468362808228,
      "learning_rate": 9.999999998667048e-05,
      "loss": 2.9103,
      "step": 152
    },
    {
      "epoch": 0.004109809820565166,
      "grad_norm": 1.4595462083816528,
      "learning_rate": 9.99999999864885e-05,
      "loss": 3.4846,
      "step": 153
    },
    {
      "epoch": 0.0041366713226603635,
      "grad_norm": 1.4289087057113647,
      "learning_rate": 9.99999999863053e-05,
      "loss": 3.3445,
      "step": 154
    },
    {
      "epoch": 0.00416353282475556,
      "grad_norm": 1.2521617412567139,
      "learning_rate": 9.999999998612086e-05,
      "loss": 3.1745,
      "step": 155
    },
    {
      "epoch": 0.0041903943268507575,
      "grad_norm": 1.2283687591552734,
      "learning_rate": 9.999999998593519e-05,
      "loss": 2.9809,
      "step": 156
    },
    {
      "epoch": 0.004217255828945955,
      "grad_norm": 1.2780406475067139,
      "learning_rate": 9.999999998574829e-05,
      "loss": 3.2447,
      "step": 157
    },
    {
      "epoch": 0.0042441173310411515,
      "grad_norm": 1.2048006057739258,
      "learning_rate": 9.999999998556015e-05,
      "loss": 3.0817,
      "step": 158
    },
    {
      "epoch": 0.004270978833136349,
      "grad_norm": 4.955997467041016,
      "learning_rate": 9.999999998537077e-05,
      "loss": 2.7906,
      "step": 159
    },
    {
      "epoch": 0.004297840335231546,
      "grad_norm": 1.3788132667541504,
      "learning_rate": 9.999999998518017e-05,
      "loss": 3.2995,
      "step": 160
    },
    {
      "epoch": 0.004324701837326743,
      "grad_norm": 1.4886622428894043,
      "learning_rate": 9.999999998498832e-05,
      "loss": 3.0405,
      "step": 161
    },
    {
      "epoch": 0.00435156333942194,
      "grad_norm": 1.3938782215118408,
      "learning_rate": 9.999999998479526e-05,
      "loss": 3.1436,
      "step": 162
    },
    {
      "epoch": 0.004378424841517138,
      "grad_norm": 1.3613883256912231,
      "learning_rate": 9.999999998460095e-05,
      "loss": 2.8424,
      "step": 163
    },
    {
      "epoch": 0.004405286343612335,
      "grad_norm": 1.4670555591583252,
      "learning_rate": 9.999999998440541e-05,
      "loss": 3.4129,
      "step": 164
    },
    {
      "epoch": 0.004432147845707532,
      "grad_norm": 1.307345986366272,
      "learning_rate": 9.999999998420864e-05,
      "loss": 3.1923,
      "step": 165
    },
    {
      "epoch": 0.004459009347802729,
      "grad_norm": 1.3759874105453491,
      "learning_rate": 9.999999998401061e-05,
      "loss": 3.2051,
      "step": 166
    },
    {
      "epoch": 0.004485870849897927,
      "grad_norm": 1.3621212244033813,
      "learning_rate": 9.999999998381138e-05,
      "loss": 3.3061,
      "step": 167
    },
    {
      "epoch": 0.004512732351993123,
      "grad_norm": 1.3702600002288818,
      "learning_rate": 9.99999999836109e-05,
      "loss": 3.1421,
      "step": 168
    },
    {
      "epoch": 0.004539593854088321,
      "grad_norm": 1.4759025573730469,
      "learning_rate": 9.999999998340919e-05,
      "loss": 3.517,
      "step": 169
    },
    {
      "epoch": 0.004566455356183518,
      "grad_norm": 1.5046229362487793,
      "learning_rate": 9.999999998320626e-05,
      "loss": 3.3313,
      "step": 170
    },
    {
      "epoch": 0.004593316858278715,
      "grad_norm": 1.4678044319152832,
      "learning_rate": 9.999999998300207e-05,
      "loss": 3.32,
      "step": 171
    },
    {
      "epoch": 0.004620178360373912,
      "grad_norm": 1.4216737747192383,
      "learning_rate": 9.999999998279666e-05,
      "loss": 2.9374,
      "step": 172
    },
    {
      "epoch": 0.0046470398624691095,
      "grad_norm": 1.5586341619491577,
      "learning_rate": 9.999999998259002e-05,
      "loss": 3.2402,
      "step": 173
    },
    {
      "epoch": 0.004673901364564306,
      "grad_norm": 1.494938850402832,
      "learning_rate": 9.999999998238214e-05,
      "loss": 3.0836,
      "step": 174
    },
    {
      "epoch": 0.0047007628666595035,
      "grad_norm": 1.4996769428253174,
      "learning_rate": 9.999999998217302e-05,
      "loss": 3.0837,
      "step": 175
    },
    {
      "epoch": 0.004727624368754701,
      "grad_norm": 1.468011736869812,
      "learning_rate": 9.999999998196268e-05,
      "loss": 3.2219,
      "step": 176
    },
    {
      "epoch": 0.0047544858708498975,
      "grad_norm": 1.4905102252960205,
      "learning_rate": 9.99999999817511e-05,
      "loss": 3.4999,
      "step": 177
    },
    {
      "epoch": 0.004781347372945095,
      "grad_norm": 1.6069837808609009,
      "learning_rate": 9.999999998153828e-05,
      "loss": 3.3729,
      "step": 178
    },
    {
      "epoch": 0.004808208875040292,
      "grad_norm": 1.5482449531555176,
      "learning_rate": 9.999999998132423e-05,
      "loss": 3.5421,
      "step": 179
    },
    {
      "epoch": 0.00483507037713549,
      "grad_norm": 1.5595391988754272,
      "learning_rate": 9.999999998110896e-05,
      "loss": 3.5452,
      "step": 180
    },
    {
      "epoch": 0.004861931879230686,
      "grad_norm": 1.505689263343811,
      "learning_rate": 9.999999998089245e-05,
      "loss": 3.4279,
      "step": 181
    },
    {
      "epoch": 0.004888793381325884,
      "grad_norm": 1.5317928791046143,
      "learning_rate": 9.999999998067468e-05,
      "loss": 3.1678,
      "step": 182
    },
    {
      "epoch": 0.004915654883421081,
      "grad_norm": 1.6715329885482788,
      "learning_rate": 9.99999999804557e-05,
      "loss": 3.8234,
      "step": 183
    },
    {
      "epoch": 0.004942516385516278,
      "grad_norm": 1.5834782123565674,
      "learning_rate": 9.999999998023549e-05,
      "loss": 3.4183,
      "step": 184
    },
    {
      "epoch": 0.004969377887611475,
      "grad_norm": 1.5933635234832764,
      "learning_rate": 9.999999998001405e-05,
      "loss": 3.1948,
      "step": 185
    },
    {
      "epoch": 0.004996239389706673,
      "grad_norm": 1.658353567123413,
      "learning_rate": 9.999999997979137e-05,
      "loss": 3.6501,
      "step": 186
    },
    {
      "epoch": 0.005023100891801869,
      "grad_norm": 1.5532222986221313,
      "learning_rate": 9.999999997956744e-05,
      "loss": 3.4074,
      "step": 187
    },
    {
      "epoch": 0.005049962393897067,
      "grad_norm": 1.7190356254577637,
      "learning_rate": 9.99999999793423e-05,
      "loss": 3.7267,
      "step": 188
    },
    {
      "epoch": 0.005076823895992264,
      "grad_norm": 1.4863312244415283,
      "learning_rate": 9.999999997911591e-05,
      "loss": 3.3407,
      "step": 189
    },
    {
      "epoch": 0.005103685398087461,
      "grad_norm": 1.798384428024292,
      "learning_rate": 9.99999999788883e-05,
      "loss": 3.5679,
      "step": 190
    },
    {
      "epoch": 0.005130546900182658,
      "grad_norm": 1.6841833591461182,
      "learning_rate": 9.999999997865944e-05,
      "loss": 3.2323,
      "step": 191
    },
    {
      "epoch": 0.005157408402277856,
      "grad_norm": 1.7028827667236328,
      "learning_rate": 9.999999997842934e-05,
      "loss": 3.6724,
      "step": 192
    },
    {
      "epoch": 0.005184269904373052,
      "grad_norm": 1.623820424079895,
      "learning_rate": 9.999999997819804e-05,
      "loss": 3.4745,
      "step": 193
    },
    {
      "epoch": 0.00521113140646825,
      "grad_norm": 1.8269304037094116,
      "learning_rate": 9.999999997796548e-05,
      "loss": 3.839,
      "step": 194
    },
    {
      "epoch": 0.005237992908563447,
      "grad_norm": 1.530657410621643,
      "learning_rate": 9.999999997773171e-05,
      "loss": 3.4414,
      "step": 195
    },
    {
      "epoch": 0.005264854410658644,
      "grad_norm": 1.580862045288086,
      "learning_rate": 9.999999997749667e-05,
      "loss": 3.2415,
      "step": 196
    },
    {
      "epoch": 0.005291715912753841,
      "grad_norm": 1.924195408821106,
      "learning_rate": 9.999999997726042e-05,
      "loss": 3.3448,
      "step": 197
    },
    {
      "epoch": 0.0053185774148490385,
      "grad_norm": 1.864040493965149,
      "learning_rate": 9.999999997702294e-05,
      "loss": 3.5684,
      "step": 198
    },
    {
      "epoch": 0.005345438916944236,
      "grad_norm": 1.8680702447891235,
      "learning_rate": 9.999999997678421e-05,
      "loss": 3.6054,
      "step": 199
    },
    {
      "epoch": 0.0053723004190394325,
      "grad_norm": 1.8232849836349487,
      "learning_rate": 9.999999997654425e-05,
      "loss": 3.4409,
      "step": 200
    },
    {
      "epoch": 0.00539916192113463,
      "grad_norm": 1.6235407590866089,
      "learning_rate": 9.999999997630307e-05,
      "loss": 3.2162,
      "step": 201
    },
    {
      "epoch": 0.005426023423229827,
      "grad_norm": 1.53889799118042,
      "learning_rate": 9.999999997606064e-05,
      "loss": 3.0395,
      "step": 202
    },
    {
      "epoch": 0.005452884925325024,
      "grad_norm": 1.5615484714508057,
      "learning_rate": 9.9999999975817e-05,
      "loss": 3.2514,
      "step": 203
    },
    {
      "epoch": 0.005479746427420221,
      "grad_norm": 1.3152657747268677,
      "learning_rate": 9.99999999755721e-05,
      "loss": 2.9712,
      "step": 204
    },
    {
      "epoch": 0.005506607929515419,
      "grad_norm": 1.3440253734588623,
      "learning_rate": 9.999999997532599e-05,
      "loss": 3.1026,
      "step": 205
    },
    {
      "epoch": 0.005533469431610615,
      "grad_norm": 1.3807686567306519,
      "learning_rate": 9.999999997507863e-05,
      "loss": 3.3082,
      "step": 206
    },
    {
      "epoch": 0.005560330933705813,
      "grad_norm": 1.3441410064697266,
      "learning_rate": 9.999999997483002e-05,
      "loss": 3.2609,
      "step": 207
    },
    {
      "epoch": 0.00558719243580101,
      "grad_norm": 1.3188197612762451,
      "learning_rate": 9.999999997458021e-05,
      "loss": 2.8421,
      "step": 208
    },
    {
      "epoch": 0.005614053937896207,
      "grad_norm": 1.566558599472046,
      "learning_rate": 9.999999997432915e-05,
      "loss": 3.1037,
      "step": 209
    },
    {
      "epoch": 0.005640915439991404,
      "grad_norm": 1.3136351108551025,
      "learning_rate": 9.999999997407685e-05,
      "loss": 3.0866,
      "step": 210
    },
    {
      "epoch": 0.005667776942086602,
      "grad_norm": 1.2555242776870728,
      "learning_rate": 9.999999997382334e-05,
      "loss": 2.9204,
      "step": 211
    },
    {
      "epoch": 0.005694638444181798,
      "grad_norm": 1.334439754486084,
      "learning_rate": 9.999999997356857e-05,
      "loss": 3.1215,
      "step": 212
    },
    {
      "epoch": 0.005721499946276996,
      "grad_norm": 1.4756368398666382,
      "learning_rate": 9.999999997331257e-05,
      "loss": 3.4476,
      "step": 213
    },
    {
      "epoch": 0.005748361448372193,
      "grad_norm": 1.2498222589492798,
      "learning_rate": 9.999999997305536e-05,
      "loss": 2.86,
      "step": 214
    },
    {
      "epoch": 0.0057752229504673906,
      "grad_norm": 1.3437496423721313,
      "learning_rate": 9.99999999727969e-05,
      "loss": 3.1574,
      "step": 215
    },
    {
      "epoch": 0.005802084452562587,
      "grad_norm": 1.3369652032852173,
      "learning_rate": 9.999999997253719e-05,
      "loss": 3.0527,
      "step": 216
    },
    {
      "epoch": 0.0058289459546577846,
      "grad_norm": 1.3278571367263794,
      "learning_rate": 9.999999997227626e-05,
      "loss": 3.1739,
      "step": 217
    },
    {
      "epoch": 0.005855807456752982,
      "grad_norm": 1.4438539743423462,
      "learning_rate": 9.999999997201412e-05,
      "loss": 3.35,
      "step": 218
    },
    {
      "epoch": 0.0058826689588481786,
      "grad_norm": 1.4686214923858643,
      "learning_rate": 9.999999997175071e-05,
      "loss": 3.2776,
      "step": 219
    },
    {
      "epoch": 0.005909530460943376,
      "grad_norm": 1.4738445281982422,
      "learning_rate": 9.999999997148608e-05,
      "loss": 3.2764,
      "step": 220
    },
    {
      "epoch": 0.005936391963038573,
      "grad_norm": 1.3048352003097534,
      "learning_rate": 9.999999997122022e-05,
      "loss": 3.007,
      "step": 221
    },
    {
      "epoch": 0.00596325346513377,
      "grad_norm": 1.4230749607086182,
      "learning_rate": 9.999999997095313e-05,
      "loss": 3.0977,
      "step": 222
    },
    {
      "epoch": 0.0059901149672289674,
      "grad_norm": 1.462730884552002,
      "learning_rate": 9.99999999706848e-05,
      "loss": 3.3268,
      "step": 223
    },
    {
      "epoch": 0.006016976469324165,
      "grad_norm": 1.6287310123443604,
      "learning_rate": 9.999999997041523e-05,
      "loss": 3.3482,
      "step": 224
    },
    {
      "epoch": 0.0060438379714193614,
      "grad_norm": 1.577708125114441,
      "learning_rate": 9.999999997014444e-05,
      "loss": 3.2947,
      "step": 225
    },
    {
      "epoch": 0.006070699473514559,
      "grad_norm": 1.4368644952774048,
      "learning_rate": 9.99999999698724e-05,
      "loss": 3.1511,
      "step": 226
    },
    {
      "epoch": 0.006097560975609756,
      "grad_norm": 1.7501544952392578,
      "learning_rate": 9.999999996959914e-05,
      "loss": 3.6171,
      "step": 227
    },
    {
      "epoch": 0.006124422477704953,
      "grad_norm": 1.531460165977478,
      "learning_rate": 9.999999996932465e-05,
      "loss": 3.3854,
      "step": 228
    },
    {
      "epoch": 0.00615128397980015,
      "grad_norm": 1.4792691469192505,
      "learning_rate": 9.99999999690489e-05,
      "loss": 3.2036,
      "step": 229
    },
    {
      "epoch": 0.006178145481895348,
      "grad_norm": 1.4752593040466309,
      "learning_rate": 9.999999996877195e-05,
      "loss": 3.0966,
      "step": 230
    },
    {
      "epoch": 0.006205006983990545,
      "grad_norm": 1.560064435005188,
      "learning_rate": 9.999999996849374e-05,
      "loss": 3.2259,
      "step": 231
    },
    {
      "epoch": 0.006231868486085742,
      "grad_norm": 1.5049961805343628,
      "learning_rate": 9.999999996821431e-05,
      "loss": 3.255,
      "step": 232
    },
    {
      "epoch": 0.006258729988180939,
      "grad_norm": 1.6931719779968262,
      "learning_rate": 9.999999996793364e-05,
      "loss": 3.2948,
      "step": 233
    },
    {
      "epoch": 0.006285591490276137,
      "grad_norm": 1.6067354679107666,
      "learning_rate": 9.999999996765175e-05,
      "loss": 3.3771,
      "step": 234
    },
    {
      "epoch": 0.006312452992371333,
      "grad_norm": 1.5622678995132446,
      "learning_rate": 9.999999996736861e-05,
      "loss": 3.2915,
      "step": 235
    },
    {
      "epoch": 0.006339314494466531,
      "grad_norm": 1.6007894277572632,
      "learning_rate": 9.999999996708425e-05,
      "loss": 3.4445,
      "step": 236
    },
    {
      "epoch": 0.006366175996561728,
      "grad_norm": 1.4729998111724854,
      "learning_rate": 9.999999996679864e-05,
      "loss": 3.3731,
      "step": 237
    },
    {
      "epoch": 0.006393037498656925,
      "grad_norm": 1.5431838035583496,
      "learning_rate": 9.99999999665118e-05,
      "loss": 3.2531,
      "step": 238
    },
    {
      "epoch": 0.006419899000752122,
      "grad_norm": 1.636124849319458,
      "learning_rate": 9.999999996622373e-05,
      "loss": 3.5058,
      "step": 239
    },
    {
      "epoch": 0.0064467605028473195,
      "grad_norm": 1.5989148616790771,
      "learning_rate": 9.999999996593443e-05,
      "loss": 3.2455,
      "step": 240
    },
    {
      "epoch": 0.006473622004942516,
      "grad_norm": 1.642051339149475,
      "learning_rate": 9.999999996564389e-05,
      "loss": 3.1518,
      "step": 241
    },
    {
      "epoch": 0.0065004835070377135,
      "grad_norm": 1.5991356372833252,
      "learning_rate": 9.999999996535212e-05,
      "loss": 3.1274,
      "step": 242
    },
    {
      "epoch": 0.006527345009132911,
      "grad_norm": 1.601693868637085,
      "learning_rate": 9.999999996505912e-05,
      "loss": 3.584,
      "step": 243
    },
    {
      "epoch": 0.0065542065112281075,
      "grad_norm": 1.5522338151931763,
      "learning_rate": 9.999999996476488e-05,
      "loss": 3.3002,
      "step": 244
    },
    {
      "epoch": 0.006581068013323305,
      "grad_norm": 1.7774641513824463,
      "learning_rate": 9.99999999644694e-05,
      "loss": 3.7639,
      "step": 245
    },
    {
      "epoch": 0.006607929515418502,
      "grad_norm": 1.7426048517227173,
      "learning_rate": 9.99999999641727e-05,
      "loss": 3.6212,
      "step": 246
    },
    {
      "epoch": 0.006634791017513699,
      "grad_norm": 1.7567298412322998,
      "learning_rate": 9.999999996387477e-05,
      "loss": 3.5256,
      "step": 247
    },
    {
      "epoch": 0.006661652519608896,
      "grad_norm": 1.7526336908340454,
      "learning_rate": 9.99999999635756e-05,
      "loss": 3.5622,
      "step": 248
    },
    {
      "epoch": 0.006688514021704094,
      "grad_norm": 1.787500262260437,
      "learning_rate": 9.999999996327518e-05,
      "loss": 3.5157,
      "step": 249
    },
    {
      "epoch": 0.006715375523799291,
      "grad_norm": 2.0876896381378174,
      "learning_rate": 9.999999996297355e-05,
      "loss": 3.3864,
      "step": 250
    },
    {
      "epoch": 0.006742237025894488,
      "grad_norm": 1.5887773036956787,
      "learning_rate": 9.999999996267068e-05,
      "loss": 2.6411,
      "step": 251
    },
    {
      "epoch": 0.006769098527989685,
      "grad_norm": 1.7187421321868896,
      "learning_rate": 9.999999996236657e-05,
      "loss": 2.9251,
      "step": 252
    },
    {
      "epoch": 0.006795960030084883,
      "grad_norm": 1.5112552642822266,
      "learning_rate": 9.999999996206122e-05,
      "loss": 3.1039,
      "step": 253
    },
    {
      "epoch": 0.006822821532180079,
      "grad_norm": 1.426690697669983,
      "learning_rate": 9.999999996175465e-05,
      "loss": 3.2291,
      "step": 254
    },
    {
      "epoch": 0.006849683034275277,
      "grad_norm": 1.2940804958343506,
      "learning_rate": 9.999999996144685e-05,
      "loss": 2.9074,
      "step": 255
    },
    {
      "epoch": 0.006876544536370474,
      "grad_norm": 1.4958122968673706,
      "learning_rate": 9.999999996113781e-05,
      "loss": 3.4205,
      "step": 256
    },
    {
      "epoch": 0.006903406038465671,
      "grad_norm": 1.314099669456482,
      "learning_rate": 9.999999996082752e-05,
      "loss": 3.1381,
      "step": 257
    },
    {
      "epoch": 0.006930267540560868,
      "grad_norm": 1.276431679725647,
      "learning_rate": 9.9999999960516e-05,
      "loss": 3.1831,
      "step": 258
    },
    {
      "epoch": 0.006957129042656066,
      "grad_norm": 1.3868881464004517,
      "learning_rate": 9.999999996020327e-05,
      "loss": 3.0969,
      "step": 259
    },
    {
      "epoch": 0.006983990544751262,
      "grad_norm": 1.3772238492965698,
      "learning_rate": 9.99999999598893e-05,
      "loss": 3.3114,
      "step": 260
    },
    {
      "epoch": 0.00701085204684646,
      "grad_norm": 1.3541710376739502,
      "learning_rate": 9.999999995957409e-05,
      "loss": 2.8867,
      "step": 261
    },
    {
      "epoch": 0.007037713548941657,
      "grad_norm": 1.481025218963623,
      "learning_rate": 9.999999995925764e-05,
      "loss": 3.2412,
      "step": 262
    },
    {
      "epoch": 0.007064575051036854,
      "grad_norm": 1.4574871063232422,
      "learning_rate": 9.999999995893995e-05,
      "loss": 3.1844,
      "step": 263
    },
    {
      "epoch": 0.007091436553132051,
      "grad_norm": 1.5566786527633667,
      "learning_rate": 9.999999995862105e-05,
      "loss": 3.2533,
      "step": 264
    },
    {
      "epoch": 0.0071182980552272485,
      "grad_norm": 1.298317551612854,
      "learning_rate": 9.999999995830091e-05,
      "loss": 3.1958,
      "step": 265
    },
    {
      "epoch": 0.007145159557322446,
      "grad_norm": 1.3086849451065063,
      "learning_rate": 9.999999995797953e-05,
      "loss": 3.1858,
      "step": 266
    },
    {
      "epoch": 0.0071720210594176425,
      "grad_norm": 1.4598180055618286,
      "learning_rate": 9.999999995765692e-05,
      "loss": 3.3412,
      "step": 267
    },
    {
      "epoch": 0.00719888256151284,
      "grad_norm": 1.3468729257583618,
      "learning_rate": 9.999999995733307e-05,
      "loss": 3.1933,
      "step": 268
    },
    {
      "epoch": 0.007225744063608037,
      "grad_norm": 1.5812733173370361,
      "learning_rate": 9.9999999957008e-05,
      "loss": 3.0884,
      "step": 269
    },
    {
      "epoch": 0.007252605565703234,
      "grad_norm": 1.3265537023544312,
      "learning_rate": 9.999999995668168e-05,
      "loss": 2.8339,
      "step": 270
    },
    {
      "epoch": 0.007279467067798431,
      "grad_norm": 1.4566750526428223,
      "learning_rate": 9.999999995635412e-05,
      "loss": 3.0427,
      "step": 271
    },
    {
      "epoch": 0.007306328569893629,
      "grad_norm": 1.4008580446243286,
      "learning_rate": 9.999999995602533e-05,
      "loss": 2.9818,
      "step": 272
    },
    {
      "epoch": 0.007333190071988825,
      "grad_norm": 1.4401434659957886,
      "learning_rate": 9.999999995569533e-05,
      "loss": 3.1373,
      "step": 273
    },
    {
      "epoch": 0.007360051574084023,
      "grad_norm": 1.3396170139312744,
      "learning_rate": 9.999999995536409e-05,
      "loss": 3.1087,
      "step": 274
    },
    {
      "epoch": 0.00738691307617922,
      "grad_norm": 1.440423846244812,
      "learning_rate": 9.99999999550316e-05,
      "loss": 3.2071,
      "step": 275
    },
    {
      "epoch": 0.007413774578274417,
      "grad_norm": 1.3632419109344482,
      "learning_rate": 9.999999995469789e-05,
      "loss": 3.024,
      "step": 276
    },
    {
      "epoch": 0.007440636080369614,
      "grad_norm": 1.4464221000671387,
      "learning_rate": 9.999999995436293e-05,
      "loss": 3.0848,
      "step": 277
    },
    {
      "epoch": 0.007467497582464812,
      "grad_norm": 1.3937352895736694,
      "learning_rate": 9.999999995402674e-05,
      "loss": 3.1426,
      "step": 278
    },
    {
      "epoch": 0.007494359084560008,
      "grad_norm": 1.3715002536773682,
      "learning_rate": 9.999999995368934e-05,
      "loss": 3.1451,
      "step": 279
    },
    {
      "epoch": 0.007521220586655206,
      "grad_norm": 1.6965054273605347,
      "learning_rate": 9.999999995335067e-05,
      "loss": 3.4319,
      "step": 280
    },
    {
      "epoch": 0.007548082088750403,
      "grad_norm": 1.5740978717803955,
      "learning_rate": 9.999999995301079e-05,
      "loss": 3.4204,
      "step": 281
    },
    {
      "epoch": 0.0075749435908456,
      "grad_norm": 1.5966700315475464,
      "learning_rate": 9.999999995266967e-05,
      "loss": 3.3832,
      "step": 282
    },
    {
      "epoch": 0.007601805092940797,
      "grad_norm": 1.4009469747543335,
      "learning_rate": 9.999999995232731e-05,
      "loss": 3.0976,
      "step": 283
    },
    {
      "epoch": 0.0076286665950359945,
      "grad_norm": 1.580001950263977,
      "learning_rate": 9.999999995198374e-05,
      "loss": 3.0962,
      "step": 284
    },
    {
      "epoch": 0.007655528097131192,
      "grad_norm": 1.5711950063705444,
      "learning_rate": 9.999999995163891e-05,
      "loss": 3.2862,
      "step": 285
    },
    {
      "epoch": 0.0076823895992263885,
      "grad_norm": 1.6567752361297607,
      "learning_rate": 9.999999995129287e-05,
      "loss": 3.387,
      "step": 286
    },
    {
      "epoch": 0.007709251101321586,
      "grad_norm": 1.5363513231277466,
      "learning_rate": 9.999999995094558e-05,
      "loss": 3.3296,
      "step": 287
    },
    {
      "epoch": 0.007736112603416783,
      "grad_norm": 1.628709316253662,
      "learning_rate": 9.999999995059706e-05,
      "loss": 3.4314,
      "step": 288
    },
    {
      "epoch": 0.00776297410551198,
      "grad_norm": 1.683313250541687,
      "learning_rate": 9.999999995024731e-05,
      "loss": 3.2991,
      "step": 289
    },
    {
      "epoch": 0.007789835607607177,
      "grad_norm": 1.7183477878570557,
      "learning_rate": 9.999999994989631e-05,
      "loss": 3.366,
      "step": 290
    },
    {
      "epoch": 0.007816697109702375,
      "grad_norm": 1.825942039489746,
      "learning_rate": 9.99999999495441e-05,
      "loss": 3.933,
      "step": 291
    },
    {
      "epoch": 0.007843558611797571,
      "grad_norm": 1.5922329425811768,
      "learning_rate": 9.999999994919063e-05,
      "loss": 3.3191,
      "step": 292
    },
    {
      "epoch": 0.00787042011389277,
      "grad_norm": 1.6675676107406616,
      "learning_rate": 9.999999994883595e-05,
      "loss": 3.7358,
      "step": 293
    },
    {
      "epoch": 0.007897281615987966,
      "grad_norm": 1.6281962394714355,
      "learning_rate": 9.999999994848003e-05,
      "loss": 3.1607,
      "step": 294
    },
    {
      "epoch": 0.007924143118083163,
      "grad_norm": 1.5776563882827759,
      "learning_rate": 9.999999994812287e-05,
      "loss": 3.4825,
      "step": 295
    },
    {
      "epoch": 0.007951004620178361,
      "grad_norm": 1.6171278953552246,
      "learning_rate": 9.999999994776447e-05,
      "loss": 3.285,
      "step": 296
    },
    {
      "epoch": 0.007977866122273558,
      "grad_norm": 1.6754804849624634,
      "learning_rate": 9.999999994740486e-05,
      "loss": 3.4483,
      "step": 297
    },
    {
      "epoch": 0.008004727624368754,
      "grad_norm": 1.6712759733200073,
      "learning_rate": 9.999999994704399e-05,
      "loss": 3.578,
      "step": 298
    },
    {
      "epoch": 0.008031589126463953,
      "grad_norm": 1.7725811004638672,
      "learning_rate": 9.999999994668191e-05,
      "loss": 3.4371,
      "step": 299
    },
    {
      "epoch": 0.00805845062855915,
      "grad_norm": 1.9692758321762085,
      "learning_rate": 9.99999999463186e-05,
      "loss": 3.8032,
      "step": 300
    },
    {
      "epoch": 0.008085312130654346,
      "grad_norm": 1.2866451740264893,
      "learning_rate": 9.999999994595403e-05,
      "loss": 3.2611,
      "step": 301
    },
    {
      "epoch": 0.008112173632749544,
      "grad_norm": 1.4625946283340454,
      "learning_rate": 9.999999994558823e-05,
      "loss": 3.0714,
      "step": 302
    },
    {
      "epoch": 0.00813903513484474,
      "grad_norm": 1.3855098485946655,
      "learning_rate": 9.99999999452212e-05,
      "loss": 3.1154,
      "step": 303
    },
    {
      "epoch": 0.008165896636939937,
      "grad_norm": 1.3074418306350708,
      "learning_rate": 9.999999994485294e-05,
      "loss": 3.2195,
      "step": 304
    },
    {
      "epoch": 0.008192758139035135,
      "grad_norm": 1.3705579042434692,
      "learning_rate": 9.999999994448346e-05,
      "loss": 3.1222,
      "step": 305
    },
    {
      "epoch": 0.008219619641130332,
      "grad_norm": 1.3016101121902466,
      "learning_rate": 9.999999994411273e-05,
      "loss": 3.0081,
      "step": 306
    },
    {
      "epoch": 0.008246481143225529,
      "grad_norm": 1.5367039442062378,
      "learning_rate": 9.999999994374077e-05,
      "loss": 3.3542,
      "step": 307
    },
    {
      "epoch": 0.008273342645320727,
      "grad_norm": 1.352465271949768,
      "learning_rate": 9.999999994336757e-05,
      "loss": 3.2663,
      "step": 308
    },
    {
      "epoch": 0.008300204147415923,
      "grad_norm": 1.0709216594696045,
      "learning_rate": 9.999999994299314e-05,
      "loss": 2.5144,
      "step": 309
    },
    {
      "epoch": 0.00832706564951112,
      "grad_norm": 1.2512985467910767,
      "learning_rate": 9.999999994261748e-05,
      "loss": 2.9406,
      "step": 310
    },
    {
      "epoch": 0.008353927151606318,
      "grad_norm": 1.4661740064620972,
      "learning_rate": 9.999999994224058e-05,
      "loss": 3.3997,
      "step": 311
    },
    {
      "epoch": 0.008380788653701515,
      "grad_norm": 1.3103265762329102,
      "learning_rate": 9.999999994186246e-05,
      "loss": 3.0447,
      "step": 312
    },
    {
      "epoch": 0.008407650155796711,
      "grad_norm": 1.5357261896133423,
      "learning_rate": 9.999999994148309e-05,
      "loss": 3.2413,
      "step": 313
    },
    {
      "epoch": 0.00843451165789191,
      "grad_norm": 1.3907216787338257,
      "learning_rate": 9.99999999411025e-05,
      "loss": 3.1204,
      "step": 314
    },
    {
      "epoch": 0.008461373159987106,
      "grad_norm": 1.3356871604919434,
      "learning_rate": 9.999999994072067e-05,
      "loss": 3.0059,
      "step": 315
    },
    {
      "epoch": 0.008488234662082303,
      "grad_norm": 1.3362263441085815,
      "learning_rate": 9.999999994033759e-05,
      "loss": 3.1497,
      "step": 316
    },
    {
      "epoch": 0.008515096164177501,
      "grad_norm": 1.3796724081039429,
      "learning_rate": 9.999999993995331e-05,
      "loss": 3.1882,
      "step": 317
    },
    {
      "epoch": 0.008541957666272698,
      "grad_norm": 1.4778923988342285,
      "learning_rate": 9.999999993956777e-05,
      "loss": 2.8893,
      "step": 318
    },
    {
      "epoch": 0.008568819168367894,
      "grad_norm": 1.4544243812561035,
      "learning_rate": 9.9999999939181e-05,
      "loss": 3.1249,
      "step": 319
    },
    {
      "epoch": 0.008595680670463093,
      "grad_norm": 1.3993685245513916,
      "learning_rate": 9.999999993879301e-05,
      "loss": 3.0868,
      "step": 320
    },
    {
      "epoch": 0.00862254217255829,
      "grad_norm": 1.455314040184021,
      "learning_rate": 9.999999993840377e-05,
      "loss": 3.2488,
      "step": 321
    },
    {
      "epoch": 0.008649403674653486,
      "grad_norm": 1.535010576248169,
      "learning_rate": 9.99999999380133e-05,
      "loss": 3.211,
      "step": 322
    },
    {
      "epoch": 0.008676265176748684,
      "grad_norm": 1.4369255304336548,
      "learning_rate": 9.999999993762161e-05,
      "loss": 3.268,
      "step": 323
    },
    {
      "epoch": 0.00870312667884388,
      "grad_norm": 2.008056163787842,
      "learning_rate": 9.999999993722867e-05,
      "loss": 2.9857,
      "step": 324
    },
    {
      "epoch": 0.008729988180939077,
      "grad_norm": 1.3951343297958374,
      "learning_rate": 9.999999993683451e-05,
      "loss": 3.2106,
      "step": 325
    },
    {
      "epoch": 0.008756849683034276,
      "grad_norm": 1.472520112991333,
      "learning_rate": 9.99999999364391e-05,
      "loss": 3.181,
      "step": 326
    },
    {
      "epoch": 0.008783711185129472,
      "grad_norm": 1.491974949836731,
      "learning_rate": 9.999999993604247e-05,
      "loss": 3.2049,
      "step": 327
    },
    {
      "epoch": 0.00881057268722467,
      "grad_norm": 1.5334579944610596,
      "learning_rate": 9.99999999356446e-05,
      "loss": 3.3579,
      "step": 328
    },
    {
      "epoch": 0.008837434189319867,
      "grad_norm": 1.509615421295166,
      "learning_rate": 9.999999993524549e-05,
      "loss": 3.5056,
      "step": 329
    },
    {
      "epoch": 0.008864295691415064,
      "grad_norm": 1.575596570968628,
      "learning_rate": 9.999999993484517e-05,
      "loss": 3.1586,
      "step": 330
    },
    {
      "epoch": 0.008891157193510262,
      "grad_norm": 1.5231512784957886,
      "learning_rate": 9.99999999344436e-05,
      "loss": 3.0471,
      "step": 331
    },
    {
      "epoch": 0.008918018695605458,
      "grad_norm": 1.5751467943191528,
      "learning_rate": 9.999999993404079e-05,
      "loss": 3.3389,
      "step": 332
    },
    {
      "epoch": 0.008944880197700655,
      "grad_norm": 1.436456561088562,
      "learning_rate": 9.999999993363676e-05,
      "loss": 3.0194,
      "step": 333
    },
    {
      "epoch": 0.008971741699795853,
      "grad_norm": 1.59831702709198,
      "learning_rate": 9.999999993323147e-05,
      "loss": 3.112,
      "step": 334
    },
    {
      "epoch": 0.00899860320189105,
      "grad_norm": 1.468543529510498,
      "learning_rate": 9.999999993282498e-05,
      "loss": 3.3302,
      "step": 335
    },
    {
      "epoch": 0.009025464703986246,
      "grad_norm": 1.5483806133270264,
      "learning_rate": 9.999999993241724e-05,
      "loss": 3.2404,
      "step": 336
    },
    {
      "epoch": 0.009052326206081445,
      "grad_norm": 1.6208584308624268,
      "learning_rate": 9.999999993200826e-05,
      "loss": 3.0803,
      "step": 337
    },
    {
      "epoch": 0.009079187708176641,
      "grad_norm": 1.4745221138000488,
      "learning_rate": 9.999999993159806e-05,
      "loss": 3.4383,
      "step": 338
    },
    {
      "epoch": 0.009106049210271838,
      "grad_norm": 1.5517362356185913,
      "learning_rate": 9.999999993118663e-05,
      "loss": 3.2555,
      "step": 339
    },
    {
      "epoch": 0.009132910712367036,
      "grad_norm": 1.5948114395141602,
      "learning_rate": 9.999999993077394e-05,
      "loss": 3.1166,
      "step": 340
    },
    {
      "epoch": 0.009159772214462233,
      "grad_norm": 1.6069583892822266,
      "learning_rate": 9.999999993036005e-05,
      "loss": 3.3697,
      "step": 341
    },
    {
      "epoch": 0.00918663371655743,
      "grad_norm": 1.6950137615203857,
      "learning_rate": 9.99999999299449e-05,
      "loss": 3.2143,
      "step": 342
    },
    {
      "epoch": 0.009213495218652628,
      "grad_norm": 1.7324154376983643,
      "learning_rate": 9.999999992952853e-05,
      "loss": 3.6276,
      "step": 343
    },
    {
      "epoch": 0.009240356720747824,
      "grad_norm": 1.6985573768615723,
      "learning_rate": 9.999999992911092e-05,
      "loss": 3.3153,
      "step": 344
    },
    {
      "epoch": 0.00926721822284302,
      "grad_norm": 1.7411456108093262,
      "learning_rate": 9.999999992869208e-05,
      "loss": 3.7364,
      "step": 345
    },
    {
      "epoch": 0.009294079724938219,
      "grad_norm": 1.7979718446731567,
      "learning_rate": 9.999999992827201e-05,
      "loss": 3.3799,
      "step": 346
    },
    {
      "epoch": 0.009320941227033416,
      "grad_norm": 1.7107107639312744,
      "learning_rate": 9.99999999278507e-05,
      "loss": 3.3509,
      "step": 347
    },
    {
      "epoch": 0.009347802729128612,
      "grad_norm": 1.7422338724136353,
      "learning_rate": 9.999999992742815e-05,
      "loss": 3.2159,
      "step": 348
    },
    {
      "epoch": 0.00937466423122381,
      "grad_norm": 1.9913054704666138,
      "learning_rate": 9.999999992700437e-05,
      "loss": 3.5324,
      "step": 349
    },
    {
      "epoch": 0.009401525733319007,
      "grad_norm": 2.0499820709228516,
      "learning_rate": 9.999999992657935e-05,
      "loss": 3.2073,
      "step": 350
    },
    {
      "epoch": 0.009428387235414204,
      "grad_norm": 1.312639832496643,
      "learning_rate": 9.999999992615312e-05,
      "loss": 3.1846,
      "step": 351
    },
    {
      "epoch": 0.009455248737509402,
      "grad_norm": 1.534393310546875,
      "learning_rate": 9.999999992572564e-05,
      "loss": 3.1861,
      "step": 352
    },
    {
      "epoch": 0.009482110239604599,
      "grad_norm": 1.592322587966919,
      "learning_rate": 9.999999992529694e-05,
      "loss": 3.202,
      "step": 353
    },
    {
      "epoch": 0.009508971741699795,
      "grad_norm": 1.6027579307556152,
      "learning_rate": 9.999999992486699e-05,
      "loss": 3.3655,
      "step": 354
    },
    {
      "epoch": 0.009535833243794993,
      "grad_norm": 1.460989236831665,
      "learning_rate": 9.99999999244358e-05,
      "loss": 3.0499,
      "step": 355
    },
    {
      "epoch": 0.00956269474589019,
      "grad_norm": 1.4667208194732666,
      "learning_rate": 9.99999999240034e-05,
      "loss": 3.0652,
      "step": 356
    },
    {
      "epoch": 0.009589556247985387,
      "grad_norm": 1.4265388250350952,
      "learning_rate": 9.999999992356974e-05,
      "loss": 3.0453,
      "step": 357
    },
    {
      "epoch": 0.009616417750080585,
      "grad_norm": 1.4427202939987183,
      "learning_rate": 9.999999992313487e-05,
      "loss": 3.1117,
      "step": 358
    },
    {
      "epoch": 0.009643279252175781,
      "grad_norm": 1.3827509880065918,
      "learning_rate": 9.999999992269875e-05,
      "loss": 3.0258,
      "step": 359
    },
    {
      "epoch": 0.00967014075427098,
      "grad_norm": 3.8355908393859863,
      "learning_rate": 9.999999992226141e-05,
      "loss": 3.1564,
      "step": 360
    },
    {
      "epoch": 0.009697002256366176,
      "grad_norm": 1.4933533668518066,
      "learning_rate": 9.999999992182283e-05,
      "loss": 3.2878,
      "step": 361
    },
    {
      "epoch": 0.009723863758461373,
      "grad_norm": 1.4010118246078491,
      "learning_rate": 9.999999992138301e-05,
      "loss": 2.9805,
      "step": 362
    },
    {
      "epoch": 0.009750725260556571,
      "grad_norm": 1.441159725189209,
      "learning_rate": 9.999999992094196e-05,
      "loss": 3.1448,
      "step": 363
    },
    {
      "epoch": 0.009777586762651768,
      "grad_norm": 1.6072075366973877,
      "learning_rate": 9.999999992049969e-05,
      "loss": 3.1391,
      "step": 364
    },
    {
      "epoch": 0.009804448264746964,
      "grad_norm": 1.3306028842926025,
      "learning_rate": 9.999999992005618e-05,
      "loss": 2.9318,
      "step": 365
    },
    {
      "epoch": 0.009831309766842163,
      "grad_norm": 1.4072214365005493,
      "learning_rate": 9.999999991961141e-05,
      "loss": 3.0531,
      "step": 366
    },
    {
      "epoch": 0.00985817126893736,
      "grad_norm": 1.4834871292114258,
      "learning_rate": 9.999999991916544e-05,
      "loss": 2.9282,
      "step": 367
    },
    {
      "epoch": 0.009885032771032556,
      "grad_norm": 1.5989537239074707,
      "learning_rate": 9.999999991871822e-05,
      "loss": 3.3957,
      "step": 368
    },
    {
      "epoch": 0.009911894273127754,
      "grad_norm": 1.4718068838119507,
      "learning_rate": 9.999999991826977e-05,
      "loss": 3.3333,
      "step": 369
    },
    {
      "epoch": 0.00993875577522295,
      "grad_norm": 1.4870935678482056,
      "learning_rate": 9.999999991782008e-05,
      "loss": 3.3692,
      "step": 370
    },
    {
      "epoch": 0.009965617277318147,
      "grad_norm": 1.3331711292266846,
      "learning_rate": 9.999999991736916e-05,
      "loss": 3.0298,
      "step": 371
    },
    {
      "epoch": 0.009992478779413345,
      "grad_norm": 1.485838770866394,
      "learning_rate": 9.999999991691702e-05,
      "loss": 3.2719,
      "step": 372
    },
    {
      "epoch": 0.010019340281508542,
      "grad_norm": 1.5010319948196411,
      "learning_rate": 9.999999991646362e-05,
      "loss": 3.0529,
      "step": 373
    },
    {
      "epoch": 0.010046201783603739,
      "grad_norm": 1.4825512170791626,
      "learning_rate": 9.999999991600902e-05,
      "loss": 3.0555,
      "step": 374
    },
    {
      "epoch": 0.010073063285698937,
      "grad_norm": 1.4355058670043945,
      "learning_rate": 9.999999991555315e-05,
      "loss": 2.9217,
      "step": 375
    },
    {
      "epoch": 0.010099924787794133,
      "grad_norm": 1.448727011680603,
      "learning_rate": 9.999999991509608e-05,
      "loss": 3.1127,
      "step": 376
    },
    {
      "epoch": 0.01012678628988933,
      "grad_norm": 1.4077942371368408,
      "learning_rate": 9.999999991463776e-05,
      "loss": 3.0063,
      "step": 377
    },
    {
      "epoch": 0.010153647791984528,
      "grad_norm": 1.4879511594772339,
      "learning_rate": 9.999999991417819e-05,
      "loss": 3.0618,
      "step": 378
    },
    {
      "epoch": 0.010180509294079725,
      "grad_norm": 1.4629594087600708,
      "learning_rate": 9.999999991371741e-05,
      "loss": 3.0791,
      "step": 379
    },
    {
      "epoch": 0.010207370796174921,
      "grad_norm": 1.5846166610717773,
      "learning_rate": 9.999999991325539e-05,
      "loss": 3.3179,
      "step": 380
    },
    {
      "epoch": 0.01023423229827012,
      "grad_norm": 1.6092251539230347,
      "learning_rate": 9.999999991279214e-05,
      "loss": 3.1621,
      "step": 381
    },
    {
      "epoch": 0.010261093800365316,
      "grad_norm": 1.3157117366790771,
      "learning_rate": 9.999999991232765e-05,
      "loss": 3.0647,
      "step": 382
    },
    {
      "epoch": 0.010287955302460513,
      "grad_norm": 1.5434112548828125,
      "learning_rate": 9.999999991186193e-05,
      "loss": 3.1533,
      "step": 383
    },
    {
      "epoch": 0.010314816804555711,
      "grad_norm": 1.445216178894043,
      "learning_rate": 9.999999991139496e-05,
      "loss": 3.2086,
      "step": 384
    },
    {
      "epoch": 0.010341678306650908,
      "grad_norm": 1.4799890518188477,
      "learning_rate": 9.999999991092678e-05,
      "loss": 3.0183,
      "step": 385
    },
    {
      "epoch": 0.010368539808746104,
      "grad_norm": 1.5840139389038086,
      "learning_rate": 9.999999991045736e-05,
      "loss": 3.3566,
      "step": 386
    },
    {
      "epoch": 0.010395401310841303,
      "grad_norm": 1.7577760219573975,
      "learning_rate": 9.99999999099867e-05,
      "loss": 3.687,
      "step": 387
    },
    {
      "epoch": 0.0104222628129365,
      "grad_norm": 1.7453808784484863,
      "learning_rate": 9.999999990951481e-05,
      "loss": 3.4915,
      "step": 388
    },
    {
      "epoch": 0.010449124315031696,
      "grad_norm": 1.5384831428527832,
      "learning_rate": 9.999999990904169e-05,
      "loss": 3.2543,
      "step": 389
    },
    {
      "epoch": 0.010475985817126894,
      "grad_norm": 1.5846140384674072,
      "learning_rate": 9.999999990856732e-05,
      "loss": 3.5913,
      "step": 390
    },
    {
      "epoch": 0.01050284731922209,
      "grad_norm": 1.6113022565841675,
      "learning_rate": 9.999999990809174e-05,
      "loss": 3.2361,
      "step": 391
    },
    {
      "epoch": 0.010529708821317287,
      "grad_norm": 1.611128330230713,
      "learning_rate": 9.999999990761491e-05,
      "loss": 3.3629,
      "step": 392
    },
    {
      "epoch": 0.010556570323412486,
      "grad_norm": 1.6922285556793213,
      "learning_rate": 9.999999990713684e-05,
      "loss": 3.4664,
      "step": 393
    },
    {
      "epoch": 0.010583431825507682,
      "grad_norm": 1.6885075569152832,
      "learning_rate": 9.999999990665756e-05,
      "loss": 3.3561,
      "step": 394
    },
    {
      "epoch": 0.01061029332760288,
      "grad_norm": 1.6231778860092163,
      "learning_rate": 9.999999990617704e-05,
      "loss": 3.2889,
      "step": 395
    },
    {
      "epoch": 0.010637154829698077,
      "grad_norm": 1.790924310684204,
      "learning_rate": 9.999999990569526e-05,
      "loss": 3.7997,
      "step": 396
    },
    {
      "epoch": 0.010664016331793274,
      "grad_norm": 1.620592713356018,
      "learning_rate": 9.999999990521227e-05,
      "loss": 3.109,
      "step": 397
    },
    {
      "epoch": 0.010690877833888472,
      "grad_norm": 1.6597546339035034,
      "learning_rate": 9.999999990472804e-05,
      "loss": 3.4153,
      "step": 398
    },
    {
      "epoch": 0.010717739335983668,
      "grad_norm": 2.0343093872070312,
      "learning_rate": 9.999999990424258e-05,
      "loss": 3.6979,
      "step": 399
    },
    {
      "epoch": 0.010744600838078865,
      "grad_norm": 1.827853798866272,
      "learning_rate": 9.999999990375589e-05,
      "loss": 3.4915,
      "step": 400
    },
    {
      "epoch": 0.010771462340174063,
      "grad_norm": 1.4068386554718018,
      "learning_rate": 9.999999990326796e-05,
      "loss": 3.139,
      "step": 401
    },
    {
      "epoch": 0.01079832384226926,
      "grad_norm": 1.4641603231430054,
      "learning_rate": 9.99999999027788e-05,
      "loss": 3.0299,
      "step": 402
    },
    {
      "epoch": 0.010825185344364456,
      "grad_norm": 1.3369860649108887,
      "learning_rate": 9.999999990228841e-05,
      "loss": 2.801,
      "step": 403
    },
    {
      "epoch": 0.010852046846459655,
      "grad_norm": 1.3419314622879028,
      "learning_rate": 9.999999990179678e-05,
      "loss": 2.8149,
      "step": 404
    },
    {
      "epoch": 0.010878908348554851,
      "grad_norm": 1.5593358278274536,
      "learning_rate": 9.999999990130392e-05,
      "loss": 3.4785,
      "step": 405
    },
    {
      "epoch": 0.010905769850650048,
      "grad_norm": 1.4393575191497803,
      "learning_rate": 9.999999990080982e-05,
      "loss": 3.0239,
      "step": 406
    },
    {
      "epoch": 0.010932631352745246,
      "grad_norm": 1.4988009929656982,
      "learning_rate": 9.999999990031447e-05,
      "loss": 3.2712,
      "step": 407
    },
    {
      "epoch": 0.010959492854840443,
      "grad_norm": 1.4504315853118896,
      "learning_rate": 9.999999989981792e-05,
      "loss": 3.1714,
      "step": 408
    },
    {
      "epoch": 0.01098635435693564,
      "grad_norm": 1.4853525161743164,
      "learning_rate": 9.999999989932012e-05,
      "loss": 3.418,
      "step": 409
    },
    {
      "epoch": 0.011013215859030838,
      "grad_norm": 1.2990763187408447,
      "learning_rate": 9.999999989882109e-05,
      "loss": 2.9569,
      "step": 410
    },
    {
      "epoch": 0.011040077361126034,
      "grad_norm": 1.3418400287628174,
      "learning_rate": 9.999999989832083e-05,
      "loss": 3.0375,
      "step": 411
    },
    {
      "epoch": 0.01106693886322123,
      "grad_norm": 1.4683629274368286,
      "learning_rate": 9.999999989781932e-05,
      "loss": 3.2594,
      "step": 412
    },
    {
      "epoch": 0.011093800365316429,
      "grad_norm": 1.411016583442688,
      "learning_rate": 9.999999989731659e-05,
      "loss": 3.2121,
      "step": 413
    },
    {
      "epoch": 0.011120661867411626,
      "grad_norm": 1.4058409929275513,
      "learning_rate": 9.999999989681262e-05,
      "loss": 3.0771,
      "step": 414
    },
    {
      "epoch": 0.011147523369506822,
      "grad_norm": 1.346800684928894,
      "learning_rate": 9.999999989630741e-05,
      "loss": 2.9472,
      "step": 415
    },
    {
      "epoch": 0.01117438487160202,
      "grad_norm": 1.4477880001068115,
      "learning_rate": 9.999999989580099e-05,
      "loss": 3.0263,
      "step": 416
    },
    {
      "epoch": 0.011201246373697217,
      "grad_norm": 1.4545361995697021,
      "learning_rate": 9.999999989529332e-05,
      "loss": 3.2445,
      "step": 417
    },
    {
      "epoch": 0.011228107875792414,
      "grad_norm": 1.5225719213485718,
      "learning_rate": 9.999999989478442e-05,
      "loss": 3.1295,
      "step": 418
    },
    {
      "epoch": 0.011254969377887612,
      "grad_norm": 1.4374668598175049,
      "learning_rate": 9.999999989427429e-05,
      "loss": 3.0734,
      "step": 419
    },
    {
      "epoch": 0.011281830879982808,
      "grad_norm": 1.5443471670150757,
      "learning_rate": 9.999999989376292e-05,
      "loss": 2.9032,
      "step": 420
    },
    {
      "epoch": 0.011308692382078005,
      "grad_norm": 1.4011187553405762,
      "learning_rate": 9.999999989325031e-05,
      "loss": 2.7283,
      "step": 421
    },
    {
      "epoch": 0.011335553884173203,
      "grad_norm": 1.5621442794799805,
      "learning_rate": 9.999999989273647e-05,
      "loss": 3.0155,
      "step": 422
    },
    {
      "epoch": 0.0113624153862684,
      "grad_norm": 1.4946808815002441,
      "learning_rate": 9.999999989222139e-05,
      "loss": 3.2491,
      "step": 423
    },
    {
      "epoch": 0.011389276888363596,
      "grad_norm": 1.5118460655212402,
      "learning_rate": 9.99999998917051e-05,
      "loss": 2.917,
      "step": 424
    },
    {
      "epoch": 0.011416138390458795,
      "grad_norm": 1.5413126945495605,
      "learning_rate": 9.999999989118755e-05,
      "loss": 3.2541,
      "step": 425
    },
    {
      "epoch": 0.011442999892553991,
      "grad_norm": 1.6237248182296753,
      "learning_rate": 9.99999998906688e-05,
      "loss": 3.2308,
      "step": 426
    },
    {
      "epoch": 0.011469861394649188,
      "grad_norm": 1.5881716012954712,
      "learning_rate": 9.999999989014879e-05,
      "loss": 3.1807,
      "step": 427
    },
    {
      "epoch": 0.011496722896744386,
      "grad_norm": 1.5346410274505615,
      "learning_rate": 9.999999988962755e-05,
      "loss": 3.1304,
      "step": 428
    },
    {
      "epoch": 0.011523584398839583,
      "grad_norm": 1.464072346687317,
      "learning_rate": 9.999999988910507e-05,
      "loss": 3.0906,
      "step": 429
    },
    {
      "epoch": 0.011550445900934781,
      "grad_norm": 1.582999587059021,
      "learning_rate": 9.999999988858137e-05,
      "loss": 3.0908,
      "step": 430
    },
    {
      "epoch": 0.011577307403029978,
      "grad_norm": 1.4797403812408447,
      "learning_rate": 9.999999988805643e-05,
      "loss": 3.1175,
      "step": 431
    },
    {
      "epoch": 0.011604168905125174,
      "grad_norm": 1.5761067867279053,
      "learning_rate": 9.999999988753027e-05,
      "loss": 3.3944,
      "step": 432
    },
    {
      "epoch": 0.011631030407220373,
      "grad_norm": 1.6513606309890747,
      "learning_rate": 9.999999988700286e-05,
      "loss": 3.3121,
      "step": 433
    },
    {
      "epoch": 0.011657891909315569,
      "grad_norm": 1.7659647464752197,
      "learning_rate": 9.99999998864742e-05,
      "loss": 3.277,
      "step": 434
    },
    {
      "epoch": 0.011684753411410766,
      "grad_norm": 1.5365813970565796,
      "learning_rate": 9.999999988594434e-05,
      "loss": 3.2994,
      "step": 435
    },
    {
      "epoch": 0.011711614913505964,
      "grad_norm": 1.5114079713821411,
      "learning_rate": 9.999999988541323e-05,
      "loss": 3.1791,
      "step": 436
    },
    {
      "epoch": 0.01173847641560116,
      "grad_norm": 1.73617422580719,
      "learning_rate": 9.999999988488088e-05,
      "loss": 3.5268,
      "step": 437
    },
    {
      "epoch": 0.011765337917696357,
      "grad_norm": 1.7058979272842407,
      "learning_rate": 9.99999998843473e-05,
      "loss": 3.1726,
      "step": 438
    },
    {
      "epoch": 0.011792199419791555,
      "grad_norm": 1.5015861988067627,
      "learning_rate": 9.99999998838125e-05,
      "loss": 3.2026,
      "step": 439
    },
    {
      "epoch": 0.011819060921886752,
      "grad_norm": 1.6483575105667114,
      "learning_rate": 9.999999988327645e-05,
      "loss": 3.3578,
      "step": 440
    },
    {
      "epoch": 0.011845922423981949,
      "grad_norm": 1.7205007076263428,
      "learning_rate": 9.999999988273918e-05,
      "loss": 3.5342,
      "step": 441
    },
    {
      "epoch": 0.011872783926077147,
      "grad_norm": 1.6439756155014038,
      "learning_rate": 9.999999988220067e-05,
      "loss": 3.3699,
      "step": 442
    },
    {
      "epoch": 0.011899645428172343,
      "grad_norm": 1.710311770439148,
      "learning_rate": 9.999999988166092e-05,
      "loss": 3.5054,
      "step": 443
    },
    {
      "epoch": 0.01192650693026754,
      "grad_norm": 1.7583166360855103,
      "learning_rate": 9.999999988111995e-05,
      "loss": 3.4383,
      "step": 444
    },
    {
      "epoch": 0.011953368432362738,
      "grad_norm": 1.5195735692977905,
      "learning_rate": 9.999999988057774e-05,
      "loss": 2.9549,
      "step": 445
    },
    {
      "epoch": 0.011980229934457935,
      "grad_norm": 1.7824205160140991,
      "learning_rate": 9.999999988003428e-05,
      "loss": 3.1322,
      "step": 446
    },
    {
      "epoch": 0.012007091436553131,
      "grad_norm": 1.7998090982437134,
      "learning_rate": 9.99999998794896e-05,
      "loss": 3.6423,
      "step": 447
    },
    {
      "epoch": 0.01203395293864833,
      "grad_norm": 1.6461634635925293,
      "learning_rate": 9.99999998789437e-05,
      "loss": 3.4975,
      "step": 448
    },
    {
      "epoch": 0.012060814440743526,
      "grad_norm": 1.8779922723770142,
      "learning_rate": 9.999999987839654e-05,
      "loss": 3.4399,
      "step": 449
    },
    {
      "epoch": 0.012087675942838723,
      "grad_norm": 1.8773419857025146,
      "learning_rate": 9.999999987784816e-05,
      "loss": 3.4724,
      "step": 450
    },
    {
      "epoch": 0.012114537444933921,
      "grad_norm": 1.4512723684310913,
      "learning_rate": 9.999999987729855e-05,
      "loss": 3.4861,
      "step": 451
    },
    {
      "epoch": 0.012141398947029118,
      "grad_norm": 1.4454329013824463,
      "learning_rate": 9.999999987674771e-05,
      "loss": 2.9476,
      "step": 452
    },
    {
      "epoch": 0.012168260449124314,
      "grad_norm": 1.5218937397003174,
      "learning_rate": 9.999999987619562e-05,
      "loss": 3.3391,
      "step": 453
    },
    {
      "epoch": 0.012195121951219513,
      "grad_norm": 1.4006235599517822,
      "learning_rate": 9.99999998756423e-05,
      "loss": 3.1386,
      "step": 454
    },
    {
      "epoch": 0.01222198345331471,
      "grad_norm": 1.5272748470306396,
      "learning_rate": 9.999999987508777e-05,
      "loss": 3.2105,
      "step": 455
    },
    {
      "epoch": 0.012248844955409906,
      "grad_norm": 1.381325125694275,
      "learning_rate": 9.999999987453198e-05,
      "loss": 3.0307,
      "step": 456
    },
    {
      "epoch": 0.012275706457505104,
      "grad_norm": 1.5673829317092896,
      "learning_rate": 9.999999987397497e-05,
      "loss": 3.2447,
      "step": 457
    },
    {
      "epoch": 0.0123025679596003,
      "grad_norm": 1.5407352447509766,
      "learning_rate": 9.99999998734167e-05,
      "loss": 3.0539,
      "step": 458
    },
    {
      "epoch": 0.012329429461695497,
      "grad_norm": 1.4541857242584229,
      "learning_rate": 9.999999987285723e-05,
      "loss": 3.212,
      "step": 459
    },
    {
      "epoch": 0.012356290963790695,
      "grad_norm": 1.3955122232437134,
      "learning_rate": 9.99999998722965e-05,
      "loss": 2.8864,
      "step": 460
    },
    {
      "epoch": 0.012383152465885892,
      "grad_norm": 1.4555715322494507,
      "learning_rate": 9.999999987173456e-05,
      "loss": 3.1965,
      "step": 461
    },
    {
      "epoch": 0.01241001396798109,
      "grad_norm": 1.4746215343475342,
      "learning_rate": 9.999999987117138e-05,
      "loss": 3.0635,
      "step": 462
    },
    {
      "epoch": 0.012436875470076287,
      "grad_norm": 1.5263087749481201,
      "learning_rate": 9.999999987060695e-05,
      "loss": 2.9506,
      "step": 463
    },
    {
      "epoch": 0.012463736972171484,
      "grad_norm": 1.4185150861740112,
      "learning_rate": 9.999999987004131e-05,
      "loss": 2.6964,
      "step": 464
    },
    {
      "epoch": 0.012490598474266682,
      "grad_norm": 1.4453799724578857,
      "learning_rate": 9.999999986947441e-05,
      "loss": 3.1973,
      "step": 465
    },
    {
      "epoch": 0.012517459976361878,
      "grad_norm": 1.4354561567306519,
      "learning_rate": 9.99999998689063e-05,
      "loss": 2.9152,
      "step": 466
    },
    {
      "epoch": 0.012544321478457075,
      "grad_norm": 1.3852994441986084,
      "learning_rate": 9.999999986833695e-05,
      "loss": 2.7374,
      "step": 467
    },
    {
      "epoch": 0.012571182980552273,
      "grad_norm": 1.4055900573730469,
      "learning_rate": 9.999999986776636e-05,
      "loss": 2.9589,
      "step": 468
    },
    {
      "epoch": 0.01259804448264747,
      "grad_norm": 1.6043518781661987,
      "learning_rate": 9.999999986719455e-05,
      "loss": 2.8888,
      "step": 469
    },
    {
      "epoch": 0.012624905984742666,
      "grad_norm": 1.423304796218872,
      "learning_rate": 9.999999986662148e-05,
      "loss": 2.9067,
      "step": 470
    },
    {
      "epoch": 0.012651767486837865,
      "grad_norm": 1.4667428731918335,
      "learning_rate": 9.99999998660472e-05,
      "loss": 3.1209,
      "step": 471
    },
    {
      "epoch": 0.012678628988933061,
      "grad_norm": 1.458530068397522,
      "learning_rate": 9.999999986547168e-05,
      "loss": 3.1869,
      "step": 472
    },
    {
      "epoch": 0.012705490491028258,
      "grad_norm": 1.5382295846939087,
      "learning_rate": 9.999999986489492e-05,
      "loss": 2.9852,
      "step": 473
    },
    {
      "epoch": 0.012732351993123456,
      "grad_norm": 1.6067780256271362,
      "learning_rate": 9.999999986431694e-05,
      "loss": 3.4255,
      "step": 474
    },
    {
      "epoch": 0.012759213495218653,
      "grad_norm": 1.3856332302093506,
      "learning_rate": 9.999999986373772e-05,
      "loss": 2.9723,
      "step": 475
    },
    {
      "epoch": 0.01278607499731385,
      "grad_norm": 1.409085750579834,
      "learning_rate": 9.999999986315725e-05,
      "loss": 2.9678,
      "step": 476
    },
    {
      "epoch": 0.012812936499409048,
      "grad_norm": 1.5785013437271118,
      "learning_rate": 9.999999986257556e-05,
      "loss": 3.0543,
      "step": 477
    },
    {
      "epoch": 0.012839798001504244,
      "grad_norm": 1.5739150047302246,
      "learning_rate": 9.999999986199264e-05,
      "loss": 3.0962,
      "step": 478
    },
    {
      "epoch": 0.01286665950359944,
      "grad_norm": 1.715063452720642,
      "learning_rate": 9.999999986140849e-05,
      "loss": 3.4156,
      "step": 479
    },
    {
      "epoch": 0.012893521005694639,
      "grad_norm": 1.5047529935836792,
      "learning_rate": 9.999999986082309e-05,
      "loss": 3.2267,
      "step": 480
    },
    {
      "epoch": 0.012920382507789836,
      "grad_norm": 1.5376882553100586,
      "learning_rate": 9.999999986023646e-05,
      "loss": 3.3192,
      "step": 481
    },
    {
      "epoch": 0.012947244009885032,
      "grad_norm": 1.6161760091781616,
      "learning_rate": 9.999999985964861e-05,
      "loss": 3.1084,
      "step": 482
    },
    {
      "epoch": 0.01297410551198023,
      "grad_norm": 1.5804671049118042,
      "learning_rate": 9.999999985905952e-05,
      "loss": 3.1579,
      "step": 483
    },
    {
      "epoch": 0.013000967014075427,
      "grad_norm": 1.6535561084747314,
      "learning_rate": 9.99999998584692e-05,
      "loss": 3.1284,
      "step": 484
    },
    {
      "epoch": 0.013027828516170624,
      "grad_norm": 1.6472439765930176,
      "learning_rate": 9.999999985787762e-05,
      "loss": 3.2893,
      "step": 485
    },
    {
      "epoch": 0.013054690018265822,
      "grad_norm": 1.5657492876052856,
      "learning_rate": 9.999999985728483e-05,
      "loss": 3.3764,
      "step": 486
    },
    {
      "epoch": 0.013081551520361018,
      "grad_norm": 1.6238974332809448,
      "learning_rate": 9.999999985669081e-05,
      "loss": 3.3152,
      "step": 487
    },
    {
      "epoch": 0.013108413022456215,
      "grad_norm": 1.579158067703247,
      "learning_rate": 9.999999985609554e-05,
      "loss": 3.5067,
      "step": 488
    },
    {
      "epoch": 0.013135274524551413,
      "grad_norm": 1.7080647945404053,
      "learning_rate": 9.999999985549906e-05,
      "loss": 3.1746,
      "step": 489
    },
    {
      "epoch": 0.01316213602664661,
      "grad_norm": 1.7463475465774536,
      "learning_rate": 9.999999985490132e-05,
      "loss": 3.3698,
      "step": 490
    },
    {
      "epoch": 0.013188997528741806,
      "grad_norm": 1.751397967338562,
      "learning_rate": 9.999999985430236e-05,
      "loss": 3.5693,
      "step": 491
    },
    {
      "epoch": 0.013215859030837005,
      "grad_norm": 1.6995598077774048,
      "learning_rate": 9.999999985370217e-05,
      "loss": 3.4366,
      "step": 492
    },
    {
      "epoch": 0.013242720532932201,
      "grad_norm": 1.7562099695205688,
      "learning_rate": 9.999999985310073e-05,
      "loss": 3.5061,
      "step": 493
    },
    {
      "epoch": 0.013269582035027398,
      "grad_norm": 1.6325514316558838,
      "learning_rate": 9.999999985249808e-05,
      "loss": 3.2136,
      "step": 494
    },
    {
      "epoch": 0.013296443537122596,
      "grad_norm": 1.7133991718292236,
      "learning_rate": 9.999999985189418e-05,
      "loss": 3.5505,
      "step": 495
    },
    {
      "epoch": 0.013323305039217793,
      "grad_norm": 1.8352411985397339,
      "learning_rate": 9.999999985128905e-05,
      "loss": 3.5474,
      "step": 496
    },
    {
      "epoch": 0.013350166541312991,
      "grad_norm": 1.6895605325698853,
      "learning_rate": 9.999999985068269e-05,
      "loss": 3.2428,
      "step": 497
    },
    {
      "epoch": 0.013377028043408188,
      "grad_norm": 1.8699710369110107,
      "learning_rate": 9.999999985007509e-05,
      "loss": 3.3019,
      "step": 498
    },
    {
      "epoch": 0.013403889545503384,
      "grad_norm": 1.8091951608657837,
      "learning_rate": 9.999999984946626e-05,
      "loss": 3.3801,
      "step": 499
    },
    {
      "epoch": 0.013430751047598583,
      "grad_norm": 1.845009207725525,
      "learning_rate": 9.999999984885619e-05,
      "loss": 3.4538,
      "step": 500
    },
    {
      "epoch": 0.013457612549693779,
      "grad_norm": 1.5051531791687012,
      "learning_rate": 9.999999984824489e-05,
      "loss": 3.2036,
      "step": 501
    },
    {
      "epoch": 0.013484474051788976,
      "grad_norm": 1.5916658639907837,
      "learning_rate": 9.999999984763236e-05,
      "loss": 2.9292,
      "step": 502
    },
    {
      "epoch": 0.013511335553884174,
      "grad_norm": 1.5836458206176758,
      "learning_rate": 9.99999998470186e-05,
      "loss": 3.3527,
      "step": 503
    },
    {
      "epoch": 0.01353819705597937,
      "grad_norm": 1.5160671472549438,
      "learning_rate": 9.999999984640359e-05,
      "loss": 2.9836,
      "step": 504
    },
    {
      "epoch": 0.013565058558074567,
      "grad_norm": 1.4878536462783813,
      "learning_rate": 9.999999984578736e-05,
      "loss": 3.1274,
      "step": 505
    },
    {
      "epoch": 0.013591920060169765,
      "grad_norm": 1.5014201402664185,
      "learning_rate": 9.999999984516989e-05,
      "loss": 2.9355,
      "step": 506
    },
    {
      "epoch": 0.013618781562264962,
      "grad_norm": 1.4217443466186523,
      "learning_rate": 9.999999984455119e-05,
      "loss": 3.0484,
      "step": 507
    },
    {
      "epoch": 0.013645643064360159,
      "grad_norm": 1.5473932027816772,
      "learning_rate": 9.999999984393126e-05,
      "loss": 3.2235,
      "step": 508
    },
    {
      "epoch": 0.013672504566455357,
      "grad_norm": 1.4834808111190796,
      "learning_rate": 9.999999984331009e-05,
      "loss": 3.1755,
      "step": 509
    },
    {
      "epoch": 0.013699366068550553,
      "grad_norm": 1.636918306350708,
      "learning_rate": 9.999999984268769e-05,
      "loss": 3.3078,
      "step": 510
    },
    {
      "epoch": 0.01372622757064575,
      "grad_norm": 1.439376711845398,
      "learning_rate": 9.999999984206406e-05,
      "loss": 3.0135,
      "step": 511
    },
    {
      "epoch": 0.013753089072740948,
      "grad_norm": 1.612460732460022,
      "learning_rate": 9.999999984143918e-05,
      "loss": 3.1161,
      "step": 512
    },
    {
      "epoch": 0.013779950574836145,
      "grad_norm": 1.2844470739364624,
      "learning_rate": 9.999999984081308e-05,
      "loss": 2.8637,
      "step": 513
    },
    {
      "epoch": 0.013806812076931341,
      "grad_norm": 1.4732033014297485,
      "learning_rate": 9.999999984018573e-05,
      "loss": 3.1481,
      "step": 514
    },
    {
      "epoch": 0.01383367357902654,
      "grad_norm": 1.3747831583023071,
      "learning_rate": 9.999999983955717e-05,
      "loss": 2.9905,
      "step": 515
    },
    {
      "epoch": 0.013860535081121736,
      "grad_norm": 1.550026297569275,
      "learning_rate": 9.999999983892737e-05,
      "loss": 3.2305,
      "step": 516
    },
    {
      "epoch": 0.013887396583216933,
      "grad_norm": 1.468211054801941,
      "learning_rate": 9.999999983829633e-05,
      "loss": 2.9263,
      "step": 517
    },
    {
      "epoch": 0.013914258085312131,
      "grad_norm": 1.5542832612991333,
      "learning_rate": 9.999999983766405e-05,
      "loss": 3.0559,
      "step": 518
    },
    {
      "epoch": 0.013941119587407328,
      "grad_norm": 1.5159426927566528,
      "learning_rate": 9.999999983703054e-05,
      "loss": 2.8209,
      "step": 519
    },
    {
      "epoch": 0.013967981089502524,
      "grad_norm": 1.4121785163879395,
      "learning_rate": 9.999999983639581e-05,
      "loss": 3.2127,
      "step": 520
    },
    {
      "epoch": 0.013994842591597723,
      "grad_norm": 1.531212329864502,
      "learning_rate": 9.999999983575983e-05,
      "loss": 3.0407,
      "step": 521
    },
    {
      "epoch": 0.01402170409369292,
      "grad_norm": 1.4030855894088745,
      "learning_rate": 9.999999983512263e-05,
      "loss": 2.8646,
      "step": 522
    },
    {
      "epoch": 0.014048565595788116,
      "grad_norm": 1.542206883430481,
      "learning_rate": 9.999999983448418e-05,
      "loss": 3.1509,
      "step": 523
    },
    {
      "epoch": 0.014075427097883314,
      "grad_norm": 1.414257526397705,
      "learning_rate": 9.999999983384452e-05,
      "loss": 3.0755,
      "step": 524
    },
    {
      "epoch": 0.01410228859997851,
      "grad_norm": 1.5378986597061157,
      "learning_rate": 9.99999998332036e-05,
      "loss": 3.0178,
      "step": 525
    },
    {
      "epoch": 0.014129150102073707,
      "grad_norm": 1.573044776916504,
      "learning_rate": 9.999999983256146e-05,
      "loss": 3.1776,
      "step": 526
    },
    {
      "epoch": 0.014156011604168905,
      "grad_norm": 1.5487502813339233,
      "learning_rate": 9.99999998319181e-05,
      "loss": 3.3601,
      "step": 527
    },
    {
      "epoch": 0.014182873106264102,
      "grad_norm": 1.622823715209961,
      "learning_rate": 9.999999983127348e-05,
      "loss": 3.1433,
      "step": 528
    },
    {
      "epoch": 0.014209734608359299,
      "grad_norm": 1.4159866571426392,
      "learning_rate": 9.999999983062765e-05,
      "loss": 3.0213,
      "step": 529
    },
    {
      "epoch": 0.014236596110454497,
      "grad_norm": 1.5231095552444458,
      "learning_rate": 9.999999982998056e-05,
      "loss": 3.0939,
      "step": 530
    },
    {
      "epoch": 0.014263457612549693,
      "grad_norm": 1.5016103982925415,
      "learning_rate": 9.999999982933226e-05,
      "loss": 3.3296,
      "step": 531
    },
    {
      "epoch": 0.014290319114644892,
      "grad_norm": 1.5221531391143799,
      "learning_rate": 9.999999982868271e-05,
      "loss": 3.2847,
      "step": 532
    },
    {
      "epoch": 0.014317180616740088,
      "grad_norm": 1.6948050260543823,
      "learning_rate": 9.999999982803193e-05,
      "loss": 3.1667,
      "step": 533
    },
    {
      "epoch": 0.014344042118835285,
      "grad_norm": 1.4808393716812134,
      "learning_rate": 9.999999982737992e-05,
      "loss": 2.9595,
      "step": 534
    },
    {
      "epoch": 0.014370903620930483,
      "grad_norm": 1.594460129737854,
      "learning_rate": 9.999999982672668e-05,
      "loss": 3.0397,
      "step": 535
    },
    {
      "epoch": 0.01439776512302568,
      "grad_norm": 1.7172738313674927,
      "learning_rate": 9.99999998260722e-05,
      "loss": 3.1509,
      "step": 536
    },
    {
      "epoch": 0.014424626625120876,
      "grad_norm": 1.5233323574066162,
      "learning_rate": 9.999999982541649e-05,
      "loss": 3.1804,
      "step": 537
    },
    {
      "epoch": 0.014451488127216075,
      "grad_norm": 1.71015465259552,
      "learning_rate": 9.999999982475954e-05,
      "loss": 3.2557,
      "step": 538
    },
    {
      "epoch": 0.014478349629311271,
      "grad_norm": 1.7213354110717773,
      "learning_rate": 9.999999982410135e-05,
      "loss": 3.474,
      "step": 539
    },
    {
      "epoch": 0.014505211131406468,
      "grad_norm": 1.5233113765716553,
      "learning_rate": 9.999999982344195e-05,
      "loss": 3.193,
      "step": 540
    },
    {
      "epoch": 0.014532072633501666,
      "grad_norm": 1.641194462776184,
      "learning_rate": 9.99999998227813e-05,
      "loss": 3.2334,
      "step": 541
    },
    {
      "epoch": 0.014558934135596863,
      "grad_norm": 1.7879207134246826,
      "learning_rate": 9.999999982211941e-05,
      "loss": 3.3306,
      "step": 542
    },
    {
      "epoch": 0.01458579563769206,
      "grad_norm": 1.6878842115402222,
      "learning_rate": 9.99999998214563e-05,
      "loss": 3.4826,
      "step": 543
    },
    {
      "epoch": 0.014612657139787258,
      "grad_norm": 1.5964375734329224,
      "learning_rate": 9.999999982079196e-05,
      "loss": 3.1481,
      "step": 544
    },
    {
      "epoch": 0.014639518641882454,
      "grad_norm": 1.7950385808944702,
      "learning_rate": 9.999999982012638e-05,
      "loss": 3.4581,
      "step": 545
    },
    {
      "epoch": 0.01466638014397765,
      "grad_norm": 1.717629075050354,
      "learning_rate": 9.999999981945955e-05,
      "loss": 3.1694,
      "step": 546
    },
    {
      "epoch": 0.014693241646072849,
      "grad_norm": 1.9837907552719116,
      "learning_rate": 9.999999981879151e-05,
      "loss": 3.4366,
      "step": 547
    },
    {
      "epoch": 0.014720103148168046,
      "grad_norm": 1.6364156007766724,
      "learning_rate": 9.999999981812223e-05,
      "loss": 3.1522,
      "step": 548
    },
    {
      "epoch": 0.014746964650263242,
      "grad_norm": 2.122551202774048,
      "learning_rate": 9.999999981745172e-05,
      "loss": 3.4088,
      "step": 549
    },
    {
      "epoch": 0.01477382615235844,
      "grad_norm": 2.0202836990356445,
      "learning_rate": 9.999999981677996e-05,
      "loss": 3.3569,
      "step": 550
    },
    {
      "epoch": 0.014800687654453637,
      "grad_norm": 1.4466760158538818,
      "learning_rate": 9.999999981610697e-05,
      "loss": 2.767,
      "step": 551
    },
    {
      "epoch": 0.014827549156548834,
      "grad_norm": 1.565651774406433,
      "learning_rate": 9.999999981543276e-05,
      "loss": 3.2593,
      "step": 552
    },
    {
      "epoch": 0.014854410658644032,
      "grad_norm": 1.4580707550048828,
      "learning_rate": 9.99999998147573e-05,
      "loss": 3.1369,
      "step": 553
    },
    {
      "epoch": 0.014881272160739228,
      "grad_norm": 1.5976176261901855,
      "learning_rate": 9.999999981408062e-05,
      "loss": 3.3074,
      "step": 554
    },
    {
      "epoch": 0.014908133662834425,
      "grad_norm": 1.3057708740234375,
      "learning_rate": 9.999999981340271e-05,
      "loss": 3.0269,
      "step": 555
    },
    {
      "epoch": 0.014934995164929623,
      "grad_norm": 1.2500848770141602,
      "learning_rate": 9.999999981272354e-05,
      "loss": 2.6988,
      "step": 556
    },
    {
      "epoch": 0.01496185666702482,
      "grad_norm": 1.383833408355713,
      "learning_rate": 9.999999981204316e-05,
      "loss": 3.1782,
      "step": 557
    },
    {
      "epoch": 0.014988718169120016,
      "grad_norm": 1.4894264936447144,
      "learning_rate": 9.999999981136154e-05,
      "loss": 3.0019,
      "step": 558
    },
    {
      "epoch": 0.015015579671215215,
      "grad_norm": 1.4194146394729614,
      "learning_rate": 9.99999998106787e-05,
      "loss": 2.7848,
      "step": 559
    },
    {
      "epoch": 0.015042441173310411,
      "grad_norm": 1.6750750541687012,
      "learning_rate": 9.99999998099946e-05,
      "loss": 2.9919,
      "step": 560
    },
    {
      "epoch": 0.015069302675405608,
      "grad_norm": 1.4904073476791382,
      "learning_rate": 9.999999980930928e-05,
      "loss": 2.9473,
      "step": 561
    },
    {
      "epoch": 0.015096164177500806,
      "grad_norm": 1.3552346229553223,
      "learning_rate": 9.999999980862274e-05,
      "loss": 2.9918,
      "step": 562
    },
    {
      "epoch": 0.015123025679596003,
      "grad_norm": 1.4699270725250244,
      "learning_rate": 9.999999980793495e-05,
      "loss": 3.0069,
      "step": 563
    },
    {
      "epoch": 0.0151498871816912,
      "grad_norm": 1.4307541847229004,
      "learning_rate": 9.999999980724592e-05,
      "loss": 3.0416,
      "step": 564
    },
    {
      "epoch": 0.015176748683786398,
      "grad_norm": 1.4240820407867432,
      "learning_rate": 9.999999980655565e-05,
      "loss": 3.0761,
      "step": 565
    },
    {
      "epoch": 0.015203610185881594,
      "grad_norm": 1.514182448387146,
      "learning_rate": 9.999999980586417e-05,
      "loss": 3.3871,
      "step": 566
    },
    {
      "epoch": 0.015230471687976792,
      "grad_norm": 1.4674021005630493,
      "learning_rate": 9.999999980517145e-05,
      "loss": 3.0696,
      "step": 567
    },
    {
      "epoch": 0.015257333190071989,
      "grad_norm": 1.5257642269134521,
      "learning_rate": 9.999999980447749e-05,
      "loss": 3.1464,
      "step": 568
    },
    {
      "epoch": 0.015284194692167186,
      "grad_norm": 1.277586579322815,
      "learning_rate": 9.99999998037823e-05,
      "loss": 2.8407,
      "step": 569
    },
    {
      "epoch": 0.015311056194262384,
      "grad_norm": 1.529081106185913,
      "learning_rate": 9.999999980308588e-05,
      "loss": 3.0011,
      "step": 570
    },
    {
      "epoch": 0.01533791769635758,
      "grad_norm": 1.4303085803985596,
      "learning_rate": 9.999999980238821e-05,
      "loss": 2.9766,
      "step": 571
    },
    {
      "epoch": 0.015364779198452777,
      "grad_norm": 1.4762227535247803,
      "learning_rate": 9.999999980168933e-05,
      "loss": 2.8978,
      "step": 572
    },
    {
      "epoch": 0.015391640700547975,
      "grad_norm": 1.5402470827102661,
      "learning_rate": 9.99999998009892e-05,
      "loss": 3.0596,
      "step": 573
    },
    {
      "epoch": 0.015418502202643172,
      "grad_norm": 1.673646092414856,
      "learning_rate": 9.999999980028785e-05,
      "loss": 3.0161,
      "step": 574
    },
    {
      "epoch": 0.015445363704738369,
      "grad_norm": 1.5393816232681274,
      "learning_rate": 9.999999979958525e-05,
      "loss": 2.9339,
      "step": 575
    },
    {
      "epoch": 0.015472225206833567,
      "grad_norm": 1.6110318899154663,
      "learning_rate": 9.999999979888141e-05,
      "loss": 3.0425,
      "step": 576
    },
    {
      "epoch": 0.015499086708928763,
      "grad_norm": 1.70695161819458,
      "learning_rate": 9.999999979817635e-05,
      "loss": 3.2202,
      "step": 577
    },
    {
      "epoch": 0.01552594821102396,
      "grad_norm": 1.6060736179351807,
      "learning_rate": 9.999999979747006e-05,
      "loss": 3.1667,
      "step": 578
    },
    {
      "epoch": 0.015552809713119158,
      "grad_norm": 1.648018717765808,
      "learning_rate": 9.999999979676254e-05,
      "loss": 3.1444,
      "step": 579
    },
    {
      "epoch": 0.015579671215214355,
      "grad_norm": 1.6265192031860352,
      "learning_rate": 9.999999979605378e-05,
      "loss": 3.0009,
      "step": 580
    },
    {
      "epoch": 0.015606532717309551,
      "grad_norm": 1.5863219499588013,
      "learning_rate": 9.999999979534378e-05,
      "loss": 3.4085,
      "step": 581
    },
    {
      "epoch": 0.01563339421940475,
      "grad_norm": 1.7096445560455322,
      "learning_rate": 9.999999979463256e-05,
      "loss": 3.1331,
      "step": 582
    },
    {
      "epoch": 0.015660255721499948,
      "grad_norm": 1.4504519701004028,
      "learning_rate": 9.99999997939201e-05,
      "loss": 3.0117,
      "step": 583
    },
    {
      "epoch": 0.015687117223595143,
      "grad_norm": 1.5081751346588135,
      "learning_rate": 9.99999997932064e-05,
      "loss": 3.2206,
      "step": 584
    },
    {
      "epoch": 0.01571397872569034,
      "grad_norm": 1.7219492197036743,
      "learning_rate": 9.999999979249146e-05,
      "loss": 3.0363,
      "step": 585
    },
    {
      "epoch": 0.01574084022778554,
      "grad_norm": 1.6603717803955078,
      "learning_rate": 9.99999997917753e-05,
      "loss": 3.2665,
      "step": 586
    },
    {
      "epoch": 0.015767701729880734,
      "grad_norm": 1.6417179107666016,
      "learning_rate": 9.99999997910579e-05,
      "loss": 2.9291,
      "step": 587
    },
    {
      "epoch": 0.015794563231975933,
      "grad_norm": 1.5284661054611206,
      "learning_rate": 9.999999979033927e-05,
      "loss": 3.2111,
      "step": 588
    },
    {
      "epoch": 0.01582142473407113,
      "grad_norm": 1.5470607280731201,
      "learning_rate": 9.999999978961941e-05,
      "loss": 2.9447,
      "step": 589
    },
    {
      "epoch": 0.015848286236166326,
      "grad_norm": 1.7079750299453735,
      "learning_rate": 9.99999997888983e-05,
      "loss": 3.3994,
      "step": 590
    },
    {
      "epoch": 0.015875147738261524,
      "grad_norm": 1.6720783710479736,
      "learning_rate": 9.999999978817598e-05,
      "loss": 3.1815,
      "step": 591
    },
    {
      "epoch": 0.015902009240356722,
      "grad_norm": 1.5435919761657715,
      "learning_rate": 9.999999978745241e-05,
      "loss": 2.8047,
      "step": 592
    },
    {
      "epoch": 0.015928870742451917,
      "grad_norm": 1.6838616132736206,
      "learning_rate": 9.999999978672761e-05,
      "loss": 3.1796,
      "step": 593
    },
    {
      "epoch": 0.015955732244547115,
      "grad_norm": 1.8546950817108154,
      "learning_rate": 9.999999978600159e-05,
      "loss": 3.3696,
      "step": 594
    },
    {
      "epoch": 0.015982593746642314,
      "grad_norm": 1.722764015197754,
      "learning_rate": 9.999999978527432e-05,
      "loss": 3.1535,
      "step": 595
    },
    {
      "epoch": 0.01600945524873751,
      "grad_norm": 1.6362978219985962,
      "learning_rate": 9.999999978454582e-05,
      "loss": 3.3235,
      "step": 596
    },
    {
      "epoch": 0.016036316750832707,
      "grad_norm": 1.7171472311019897,
      "learning_rate": 9.99999997838161e-05,
      "loss": 3.2714,
      "step": 597
    },
    {
      "epoch": 0.016063178252927905,
      "grad_norm": 1.8612757921218872,
      "learning_rate": 9.999999978308511e-05,
      "loss": 3.3938,
      "step": 598
    },
    {
      "epoch": 0.0160900397550231,
      "grad_norm": 2.0596330165863037,
      "learning_rate": 9.999999978235291e-05,
      "loss": 3.3465,
      "step": 599
    },
    {
      "epoch": 0.0161169012571183,
      "grad_norm": 1.7264341115951538,
      "learning_rate": 9.999999978161947e-05,
      "loss": 3.2179,
      "step": 600
    },
    {
      "epoch": 0.016143762759213497,
      "grad_norm": 1.4845199584960938,
      "learning_rate": 9.99999997808848e-05,
      "loss": 3.2596,
      "step": 601
    },
    {
      "epoch": 0.01617062426130869,
      "grad_norm": 1.5728617906570435,
      "learning_rate": 9.999999978014892e-05,
      "loss": 3.0341,
      "step": 602
    },
    {
      "epoch": 0.01619748576340389,
      "grad_norm": 1.4966508150100708,
      "learning_rate": 9.999999977941177e-05,
      "loss": 3.1194,
      "step": 603
    },
    {
      "epoch": 0.016224347265499088,
      "grad_norm": 1.623679518699646,
      "learning_rate": 9.99999997786734e-05,
      "loss": 2.9306,
      "step": 604
    },
    {
      "epoch": 0.016251208767594283,
      "grad_norm": 1.5571153163909912,
      "learning_rate": 9.99999997779338e-05,
      "loss": 3.1209,
      "step": 605
    },
    {
      "epoch": 0.01627807026968948,
      "grad_norm": 1.5426580905914307,
      "learning_rate": 9.999999977719296e-05,
      "loss": 3.2077,
      "step": 606
    },
    {
      "epoch": 0.01630493177178468,
      "grad_norm": 1.5416814088821411,
      "learning_rate": 9.999999977645089e-05,
      "loss": 3.2349,
      "step": 607
    },
    {
      "epoch": 0.016331793273879874,
      "grad_norm": 1.580945611000061,
      "learning_rate": 9.999999977570758e-05,
      "loss": 2.8518,
      "step": 608
    },
    {
      "epoch": 0.016358654775975073,
      "grad_norm": 1.3710176944732666,
      "learning_rate": 9.999999977496304e-05,
      "loss": 2.9742,
      "step": 609
    },
    {
      "epoch": 0.01638551627807027,
      "grad_norm": 1.4292913675308228,
      "learning_rate": 9.999999977421727e-05,
      "loss": 2.9606,
      "step": 610
    },
    {
      "epoch": 0.016412377780165466,
      "grad_norm": 1.366773247718811,
      "learning_rate": 9.999999977347026e-05,
      "loss": 2.9488,
      "step": 611
    },
    {
      "epoch": 0.016439239282260664,
      "grad_norm": 1.4914782047271729,
      "learning_rate": 9.999999977272204e-05,
      "loss": 3.087,
      "step": 612
    },
    {
      "epoch": 0.016466100784355862,
      "grad_norm": 1.4016969203948975,
      "learning_rate": 9.999999977197256e-05,
      "loss": 2.9716,
      "step": 613
    },
    {
      "epoch": 0.016492962286451057,
      "grad_norm": 1.3989452123641968,
      "learning_rate": 9.999999977122184e-05,
      "loss": 3.0153,
      "step": 614
    },
    {
      "epoch": 0.016519823788546256,
      "grad_norm": 1.3405635356903076,
      "learning_rate": 9.99999997704699e-05,
      "loss": 2.6192,
      "step": 615
    },
    {
      "epoch": 0.016546685290641454,
      "grad_norm": 1.4095908403396606,
      "learning_rate": 9.999999976971672e-05,
      "loss": 3.0062,
      "step": 616
    },
    {
      "epoch": 0.01657354679273665,
      "grad_norm": 1.4745733737945557,
      "learning_rate": 9.999999976896232e-05,
      "loss": 3.125,
      "step": 617
    },
    {
      "epoch": 0.016600408294831847,
      "grad_norm": 1.4414845705032349,
      "learning_rate": 9.999999976820667e-05,
      "loss": 2.9896,
      "step": 618
    },
    {
      "epoch": 0.016627269796927045,
      "grad_norm": 1.6898787021636963,
      "learning_rate": 9.99999997674498e-05,
      "loss": 3.1397,
      "step": 619
    },
    {
      "epoch": 0.01665413129902224,
      "grad_norm": 1.5939667224884033,
      "learning_rate": 9.99999997666917e-05,
      "loss": 3.0917,
      "step": 620
    },
    {
      "epoch": 0.01668099280111744,
      "grad_norm": 1.5486644506454468,
      "learning_rate": 9.999999976593236e-05,
      "loss": 3.0241,
      "step": 621
    },
    {
      "epoch": 0.016707854303212637,
      "grad_norm": 1.4304726123809814,
      "learning_rate": 9.999999976517178e-05,
      "loss": 3.0435,
      "step": 622
    },
    {
      "epoch": 0.01673471580530783,
      "grad_norm": 1.4396814107894897,
      "learning_rate": 9.999999976440997e-05,
      "loss": 3.3262,
      "step": 623
    },
    {
      "epoch": 0.01676157730740303,
      "grad_norm": 1.3595683574676514,
      "learning_rate": 9.999999976364691e-05,
      "loss": 2.9957,
      "step": 624
    },
    {
      "epoch": 0.016788438809498228,
      "grad_norm": 1.4666507244110107,
      "learning_rate": 9.999999976288263e-05,
      "loss": 3.1825,
      "step": 625
    },
    {
      "epoch": 0.016815300311593423,
      "grad_norm": 1.5233851671218872,
      "learning_rate": 9.999999976211713e-05,
      "loss": 3.266,
      "step": 626
    },
    {
      "epoch": 0.01684216181368862,
      "grad_norm": 1.5205273628234863,
      "learning_rate": 9.999999976135038e-05,
      "loss": 3.114,
      "step": 627
    },
    {
      "epoch": 0.01686902331578382,
      "grad_norm": 1.4155679941177368,
      "learning_rate": 9.99999997605824e-05,
      "loss": 3.0824,
      "step": 628
    },
    {
      "epoch": 0.016895884817879014,
      "grad_norm": 1.4930715560913086,
      "learning_rate": 9.999999975981319e-05,
      "loss": 3.0347,
      "step": 629
    },
    {
      "epoch": 0.016922746319974213,
      "grad_norm": 1.642045021057129,
      "learning_rate": 9.999999975904275e-05,
      "loss": 3.339,
      "step": 630
    },
    {
      "epoch": 0.01694960782206941,
      "grad_norm": 1.6314470767974854,
      "learning_rate": 9.999999975827107e-05,
      "loss": 3.4392,
      "step": 631
    },
    {
      "epoch": 0.016976469324164606,
      "grad_norm": 1.479723572731018,
      "learning_rate": 9.999999975749816e-05,
      "loss": 3.4092,
      "step": 632
    },
    {
      "epoch": 0.017003330826259804,
      "grad_norm": 1.4937212467193604,
      "learning_rate": 9.999999975672401e-05,
      "loss": 3.1713,
      "step": 633
    },
    {
      "epoch": 0.017030192328355002,
      "grad_norm": 1.5711121559143066,
      "learning_rate": 9.999999975594863e-05,
      "loss": 3.2504,
      "step": 634
    },
    {
      "epoch": 0.017057053830450197,
      "grad_norm": 1.7083232402801514,
      "learning_rate": 9.9999999755172e-05,
      "loss": 3.5004,
      "step": 635
    },
    {
      "epoch": 0.017083915332545396,
      "grad_norm": 1.6591018438339233,
      "learning_rate": 9.999999975439416e-05,
      "loss": 3.2098,
      "step": 636
    },
    {
      "epoch": 0.017110776834640594,
      "grad_norm": 1.6006439924240112,
      "learning_rate": 9.999999975361507e-05,
      "loss": 3.1391,
      "step": 637
    },
    {
      "epoch": 0.01713763833673579,
      "grad_norm": 1.6198042631149292,
      "learning_rate": 9.999999975283476e-05,
      "loss": 3.303,
      "step": 638
    },
    {
      "epoch": 0.017164499838830987,
      "grad_norm": 1.6083146333694458,
      "learning_rate": 9.999999975205321e-05,
      "loss": 3.3684,
      "step": 639
    },
    {
      "epoch": 0.017191361340926185,
      "grad_norm": 1.8380709886550903,
      "learning_rate": 9.999999975127043e-05,
      "loss": 3.5626,
      "step": 640
    },
    {
      "epoch": 0.01721822284302138,
      "grad_norm": 1.5533370971679688,
      "learning_rate": 9.999999975048642e-05,
      "loss": 3.2367,
      "step": 641
    },
    {
      "epoch": 0.01724508434511658,
      "grad_norm": 1.57891845703125,
      "learning_rate": 9.999999974970116e-05,
      "loss": 3.0813,
      "step": 642
    },
    {
      "epoch": 0.017271945847211777,
      "grad_norm": 1.556593418121338,
      "learning_rate": 9.999999974891468e-05,
      "loss": 3.3402,
      "step": 643
    },
    {
      "epoch": 0.01729880734930697,
      "grad_norm": 1.5907540321350098,
      "learning_rate": 9.999999974812697e-05,
      "loss": 3.3009,
      "step": 644
    },
    {
      "epoch": 0.01732566885140217,
      "grad_norm": 1.5855154991149902,
      "learning_rate": 9.9999999747338e-05,
      "loss": 3.4019,
      "step": 645
    },
    {
      "epoch": 0.017352530353497368,
      "grad_norm": 1.5891433954238892,
      "learning_rate": 9.999999974654781e-05,
      "loss": 3.3439,
      "step": 646
    },
    {
      "epoch": 0.017379391855592563,
      "grad_norm": 1.7367231845855713,
      "learning_rate": 9.99999997457564e-05,
      "loss": 3.4537,
      "step": 647
    },
    {
      "epoch": 0.01740625335768776,
      "grad_norm": 1.702995777130127,
      "learning_rate": 9.999999974496375e-05,
      "loss": 3.3761,
      "step": 648
    },
    {
      "epoch": 0.01743311485978296,
      "grad_norm": 1.714220643043518,
      "learning_rate": 9.999999974416986e-05,
      "loss": 3.4498,
      "step": 649
    },
    {
      "epoch": 0.017459976361878154,
      "grad_norm": 1.9178133010864258,
      "learning_rate": 9.999999974337474e-05,
      "loss": 3.5324,
      "step": 650
    },
    {
      "epoch": 0.017486837863973353,
      "grad_norm": 1.398617148399353,
      "learning_rate": 9.999999974257838e-05,
      "loss": 3.1473,
      "step": 651
    },
    {
      "epoch": 0.01751369936606855,
      "grad_norm": 1.5064873695373535,
      "learning_rate": 9.99999997417808e-05,
      "loss": 3.0855,
      "step": 652
    },
    {
      "epoch": 0.01754056086816375,
      "grad_norm": 1.5795984268188477,
      "learning_rate": 9.999999974098198e-05,
      "loss": 2.8589,
      "step": 653
    },
    {
      "epoch": 0.017567422370258944,
      "grad_norm": 1.505240797996521,
      "learning_rate": 9.999999974018193e-05,
      "loss": 2.99,
      "step": 654
    },
    {
      "epoch": 0.017594283872354143,
      "grad_norm": 1.434888243675232,
      "learning_rate": 9.999999973938064e-05,
      "loss": 2.9245,
      "step": 655
    },
    {
      "epoch": 0.01762114537444934,
      "grad_norm": 1.3589751720428467,
      "learning_rate": 9.999999973857811e-05,
      "loss": 2.6756,
      "step": 656
    },
    {
      "epoch": 0.017648006876544536,
      "grad_norm": 1.432234764099121,
      "learning_rate": 9.999999973777435e-05,
      "loss": 3.0035,
      "step": 657
    },
    {
      "epoch": 0.017674868378639734,
      "grad_norm": 1.450545072555542,
      "learning_rate": 9.999999973696936e-05,
      "loss": 3.1422,
      "step": 658
    },
    {
      "epoch": 0.017701729880734932,
      "grad_norm": 1.426239013671875,
      "learning_rate": 9.999999973616315e-05,
      "loss": 3.1334,
      "step": 659
    },
    {
      "epoch": 0.017728591382830127,
      "grad_norm": 1.541068196296692,
      "learning_rate": 9.999999973535567e-05,
      "loss": 3.0724,
      "step": 660
    },
    {
      "epoch": 0.017755452884925325,
      "grad_norm": 1.3603521585464478,
      "learning_rate": 9.9999999734547e-05,
      "loss": 2.9346,
      "step": 661
    },
    {
      "epoch": 0.017782314387020524,
      "grad_norm": 1.3395049571990967,
      "learning_rate": 9.999999973373705e-05,
      "loss": 3.0172,
      "step": 662
    },
    {
      "epoch": 0.01780917588911572,
      "grad_norm": 1.4169708490371704,
      "learning_rate": 9.999999973292591e-05,
      "loss": 3.08,
      "step": 663
    },
    {
      "epoch": 0.017836037391210917,
      "grad_norm": 1.4493162631988525,
      "learning_rate": 9.999999973211352e-05,
      "loss": 3.0241,
      "step": 664
    },
    {
      "epoch": 0.017862898893306115,
      "grad_norm": 1.4321480989456177,
      "learning_rate": 9.99999997312999e-05,
      "loss": 2.7355,
      "step": 665
    },
    {
      "epoch": 0.01788976039540131,
      "grad_norm": 1.4396843910217285,
      "learning_rate": 9.999999973048504e-05,
      "loss": 3.1973,
      "step": 666
    },
    {
      "epoch": 0.01791662189749651,
      "grad_norm": 1.5023831129074097,
      "learning_rate": 9.999999972966894e-05,
      "loss": 2.8972,
      "step": 667
    },
    {
      "epoch": 0.017943483399591707,
      "grad_norm": 1.574151635169983,
      "learning_rate": 9.999999972885161e-05,
      "loss": 3.25,
      "step": 668
    },
    {
      "epoch": 0.0179703449016869,
      "grad_norm": 1.382725477218628,
      "learning_rate": 9.999999972803304e-05,
      "loss": 2.8051,
      "step": 669
    },
    {
      "epoch": 0.0179972064037821,
      "grad_norm": 1.4401991367340088,
      "learning_rate": 9.999999972721326e-05,
      "loss": 3.0833,
      "step": 670
    },
    {
      "epoch": 0.018024067905877298,
      "grad_norm": 1.5565216541290283,
      "learning_rate": 9.999999972639222e-05,
      "loss": 2.9725,
      "step": 671
    },
    {
      "epoch": 0.018050929407972493,
      "grad_norm": 1.396734595298767,
      "learning_rate": 9.999999972556998e-05,
      "loss": 2.8631,
      "step": 672
    },
    {
      "epoch": 0.01807779091006769,
      "grad_norm": 1.5907701253890991,
      "learning_rate": 9.999999972474647e-05,
      "loss": 3.4372,
      "step": 673
    },
    {
      "epoch": 0.01810465241216289,
      "grad_norm": 1.479804515838623,
      "learning_rate": 9.999999972392173e-05,
      "loss": 2.9795,
      "step": 674
    },
    {
      "epoch": 0.018131513914258084,
      "grad_norm": 1.3441416025161743,
      "learning_rate": 9.999999972309577e-05,
      "loss": 2.8914,
      "step": 675
    },
    {
      "epoch": 0.018158375416353283,
      "grad_norm": 1.435198426246643,
      "learning_rate": 9.99999997222686e-05,
      "loss": 3.1491,
      "step": 676
    },
    {
      "epoch": 0.01818523691844848,
      "grad_norm": 1.602691411972046,
      "learning_rate": 9.999999972144016e-05,
      "loss": 3.148,
      "step": 677
    },
    {
      "epoch": 0.018212098420543676,
      "grad_norm": 1.5204836130142212,
      "learning_rate": 9.99999997206105e-05,
      "loss": 3.2252,
      "step": 678
    },
    {
      "epoch": 0.018238959922638874,
      "grad_norm": 1.5765471458435059,
      "learning_rate": 9.999999971977959e-05,
      "loss": 2.9243,
      "step": 679
    },
    {
      "epoch": 0.018265821424734072,
      "grad_norm": 1.6346088647842407,
      "learning_rate": 9.999999971894745e-05,
      "loss": 3.3296,
      "step": 680
    },
    {
      "epoch": 0.018292682926829267,
      "grad_norm": 1.4339308738708496,
      "learning_rate": 9.99999997181141e-05,
      "loss": 3.091,
      "step": 681
    },
    {
      "epoch": 0.018319544428924465,
      "grad_norm": 1.6199924945831299,
      "learning_rate": 9.999999971727949e-05,
      "loss": 3.2955,
      "step": 682
    },
    {
      "epoch": 0.018346405931019664,
      "grad_norm": 1.4859867095947266,
      "learning_rate": 9.999999971644366e-05,
      "loss": 3.166,
      "step": 683
    },
    {
      "epoch": 0.01837326743311486,
      "grad_norm": 1.512736439704895,
      "learning_rate": 9.999999971560659e-05,
      "loss": 3.2106,
      "step": 684
    },
    {
      "epoch": 0.018400128935210057,
      "grad_norm": 1.5425671339035034,
      "learning_rate": 9.99999997147683e-05,
      "loss": 3.2207,
      "step": 685
    },
    {
      "epoch": 0.018426990437305255,
      "grad_norm": 1.5040771961212158,
      "learning_rate": 9.999999971392876e-05,
      "loss": 2.9827,
      "step": 686
    },
    {
      "epoch": 0.01845385193940045,
      "grad_norm": 1.5521918535232544,
      "learning_rate": 9.9999999713088e-05,
      "loss": 3.3105,
      "step": 687
    },
    {
      "epoch": 0.01848071344149565,
      "grad_norm": 1.67476487159729,
      "learning_rate": 9.9999999712246e-05,
      "loss": 2.9965,
      "step": 688
    },
    {
      "epoch": 0.018507574943590847,
      "grad_norm": 1.6089451313018799,
      "learning_rate": 9.999999971140276e-05,
      "loss": 3.2392,
      "step": 689
    },
    {
      "epoch": 0.01853443644568604,
      "grad_norm": 1.599771499633789,
      "learning_rate": 9.999999971055829e-05,
      "loss": 2.8795,
      "step": 690
    },
    {
      "epoch": 0.01856129794778124,
      "grad_norm": 1.6681711673736572,
      "learning_rate": 9.999999970971258e-05,
      "loss": 3.567,
      "step": 691
    },
    {
      "epoch": 0.018588159449876438,
      "grad_norm": 1.7651498317718506,
      "learning_rate": 9.999999970886565e-05,
      "loss": 3.2726,
      "step": 692
    },
    {
      "epoch": 0.018615020951971633,
      "grad_norm": 1.6382040977478027,
      "learning_rate": 9.999999970801748e-05,
      "loss": 3.2873,
      "step": 693
    },
    {
      "epoch": 0.01864188245406683,
      "grad_norm": 1.6960110664367676,
      "learning_rate": 9.999999970716808e-05,
      "loss": 3.1896,
      "step": 694
    },
    {
      "epoch": 0.01866874395616203,
      "grad_norm": 1.8641104698181152,
      "learning_rate": 9.999999970631743e-05,
      "loss": 3.4889,
      "step": 695
    },
    {
      "epoch": 0.018695605458257224,
      "grad_norm": 1.672577142715454,
      "learning_rate": 9.999999970546556e-05,
      "loss": 3.5504,
      "step": 696
    },
    {
      "epoch": 0.018722466960352423,
      "grad_norm": 1.8528612852096558,
      "learning_rate": 9.999999970461247e-05,
      "loss": 3.5622,
      "step": 697
    },
    {
      "epoch": 0.01874932846244762,
      "grad_norm": 1.8413525819778442,
      "learning_rate": 9.999999970375813e-05,
      "loss": 3.7652,
      "step": 698
    },
    {
      "epoch": 0.018776189964542816,
      "grad_norm": 1.9651743173599243,
      "learning_rate": 9.999999970290256e-05,
      "loss": 3.6158,
      "step": 699
    },
    {
      "epoch": 0.018803051466638014,
      "grad_norm": 1.8817933797836304,
      "learning_rate": 9.999999970204576e-05,
      "loss": 3.3605,
      "step": 700
    },
    {
      "epoch": 0.018829912968733212,
      "grad_norm": 1.4275424480438232,
      "learning_rate": 9.999999970118772e-05,
      "loss": 3.0604,
      "step": 701
    },
    {
      "epoch": 0.018856774470828407,
      "grad_norm": 1.4607408046722412,
      "learning_rate": 9.999999970032844e-05,
      "loss": 3.1583,
      "step": 702
    },
    {
      "epoch": 0.018883635972923606,
      "grad_norm": 1.4945420026779175,
      "learning_rate": 9.999999969946794e-05,
      "loss": 3.4831,
      "step": 703
    },
    {
      "epoch": 0.018910497475018804,
      "grad_norm": 1.649383306503296,
      "learning_rate": 9.99999996986062e-05,
      "loss": 3.1655,
      "step": 704
    },
    {
      "epoch": 0.018937358977114,
      "grad_norm": 1.4347844123840332,
      "learning_rate": 9.999999969774322e-05,
      "loss": 3.0471,
      "step": 705
    },
    {
      "epoch": 0.018964220479209197,
      "grad_norm": 1.4316843748092651,
      "learning_rate": 9.9999999696879e-05,
      "loss": 2.8433,
      "step": 706
    },
    {
      "epoch": 0.018991081981304395,
      "grad_norm": 1.3896877765655518,
      "learning_rate": 9.999999969601357e-05,
      "loss": 2.913,
      "step": 707
    },
    {
      "epoch": 0.01901794348339959,
      "grad_norm": 1.5291829109191895,
      "learning_rate": 9.99999996951469e-05,
      "loss": 3.0895,
      "step": 708
    },
    {
      "epoch": 0.01904480498549479,
      "grad_norm": 1.4520114660263062,
      "learning_rate": 9.999999969427898e-05,
      "loss": 2.9309,
      "step": 709
    },
    {
      "epoch": 0.019071666487589987,
      "grad_norm": 1.4198780059814453,
      "learning_rate": 9.999999969340986e-05,
      "loss": 3.0022,
      "step": 710
    },
    {
      "epoch": 0.01909852798968518,
      "grad_norm": 1.501908779144287,
      "learning_rate": 9.999999969253947e-05,
      "loss": 3.2037,
      "step": 711
    },
    {
      "epoch": 0.01912538949178038,
      "grad_norm": 1.5056076049804688,
      "learning_rate": 9.999999969166786e-05,
      "loss": 3.1124,
      "step": 712
    },
    {
      "epoch": 0.019152250993875578,
      "grad_norm": 1.327278733253479,
      "learning_rate": 9.9999999690795e-05,
      "loss": 2.8448,
      "step": 713
    },
    {
      "epoch": 0.019179112495970773,
      "grad_norm": 1.3927946090698242,
      "learning_rate": 9.999999968992093e-05,
      "loss": 3.0786,
      "step": 714
    },
    {
      "epoch": 0.01920597399806597,
      "grad_norm": 1.6739174127578735,
      "learning_rate": 9.999999968904563e-05,
      "loss": 3.3298,
      "step": 715
    },
    {
      "epoch": 0.01923283550016117,
      "grad_norm": 1.6534708738327026,
      "learning_rate": 9.99999996881691e-05,
      "loss": 3.4348,
      "step": 716
    },
    {
      "epoch": 0.019259697002256364,
      "grad_norm": 1.492730736732483,
      "learning_rate": 9.99999996872913e-05,
      "loss": 3.0935,
      "step": 717
    },
    {
      "epoch": 0.019286558504351563,
      "grad_norm": 1.4301120042800903,
      "learning_rate": 9.99999996864123e-05,
      "loss": 2.872,
      "step": 718
    },
    {
      "epoch": 0.01931342000644676,
      "grad_norm": 1.351636290550232,
      "learning_rate": 9.999999968553204e-05,
      "loss": 2.7152,
      "step": 719
    },
    {
      "epoch": 0.01934028150854196,
      "grad_norm": 1.4593547582626343,
      "learning_rate": 9.999999968465057e-05,
      "loss": 3.1069,
      "step": 720
    },
    {
      "epoch": 0.019367143010637154,
      "grad_norm": 1.588448405265808,
      "learning_rate": 9.999999968376786e-05,
      "loss": 3.278,
      "step": 721
    },
    {
      "epoch": 0.019394004512732353,
      "grad_norm": 1.5615174770355225,
      "learning_rate": 9.999999968288391e-05,
      "loss": 3.0554,
      "step": 722
    },
    {
      "epoch": 0.01942086601482755,
      "grad_norm": 1.5523321628570557,
      "learning_rate": 9.999999968199872e-05,
      "loss": 3.3334,
      "step": 723
    },
    {
      "epoch": 0.019447727516922746,
      "grad_norm": 1.4447002410888672,
      "learning_rate": 9.999999968111231e-05,
      "loss": 3.0282,
      "step": 724
    },
    {
      "epoch": 0.019474589019017944,
      "grad_norm": 1.560368299484253,
      "learning_rate": 9.999999968022467e-05,
      "loss": 2.9756,
      "step": 725
    },
    {
      "epoch": 0.019501450521113142,
      "grad_norm": 1.681187629699707,
      "learning_rate": 9.999999967933577e-05,
      "loss": 3.496,
      "step": 726
    },
    {
      "epoch": 0.019528312023208337,
      "grad_norm": 1.6926003694534302,
      "learning_rate": 9.999999967844567e-05,
      "loss": 3.3941,
      "step": 727
    },
    {
      "epoch": 0.019555173525303535,
      "grad_norm": 1.4728666543960571,
      "learning_rate": 9.999999967755431e-05,
      "loss": 3.0974,
      "step": 728
    },
    {
      "epoch": 0.019582035027398734,
      "grad_norm": 1.5718094110488892,
      "learning_rate": 9.999999967666174e-05,
      "loss": 3.3441,
      "step": 729
    },
    {
      "epoch": 0.01960889652949393,
      "grad_norm": 1.541793704032898,
      "learning_rate": 9.999999967576792e-05,
      "loss": 3.2327,
      "step": 730
    },
    {
      "epoch": 0.019635758031589127,
      "grad_norm": 1.555477261543274,
      "learning_rate": 9.999999967487287e-05,
      "loss": 3.1065,
      "step": 731
    },
    {
      "epoch": 0.019662619533684325,
      "grad_norm": 1.6656402349472046,
      "learning_rate": 9.999999967397659e-05,
      "loss": 3.0119,
      "step": 732
    },
    {
      "epoch": 0.01968948103577952,
      "grad_norm": 1.548744559288025,
      "learning_rate": 9.999999967307907e-05,
      "loss": 3.1914,
      "step": 733
    },
    {
      "epoch": 0.01971634253787472,
      "grad_norm": 1.5926529169082642,
      "learning_rate": 9.999999967218032e-05,
      "loss": 3.1765,
      "step": 734
    },
    {
      "epoch": 0.019743204039969917,
      "grad_norm": 1.5121639966964722,
      "learning_rate": 9.999999967128033e-05,
      "loss": 2.8757,
      "step": 735
    },
    {
      "epoch": 0.01977006554206511,
      "grad_norm": 1.5775986909866333,
      "learning_rate": 9.999999967037911e-05,
      "loss": 3.1299,
      "step": 736
    },
    {
      "epoch": 0.01979692704416031,
      "grad_norm": 1.5838713645935059,
      "learning_rate": 9.999999966947666e-05,
      "loss": 3.2553,
      "step": 737
    },
    {
      "epoch": 0.019823788546255508,
      "grad_norm": 1.5859333276748657,
      "learning_rate": 9.999999966857298e-05,
      "loss": 3.4363,
      "step": 738
    },
    {
      "epoch": 0.019850650048350703,
      "grad_norm": 1.6026620864868164,
      "learning_rate": 9.999999966766805e-05,
      "loss": 3.2706,
      "step": 739
    },
    {
      "epoch": 0.0198775115504459,
      "grad_norm": 1.58791184425354,
      "learning_rate": 9.99999996667619e-05,
      "loss": 3.2601,
      "step": 740
    },
    {
      "epoch": 0.0199043730525411,
      "grad_norm": 1.5763542652130127,
      "learning_rate": 9.999999966585452e-05,
      "loss": 3.1218,
      "step": 741
    },
    {
      "epoch": 0.019931234554636294,
      "grad_norm": 1.6854625940322876,
      "learning_rate": 9.999999966494589e-05,
      "loss": 3.4081,
      "step": 742
    },
    {
      "epoch": 0.019958096056731493,
      "grad_norm": 1.802384614944458,
      "learning_rate": 9.999999966403604e-05,
      "loss": 3.1813,
      "step": 743
    },
    {
      "epoch": 0.01998495755882669,
      "grad_norm": 1.7802817821502686,
      "learning_rate": 9.999999966312495e-05,
      "loss": 3.4648,
      "step": 744
    },
    {
      "epoch": 0.020011819060921886,
      "grad_norm": 1.629107117652893,
      "learning_rate": 9.999999966221263e-05,
      "loss": 3.173,
      "step": 745
    },
    {
      "epoch": 0.020038680563017084,
      "grad_norm": 1.7447935342788696,
      "learning_rate": 9.999999966129907e-05,
      "loss": 2.9959,
      "step": 746
    },
    {
      "epoch": 0.020065542065112282,
      "grad_norm": 1.965346097946167,
      "learning_rate": 9.999999966038429e-05,
      "loss": 3.7127,
      "step": 747
    },
    {
      "epoch": 0.020092403567207477,
      "grad_norm": 1.9311946630477905,
      "learning_rate": 9.999999965946826e-05,
      "loss": 3.7032,
      "step": 748
    },
    {
      "epoch": 0.020119265069302675,
      "grad_norm": 2.012167453765869,
      "learning_rate": 9.999999965855101e-05,
      "loss": 3.0979,
      "step": 749
    },
    {
      "epoch": 0.020146126571397874,
      "grad_norm": 1.8708699941635132,
      "learning_rate": 9.999999965763251e-05,
      "loss": 3.4835,
      "step": 750
    },
    {
      "epoch": 0.02017298807349307,
      "grad_norm": 1.5525647401809692,
      "learning_rate": 9.999999965671278e-05,
      "loss": 3.4838,
      "step": 751
    },
    {
      "epoch": 0.020199849575588267,
      "grad_norm": 1.684830904006958,
      "learning_rate": 9.999999965579184e-05,
      "loss": 3.4157,
      "step": 752
    },
    {
      "epoch": 0.020226711077683465,
      "grad_norm": 1.487632155418396,
      "learning_rate": 9.999999965486964e-05,
      "loss": 3.3135,
      "step": 753
    },
    {
      "epoch": 0.02025357257977866,
      "grad_norm": 1.4215151071548462,
      "learning_rate": 9.999999965394623e-05,
      "loss": 2.806,
      "step": 754
    },
    {
      "epoch": 0.02028043408187386,
      "grad_norm": 1.4384040832519531,
      "learning_rate": 9.999999965302156e-05,
      "loss": 2.725,
      "step": 755
    },
    {
      "epoch": 0.020307295583969057,
      "grad_norm": 1.519808292388916,
      "learning_rate": 9.999999965209566e-05,
      "loss": 3.0673,
      "step": 756
    },
    {
      "epoch": 0.02033415708606425,
      "grad_norm": 1.5362298488616943,
      "learning_rate": 9.999999965116853e-05,
      "loss": 2.9519,
      "step": 757
    },
    {
      "epoch": 0.02036101858815945,
      "grad_norm": 1.3809818029403687,
      "learning_rate": 9.999999965024018e-05,
      "loss": 2.9788,
      "step": 758
    },
    {
      "epoch": 0.020387880090254648,
      "grad_norm": 1.4069761037826538,
      "learning_rate": 9.999999964931057e-05,
      "loss": 3.0046,
      "step": 759
    },
    {
      "epoch": 0.020414741592349843,
      "grad_norm": 1.5650560855865479,
      "learning_rate": 9.999999964837976e-05,
      "loss": 3.0619,
      "step": 760
    },
    {
      "epoch": 0.02044160309444504,
      "grad_norm": 1.410232663154602,
      "learning_rate": 9.999999964744768e-05,
      "loss": 2.9326,
      "step": 761
    },
    {
      "epoch": 0.02046846459654024,
      "grad_norm": 1.459146499633789,
      "learning_rate": 9.99999996465144e-05,
      "loss": 3.0465,
      "step": 762
    },
    {
      "epoch": 0.020495326098635434,
      "grad_norm": 1.385705590248108,
      "learning_rate": 9.999999964557987e-05,
      "loss": 3.0244,
      "step": 763
    },
    {
      "epoch": 0.020522187600730633,
      "grad_norm": 1.380672812461853,
      "learning_rate": 9.999999964464411e-05,
      "loss": 2.7153,
      "step": 764
    },
    {
      "epoch": 0.02054904910282583,
      "grad_norm": 1.549020528793335,
      "learning_rate": 9.999999964370712e-05,
      "loss": 3.2815,
      "step": 765
    },
    {
      "epoch": 0.020575910604921026,
      "grad_norm": 1.4888899326324463,
      "learning_rate": 9.999999964276887e-05,
      "loss": 3.1887,
      "step": 766
    },
    {
      "epoch": 0.020602772107016224,
      "grad_norm": 1.613270878791809,
      "learning_rate": 9.999999964182943e-05,
      "loss": 3.2248,
      "step": 767
    },
    {
      "epoch": 0.020629633609111422,
      "grad_norm": 1.4555995464324951,
      "learning_rate": 9.999999964088872e-05,
      "loss": 3.1396,
      "step": 768
    },
    {
      "epoch": 0.020656495111206617,
      "grad_norm": 1.5034013986587524,
      "learning_rate": 9.999999963994679e-05,
      "loss": 3.0802,
      "step": 769
    },
    {
      "epoch": 0.020683356613301816,
      "grad_norm": 1.5299456119537354,
      "learning_rate": 9.999999963900362e-05,
      "loss": 3.0,
      "step": 770
    },
    {
      "epoch": 0.020710218115397014,
      "grad_norm": 1.523677110671997,
      "learning_rate": 9.999999963805922e-05,
      "loss": 2.9568,
      "step": 771
    },
    {
      "epoch": 0.02073707961749221,
      "grad_norm": 1.3994543552398682,
      "learning_rate": 9.99999996371136e-05,
      "loss": 2.8977,
      "step": 772
    },
    {
      "epoch": 0.020763941119587407,
      "grad_norm": 1.4499335289001465,
      "learning_rate": 9.999999963616673e-05,
      "loss": 3.0254,
      "step": 773
    },
    {
      "epoch": 0.020790802621682605,
      "grad_norm": 1.4630128145217896,
      "learning_rate": 9.999999963521863e-05,
      "loss": 3.2021,
      "step": 774
    },
    {
      "epoch": 0.0208176641237778,
      "grad_norm": 1.4782243967056274,
      "learning_rate": 9.99999996342693e-05,
      "loss": 3.0726,
      "step": 775
    },
    {
      "epoch": 0.020844525625873,
      "grad_norm": 1.5138225555419922,
      "learning_rate": 9.999999963331872e-05,
      "loss": 3.1348,
      "step": 776
    },
    {
      "epoch": 0.020871387127968197,
      "grad_norm": 1.437690019607544,
      "learning_rate": 9.999999963236693e-05,
      "loss": 2.9476,
      "step": 777
    },
    {
      "epoch": 0.02089824863006339,
      "grad_norm": 1.5484408140182495,
      "learning_rate": 9.999999963141389e-05,
      "loss": 3.2604,
      "step": 778
    },
    {
      "epoch": 0.02092511013215859,
      "grad_norm": 1.4620468616485596,
      "learning_rate": 9.999999963045963e-05,
      "loss": 2.8977,
      "step": 779
    },
    {
      "epoch": 0.020951971634253788,
      "grad_norm": 1.5116389989852905,
      "learning_rate": 9.999999962950412e-05,
      "loss": 3.4297,
      "step": 780
    },
    {
      "epoch": 0.020978833136348983,
      "grad_norm": 1.4593900442123413,
      "learning_rate": 9.999999962854738e-05,
      "loss": 2.9795,
      "step": 781
    },
    {
      "epoch": 0.02100569463844418,
      "grad_norm": 1.7128779888153076,
      "learning_rate": 9.999999962758942e-05,
      "loss": 3.16,
      "step": 782
    },
    {
      "epoch": 0.02103255614053938,
      "grad_norm": 1.6186964511871338,
      "learning_rate": 9.999999962663021e-05,
      "loss": 3.3634,
      "step": 783
    },
    {
      "epoch": 0.021059417642634574,
      "grad_norm": 1.6430516242980957,
      "learning_rate": 9.999999962566979e-05,
      "loss": 3.1656,
      "step": 784
    },
    {
      "epoch": 0.021086279144729773,
      "grad_norm": 1.5856648683547974,
      "learning_rate": 9.999999962470811e-05,
      "loss": 3.1166,
      "step": 785
    },
    {
      "epoch": 0.02111314064682497,
      "grad_norm": 1.5856009721755981,
      "learning_rate": 9.99999996237452e-05,
      "loss": 2.9903,
      "step": 786
    },
    {
      "epoch": 0.02114000214892017,
      "grad_norm": 1.6610063314437866,
      "learning_rate": 9.999999962278106e-05,
      "loss": 2.9762,
      "step": 787
    },
    {
      "epoch": 0.021166863651015364,
      "grad_norm": 1.5724165439605713,
      "learning_rate": 9.99999996218157e-05,
      "loss": 3.2502,
      "step": 788
    },
    {
      "epoch": 0.021193725153110562,
      "grad_norm": 1.6868854761123657,
      "learning_rate": 9.99999996208491e-05,
      "loss": 3.6198,
      "step": 789
    },
    {
      "epoch": 0.02122058665520576,
      "grad_norm": 1.633146047592163,
      "learning_rate": 9.999999961988126e-05,
      "loss": 3.4037,
      "step": 790
    },
    {
      "epoch": 0.021247448157300956,
      "grad_norm": 1.6730390787124634,
      "learning_rate": 9.999999961891217e-05,
      "loss": 3.2181,
      "step": 791
    },
    {
      "epoch": 0.021274309659396154,
      "grad_norm": 1.7232856750488281,
      "learning_rate": 9.999999961794188e-05,
      "loss": 3.3526,
      "step": 792
    },
    {
      "epoch": 0.021301171161491352,
      "grad_norm": 1.7128450870513916,
      "learning_rate": 9.999999961697032e-05,
      "loss": 3.4663,
      "step": 793
    },
    {
      "epoch": 0.021328032663586547,
      "grad_norm": 1.6938360929489136,
      "learning_rate": 9.999999961599756e-05,
      "loss": 3.4175,
      "step": 794
    },
    {
      "epoch": 0.021354894165681745,
      "grad_norm": 1.680937647819519,
      "learning_rate": 9.999999961502356e-05,
      "loss": 3.4235,
      "step": 795
    },
    {
      "epoch": 0.021381755667776944,
      "grad_norm": 1.6139475107192993,
      "learning_rate": 9.999999961404832e-05,
      "loss": 3.0056,
      "step": 796
    },
    {
      "epoch": 0.02140861716987214,
      "grad_norm": 2.0173404216766357,
      "learning_rate": 9.999999961307184e-05,
      "loss": 3.5415,
      "step": 797
    },
    {
      "epoch": 0.021435478671967337,
      "grad_norm": 1.920420527458191,
      "learning_rate": 9.999999961209412e-05,
      "loss": 3.5341,
      "step": 798
    },
    {
      "epoch": 0.021462340174062535,
      "grad_norm": 2.095407247543335,
      "learning_rate": 9.99999996111152e-05,
      "loss": 3.5424,
      "step": 799
    },
    {
      "epoch": 0.02148920167615773,
      "grad_norm": 1.6919872760772705,
      "learning_rate": 9.999999961013503e-05,
      "loss": 2.8392,
      "step": 800
    },
    {
      "epoch": 0.021516063178252928,
      "grad_norm": 1.4953211545944214,
      "learning_rate": 9.99999996091536e-05,
      "loss": 3.282,
      "step": 801
    },
    {
      "epoch": 0.021542924680348127,
      "grad_norm": 1.3652949333190918,
      "learning_rate": 9.999999960817097e-05,
      "loss": 2.9762,
      "step": 802
    },
    {
      "epoch": 0.02156978618244332,
      "grad_norm": 1.412086009979248,
      "learning_rate": 9.999999960718709e-05,
      "loss": 2.949,
      "step": 803
    },
    {
      "epoch": 0.02159664768453852,
      "grad_norm": 1.33501398563385,
      "learning_rate": 9.999999960620197e-05,
      "loss": 2.753,
      "step": 804
    },
    {
      "epoch": 0.021623509186633718,
      "grad_norm": 1.3840041160583496,
      "learning_rate": 9.999999960521563e-05,
      "loss": 2.8277,
      "step": 805
    },
    {
      "epoch": 0.021650370688728913,
      "grad_norm": 1.3337507247924805,
      "learning_rate": 9.999999960422806e-05,
      "loss": 2.618,
      "step": 806
    },
    {
      "epoch": 0.02167723219082411,
      "grad_norm": 1.3044178485870361,
      "learning_rate": 9.999999960323924e-05,
      "loss": 2.8496,
      "step": 807
    },
    {
      "epoch": 0.02170409369291931,
      "grad_norm": 1.5015467405319214,
      "learning_rate": 9.999999960224919e-05,
      "loss": 3.1227,
      "step": 808
    },
    {
      "epoch": 0.021730955195014504,
      "grad_norm": 1.524622917175293,
      "learning_rate": 9.999999960125792e-05,
      "loss": 2.9861,
      "step": 809
    },
    {
      "epoch": 0.021757816697109703,
      "grad_norm": 1.4448390007019043,
      "learning_rate": 9.99999996002654e-05,
      "loss": 2.9704,
      "step": 810
    },
    {
      "epoch": 0.0217846781992049,
      "grad_norm": 1.434091567993164,
      "learning_rate": 9.999999959927167e-05,
      "loss": 2.9119,
      "step": 811
    },
    {
      "epoch": 0.021811539701300096,
      "grad_norm": 1.6284412145614624,
      "learning_rate": 9.999999959827669e-05,
      "loss": 3.1126,
      "step": 812
    },
    {
      "epoch": 0.021838401203395294,
      "grad_norm": 1.4972847700119019,
      "learning_rate": 9.999999959728045e-05,
      "loss": 3.0218,
      "step": 813
    },
    {
      "epoch": 0.021865262705490492,
      "grad_norm": 1.459935188293457,
      "learning_rate": 9.999999959628301e-05,
      "loss": 3.062,
      "step": 814
    },
    {
      "epoch": 0.021892124207585687,
      "grad_norm": 1.628962755203247,
      "learning_rate": 9.999999959528434e-05,
      "loss": 3.327,
      "step": 815
    },
    {
      "epoch": 0.021918985709680885,
      "grad_norm": 1.4703344106674194,
      "learning_rate": 9.999999959428442e-05,
      "loss": 3.0919,
      "step": 816
    },
    {
      "epoch": 0.021945847211776084,
      "grad_norm": 1.5622847080230713,
      "learning_rate": 9.999999959328328e-05,
      "loss": 3.1916,
      "step": 817
    },
    {
      "epoch": 0.02197270871387128,
      "grad_norm": 1.4799901247024536,
      "learning_rate": 9.999999959228089e-05,
      "loss": 3.0399,
      "step": 818
    },
    {
      "epoch": 0.021999570215966477,
      "grad_norm": 1.5466630458831787,
      "learning_rate": 9.999999959127728e-05,
      "loss": 3.006,
      "step": 819
    },
    {
      "epoch": 0.022026431718061675,
      "grad_norm": 1.5267852544784546,
      "learning_rate": 9.999999959027241e-05,
      "loss": 3.1746,
      "step": 820
    },
    {
      "epoch": 0.02205329322015687,
      "grad_norm": 1.6994048357009888,
      "learning_rate": 9.999999958926633e-05,
      "loss": 3.2067,
      "step": 821
    },
    {
      "epoch": 0.02208015472225207,
      "grad_norm": 1.4320420026779175,
      "learning_rate": 9.999999958825902e-05,
      "loss": 2.9041,
      "step": 822
    },
    {
      "epoch": 0.022107016224347267,
      "grad_norm": 1.6223822832107544,
      "learning_rate": 9.999999958725049e-05,
      "loss": 3.3279,
      "step": 823
    },
    {
      "epoch": 0.02213387772644246,
      "grad_norm": 1.6538143157958984,
      "learning_rate": 9.999999958624069e-05,
      "loss": 3.2987,
      "step": 824
    },
    {
      "epoch": 0.02216073922853766,
      "grad_norm": 1.4128156900405884,
      "learning_rate": 9.999999958522968e-05,
      "loss": 2.9321,
      "step": 825
    },
    {
      "epoch": 0.022187600730632858,
      "grad_norm": 1.4784235954284668,
      "learning_rate": 9.999999958421742e-05,
      "loss": 2.7364,
      "step": 826
    },
    {
      "epoch": 0.022214462232728053,
      "grad_norm": 1.5601043701171875,
      "learning_rate": 9.999999958320394e-05,
      "loss": 3.172,
      "step": 827
    },
    {
      "epoch": 0.02224132373482325,
      "grad_norm": 1.6303744316101074,
      "learning_rate": 9.999999958218921e-05,
      "loss": 3.1838,
      "step": 828
    },
    {
      "epoch": 0.02226818523691845,
      "grad_norm": 1.524660587310791,
      "learning_rate": 9.999999958117326e-05,
      "loss": 3.0486,
      "step": 829
    },
    {
      "epoch": 0.022295046739013644,
      "grad_norm": 2.108213186264038,
      "learning_rate": 9.999999958015608e-05,
      "loss": 2.8753,
      "step": 830
    },
    {
      "epoch": 0.022321908241108843,
      "grad_norm": 1.5068790912628174,
      "learning_rate": 9.999999957913766e-05,
      "loss": 3.1045,
      "step": 831
    },
    {
      "epoch": 0.02234876974320404,
      "grad_norm": 1.752556324005127,
      "learning_rate": 9.9999999578118e-05,
      "loss": 3.3558,
      "step": 832
    },
    {
      "epoch": 0.022375631245299236,
      "grad_norm": 1.6734263896942139,
      "learning_rate": 9.999999957709711e-05,
      "loss": 3.4312,
      "step": 833
    },
    {
      "epoch": 0.022402492747394434,
      "grad_norm": 1.5546562671661377,
      "learning_rate": 9.9999999576075e-05,
      "loss": 2.9735,
      "step": 834
    },
    {
      "epoch": 0.022429354249489632,
      "grad_norm": 1.7671935558319092,
      "learning_rate": 9.999999957505164e-05,
      "loss": 3.2829,
      "step": 835
    },
    {
      "epoch": 0.022456215751584827,
      "grad_norm": 1.6296780109405518,
      "learning_rate": 9.999999957402705e-05,
      "loss": 3.337,
      "step": 836
    },
    {
      "epoch": 0.022483077253680026,
      "grad_norm": 1.602173924446106,
      "learning_rate": 9.999999957300124e-05,
      "loss": 3.2137,
      "step": 837
    },
    {
      "epoch": 0.022509938755775224,
      "grad_norm": 1.4791547060012817,
      "learning_rate": 9.999999957197417e-05,
      "loss": 3.0097,
      "step": 838
    },
    {
      "epoch": 0.02253680025787042,
      "grad_norm": 1.6333667039871216,
      "learning_rate": 9.999999957094587e-05,
      "loss": 3.162,
      "step": 839
    },
    {
      "epoch": 0.022563661759965617,
      "grad_norm": 1.7177056074142456,
      "learning_rate": 9.999999956991636e-05,
      "loss": 3.2876,
      "step": 840
    },
    {
      "epoch": 0.022590523262060815,
      "grad_norm": 1.5853782892227173,
      "learning_rate": 9.999999956888559e-05,
      "loss": 3.2918,
      "step": 841
    },
    {
      "epoch": 0.02261738476415601,
      "grad_norm": 1.6492536067962646,
      "learning_rate": 9.999999956785361e-05,
      "loss": 3.1207,
      "step": 842
    },
    {
      "epoch": 0.02264424626625121,
      "grad_norm": 1.7022151947021484,
      "learning_rate": 9.999999956682038e-05,
      "loss": 3.3891,
      "step": 843
    },
    {
      "epoch": 0.022671107768346407,
      "grad_norm": 1.8401191234588623,
      "learning_rate": 9.999999956578591e-05,
      "loss": 3.5495,
      "step": 844
    },
    {
      "epoch": 0.0226979692704416,
      "grad_norm": 2.1895368099212646,
      "learning_rate": 9.999999956475024e-05,
      "loss": 3.2325,
      "step": 845
    },
    {
      "epoch": 0.0227248307725368,
      "grad_norm": 1.718263030052185,
      "learning_rate": 9.999999956371331e-05,
      "loss": 3.3796,
      "step": 846
    },
    {
      "epoch": 0.022751692274631998,
      "grad_norm": 1.8245230913162231,
      "learning_rate": 9.999999956267515e-05,
      "loss": 3.2055,
      "step": 847
    },
    {
      "epoch": 0.022778553776727193,
      "grad_norm": 1.75575852394104,
      "learning_rate": 9.999999956163575e-05,
      "loss": 3.2328,
      "step": 848
    },
    {
      "epoch": 0.02280541527882239,
      "grad_norm": 2.0926082134246826,
      "learning_rate": 9.999999956059513e-05,
      "loss": 3.7155,
      "step": 849
    },
    {
      "epoch": 0.02283227678091759,
      "grad_norm": 2.1530840396881104,
      "learning_rate": 9.999999955955327e-05,
      "loss": 3.1956,
      "step": 850
    },
    {
      "epoch": 0.022859138283012784,
      "grad_norm": 1.3932898044586182,
      "learning_rate": 9.999999955851017e-05,
      "loss": 3.1861,
      "step": 851
    },
    {
      "epoch": 0.022885999785107983,
      "grad_norm": 1.542462944984436,
      "learning_rate": 9.999999955746584e-05,
      "loss": 2.8578,
      "step": 852
    },
    {
      "epoch": 0.02291286128720318,
      "grad_norm": 1.531349778175354,
      "learning_rate": 9.999999955642028e-05,
      "loss": 3.432,
      "step": 853
    },
    {
      "epoch": 0.022939722789298376,
      "grad_norm": 1.368952751159668,
      "learning_rate": 9.999999955537349e-05,
      "loss": 2.9651,
      "step": 854
    },
    {
      "epoch": 0.022966584291393574,
      "grad_norm": 1.5063704252243042,
      "learning_rate": 9.999999955432546e-05,
      "loss": 3.0688,
      "step": 855
    },
    {
      "epoch": 0.022993445793488772,
      "grad_norm": 1.3633675575256348,
      "learning_rate": 9.99999995532762e-05,
      "loss": 2.7157,
      "step": 856
    },
    {
      "epoch": 0.02302030729558397,
      "grad_norm": 1.3414454460144043,
      "learning_rate": 9.99999995522257e-05,
      "loss": 2.8958,
      "step": 857
    },
    {
      "epoch": 0.023047168797679166,
      "grad_norm": 1.3742226362228394,
      "learning_rate": 9.999999955117396e-05,
      "loss": 2.9802,
      "step": 858
    },
    {
      "epoch": 0.023074030299774364,
      "grad_norm": 1.314835548400879,
      "learning_rate": 9.9999999550121e-05,
      "loss": 2.9829,
      "step": 859
    },
    {
      "epoch": 0.023100891801869562,
      "grad_norm": 1.4580899477005005,
      "learning_rate": 9.999999954906681e-05,
      "loss": 2.8899,
      "step": 860
    },
    {
      "epoch": 0.023127753303964757,
      "grad_norm": 1.5694551467895508,
      "learning_rate": 9.999999954801138e-05,
      "loss": 2.7876,
      "step": 861
    },
    {
      "epoch": 0.023154614806059955,
      "grad_norm": 1.5796012878417969,
      "learning_rate": 9.999999954695472e-05,
      "loss": 2.9759,
      "step": 862
    },
    {
      "epoch": 0.023181476308155154,
      "grad_norm": 1.4369691610336304,
      "learning_rate": 9.99999995458968e-05,
      "loss": 2.892,
      "step": 863
    },
    {
      "epoch": 0.02320833781025035,
      "grad_norm": 1.4166815280914307,
      "learning_rate": 9.999999954483768e-05,
      "loss": 2.8314,
      "step": 864
    },
    {
      "epoch": 0.023235199312345547,
      "grad_norm": 1.4398976564407349,
      "learning_rate": 9.999999954377731e-05,
      "loss": 2.7829,
      "step": 865
    },
    {
      "epoch": 0.023262060814440745,
      "grad_norm": 1.561009168624878,
      "learning_rate": 9.999999954271572e-05,
      "loss": 3.2241,
      "step": 866
    },
    {
      "epoch": 0.02328892231653594,
      "grad_norm": 1.3493255376815796,
      "learning_rate": 9.999999954165287e-05,
      "loss": 2.895,
      "step": 867
    },
    {
      "epoch": 0.023315783818631138,
      "grad_norm": 1.5968419313430786,
      "learning_rate": 9.999999954058881e-05,
      "loss": 3.0864,
      "step": 868
    },
    {
      "epoch": 0.023342645320726337,
      "grad_norm": 1.4958103895187378,
      "learning_rate": 9.999999953952351e-05,
      "loss": 3.1792,
      "step": 869
    },
    {
      "epoch": 0.02336950682282153,
      "grad_norm": 1.609138011932373,
      "learning_rate": 9.999999953845698e-05,
      "loss": 3.1673,
      "step": 870
    },
    {
      "epoch": 0.02339636832491673,
      "grad_norm": 1.5247716903686523,
      "learning_rate": 9.99999995373892e-05,
      "loss": 3.1263,
      "step": 871
    },
    {
      "epoch": 0.023423229827011928,
      "grad_norm": 1.4824477434158325,
      "learning_rate": 9.99999995363202e-05,
      "loss": 2.8878,
      "step": 872
    },
    {
      "epoch": 0.023450091329107123,
      "grad_norm": 1.5343728065490723,
      "learning_rate": 9.999999953524997e-05,
      "loss": 3.1241,
      "step": 873
    },
    {
      "epoch": 0.02347695283120232,
      "grad_norm": 1.653027057647705,
      "learning_rate": 9.99999995341785e-05,
      "loss": 3.4166,
      "step": 874
    },
    {
      "epoch": 0.02350381433329752,
      "grad_norm": 1.5273804664611816,
      "learning_rate": 9.99999995331058e-05,
      "loss": 3.1788,
      "step": 875
    },
    {
      "epoch": 0.023530675835392714,
      "grad_norm": 1.5059871673583984,
      "learning_rate": 9.999999953203187e-05,
      "loss": 2.8871,
      "step": 876
    },
    {
      "epoch": 0.023557537337487913,
      "grad_norm": 1.7065390348434448,
      "learning_rate": 9.999999953095669e-05,
      "loss": 3.3388,
      "step": 877
    },
    {
      "epoch": 0.02358439883958311,
      "grad_norm": 1.5093010663986206,
      "learning_rate": 9.999999952988028e-05,
      "loss": 2.92,
      "step": 878
    },
    {
      "epoch": 0.023611260341678306,
      "grad_norm": 1.479245901107788,
      "learning_rate": 9.999999952880265e-05,
      "loss": 2.8458,
      "step": 879
    },
    {
      "epoch": 0.023638121843773504,
      "grad_norm": 1.303599238395691,
      "learning_rate": 9.999999952772377e-05,
      "loss": 2.6848,
      "step": 880
    },
    {
      "epoch": 0.023664983345868702,
      "grad_norm": 1.4369533061981201,
      "learning_rate": 9.999999952664368e-05,
      "loss": 2.9162,
      "step": 881
    },
    {
      "epoch": 0.023691844847963897,
      "grad_norm": 1.4322164058685303,
      "learning_rate": 9.999999952556234e-05,
      "loss": 2.807,
      "step": 882
    },
    {
      "epoch": 0.023718706350059095,
      "grad_norm": 1.5332303047180176,
      "learning_rate": 9.999999952447976e-05,
      "loss": 3.1427,
      "step": 883
    },
    {
      "epoch": 0.023745567852154294,
      "grad_norm": 1.5984649658203125,
      "learning_rate": 9.999999952339596e-05,
      "loss": 3.343,
      "step": 884
    },
    {
      "epoch": 0.02377242935424949,
      "grad_norm": 1.5730228424072266,
      "learning_rate": 9.999999952231092e-05,
      "loss": 3.3833,
      "step": 885
    },
    {
      "epoch": 0.023799290856344687,
      "grad_norm": 1.5562658309936523,
      "learning_rate": 9.999999952122465e-05,
      "loss": 3.1578,
      "step": 886
    },
    {
      "epoch": 0.023826152358439885,
      "grad_norm": 1.659433126449585,
      "learning_rate": 9.999999952013714e-05,
      "loss": 3.2602,
      "step": 887
    },
    {
      "epoch": 0.02385301386053508,
      "grad_norm": 1.6053160429000854,
      "learning_rate": 9.99999995190484e-05,
      "loss": 3.1533,
      "step": 888
    },
    {
      "epoch": 0.02387987536263028,
      "grad_norm": 1.509786605834961,
      "learning_rate": 9.999999951795842e-05,
      "loss": 3.0258,
      "step": 889
    },
    {
      "epoch": 0.023906736864725477,
      "grad_norm": 1.519740104675293,
      "learning_rate": 9.999999951686721e-05,
      "loss": 3.0771,
      "step": 890
    },
    {
      "epoch": 0.02393359836682067,
      "grad_norm": 1.705270767211914,
      "learning_rate": 9.999999951577477e-05,
      "loss": 3.0579,
      "step": 891
    },
    {
      "epoch": 0.02396045986891587,
      "grad_norm": 1.7090507745742798,
      "learning_rate": 9.999999951468108e-05,
      "loss": 3.5565,
      "step": 892
    },
    {
      "epoch": 0.023987321371011068,
      "grad_norm": 1.6336791515350342,
      "learning_rate": 9.999999951358618e-05,
      "loss": 3.1644,
      "step": 893
    },
    {
      "epoch": 0.024014182873106263,
      "grad_norm": 1.828244924545288,
      "learning_rate": 9.999999951249004e-05,
      "loss": 3.2391,
      "step": 894
    },
    {
      "epoch": 0.02404104437520146,
      "grad_norm": 1.870465874671936,
      "learning_rate": 9.999999951139265e-05,
      "loss": 3.3868,
      "step": 895
    },
    {
      "epoch": 0.02406790587729666,
      "grad_norm": 1.7341759204864502,
      "learning_rate": 9.999999951029405e-05,
      "loss": 3.0196,
      "step": 896
    },
    {
      "epoch": 0.024094767379391854,
      "grad_norm": 1.7239795923233032,
      "learning_rate": 9.99999995091942e-05,
      "loss": 3.5867,
      "step": 897
    },
    {
      "epoch": 0.024121628881487053,
      "grad_norm": 1.9196280241012573,
      "learning_rate": 9.999999950809313e-05,
      "loss": 3.5105,
      "step": 898
    },
    {
      "epoch": 0.02414849038358225,
      "grad_norm": 1.9287900924682617,
      "learning_rate": 9.999999950699082e-05,
      "loss": 3.3439,
      "step": 899
    },
    {
      "epoch": 0.024175351885677446,
      "grad_norm": 1.782228708267212,
      "learning_rate": 9.999999950588727e-05,
      "loss": 3.1187,
      "step": 900
    },
    {
      "epoch": 0.024202213387772644,
      "grad_norm": 1.4669206142425537,
      "learning_rate": 9.999999950478249e-05,
      "loss": 3.1636,
      "step": 901
    },
    {
      "epoch": 0.024229074889867842,
      "grad_norm": 1.4358468055725098,
      "learning_rate": 9.999999950367647e-05,
      "loss": 2.9277,
      "step": 902
    },
    {
      "epoch": 0.024255936391963037,
      "grad_norm": 1.6136616468429565,
      "learning_rate": 9.999999950256923e-05,
      "loss": 3.2609,
      "step": 903
    },
    {
      "epoch": 0.024282797894058235,
      "grad_norm": 1.5683788061141968,
      "learning_rate": 9.999999950146075e-05,
      "loss": 3.3504,
      "step": 904
    },
    {
      "epoch": 0.024309659396153434,
      "grad_norm": 1.3870428800582886,
      "learning_rate": 9.999999950035104e-05,
      "loss": 2.9261,
      "step": 905
    },
    {
      "epoch": 0.02433652089824863,
      "grad_norm": 1.3878294229507446,
      "learning_rate": 9.999999949924008e-05,
      "loss": 2.8212,
      "step": 906
    },
    {
      "epoch": 0.024363382400343827,
      "grad_norm": 1.377278208732605,
      "learning_rate": 9.99999994981279e-05,
      "loss": 3.0824,
      "step": 907
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 1.3985542058944702,
      "learning_rate": 9.999999949701448e-05,
      "loss": 2.8544,
      "step": 908
    },
    {
      "epoch": 0.02441710540453422,
      "grad_norm": 1.439070701599121,
      "learning_rate": 9.999999949589984e-05,
      "loss": 3.0956,
      "step": 909
    },
    {
      "epoch": 0.02444396690662942,
      "grad_norm": 1.6687121391296387,
      "learning_rate": 9.999999949478396e-05,
      "loss": 3.2903,
      "step": 910
    },
    {
      "epoch": 0.024470828408724617,
      "grad_norm": 1.4199494123458862,
      "learning_rate": 9.999999949366684e-05,
      "loss": 3.1253,
      "step": 911
    },
    {
      "epoch": 0.02449768991081981,
      "grad_norm": 1.4392849206924438,
      "learning_rate": 9.999999949254849e-05,
      "loss": 2.8105,
      "step": 912
    },
    {
      "epoch": 0.02452455141291501,
      "grad_norm": 1.5232583284378052,
      "learning_rate": 9.999999949142891e-05,
      "loss": 2.9298,
      "step": 913
    },
    {
      "epoch": 0.024551412915010208,
      "grad_norm": 1.428612470626831,
      "learning_rate": 9.999999949030809e-05,
      "loss": 3.15,
      "step": 914
    },
    {
      "epoch": 0.024578274417105403,
      "grad_norm": 1.459532380104065,
      "learning_rate": 9.999999948918604e-05,
      "loss": 2.9516,
      "step": 915
    },
    {
      "epoch": 0.0246051359192006,
      "grad_norm": 1.4741835594177246,
      "learning_rate": 9.999999948806275e-05,
      "loss": 2.9968,
      "step": 916
    },
    {
      "epoch": 0.0246319974212958,
      "grad_norm": 1.3959861993789673,
      "learning_rate": 9.999999948693823e-05,
      "loss": 2.7548,
      "step": 917
    },
    {
      "epoch": 0.024658858923390994,
      "grad_norm": 1.5814383029937744,
      "learning_rate": 9.999999948581248e-05,
      "loss": 3.071,
      "step": 918
    },
    {
      "epoch": 0.024685720425486193,
      "grad_norm": 1.3421992063522339,
      "learning_rate": 9.99999994846855e-05,
      "loss": 2.8023,
      "step": 919
    },
    {
      "epoch": 0.02471258192758139,
      "grad_norm": 1.5291359424591064,
      "learning_rate": 9.999999948355727e-05,
      "loss": 2.9152,
      "step": 920
    },
    {
      "epoch": 0.024739443429676586,
      "grad_norm": 1.5606157779693604,
      "learning_rate": 9.999999948242781e-05,
      "loss": 2.769,
      "step": 921
    },
    {
      "epoch": 0.024766304931771784,
      "grad_norm": 1.4144341945648193,
      "learning_rate": 9.999999948129714e-05,
      "loss": 2.9678,
      "step": 922
    },
    {
      "epoch": 0.024793166433866982,
      "grad_norm": 1.4727364778518677,
      "learning_rate": 9.99999994801652e-05,
      "loss": 2.9027,
      "step": 923
    },
    {
      "epoch": 0.02482002793596218,
      "grad_norm": 1.5003682374954224,
      "learning_rate": 9.999999947903205e-05,
      "loss": 2.8261,
      "step": 924
    },
    {
      "epoch": 0.024846889438057376,
      "grad_norm": 1.4803307056427002,
      "learning_rate": 9.999999947789766e-05,
      "loss": 2.959,
      "step": 925
    },
    {
      "epoch": 0.024873750940152574,
      "grad_norm": 1.6897430419921875,
      "learning_rate": 9.999999947676206e-05,
      "loss": 3.5019,
      "step": 926
    },
    {
      "epoch": 0.024900612442247772,
      "grad_norm": 1.4898759126663208,
      "learning_rate": 9.999999947562519e-05,
      "loss": 2.921,
      "step": 927
    },
    {
      "epoch": 0.024927473944342967,
      "grad_norm": 1.496799349784851,
      "learning_rate": 9.99999994744871e-05,
      "loss": 2.8998,
      "step": 928
    },
    {
      "epoch": 0.024954335446438165,
      "grad_norm": 1.5128743648529053,
      "learning_rate": 9.999999947334778e-05,
      "loss": 3.0422,
      "step": 929
    },
    {
      "epoch": 0.024981196948533364,
      "grad_norm": 1.6503703594207764,
      "learning_rate": 9.999999947220722e-05,
      "loss": 3.1786,
      "step": 930
    },
    {
      "epoch": 0.02500805845062856,
      "grad_norm": 1.4745047092437744,
      "learning_rate": 9.999999947106544e-05,
      "loss": 3.3131,
      "step": 931
    },
    {
      "epoch": 0.025034919952723757,
      "grad_norm": 1.529066801071167,
      "learning_rate": 9.999999946992241e-05,
      "loss": 2.9621,
      "step": 932
    },
    {
      "epoch": 0.025061781454818955,
      "grad_norm": 1.4728082418441772,
      "learning_rate": 9.999999946877816e-05,
      "loss": 2.8143,
      "step": 933
    },
    {
      "epoch": 0.02508864295691415,
      "grad_norm": 1.5155495405197144,
      "learning_rate": 9.999999946763266e-05,
      "loss": 3.0541,
      "step": 934
    },
    {
      "epoch": 0.025115504459009348,
      "grad_norm": 1.4486238956451416,
      "learning_rate": 9.999999946648594e-05,
      "loss": 3.2015,
      "step": 935
    },
    {
      "epoch": 0.025142365961104546,
      "grad_norm": 1.548398733139038,
      "learning_rate": 9.999999946533798e-05,
      "loss": 2.9229,
      "step": 936
    },
    {
      "epoch": 0.02516922746319974,
      "grad_norm": 1.5989586114883423,
      "learning_rate": 9.999999946418878e-05,
      "loss": 3.4691,
      "step": 937
    },
    {
      "epoch": 0.02519608896529494,
      "grad_norm": 1.5218377113342285,
      "learning_rate": 9.999999946303835e-05,
      "loss": 3.1763,
      "step": 938
    },
    {
      "epoch": 0.025222950467390138,
      "grad_norm": 1.7632472515106201,
      "learning_rate": 9.99999994618867e-05,
      "loss": 3.0938,
      "step": 939
    },
    {
      "epoch": 0.025249811969485333,
      "grad_norm": 1.6770436763763428,
      "learning_rate": 9.999999946073381e-05,
      "loss": 3.4123,
      "step": 940
    },
    {
      "epoch": 0.02527667347158053,
      "grad_norm": 1.524204969406128,
      "learning_rate": 9.999999945957968e-05,
      "loss": 3.1141,
      "step": 941
    },
    {
      "epoch": 0.02530353497367573,
      "grad_norm": 1.7641979455947876,
      "learning_rate": 9.999999945842431e-05,
      "loss": 3.4322,
      "step": 942
    },
    {
      "epoch": 0.025330396475770924,
      "grad_norm": 1.7929165363311768,
      "learning_rate": 9.999999945726772e-05,
      "loss": 3.2306,
      "step": 943
    },
    {
      "epoch": 0.025357257977866123,
      "grad_norm": 1.5453864336013794,
      "learning_rate": 9.999999945610989e-05,
      "loss": 3.1985,
      "step": 944
    },
    {
      "epoch": 0.02538411947996132,
      "grad_norm": 1.512291431427002,
      "learning_rate": 9.999999945495084e-05,
      "loss": 2.8978,
      "step": 945
    },
    {
      "epoch": 0.025410980982056516,
      "grad_norm": 1.6914894580841064,
      "learning_rate": 9.999999945379052e-05,
      "loss": 3.2245,
      "step": 946
    },
    {
      "epoch": 0.025437842484151714,
      "grad_norm": 1.6791399717330933,
      "learning_rate": 9.999999945262902e-05,
      "loss": 3.2164,
      "step": 947
    },
    {
      "epoch": 0.025464703986246912,
      "grad_norm": 1.818323016166687,
      "learning_rate": 9.999999945146624e-05,
      "loss": 3.3281,
      "step": 948
    },
    {
      "epoch": 0.025491565488342107,
      "grad_norm": 1.8089464902877808,
      "learning_rate": 9.999999945030225e-05,
      "loss": 3.618,
      "step": 949
    },
    {
      "epoch": 0.025518426990437305,
      "grad_norm": 2.034975290298462,
      "learning_rate": 9.999999944913702e-05,
      "loss": 3.4737,
      "step": 950
    },
    {
      "epoch": 0.025545288492532504,
      "grad_norm": 1.1866132020950317,
      "learning_rate": 9.999999944797055e-05,
      "loss": 3.2219,
      "step": 951
    },
    {
      "epoch": 0.0255721499946277,
      "grad_norm": 1.416660189628601,
      "learning_rate": 9.999999944680285e-05,
      "loss": 2.9487,
      "step": 952
    },
    {
      "epoch": 0.025599011496722897,
      "grad_norm": 1.3192981481552124,
      "learning_rate": 9.999999944563392e-05,
      "loss": 3.0613,
      "step": 953
    },
    {
      "epoch": 0.025625872998818095,
      "grad_norm": 1.4292821884155273,
      "learning_rate": 9.999999944446376e-05,
      "loss": 2.9264,
      "step": 954
    },
    {
      "epoch": 0.02565273450091329,
      "grad_norm": 1.3824403285980225,
      "learning_rate": 9.999999944329234e-05,
      "loss": 2.8121,
      "step": 955
    },
    {
      "epoch": 0.02567959600300849,
      "grad_norm": 1.3949388265609741,
      "learning_rate": 9.999999944211972e-05,
      "loss": 2.799,
      "step": 956
    },
    {
      "epoch": 0.025706457505103687,
      "grad_norm": 1.4141530990600586,
      "learning_rate": 9.999999944094586e-05,
      "loss": 3.0121,
      "step": 957
    },
    {
      "epoch": 0.02573331900719888,
      "grad_norm": 1.291210651397705,
      "learning_rate": 9.999999943977075e-05,
      "loss": 2.8466,
      "step": 958
    },
    {
      "epoch": 0.02576018050929408,
      "grad_norm": 1.3864771127700806,
      "learning_rate": 9.999999943859441e-05,
      "loss": 2.8567,
      "step": 959
    },
    {
      "epoch": 0.025787042011389278,
      "grad_norm": 1.444599986076355,
      "learning_rate": 9.999999943741685e-05,
      "loss": 3.0129,
      "step": 960
    },
    {
      "epoch": 0.025813903513484473,
      "grad_norm": 1.3651354312896729,
      "learning_rate": 9.999999943623805e-05,
      "loss": 2.8797,
      "step": 961
    },
    {
      "epoch": 0.02584076501557967,
      "grad_norm": 1.4720170497894287,
      "learning_rate": 9.999999943505802e-05,
      "loss": 3.1372,
      "step": 962
    },
    {
      "epoch": 0.02586762651767487,
      "grad_norm": 1.3237793445587158,
      "learning_rate": 9.999999943387674e-05,
      "loss": 2.8831,
      "step": 963
    },
    {
      "epoch": 0.025894488019770064,
      "grad_norm": 1.3659074306488037,
      "learning_rate": 9.999999943269424e-05,
      "loss": 2.9116,
      "step": 964
    },
    {
      "epoch": 0.025921349521865263,
      "grad_norm": 1.487601637840271,
      "learning_rate": 9.999999943151051e-05,
      "loss": 3.1031,
      "step": 965
    },
    {
      "epoch": 0.02594821102396046,
      "grad_norm": 1.4030911922454834,
      "learning_rate": 9.999999943032554e-05,
      "loss": 2.7534,
      "step": 966
    },
    {
      "epoch": 0.025975072526055656,
      "grad_norm": 1.4361507892608643,
      "learning_rate": 9.999999942913933e-05,
      "loss": 3.0331,
      "step": 967
    },
    {
      "epoch": 0.026001934028150854,
      "grad_norm": 1.2849395275115967,
      "learning_rate": 9.99999994279519e-05,
      "loss": 2.9815,
      "step": 968
    },
    {
      "epoch": 0.026028795530246052,
      "grad_norm": 1.471046805381775,
      "learning_rate": 9.999999942676322e-05,
      "loss": 3.2367,
      "step": 969
    },
    {
      "epoch": 0.026055657032341247,
      "grad_norm": 1.5335110425949097,
      "learning_rate": 9.999999942557332e-05,
      "loss": 3.1871,
      "step": 970
    },
    {
      "epoch": 0.026082518534436445,
      "grad_norm": 1.4752976894378662,
      "learning_rate": 9.999999942438219e-05,
      "loss": 2.9454,
      "step": 971
    },
    {
      "epoch": 0.026109380036531644,
      "grad_norm": 1.4635950326919556,
      "learning_rate": 9.99999994231898e-05,
      "loss": 2.9933,
      "step": 972
    },
    {
      "epoch": 0.02613624153862684,
      "grad_norm": 1.5313303470611572,
      "learning_rate": 9.99999994219962e-05,
      "loss": 3.1916,
      "step": 973
    },
    {
      "epoch": 0.026163103040722037,
      "grad_norm": 1.5451520681381226,
      "learning_rate": 9.999999942080136e-05,
      "loss": 3.0727,
      "step": 974
    },
    {
      "epoch": 0.026189964542817235,
      "grad_norm": 1.551936149597168,
      "learning_rate": 9.999999941960529e-05,
      "loss": 3.007,
      "step": 975
    },
    {
      "epoch": 0.02621682604491243,
      "grad_norm": 1.4487463235855103,
      "learning_rate": 9.999999941840798e-05,
      "loss": 3.1173,
      "step": 976
    },
    {
      "epoch": 0.02624368754700763,
      "grad_norm": 1.5345468521118164,
      "learning_rate": 9.999999941720945e-05,
      "loss": 3.3884,
      "step": 977
    },
    {
      "epoch": 0.026270549049102827,
      "grad_norm": 1.5212229490280151,
      "learning_rate": 9.999999941600967e-05,
      "loss": 3.0135,
      "step": 978
    },
    {
      "epoch": 0.02629741055119802,
      "grad_norm": 1.5727816820144653,
      "learning_rate": 9.999999941480866e-05,
      "loss": 3.2597,
      "step": 979
    },
    {
      "epoch": 0.02632427205329322,
      "grad_norm": 1.4715198278427124,
      "learning_rate": 9.999999941360643e-05,
      "loss": 3.0167,
      "step": 980
    },
    {
      "epoch": 0.026351133555388418,
      "grad_norm": 1.5815037488937378,
      "learning_rate": 9.999999941240295e-05,
      "loss": 3.3436,
      "step": 981
    },
    {
      "epoch": 0.026377995057483613,
      "grad_norm": 1.573350191116333,
      "learning_rate": 9.999999941119824e-05,
      "loss": 2.8999,
      "step": 982
    },
    {
      "epoch": 0.02640485655957881,
      "grad_norm": 1.422698736190796,
      "learning_rate": 9.99999994099923e-05,
      "loss": 2.8865,
      "step": 983
    },
    {
      "epoch": 0.02643171806167401,
      "grad_norm": 1.5705293416976929,
      "learning_rate": 9.999999940878512e-05,
      "loss": 3.0761,
      "step": 984
    },
    {
      "epoch": 0.026458579563769204,
      "grad_norm": 1.4294575452804565,
      "learning_rate": 9.99999994075767e-05,
      "loss": 2.8289,
      "step": 985
    },
    {
      "epoch": 0.026485441065864403,
      "grad_norm": 1.5163748264312744,
      "learning_rate": 9.999999940636706e-05,
      "loss": 2.8431,
      "step": 986
    },
    {
      "epoch": 0.0265123025679596,
      "grad_norm": 1.6206839084625244,
      "learning_rate": 9.999999940515618e-05,
      "loss": 3.3664,
      "step": 987
    },
    {
      "epoch": 0.026539164070054796,
      "grad_norm": 1.7007125616073608,
      "learning_rate": 9.999999940394407e-05,
      "loss": 3.2286,
      "step": 988
    },
    {
      "epoch": 0.026566025572149994,
      "grad_norm": 1.630654215812683,
      "learning_rate": 9.999999940273073e-05,
      "loss": 3.2184,
      "step": 989
    },
    {
      "epoch": 0.026592887074245192,
      "grad_norm": 1.6554126739501953,
      "learning_rate": 9.999999940151614e-05,
      "loss": 3.322,
      "step": 990
    },
    {
      "epoch": 0.026619748576340387,
      "grad_norm": 1.6082969903945923,
      "learning_rate": 9.999999940030034e-05,
      "loss": 3.2394,
      "step": 991
    },
    {
      "epoch": 0.026646610078435586,
      "grad_norm": 1.7029058933258057,
      "learning_rate": 9.999999939908328e-05,
      "loss": 3.2544,
      "step": 992
    },
    {
      "epoch": 0.026673471580530784,
      "grad_norm": 1.6000038385391235,
      "learning_rate": 9.9999999397865e-05,
      "loss": 3.1424,
      "step": 993
    },
    {
      "epoch": 0.026700333082625982,
      "grad_norm": 1.7077949047088623,
      "learning_rate": 9.99999993966455e-05,
      "loss": 3.2201,
      "step": 994
    },
    {
      "epoch": 0.026727194584721177,
      "grad_norm": 1.7065461874008179,
      "learning_rate": 9.999999939542475e-05,
      "loss": 3.336,
      "step": 995
    },
    {
      "epoch": 0.026754056086816375,
      "grad_norm": 1.8289830684661865,
      "learning_rate": 9.999999939420277e-05,
      "loss": 3.5807,
      "step": 996
    },
    {
      "epoch": 0.026780917588911574,
      "grad_norm": 1.6227738857269287,
      "learning_rate": 9.999999939297955e-05,
      "loss": 3.0879,
      "step": 997
    },
    {
      "epoch": 0.02680777909100677,
      "grad_norm": 1.790627121925354,
      "learning_rate": 9.99999993917551e-05,
      "loss": 3.353,
      "step": 998
    },
    {
      "epoch": 0.026834640593101967,
      "grad_norm": 1.885088562965393,
      "learning_rate": 9.999999939052941e-05,
      "loss": 3.4586,
      "step": 999
    },
    {
      "epoch": 0.026861502095197165,
      "grad_norm": 1.8984928131103516,
      "learning_rate": 9.999999938930252e-05,
      "loss": 3.5962,
      "step": 1000
    },
    {
      "epoch": 0.02688836359729236,
      "grad_norm": 1.2081084251403809,
      "learning_rate": 9.999999938807436e-05,
      "loss": 3.2575,
      "step": 1001
    },
    {
      "epoch": 0.026915225099387558,
      "grad_norm": 1.3805323839187622,
      "learning_rate": 9.999999938684497e-05,
      "loss": 3.2441,
      "step": 1002
    },
    {
      "epoch": 0.026942086601482756,
      "grad_norm": 1.7433286905288696,
      "learning_rate": 9.999999938561435e-05,
      "loss": 2.8775,
      "step": 1003
    },
    {
      "epoch": 0.02696894810357795,
      "grad_norm": 1.3275337219238281,
      "learning_rate": 9.999999938438252e-05,
      "loss": 2.8907,
      "step": 1004
    },
    {
      "epoch": 0.02699580960567315,
      "grad_norm": 1.4613322019577026,
      "learning_rate": 9.999999938314943e-05,
      "loss": 2.7962,
      "step": 1005
    },
    {
      "epoch": 0.027022671107768348,
      "grad_norm": 1.485374927520752,
      "learning_rate": 9.999999938191512e-05,
      "loss": 2.8892,
      "step": 1006
    },
    {
      "epoch": 0.027049532609863543,
      "grad_norm": 1.3549524545669556,
      "learning_rate": 9.999999938067956e-05,
      "loss": 3.1675,
      "step": 1007
    },
    {
      "epoch": 0.02707639411195874,
      "grad_norm": 1.3601874113082886,
      "learning_rate": 9.999999937944277e-05,
      "loss": 2.8808,
      "step": 1008
    },
    {
      "epoch": 0.02710325561405394,
      "grad_norm": 1.3621183633804321,
      "learning_rate": 9.999999937820475e-05,
      "loss": 2.7581,
      "step": 1009
    },
    {
      "epoch": 0.027130117116149134,
      "grad_norm": 1.3292536735534668,
      "learning_rate": 9.999999937696549e-05,
      "loss": 2.9433,
      "step": 1010
    },
    {
      "epoch": 0.027156978618244332,
      "grad_norm": 1.4130446910858154,
      "learning_rate": 9.9999999375725e-05,
      "loss": 3.0202,
      "step": 1011
    },
    {
      "epoch": 0.02718384012033953,
      "grad_norm": 1.4119086265563965,
      "learning_rate": 9.999999937448329e-05,
      "loss": 2.8933,
      "step": 1012
    },
    {
      "epoch": 0.027210701622434726,
      "grad_norm": 1.286368727684021,
      "learning_rate": 9.999999937324033e-05,
      "loss": 2.7099,
      "step": 1013
    },
    {
      "epoch": 0.027237563124529924,
      "grad_norm": 1.292993426322937,
      "learning_rate": 9.999999937199614e-05,
      "loss": 2.7148,
      "step": 1014
    },
    {
      "epoch": 0.027264424626625122,
      "grad_norm": 1.2486885786056519,
      "learning_rate": 9.999999937075073e-05,
      "loss": 2.8137,
      "step": 1015
    },
    {
      "epoch": 0.027291286128720317,
      "grad_norm": 1.3911099433898926,
      "learning_rate": 9.999999936950408e-05,
      "loss": 2.7745,
      "step": 1016
    },
    {
      "epoch": 0.027318147630815515,
      "grad_norm": 1.5100406408309937,
      "learning_rate": 9.999999936825617e-05,
      "loss": 2.9737,
      "step": 1017
    },
    {
      "epoch": 0.027345009132910714,
      "grad_norm": 1.372920274734497,
      "learning_rate": 9.999999936700705e-05,
      "loss": 2.7351,
      "step": 1018
    },
    {
      "epoch": 0.02737187063500591,
      "grad_norm": 1.4184085130691528,
      "learning_rate": 9.99999993657567e-05,
      "loss": 2.9937,
      "step": 1019
    },
    {
      "epoch": 0.027398732137101107,
      "grad_norm": 1.5382386445999146,
      "learning_rate": 9.999999936450512e-05,
      "loss": 2.7445,
      "step": 1020
    },
    {
      "epoch": 0.027425593639196305,
      "grad_norm": 1.5459246635437012,
      "learning_rate": 9.999999936325228e-05,
      "loss": 3.0237,
      "step": 1021
    },
    {
      "epoch": 0.0274524551412915,
      "grad_norm": 1.3898468017578125,
      "learning_rate": 9.999999936199823e-05,
      "loss": 2.7733,
      "step": 1022
    },
    {
      "epoch": 0.027479316643386698,
      "grad_norm": 1.4596545696258545,
      "learning_rate": 9.999999936074294e-05,
      "loss": 3.051,
      "step": 1023
    },
    {
      "epoch": 0.027506178145481897,
      "grad_norm": 1.5145318508148193,
      "learning_rate": 9.999999935948642e-05,
      "loss": 3.1378,
      "step": 1024
    },
    {
      "epoch": 0.02753303964757709,
      "grad_norm": 1.5298538208007812,
      "learning_rate": 9.999999935822866e-05,
      "loss": 3.0653,
      "step": 1025
    },
    {
      "epoch": 0.02755990114967229,
      "grad_norm": 1.523829698562622,
      "learning_rate": 9.999999935696967e-05,
      "loss": 3.0143,
      "step": 1026
    },
    {
      "epoch": 0.027586762651767488,
      "grad_norm": 1.5662864446640015,
      "learning_rate": 9.999999935570944e-05,
      "loss": 2.9266,
      "step": 1027
    },
    {
      "epoch": 0.027613624153862683,
      "grad_norm": 1.5643287897109985,
      "learning_rate": 9.999999935444798e-05,
      "loss": 3.0891,
      "step": 1028
    },
    {
      "epoch": 0.02764048565595788,
      "grad_norm": 1.552836537361145,
      "learning_rate": 9.999999935318529e-05,
      "loss": 3.0627,
      "step": 1029
    },
    {
      "epoch": 0.02766734715805308,
      "grad_norm": 1.5275644063949585,
      "learning_rate": 9.999999935192137e-05,
      "loss": 3.2854,
      "step": 1030
    },
    {
      "epoch": 0.027694208660148274,
      "grad_norm": 1.5676196813583374,
      "learning_rate": 9.99999993506562e-05,
      "loss": 3.3264,
      "step": 1031
    },
    {
      "epoch": 0.027721070162243473,
      "grad_norm": 1.6966971158981323,
      "learning_rate": 9.999999934938981e-05,
      "loss": 3.1521,
      "step": 1032
    },
    {
      "epoch": 0.02774793166433867,
      "grad_norm": 1.573509931564331,
      "learning_rate": 9.999999934812219e-05,
      "loss": 3.2024,
      "step": 1033
    },
    {
      "epoch": 0.027774793166433866,
      "grad_norm": 1.5276821851730347,
      "learning_rate": 9.99999993468533e-05,
      "loss": 3.3231,
      "step": 1034
    },
    {
      "epoch": 0.027801654668529064,
      "grad_norm": 1.4477492570877075,
      "learning_rate": 9.999999934558322e-05,
      "loss": 2.9656,
      "step": 1035
    },
    {
      "epoch": 0.027828516170624262,
      "grad_norm": 1.5181984901428223,
      "learning_rate": 9.99999993443119e-05,
      "loss": 3.2101,
      "step": 1036
    },
    {
      "epoch": 0.027855377672719457,
      "grad_norm": 1.5033587217330933,
      "learning_rate": 9.999999934303933e-05,
      "loss": 2.9541,
      "step": 1037
    },
    {
      "epoch": 0.027882239174814655,
      "grad_norm": 1.6101270914077759,
      "learning_rate": 9.999999934176554e-05,
      "loss": 3.1678,
      "step": 1038
    },
    {
      "epoch": 0.027909100676909854,
      "grad_norm": 1.7118161916732788,
      "learning_rate": 9.99999993404905e-05,
      "loss": 3.107,
      "step": 1039
    },
    {
      "epoch": 0.02793596217900505,
      "grad_norm": 1.597589373588562,
      "learning_rate": 9.999999933921425e-05,
      "loss": 3.1001,
      "step": 1040
    },
    {
      "epoch": 0.027962823681100247,
      "grad_norm": 1.608934760093689,
      "learning_rate": 9.999999933793675e-05,
      "loss": 2.9767,
      "step": 1041
    },
    {
      "epoch": 0.027989685183195445,
      "grad_norm": 1.5927395820617676,
      "learning_rate": 9.999999933665801e-05,
      "loss": 3.231,
      "step": 1042
    },
    {
      "epoch": 0.02801654668529064,
      "grad_norm": 1.8898730278015137,
      "learning_rate": 9.999999933537804e-05,
      "loss": 3.8677,
      "step": 1043
    },
    {
      "epoch": 0.02804340818738584,
      "grad_norm": 1.6011654138565063,
      "learning_rate": 9.999999933409685e-05,
      "loss": 3.1195,
      "step": 1044
    },
    {
      "epoch": 0.028070269689481037,
      "grad_norm": 1.7558504343032837,
      "learning_rate": 9.99999993328144e-05,
      "loss": 3.2243,
      "step": 1045
    },
    {
      "epoch": 0.02809713119157623,
      "grad_norm": 1.6238560676574707,
      "learning_rate": 9.999999933153075e-05,
      "loss": 3.3974,
      "step": 1046
    },
    {
      "epoch": 0.02812399269367143,
      "grad_norm": 1.8154619932174683,
      "learning_rate": 9.999999933024584e-05,
      "loss": 3.2551,
      "step": 1047
    },
    {
      "epoch": 0.028150854195766628,
      "grad_norm": 1.7730555534362793,
      "learning_rate": 9.999999932895971e-05,
      "loss": 3.384,
      "step": 1048
    },
    {
      "epoch": 0.028177715697861823,
      "grad_norm": 1.717028260231018,
      "learning_rate": 9.999999932767235e-05,
      "loss": 3.1164,
      "step": 1049
    },
    {
      "epoch": 0.02820457719995702,
      "grad_norm": 2.2091801166534424,
      "learning_rate": 9.999999932638376e-05,
      "loss": 3.6432,
      "step": 1050
    },
    {
      "epoch": 0.02823143870205222,
      "grad_norm": 1.4108836650848389,
      "learning_rate": 9.999999932509392e-05,
      "loss": 2.8067,
      "step": 1051
    },
    {
      "epoch": 0.028258300204147414,
      "grad_norm": 1.5172497034072876,
      "learning_rate": 9.999999932380284e-05,
      "loss": 2.8339,
      "step": 1052
    },
    {
      "epoch": 0.028285161706242613,
      "grad_norm": 1.5175940990447998,
      "learning_rate": 9.999999932251054e-05,
      "loss": 3.0708,
      "step": 1053
    },
    {
      "epoch": 0.02831202320833781,
      "grad_norm": 1.5778652429580688,
      "learning_rate": 9.9999999321217e-05,
      "loss": 2.9358,
      "step": 1054
    },
    {
      "epoch": 0.028338884710433006,
      "grad_norm": 1.3586126565933228,
      "learning_rate": 9.999999931992224e-05,
      "loss": 2.7746,
      "step": 1055
    },
    {
      "epoch": 0.028365746212528204,
      "grad_norm": 1.4578770399093628,
      "learning_rate": 9.999999931862624e-05,
      "loss": 3.1832,
      "step": 1056
    },
    {
      "epoch": 0.028392607714623402,
      "grad_norm": 1.4358021020889282,
      "learning_rate": 9.999999931732899e-05,
      "loss": 2.9011,
      "step": 1057
    },
    {
      "epoch": 0.028419469216718597,
      "grad_norm": 1.2991894483566284,
      "learning_rate": 9.999999931603052e-05,
      "loss": 2.7933,
      "step": 1058
    },
    {
      "epoch": 0.028446330718813796,
      "grad_norm": 1.4651480913162231,
      "learning_rate": 9.999999931473083e-05,
      "loss": 2.9399,
      "step": 1059
    },
    {
      "epoch": 0.028473192220908994,
      "grad_norm": 1.494518756866455,
      "learning_rate": 9.999999931342989e-05,
      "loss": 2.9017,
      "step": 1060
    },
    {
      "epoch": 0.028500053723004192,
      "grad_norm": 1.462116003036499,
      "learning_rate": 9.99999993121277e-05,
      "loss": 2.8853,
      "step": 1061
    },
    {
      "epoch": 0.028526915225099387,
      "grad_norm": 1.2462494373321533,
      "learning_rate": 9.99999993108243e-05,
      "loss": 2.5481,
      "step": 1062
    },
    {
      "epoch": 0.028553776727194585,
      "grad_norm": 1.4719215631484985,
      "learning_rate": 9.999999930951967e-05,
      "loss": 3.06,
      "step": 1063
    },
    {
      "epoch": 0.028580638229289784,
      "grad_norm": 1.359512209892273,
      "learning_rate": 9.99999993082138e-05,
      "loss": 2.851,
      "step": 1064
    },
    {
      "epoch": 0.02860749973138498,
      "grad_norm": 1.4545495510101318,
      "learning_rate": 9.999999930690669e-05,
      "loss": 2.9734,
      "step": 1065
    },
    {
      "epoch": 0.028634361233480177,
      "grad_norm": 1.3730028867721558,
      "learning_rate": 9.999999930559836e-05,
      "loss": 2.8539,
      "step": 1066
    },
    {
      "epoch": 0.028661222735575375,
      "grad_norm": 1.6204618215560913,
      "learning_rate": 9.999999930428879e-05,
      "loss": 3.3271,
      "step": 1067
    },
    {
      "epoch": 0.02868808423767057,
      "grad_norm": 1.4270602464675903,
      "learning_rate": 9.999999930297796e-05,
      "loss": 3.167,
      "step": 1068
    },
    {
      "epoch": 0.028714945739765768,
      "grad_norm": 1.3054600954055786,
      "learning_rate": 9.999999930166593e-05,
      "loss": 2.8688,
      "step": 1069
    },
    {
      "epoch": 0.028741807241860966,
      "grad_norm": 1.5023961067199707,
      "learning_rate": 9.999999930035265e-05,
      "loss": 3.0314,
      "step": 1070
    },
    {
      "epoch": 0.02876866874395616,
      "grad_norm": 1.6203995943069458,
      "learning_rate": 9.999999929903813e-05,
      "loss": 2.913,
      "step": 1071
    },
    {
      "epoch": 0.02879553024605136,
      "grad_norm": 1.333970546722412,
      "learning_rate": 9.99999992977224e-05,
      "loss": 2.9784,
      "step": 1072
    },
    {
      "epoch": 0.028822391748146558,
      "grad_norm": 1.5159754753112793,
      "learning_rate": 9.999999929640542e-05,
      "loss": 2.9283,
      "step": 1073
    },
    {
      "epoch": 0.028849253250241753,
      "grad_norm": 1.4754468202590942,
      "learning_rate": 9.99999992950872e-05,
      "loss": 3.3527,
      "step": 1074
    },
    {
      "epoch": 0.02887611475233695,
      "grad_norm": 1.4421638250350952,
      "learning_rate": 9.999999929376777e-05,
      "loss": 2.8453,
      "step": 1075
    },
    {
      "epoch": 0.02890297625443215,
      "grad_norm": 1.4105534553527832,
      "learning_rate": 9.99999992924471e-05,
      "loss": 2.9349,
      "step": 1076
    },
    {
      "epoch": 0.028929837756527344,
      "grad_norm": 1.4494372606277466,
      "learning_rate": 9.999999929112518e-05,
      "loss": 2.9975,
      "step": 1077
    },
    {
      "epoch": 0.028956699258622542,
      "grad_norm": 1.558030605316162,
      "learning_rate": 9.999999928980203e-05,
      "loss": 3.5427,
      "step": 1078
    },
    {
      "epoch": 0.02898356076071774,
      "grad_norm": 1.4982047080993652,
      "learning_rate": 9.999999928847767e-05,
      "loss": 3.0858,
      "step": 1079
    },
    {
      "epoch": 0.029010422262812936,
      "grad_norm": 1.428868293762207,
      "learning_rate": 9.999999928715205e-05,
      "loss": 2.9926,
      "step": 1080
    },
    {
      "epoch": 0.029037283764908134,
      "grad_norm": 1.652311086654663,
      "learning_rate": 9.99999992858252e-05,
      "loss": 3.0683,
      "step": 1081
    },
    {
      "epoch": 0.029064145267003332,
      "grad_norm": 1.5730986595153809,
      "learning_rate": 9.999999928449713e-05,
      "loss": 3.17,
      "step": 1082
    },
    {
      "epoch": 0.029091006769098527,
      "grad_norm": 1.486251950263977,
      "learning_rate": 9.999999928316781e-05,
      "loss": 2.8794,
      "step": 1083
    },
    {
      "epoch": 0.029117868271193725,
      "grad_norm": 1.6299165487289429,
      "learning_rate": 9.999999928183727e-05,
      "loss": 3.1198,
      "step": 1084
    },
    {
      "epoch": 0.029144729773288924,
      "grad_norm": 1.542945146560669,
      "learning_rate": 9.999999928050549e-05,
      "loss": 3.2898,
      "step": 1085
    },
    {
      "epoch": 0.02917159127538412,
      "grad_norm": 1.529078722000122,
      "learning_rate": 9.999999927917247e-05,
      "loss": 3.0489,
      "step": 1086
    },
    {
      "epoch": 0.029198452777479317,
      "grad_norm": 1.5292929410934448,
      "learning_rate": 9.999999927783823e-05,
      "loss": 3.3094,
      "step": 1087
    },
    {
      "epoch": 0.029225314279574515,
      "grad_norm": 1.4960463047027588,
      "learning_rate": 9.999999927650275e-05,
      "loss": 3.3103,
      "step": 1088
    },
    {
      "epoch": 0.02925217578166971,
      "grad_norm": 1.602150797843933,
      "learning_rate": 9.999999927516603e-05,
      "loss": 3.2901,
      "step": 1089
    },
    {
      "epoch": 0.029279037283764908,
      "grad_norm": 1.5850414037704468,
      "learning_rate": 9.999999927382809e-05,
      "loss": 3.1176,
      "step": 1090
    },
    {
      "epoch": 0.029305898785860107,
      "grad_norm": 1.6117327213287354,
      "learning_rate": 9.99999992724889e-05,
      "loss": 3.1988,
      "step": 1091
    },
    {
      "epoch": 0.0293327602879553,
      "grad_norm": 1.6982961893081665,
      "learning_rate": 9.999999927114848e-05,
      "loss": 3.0294,
      "step": 1092
    },
    {
      "epoch": 0.0293596217900505,
      "grad_norm": 1.521851658821106,
      "learning_rate": 9.999999926980684e-05,
      "loss": 3.2821,
      "step": 1093
    },
    {
      "epoch": 0.029386483292145698,
      "grad_norm": 1.780248761177063,
      "learning_rate": 9.999999926846396e-05,
      "loss": 3.6807,
      "step": 1094
    },
    {
      "epoch": 0.029413344794240893,
      "grad_norm": 1.5586732625961304,
      "learning_rate": 9.999999926711983e-05,
      "loss": 3.1557,
      "step": 1095
    },
    {
      "epoch": 0.02944020629633609,
      "grad_norm": 1.6442537307739258,
      "learning_rate": 9.999999926577449e-05,
      "loss": 3.1615,
      "step": 1096
    },
    {
      "epoch": 0.02946706779843129,
      "grad_norm": 1.8581600189208984,
      "learning_rate": 9.999999926442789e-05,
      "loss": 3.6989,
      "step": 1097
    },
    {
      "epoch": 0.029493929300526484,
      "grad_norm": 2.0552151203155518,
      "learning_rate": 9.999999926308008e-05,
      "loss": 3.532,
      "step": 1098
    },
    {
      "epoch": 0.029520790802621683,
      "grad_norm": 1.9782418012619019,
      "learning_rate": 9.999999926173102e-05,
      "loss": 3.4415,
      "step": 1099
    },
    {
      "epoch": 0.02954765230471688,
      "grad_norm": 1.8358986377716064,
      "learning_rate": 9.999999926038073e-05,
      "loss": 3.2857,
      "step": 1100
    },
    {
      "epoch": 0.029574513806812076,
      "grad_norm": 1.302351951599121,
      "learning_rate": 9.99999992590292e-05,
      "loss": 3.3119,
      "step": 1101
    },
    {
      "epoch": 0.029601375308907274,
      "grad_norm": 1.38497793674469,
      "learning_rate": 9.999999925767647e-05,
      "loss": 3.249,
      "step": 1102
    },
    {
      "epoch": 0.029628236811002472,
      "grad_norm": 1.4507490396499634,
      "learning_rate": 9.999999925632248e-05,
      "loss": 2.9888,
      "step": 1103
    },
    {
      "epoch": 0.029655098313097667,
      "grad_norm": 1.4330673217773438,
      "learning_rate": 9.999999925496725e-05,
      "loss": 2.9996,
      "step": 1104
    },
    {
      "epoch": 0.029681959815192865,
      "grad_norm": 1.317047357559204,
      "learning_rate": 9.99999992536108e-05,
      "loss": 2.8336,
      "step": 1105
    },
    {
      "epoch": 0.029708821317288064,
      "grad_norm": 1.5266220569610596,
      "learning_rate": 9.999999925225312e-05,
      "loss": 2.9996,
      "step": 1106
    },
    {
      "epoch": 0.02973568281938326,
      "grad_norm": 1.4145293235778809,
      "learning_rate": 9.999999925089418e-05,
      "loss": 3.0004,
      "step": 1107
    },
    {
      "epoch": 0.029762544321478457,
      "grad_norm": 1.4431809186935425,
      "learning_rate": 9.999999924953403e-05,
      "loss": 2.85,
      "step": 1108
    },
    {
      "epoch": 0.029789405823573655,
      "grad_norm": 1.3787111043930054,
      "learning_rate": 9.999999924817265e-05,
      "loss": 2.9795,
      "step": 1109
    },
    {
      "epoch": 0.02981626732566885,
      "grad_norm": 1.3092586994171143,
      "learning_rate": 9.999999924681003e-05,
      "loss": 3.0196,
      "step": 1110
    },
    {
      "epoch": 0.02984312882776405,
      "grad_norm": 1.555515170097351,
      "learning_rate": 9.999999924544617e-05,
      "loss": 2.9242,
      "step": 1111
    },
    {
      "epoch": 0.029869990329859247,
      "grad_norm": 1.4019018411636353,
      "learning_rate": 9.999999924408108e-05,
      "loss": 3.1464,
      "step": 1112
    },
    {
      "epoch": 0.02989685183195444,
      "grad_norm": 1.4033538103103638,
      "learning_rate": 9.999999924271475e-05,
      "loss": 3.0895,
      "step": 1113
    },
    {
      "epoch": 0.02992371333404964,
      "grad_norm": 1.3757456541061401,
      "learning_rate": 9.99999992413472e-05,
      "loss": 2.8177,
      "step": 1114
    },
    {
      "epoch": 0.029950574836144838,
      "grad_norm": 1.557436227798462,
      "learning_rate": 9.99999992399784e-05,
      "loss": 3.0616,
      "step": 1115
    },
    {
      "epoch": 0.029977436338240033,
      "grad_norm": 1.4687376022338867,
      "learning_rate": 9.999999923860838e-05,
      "loss": 3.2204,
      "step": 1116
    },
    {
      "epoch": 0.03000429784033523,
      "grad_norm": 1.4528030157089233,
      "learning_rate": 9.999999923723712e-05,
      "loss": 3.104,
      "step": 1117
    },
    {
      "epoch": 0.03003115934243043,
      "grad_norm": 1.4383186101913452,
      "learning_rate": 9.999999923586463e-05,
      "loss": 3.0886,
      "step": 1118
    },
    {
      "epoch": 0.030058020844525624,
      "grad_norm": 1.5394346714019775,
      "learning_rate": 9.99999992344909e-05,
      "loss": 3.123,
      "step": 1119
    },
    {
      "epoch": 0.030084882346620823,
      "grad_norm": 1.4198474884033203,
      "learning_rate": 9.999999923311594e-05,
      "loss": 2.8897,
      "step": 1120
    },
    {
      "epoch": 0.03011174384871602,
      "grad_norm": 1.4363526105880737,
      "learning_rate": 9.999999923173975e-05,
      "loss": 2.726,
      "step": 1121
    },
    {
      "epoch": 0.030138605350811216,
      "grad_norm": 1.4953033924102783,
      "learning_rate": 9.999999923036231e-05,
      "loss": 2.732,
      "step": 1122
    },
    {
      "epoch": 0.030165466852906414,
      "grad_norm": 1.4169050455093384,
      "learning_rate": 9.999999922898367e-05,
      "loss": 3.1504,
      "step": 1123
    },
    {
      "epoch": 0.030192328355001612,
      "grad_norm": 1.4073117971420288,
      "learning_rate": 9.999999922760377e-05,
      "loss": 2.8981,
      "step": 1124
    },
    {
      "epoch": 0.030219189857096807,
      "grad_norm": 1.4770647287368774,
      "learning_rate": 9.999999922622263e-05,
      "loss": 3.1128,
      "step": 1125
    },
    {
      "epoch": 0.030246051359192005,
      "grad_norm": 1.5720592737197876,
      "learning_rate": 9.999999922484027e-05,
      "loss": 3.1258,
      "step": 1126
    },
    {
      "epoch": 0.030272912861287204,
      "grad_norm": 1.4397382736206055,
      "learning_rate": 9.999999922345668e-05,
      "loss": 2.8535,
      "step": 1127
    },
    {
      "epoch": 0.0302997743633824,
      "grad_norm": 1.7102172374725342,
      "learning_rate": 9.999999922207185e-05,
      "loss": 3.1901,
      "step": 1128
    },
    {
      "epoch": 0.030326635865477597,
      "grad_norm": 1.5595030784606934,
      "learning_rate": 9.99999992206858e-05,
      "loss": 3.0476,
      "step": 1129
    },
    {
      "epoch": 0.030353497367572795,
      "grad_norm": 1.5575852394104004,
      "learning_rate": 9.999999921929848e-05,
      "loss": 3.0828,
      "step": 1130
    },
    {
      "epoch": 0.030380358869667994,
      "grad_norm": 1.4322677850723267,
      "learning_rate": 9.999999921790995e-05,
      "loss": 3.2307,
      "step": 1131
    },
    {
      "epoch": 0.03040722037176319,
      "grad_norm": 1.57375967502594,
      "learning_rate": 9.99999992165202e-05,
      "loss": 2.9122,
      "step": 1132
    },
    {
      "epoch": 0.030434081873858387,
      "grad_norm": 1.4671905040740967,
      "learning_rate": 9.999999921512919e-05,
      "loss": 2.8622,
      "step": 1133
    },
    {
      "epoch": 0.030460943375953585,
      "grad_norm": 1.4324138164520264,
      "learning_rate": 9.999999921373696e-05,
      "loss": 2.8266,
      "step": 1134
    },
    {
      "epoch": 0.03048780487804878,
      "grad_norm": 1.4229416847229004,
      "learning_rate": 9.99999992123435e-05,
      "loss": 2.8596,
      "step": 1135
    },
    {
      "epoch": 0.030514666380143978,
      "grad_norm": 1.6876296997070312,
      "learning_rate": 9.999999921094881e-05,
      "loss": 3.2123,
      "step": 1136
    },
    {
      "epoch": 0.030541527882239176,
      "grad_norm": 1.5105502605438232,
      "learning_rate": 9.999999920955287e-05,
      "loss": 3.0074,
      "step": 1137
    },
    {
      "epoch": 0.03056838938433437,
      "grad_norm": 1.645257830619812,
      "learning_rate": 9.99999992081557e-05,
      "loss": 3.1036,
      "step": 1138
    },
    {
      "epoch": 0.03059525088642957,
      "grad_norm": 1.5362334251403809,
      "learning_rate": 9.99999992067573e-05,
      "loss": 3.2537,
      "step": 1139
    },
    {
      "epoch": 0.030622112388524768,
      "grad_norm": 1.575076699256897,
      "learning_rate": 9.999999920535766e-05,
      "loss": 3.1689,
      "step": 1140
    },
    {
      "epoch": 0.030648973890619963,
      "grad_norm": 1.5552635192871094,
      "learning_rate": 9.99999992039568e-05,
      "loss": 3.0966,
      "step": 1141
    },
    {
      "epoch": 0.03067583539271516,
      "grad_norm": 1.720339059829712,
      "learning_rate": 9.99999992025547e-05,
      "loss": 3.1505,
      "step": 1142
    },
    {
      "epoch": 0.03070269689481036,
      "grad_norm": 1.7185444831848145,
      "learning_rate": 9.999999920115136e-05,
      "loss": 3.1868,
      "step": 1143
    },
    {
      "epoch": 0.030729558396905554,
      "grad_norm": 1.6548991203308105,
      "learning_rate": 9.999999919974679e-05,
      "loss": 3.1969,
      "step": 1144
    },
    {
      "epoch": 0.030756419899000752,
      "grad_norm": 1.5805326700210571,
      "learning_rate": 9.999999919834099e-05,
      "loss": 3.292,
      "step": 1145
    },
    {
      "epoch": 0.03078328140109595,
      "grad_norm": 1.8295358419418335,
      "learning_rate": 9.999999919693395e-05,
      "loss": 3.3623,
      "step": 1146
    },
    {
      "epoch": 0.030810142903191146,
      "grad_norm": 1.737511157989502,
      "learning_rate": 9.999999919552568e-05,
      "loss": 2.9244,
      "step": 1147
    },
    {
      "epoch": 0.030837004405286344,
      "grad_norm": 1.8514723777770996,
      "learning_rate": 9.999999919411618e-05,
      "loss": 3.5062,
      "step": 1148
    },
    {
      "epoch": 0.030863865907381542,
      "grad_norm": 1.9149837493896484,
      "learning_rate": 9.999999919270544e-05,
      "loss": 3.6376,
      "step": 1149
    },
    {
      "epoch": 0.030890727409476737,
      "grad_norm": 1.8508201837539673,
      "learning_rate": 9.999999919129347e-05,
      "loss": 3.1467,
      "step": 1150
    },
    {
      "epoch": 0.030917588911571935,
      "grad_norm": 1.4321824312210083,
      "learning_rate": 9.999999918988027e-05,
      "loss": 3.0066,
      "step": 1151
    },
    {
      "epoch": 0.030944450413667134,
      "grad_norm": 1.5123839378356934,
      "learning_rate": 9.999999918846583e-05,
      "loss": 2.9124,
      "step": 1152
    },
    {
      "epoch": 0.03097131191576233,
      "grad_norm": 1.4272949695587158,
      "learning_rate": 9.999999918705016e-05,
      "loss": 2.8225,
      "step": 1153
    },
    {
      "epoch": 0.030998173417857527,
      "grad_norm": 1.405633568763733,
      "learning_rate": 9.999999918563326e-05,
      "loss": 2.7074,
      "step": 1154
    },
    {
      "epoch": 0.031025034919952725,
      "grad_norm": 1.487655520439148,
      "learning_rate": 9.99999991842151e-05,
      "loss": 3.0166,
      "step": 1155
    },
    {
      "epoch": 0.03105189642204792,
      "grad_norm": 1.4216954708099365,
      "learning_rate": 9.999999918279575e-05,
      "loss": 3.0085,
      "step": 1156
    },
    {
      "epoch": 0.031078757924143118,
      "grad_norm": 1.403308629989624,
      "learning_rate": 9.999999918137512e-05,
      "loss": 3.0478,
      "step": 1157
    },
    {
      "epoch": 0.031105619426238316,
      "grad_norm": 1.4875620603561401,
      "learning_rate": 9.999999917995328e-05,
      "loss": 2.9702,
      "step": 1158
    },
    {
      "epoch": 0.03113248092833351,
      "grad_norm": 1.4649858474731445,
      "learning_rate": 9.999999917853023e-05,
      "loss": 3.2984,
      "step": 1159
    },
    {
      "epoch": 0.03115934243042871,
      "grad_norm": 1.4901381731033325,
      "learning_rate": 9.999999917710591e-05,
      "loss": 3.0888,
      "step": 1160
    },
    {
      "epoch": 0.031186203932523908,
      "grad_norm": 1.4207940101623535,
      "learning_rate": 9.999999917568037e-05,
      "loss": 2.7057,
      "step": 1161
    },
    {
      "epoch": 0.031213065434619103,
      "grad_norm": 1.3899128437042236,
      "learning_rate": 9.999999917425359e-05,
      "loss": 2.9937,
      "step": 1162
    },
    {
      "epoch": 0.0312399269367143,
      "grad_norm": 1.36077082157135,
      "learning_rate": 9.999999917282559e-05,
      "loss": 2.7426,
      "step": 1163
    },
    {
      "epoch": 0.0312667884388095,
      "grad_norm": 1.2691547870635986,
      "learning_rate": 9.999999917139633e-05,
      "loss": 2.6888,
      "step": 1164
    },
    {
      "epoch": 0.031293649940904694,
      "grad_norm": 1.309258222579956,
      "learning_rate": 9.999999916996586e-05,
      "loss": 2.7449,
      "step": 1165
    },
    {
      "epoch": 0.031320511442999896,
      "grad_norm": 1.3002082109451294,
      "learning_rate": 9.999999916853415e-05,
      "loss": 2.8566,
      "step": 1166
    },
    {
      "epoch": 0.03134737294509509,
      "grad_norm": 1.4026422500610352,
      "learning_rate": 9.99999991671012e-05,
      "loss": 3.1578,
      "step": 1167
    },
    {
      "epoch": 0.031374234447190286,
      "grad_norm": 1.3621559143066406,
      "learning_rate": 9.999999916566702e-05,
      "loss": 2.9677,
      "step": 1168
    },
    {
      "epoch": 0.03140109594928549,
      "grad_norm": 1.408431887626648,
      "learning_rate": 9.999999916423162e-05,
      "loss": 2.8851,
      "step": 1169
    },
    {
      "epoch": 0.03142795745138068,
      "grad_norm": 1.5579081773757935,
      "learning_rate": 9.999999916279497e-05,
      "loss": 3.2076,
      "step": 1170
    },
    {
      "epoch": 0.03145481895347588,
      "grad_norm": 1.422640085220337,
      "learning_rate": 9.999999916135709e-05,
      "loss": 2.9792,
      "step": 1171
    },
    {
      "epoch": 0.03148168045557108,
      "grad_norm": 1.508641004562378,
      "learning_rate": 9.999999915991797e-05,
      "loss": 3.1418,
      "step": 1172
    },
    {
      "epoch": 0.031508541957666274,
      "grad_norm": 1.4941315650939941,
      "learning_rate": 9.999999915847764e-05,
      "loss": 2.9193,
      "step": 1173
    },
    {
      "epoch": 0.03153540345976147,
      "grad_norm": 1.4279118776321411,
      "learning_rate": 9.999999915703607e-05,
      "loss": 2.7299,
      "step": 1174
    },
    {
      "epoch": 0.03156226496185667,
      "grad_norm": 1.6342811584472656,
      "learning_rate": 9.999999915559324e-05,
      "loss": 3.1708,
      "step": 1175
    },
    {
      "epoch": 0.031589126463951865,
      "grad_norm": 1.4493217468261719,
      "learning_rate": 9.999999915414919e-05,
      "loss": 2.9587,
      "step": 1176
    },
    {
      "epoch": 0.03161598796604706,
      "grad_norm": 1.4111332893371582,
      "learning_rate": 9.999999915270392e-05,
      "loss": 2.8108,
      "step": 1177
    },
    {
      "epoch": 0.03164284946814226,
      "grad_norm": 1.5913037061691284,
      "learning_rate": 9.999999915125739e-05,
      "loss": 2.9306,
      "step": 1178
    },
    {
      "epoch": 0.03166971097023746,
      "grad_norm": 1.4138201475143433,
      "learning_rate": 9.999999914980966e-05,
      "loss": 2.9746,
      "step": 1179
    },
    {
      "epoch": 0.03169657247233265,
      "grad_norm": 1.542961835861206,
      "learning_rate": 9.999999914836066e-05,
      "loss": 3.1946,
      "step": 1180
    },
    {
      "epoch": 0.03172343397442785,
      "grad_norm": 1.4823840856552124,
      "learning_rate": 9.999999914691045e-05,
      "loss": 2.7606,
      "step": 1181
    },
    {
      "epoch": 0.03175029547652305,
      "grad_norm": 1.477311611175537,
      "learning_rate": 9.999999914545901e-05,
      "loss": 3.1212,
      "step": 1182
    },
    {
      "epoch": 0.03177715697861824,
      "grad_norm": 1.6379215717315674,
      "learning_rate": 9.999999914400633e-05,
      "loss": 2.98,
      "step": 1183
    },
    {
      "epoch": 0.031804018480713445,
      "grad_norm": 1.4688607454299927,
      "learning_rate": 9.99999991425524e-05,
      "loss": 2.9571,
      "step": 1184
    },
    {
      "epoch": 0.03183087998280864,
      "grad_norm": 1.4939223527908325,
      "learning_rate": 9.999999914109725e-05,
      "loss": 3.0431,
      "step": 1185
    },
    {
      "epoch": 0.031857741484903834,
      "grad_norm": 1.5596034526824951,
      "learning_rate": 9.999999913964087e-05,
      "loss": 3.1911,
      "step": 1186
    },
    {
      "epoch": 0.031884602986999036,
      "grad_norm": 1.693818211555481,
      "learning_rate": 9.999999913818326e-05,
      "loss": 3.0475,
      "step": 1187
    },
    {
      "epoch": 0.03191146448909423,
      "grad_norm": 1.6540929079055786,
      "learning_rate": 9.99999991367244e-05,
      "loss": 3.1208,
      "step": 1188
    },
    {
      "epoch": 0.031938325991189426,
      "grad_norm": 1.4901649951934814,
      "learning_rate": 9.999999913526433e-05,
      "loss": 2.9145,
      "step": 1189
    },
    {
      "epoch": 0.03196518749328463,
      "grad_norm": 1.585123896598816,
      "learning_rate": 9.9999999133803e-05,
      "loss": 3.1728,
      "step": 1190
    },
    {
      "epoch": 0.03199204899537982,
      "grad_norm": 1.743322730064392,
      "learning_rate": 9.999999913234044e-05,
      "loss": 3.4658,
      "step": 1191
    },
    {
      "epoch": 0.03201891049747502,
      "grad_norm": 1.6741468906402588,
      "learning_rate": 9.999999913087666e-05,
      "loss": 3.4217,
      "step": 1192
    },
    {
      "epoch": 0.03204577199957022,
      "grad_norm": 1.5321619510650635,
      "learning_rate": 9.999999912941164e-05,
      "loss": 3.0894,
      "step": 1193
    },
    {
      "epoch": 0.032072633501665414,
      "grad_norm": 1.6068675518035889,
      "learning_rate": 9.999999912794538e-05,
      "loss": 2.8289,
      "step": 1194
    },
    {
      "epoch": 0.03209949500376061,
      "grad_norm": 1.7000651359558105,
      "learning_rate": 9.99999991264779e-05,
      "loss": 3.2203,
      "step": 1195
    },
    {
      "epoch": 0.03212635650585581,
      "grad_norm": 1.652759313583374,
      "learning_rate": 9.999999912500917e-05,
      "loss": 3.2407,
      "step": 1196
    },
    {
      "epoch": 0.032153218007951005,
      "grad_norm": 1.8210004568099976,
      "learning_rate": 9.999999912353921e-05,
      "loss": 3.2567,
      "step": 1197
    },
    {
      "epoch": 0.0321800795100462,
      "grad_norm": 1.9404109716415405,
      "learning_rate": 9.999999912206803e-05,
      "loss": 3.3876,
      "step": 1198
    },
    {
      "epoch": 0.0322069410121414,
      "grad_norm": 1.9805941581726074,
      "learning_rate": 9.999999912059561e-05,
      "loss": 3.4857,
      "step": 1199
    },
    {
      "epoch": 0.0322338025142366,
      "grad_norm": 1.789170742034912,
      "learning_rate": 9.999999911912195e-05,
      "loss": 3.3542,
      "step": 1200
    },
    {
      "epoch": 0.03226066401633179,
      "grad_norm": 1.601011872291565,
      "learning_rate": 9.999999911764706e-05,
      "loss": 3.0094,
      "step": 1201
    },
    {
      "epoch": 0.03228752551842699,
      "grad_norm": 1.5233467817306519,
      "learning_rate": 9.999999911617094e-05,
      "loss": 2.9386,
      "step": 1202
    },
    {
      "epoch": 0.03231438702052219,
      "grad_norm": 1.4733026027679443,
      "learning_rate": 9.999999911469358e-05,
      "loss": 2.6907,
      "step": 1203
    },
    {
      "epoch": 0.03234124852261738,
      "grad_norm": 1.5229771137237549,
      "learning_rate": 9.999999911321499e-05,
      "loss": 2.9349,
      "step": 1204
    },
    {
      "epoch": 0.032368110024712585,
      "grad_norm": 1.4411981105804443,
      "learning_rate": 9.999999911173516e-05,
      "loss": 2.7616,
      "step": 1205
    },
    {
      "epoch": 0.03239497152680778,
      "grad_norm": 1.3132145404815674,
      "learning_rate": 9.99999991102541e-05,
      "loss": 2.7224,
      "step": 1206
    },
    {
      "epoch": 0.032421833028902974,
      "grad_norm": 1.500889778137207,
      "learning_rate": 9.999999910877182e-05,
      "loss": 3.0192,
      "step": 1207
    },
    {
      "epoch": 0.032448694530998176,
      "grad_norm": 1.2993311882019043,
      "learning_rate": 9.999999910728829e-05,
      "loss": 2.7571,
      "step": 1208
    },
    {
      "epoch": 0.03247555603309337,
      "grad_norm": 1.5036752223968506,
      "learning_rate": 9.999999910580353e-05,
      "loss": 3.0489,
      "step": 1209
    },
    {
      "epoch": 0.032502417535188566,
      "grad_norm": 1.3930540084838867,
      "learning_rate": 9.999999910431754e-05,
      "loss": 2.6477,
      "step": 1210
    },
    {
      "epoch": 0.03252927903728377,
      "grad_norm": 1.3620692491531372,
      "learning_rate": 9.999999910283031e-05,
      "loss": 2.7436,
      "step": 1211
    },
    {
      "epoch": 0.03255614053937896,
      "grad_norm": 1.5991593599319458,
      "learning_rate": 9.999999910134186e-05,
      "loss": 2.979,
      "step": 1212
    },
    {
      "epoch": 0.03258300204147416,
      "grad_norm": 1.58819580078125,
      "learning_rate": 9.999999909985216e-05,
      "loss": 3.0439,
      "step": 1213
    },
    {
      "epoch": 0.03260986354356936,
      "grad_norm": 1.4783276319503784,
      "learning_rate": 9.999999909836123e-05,
      "loss": 2.8303,
      "step": 1214
    },
    {
      "epoch": 0.032636725045664554,
      "grad_norm": 1.3597859144210815,
      "learning_rate": 9.999999909686907e-05,
      "loss": 2.6549,
      "step": 1215
    },
    {
      "epoch": 0.03266358654775975,
      "grad_norm": 1.3881914615631104,
      "learning_rate": 9.999999909537567e-05,
      "loss": 2.9567,
      "step": 1216
    },
    {
      "epoch": 0.03269044804985495,
      "grad_norm": 1.3996658325195312,
      "learning_rate": 9.999999909388105e-05,
      "loss": 2.7215,
      "step": 1217
    },
    {
      "epoch": 0.032717309551950145,
      "grad_norm": 1.3187663555145264,
      "learning_rate": 9.999999909238517e-05,
      "loss": 2.8384,
      "step": 1218
    },
    {
      "epoch": 0.03274417105404534,
      "grad_norm": 1.509340763092041,
      "learning_rate": 9.999999909088808e-05,
      "loss": 2.9029,
      "step": 1219
    },
    {
      "epoch": 0.03277103255614054,
      "grad_norm": 1.4753671884536743,
      "learning_rate": 9.999999908938975e-05,
      "loss": 3.2016,
      "step": 1220
    },
    {
      "epoch": 0.03279789405823574,
      "grad_norm": 1.4223262071609497,
      "learning_rate": 9.99999990878902e-05,
      "loss": 2.8779,
      "step": 1221
    },
    {
      "epoch": 0.03282475556033093,
      "grad_norm": 1.3905526399612427,
      "learning_rate": 9.99999990863894e-05,
      "loss": 2.8502,
      "step": 1222
    },
    {
      "epoch": 0.03285161706242613,
      "grad_norm": 1.6468901634216309,
      "learning_rate": 9.999999908488736e-05,
      "loss": 3.287,
      "step": 1223
    },
    {
      "epoch": 0.03287847856452133,
      "grad_norm": 1.385778546333313,
      "learning_rate": 9.99999990833841e-05,
      "loss": 2.8047,
      "step": 1224
    },
    {
      "epoch": 0.03290534006661652,
      "grad_norm": 1.3317996263504028,
      "learning_rate": 9.999999908187961e-05,
      "loss": 2.9318,
      "step": 1225
    },
    {
      "epoch": 0.032932201568711725,
      "grad_norm": 1.4120289087295532,
      "learning_rate": 9.999999908037387e-05,
      "loss": 3.1235,
      "step": 1226
    },
    {
      "epoch": 0.03295906307080692,
      "grad_norm": 1.570051670074463,
      "learning_rate": 9.999999907886691e-05,
      "loss": 3.1125,
      "step": 1227
    },
    {
      "epoch": 0.032985924572902114,
      "grad_norm": 1.5419656038284302,
      "learning_rate": 9.999999907735871e-05,
      "loss": 2.9827,
      "step": 1228
    },
    {
      "epoch": 0.033012786074997316,
      "grad_norm": 1.5014476776123047,
      "learning_rate": 9.999999907584927e-05,
      "loss": 3.1199,
      "step": 1229
    },
    {
      "epoch": 0.03303964757709251,
      "grad_norm": 1.653408169746399,
      "learning_rate": 9.99999990743386e-05,
      "loss": 3.2555,
      "step": 1230
    },
    {
      "epoch": 0.033066509079187706,
      "grad_norm": 1.3667315244674683,
      "learning_rate": 9.99999990728267e-05,
      "loss": 2.9395,
      "step": 1231
    },
    {
      "epoch": 0.03309337058128291,
      "grad_norm": 1.576900839805603,
      "learning_rate": 9.999999907131357e-05,
      "loss": 3.0171,
      "step": 1232
    },
    {
      "epoch": 0.0331202320833781,
      "grad_norm": 1.6688435077667236,
      "learning_rate": 9.99999990697992e-05,
      "loss": 3.283,
      "step": 1233
    },
    {
      "epoch": 0.0331470935854733,
      "grad_norm": 1.5353676080703735,
      "learning_rate": 9.999999906828359e-05,
      "loss": 3.0857,
      "step": 1234
    },
    {
      "epoch": 0.0331739550875685,
      "grad_norm": 1.4716721773147583,
      "learning_rate": 9.999999906676676e-05,
      "loss": 2.8933,
      "step": 1235
    },
    {
      "epoch": 0.033200816589663694,
      "grad_norm": 1.624703288078308,
      "learning_rate": 9.999999906524868e-05,
      "loss": 3.1667,
      "step": 1236
    },
    {
      "epoch": 0.03322767809175889,
      "grad_norm": 1.4201050996780396,
      "learning_rate": 9.999999906372939e-05,
      "loss": 3.0671,
      "step": 1237
    },
    {
      "epoch": 0.03325453959385409,
      "grad_norm": 1.5696899890899658,
      "learning_rate": 9.999999906220885e-05,
      "loss": 2.8465,
      "step": 1238
    },
    {
      "epoch": 0.033281401095949285,
      "grad_norm": 1.5536811351776123,
      "learning_rate": 9.999999906068709e-05,
      "loss": 3.101,
      "step": 1239
    },
    {
      "epoch": 0.03330826259804448,
      "grad_norm": 1.5983662605285645,
      "learning_rate": 9.999999905916408e-05,
      "loss": 3.286,
      "step": 1240
    },
    {
      "epoch": 0.03333512410013968,
      "grad_norm": 1.7818171977996826,
      "learning_rate": 9.999999905763984e-05,
      "loss": 3.3457,
      "step": 1241
    },
    {
      "epoch": 0.03336198560223488,
      "grad_norm": 1.4938325881958008,
      "learning_rate": 9.999999905611437e-05,
      "loss": 3.1784,
      "step": 1242
    },
    {
      "epoch": 0.03338884710433007,
      "grad_norm": 1.4970808029174805,
      "learning_rate": 9.999999905458765e-05,
      "loss": 3.1908,
      "step": 1243
    },
    {
      "epoch": 0.03341570860642527,
      "grad_norm": 1.6312114000320435,
      "learning_rate": 9.999999905305972e-05,
      "loss": 3.2914,
      "step": 1244
    },
    {
      "epoch": 0.03344257010852047,
      "grad_norm": 1.6739710569381714,
      "learning_rate": 9.999999905153055e-05,
      "loss": 3.2133,
      "step": 1245
    },
    {
      "epoch": 0.03346943161061566,
      "grad_norm": 1.586802363395691,
      "learning_rate": 9.999999905000015e-05,
      "loss": 3.1242,
      "step": 1246
    },
    {
      "epoch": 0.033496293112710865,
      "grad_norm": 1.670663595199585,
      "learning_rate": 9.999999904846851e-05,
      "loss": 3.2219,
      "step": 1247
    },
    {
      "epoch": 0.03352315461480606,
      "grad_norm": 1.7760072946548462,
      "learning_rate": 9.999999904693564e-05,
      "loss": 3.0606,
      "step": 1248
    },
    {
      "epoch": 0.033550016116901255,
      "grad_norm": 2.1186087131500244,
      "learning_rate": 9.999999904540152e-05,
      "loss": 3.6282,
      "step": 1249
    },
    {
      "epoch": 0.033576877618996456,
      "grad_norm": 1.7062263488769531,
      "learning_rate": 9.999999904386619e-05,
      "loss": 3.3226,
      "step": 1250
    },
    {
      "epoch": 0.03360373912109165,
      "grad_norm": 1.3062018156051636,
      "learning_rate": 9.99999990423296e-05,
      "loss": 2.91,
      "step": 1251
    },
    {
      "epoch": 0.033630600623186846,
      "grad_norm": 1.5175307989120483,
      "learning_rate": 9.99999990407918e-05,
      "loss": 2.8442,
      "step": 1252
    },
    {
      "epoch": 0.03365746212528205,
      "grad_norm": 1.4948550462722778,
      "learning_rate": 9.999999903925276e-05,
      "loss": 3.1676,
      "step": 1253
    },
    {
      "epoch": 0.03368432362737724,
      "grad_norm": 1.4918733835220337,
      "learning_rate": 9.999999903771249e-05,
      "loss": 2.6735,
      "step": 1254
    },
    {
      "epoch": 0.03371118512947244,
      "grad_norm": 1.4677417278289795,
      "learning_rate": 9.999999903617097e-05,
      "loss": 3.114,
      "step": 1255
    },
    {
      "epoch": 0.03373804663156764,
      "grad_norm": 1.4189213514328003,
      "learning_rate": 9.999999903462822e-05,
      "loss": 3.101,
      "step": 1256
    },
    {
      "epoch": 0.033764908133662834,
      "grad_norm": 1.2749756574630737,
      "learning_rate": 9.999999903308426e-05,
      "loss": 2.7508,
      "step": 1257
    },
    {
      "epoch": 0.03379176963575803,
      "grad_norm": 1.621442437171936,
      "learning_rate": 9.999999903153904e-05,
      "loss": 3.0446,
      "step": 1258
    },
    {
      "epoch": 0.03381863113785323,
      "grad_norm": 1.3713841438293457,
      "learning_rate": 9.999999902999259e-05,
      "loss": 2.8511,
      "step": 1259
    },
    {
      "epoch": 0.033845492639948425,
      "grad_norm": 1.3999048471450806,
      "learning_rate": 9.999999902844492e-05,
      "loss": 3.0518,
      "step": 1260
    },
    {
      "epoch": 0.03387235414204362,
      "grad_norm": 1.4194591045379639,
      "learning_rate": 9.9999999026896e-05,
      "loss": 3.0724,
      "step": 1261
    },
    {
      "epoch": 0.03389921564413882,
      "grad_norm": 1.4514492750167847,
      "learning_rate": 9.999999902534586e-05,
      "loss": 3.3577,
      "step": 1262
    },
    {
      "epoch": 0.03392607714623402,
      "grad_norm": 1.4304295778274536,
      "learning_rate": 9.999999902379449e-05,
      "loss": 2.897,
      "step": 1263
    },
    {
      "epoch": 0.03395293864832921,
      "grad_norm": 1.552476406097412,
      "learning_rate": 9.999999902224186e-05,
      "loss": 3.0367,
      "step": 1264
    },
    {
      "epoch": 0.033979800150424413,
      "grad_norm": 1.3050061464309692,
      "learning_rate": 9.999999902068802e-05,
      "loss": 2.7819,
      "step": 1265
    },
    {
      "epoch": 0.03400666165251961,
      "grad_norm": 1.317513346672058,
      "learning_rate": 9.999999901913294e-05,
      "loss": 2.7208,
      "step": 1266
    },
    {
      "epoch": 0.0340335231546148,
      "grad_norm": 1.4023017883300781,
      "learning_rate": 9.999999901757663e-05,
      "loss": 2.8803,
      "step": 1267
    },
    {
      "epoch": 0.034060384656710005,
      "grad_norm": 1.3437764644622803,
      "learning_rate": 9.999999901601908e-05,
      "loss": 2.7362,
      "step": 1268
    },
    {
      "epoch": 0.0340872461588052,
      "grad_norm": 1.3950257301330566,
      "learning_rate": 9.99999990144603e-05,
      "loss": 2.9147,
      "step": 1269
    },
    {
      "epoch": 0.034114107660900395,
      "grad_norm": 1.3344415426254272,
      "learning_rate": 9.999999901290028e-05,
      "loss": 2.9881,
      "step": 1270
    },
    {
      "epoch": 0.034140969162995596,
      "grad_norm": 1.3639448881149292,
      "learning_rate": 9.999999901133903e-05,
      "loss": 3.0179,
      "step": 1271
    },
    {
      "epoch": 0.03416783066509079,
      "grad_norm": 1.5759127140045166,
      "learning_rate": 9.999999900977654e-05,
      "loss": 3.1648,
      "step": 1272
    },
    {
      "epoch": 0.034194692167185986,
      "grad_norm": 1.438621163368225,
      "learning_rate": 9.999999900821284e-05,
      "loss": 3.3035,
      "step": 1273
    },
    {
      "epoch": 0.03422155366928119,
      "grad_norm": 1.6033320426940918,
      "learning_rate": 9.999999900664789e-05,
      "loss": 3.0717,
      "step": 1274
    },
    {
      "epoch": 0.03424841517137638,
      "grad_norm": 1.4944597482681274,
      "learning_rate": 9.99999990050817e-05,
      "loss": 3.1131,
      "step": 1275
    },
    {
      "epoch": 0.03427527667347158,
      "grad_norm": 1.5210902690887451,
      "learning_rate": 9.999999900351429e-05,
      "loss": 2.9868,
      "step": 1276
    },
    {
      "epoch": 0.03430213817556678,
      "grad_norm": 1.703231930732727,
      "learning_rate": 9.999999900194564e-05,
      "loss": 3.0296,
      "step": 1277
    },
    {
      "epoch": 0.034328999677661974,
      "grad_norm": 1.466875433921814,
      "learning_rate": 9.999999900037575e-05,
      "loss": 2.7917,
      "step": 1278
    },
    {
      "epoch": 0.03435586117975717,
      "grad_norm": 1.3921716213226318,
      "learning_rate": 9.999999899880464e-05,
      "loss": 2.8462,
      "step": 1279
    },
    {
      "epoch": 0.03438272268185237,
      "grad_norm": 1.496484637260437,
      "learning_rate": 9.999999899723228e-05,
      "loss": 2.8478,
      "step": 1280
    },
    {
      "epoch": 0.034409584183947566,
      "grad_norm": 1.4175546169281006,
      "learning_rate": 9.99999989956587e-05,
      "loss": 3.0167,
      "step": 1281
    },
    {
      "epoch": 0.03443644568604276,
      "grad_norm": 1.5685076713562012,
      "learning_rate": 9.999999899408387e-05,
      "loss": 2.8429,
      "step": 1282
    },
    {
      "epoch": 0.03446330718813796,
      "grad_norm": 1.6189390420913696,
      "learning_rate": 9.999999899250783e-05,
      "loss": 3.1261,
      "step": 1283
    },
    {
      "epoch": 0.03449016869023316,
      "grad_norm": 1.6355414390563965,
      "learning_rate": 9.999999899093054e-05,
      "loss": 3.1554,
      "step": 1284
    },
    {
      "epoch": 0.03451703019232835,
      "grad_norm": 1.6216799020767212,
      "learning_rate": 9.999999898935202e-05,
      "loss": 3.1108,
      "step": 1285
    },
    {
      "epoch": 0.034543891694423554,
      "grad_norm": 1.700219988822937,
      "learning_rate": 9.999999898777225e-05,
      "loss": 3.1269,
      "step": 1286
    },
    {
      "epoch": 0.03457075319651875,
      "grad_norm": 1.6669987440109253,
      "learning_rate": 9.999999898619127e-05,
      "loss": 3.1803,
      "step": 1287
    },
    {
      "epoch": 0.03459761469861394,
      "grad_norm": 1.504661202430725,
      "learning_rate": 9.999999898460905e-05,
      "loss": 3.1455,
      "step": 1288
    },
    {
      "epoch": 0.034624476200709145,
      "grad_norm": 1.5908552408218384,
      "learning_rate": 9.999999898302559e-05,
      "loss": 3.1618,
      "step": 1289
    },
    {
      "epoch": 0.03465133770280434,
      "grad_norm": 1.505043864250183,
      "learning_rate": 9.999999898144091e-05,
      "loss": 3.1347,
      "step": 1290
    },
    {
      "epoch": 0.034678199204899535,
      "grad_norm": 1.5669896602630615,
      "learning_rate": 9.999999897985498e-05,
      "loss": 3.2465,
      "step": 1291
    },
    {
      "epoch": 0.034705060706994736,
      "grad_norm": 1.8743952512741089,
      "learning_rate": 9.999999897826781e-05,
      "loss": 3.258,
      "step": 1292
    },
    {
      "epoch": 0.03473192220908993,
      "grad_norm": 1.6266064643859863,
      "learning_rate": 9.999999897667943e-05,
      "loss": 3.3005,
      "step": 1293
    },
    {
      "epoch": 0.034758783711185126,
      "grad_norm": 1.530107021331787,
      "learning_rate": 9.99999989750898e-05,
      "loss": 3.2074,
      "step": 1294
    },
    {
      "epoch": 0.03478564521328033,
      "grad_norm": 1.5905359983444214,
      "learning_rate": 9.999999897349895e-05,
      "loss": 3.1991,
      "step": 1295
    },
    {
      "epoch": 0.03481250671537552,
      "grad_norm": 1.6169246435165405,
      "learning_rate": 9.999999897190687e-05,
      "loss": 3.0985,
      "step": 1296
    },
    {
      "epoch": 0.03483936821747072,
      "grad_norm": 1.626260757446289,
      "learning_rate": 9.999999897031354e-05,
      "loss": 3.258,
      "step": 1297
    },
    {
      "epoch": 0.03486622971956592,
      "grad_norm": 1.6810840368270874,
      "learning_rate": 9.999999896871897e-05,
      "loss": 3.2958,
      "step": 1298
    },
    {
      "epoch": 0.034893091221661114,
      "grad_norm": 1.7200833559036255,
      "learning_rate": 9.999999896712319e-05,
      "loss": 3.2765,
      "step": 1299
    },
    {
      "epoch": 0.03491995272375631,
      "grad_norm": 1.8822802305221558,
      "learning_rate": 9.999999896552616e-05,
      "loss": 3.6315,
      "step": 1300
    },
    {
      "epoch": 0.03494681422585151,
      "grad_norm": 1.3117274045944214,
      "learning_rate": 9.99999989639279e-05,
      "loss": 3.141,
      "step": 1301
    },
    {
      "epoch": 0.034973675727946706,
      "grad_norm": 1.3925999402999878,
      "learning_rate": 9.999999896232841e-05,
      "loss": 2.7824,
      "step": 1302
    },
    {
      "epoch": 0.03500053723004191,
      "grad_norm": 1.2862520217895508,
      "learning_rate": 9.999999896072767e-05,
      "loss": 2.5686,
      "step": 1303
    },
    {
      "epoch": 0.0350273987321371,
      "grad_norm": 1.3466994762420654,
      "learning_rate": 9.999999895912572e-05,
      "loss": 3.005,
      "step": 1304
    },
    {
      "epoch": 0.0350542602342323,
      "grad_norm": 1.3251762390136719,
      "learning_rate": 9.999999895752253e-05,
      "loss": 3.1006,
      "step": 1305
    },
    {
      "epoch": 0.0350811217363275,
      "grad_norm": 1.561327338218689,
      "learning_rate": 9.99999989559181e-05,
      "loss": 2.9555,
      "step": 1306
    },
    {
      "epoch": 0.035107983238422694,
      "grad_norm": 1.4351792335510254,
      "learning_rate": 9.999999895431242e-05,
      "loss": 3.1577,
      "step": 1307
    },
    {
      "epoch": 0.03513484474051789,
      "grad_norm": 1.4175413846969604,
      "learning_rate": 9.999999895270554e-05,
      "loss": 2.973,
      "step": 1308
    },
    {
      "epoch": 0.03516170624261309,
      "grad_norm": 1.4891241788864136,
      "learning_rate": 9.999999895109741e-05,
      "loss": 3.0839,
      "step": 1309
    },
    {
      "epoch": 0.035188567744708285,
      "grad_norm": 1.434227705001831,
      "learning_rate": 9.999999894948803e-05,
      "loss": 3.0451,
      "step": 1310
    },
    {
      "epoch": 0.03521542924680348,
      "grad_norm": 1.4105538129806519,
      "learning_rate": 9.999999894787745e-05,
      "loss": 2.9684,
      "step": 1311
    },
    {
      "epoch": 0.03524229074889868,
      "grad_norm": 1.4166136980056763,
      "learning_rate": 9.999999894626562e-05,
      "loss": 2.9175,
      "step": 1312
    },
    {
      "epoch": 0.035269152250993877,
      "grad_norm": 1.3854029178619385,
      "learning_rate": 9.999999894465254e-05,
      "loss": 2.9375,
      "step": 1313
    },
    {
      "epoch": 0.03529601375308907,
      "grad_norm": 1.3970662355422974,
      "learning_rate": 9.999999894303825e-05,
      "loss": 2.8387,
      "step": 1314
    },
    {
      "epoch": 0.03532287525518427,
      "grad_norm": 1.337108850479126,
      "learning_rate": 9.999999894142272e-05,
      "loss": 2.8672,
      "step": 1315
    },
    {
      "epoch": 0.03534973675727947,
      "grad_norm": 1.4422601461410522,
      "learning_rate": 9.999999893980596e-05,
      "loss": 2.8305,
      "step": 1316
    },
    {
      "epoch": 0.03537659825937466,
      "grad_norm": 1.3969252109527588,
      "learning_rate": 9.999999893818795e-05,
      "loss": 2.903,
      "step": 1317
    },
    {
      "epoch": 0.035403459761469865,
      "grad_norm": 1.3909648656845093,
      "learning_rate": 9.999999893656873e-05,
      "loss": 2.7991,
      "step": 1318
    },
    {
      "epoch": 0.03543032126356506,
      "grad_norm": 1.3882468938827515,
      "learning_rate": 9.999999893494826e-05,
      "loss": 2.9512,
      "step": 1319
    },
    {
      "epoch": 0.035457182765660254,
      "grad_norm": 1.487926721572876,
      "learning_rate": 9.999999893332656e-05,
      "loss": 2.8579,
      "step": 1320
    },
    {
      "epoch": 0.035484044267755456,
      "grad_norm": 1.5172538757324219,
      "learning_rate": 9.999999893170362e-05,
      "loss": 3.0704,
      "step": 1321
    },
    {
      "epoch": 0.03551090576985065,
      "grad_norm": 1.4259088039398193,
      "learning_rate": 9.999999893007946e-05,
      "loss": 2.6893,
      "step": 1322
    },
    {
      "epoch": 0.035537767271945846,
      "grad_norm": 1.4706926345825195,
      "learning_rate": 9.999999892845406e-05,
      "loss": 3.0833,
      "step": 1323
    },
    {
      "epoch": 0.03556462877404105,
      "grad_norm": 1.39505934715271,
      "learning_rate": 9.999999892682742e-05,
      "loss": 2.9461,
      "step": 1324
    },
    {
      "epoch": 0.03559149027613624,
      "grad_norm": 1.4205573797225952,
      "learning_rate": 9.999999892519956e-05,
      "loss": 3.0997,
      "step": 1325
    },
    {
      "epoch": 0.03561835177823144,
      "grad_norm": 1.408349871635437,
      "learning_rate": 9.999999892357045e-05,
      "loss": 3.0815,
      "step": 1326
    },
    {
      "epoch": 0.03564521328032664,
      "grad_norm": 1.4647812843322754,
      "learning_rate": 9.999999892194012e-05,
      "loss": 2.9036,
      "step": 1327
    },
    {
      "epoch": 0.035672074782421834,
      "grad_norm": 1.5610274076461792,
      "learning_rate": 9.999999892030855e-05,
      "loss": 3.0536,
      "step": 1328
    },
    {
      "epoch": 0.03569893628451703,
      "grad_norm": 1.4923373460769653,
      "learning_rate": 9.999999891867574e-05,
      "loss": 3.1588,
      "step": 1329
    },
    {
      "epoch": 0.03572579778661223,
      "grad_norm": 1.4299325942993164,
      "learning_rate": 9.999999891704171e-05,
      "loss": 2.7535,
      "step": 1330
    },
    {
      "epoch": 0.035752659288707425,
      "grad_norm": 1.51185142993927,
      "learning_rate": 9.999999891540644e-05,
      "loss": 3.0361,
      "step": 1331
    },
    {
      "epoch": 0.03577952079080262,
      "grad_norm": 1.6059998273849487,
      "learning_rate": 9.999999891376993e-05,
      "loss": 2.9251,
      "step": 1332
    },
    {
      "epoch": 0.03580638229289782,
      "grad_norm": 1.5953091382980347,
      "learning_rate": 9.999999891213219e-05,
      "loss": 3.2808,
      "step": 1333
    },
    {
      "epoch": 0.03583324379499302,
      "grad_norm": 1.534511685371399,
      "learning_rate": 9.999999891049322e-05,
      "loss": 3.1435,
      "step": 1334
    },
    {
      "epoch": 0.03586010529708821,
      "grad_norm": 1.5457303524017334,
      "learning_rate": 9.999999890885302e-05,
      "loss": 3.3197,
      "step": 1335
    },
    {
      "epoch": 0.03588696679918341,
      "grad_norm": 1.5344340801239014,
      "learning_rate": 9.999999890721156e-05,
      "loss": 3.2265,
      "step": 1336
    },
    {
      "epoch": 0.03591382830127861,
      "grad_norm": 1.5263392925262451,
      "learning_rate": 9.99999989055689e-05,
      "loss": 3.2468,
      "step": 1337
    },
    {
      "epoch": 0.0359406898033738,
      "grad_norm": 1.5331720113754272,
      "learning_rate": 9.999999890392499e-05,
      "loss": 2.9191,
      "step": 1338
    },
    {
      "epoch": 0.035967551305469005,
      "grad_norm": 1.6845088005065918,
      "learning_rate": 9.999999890227985e-05,
      "loss": 3.4195,
      "step": 1339
    },
    {
      "epoch": 0.0359944128075642,
      "grad_norm": 1.6606717109680176,
      "learning_rate": 9.999999890063348e-05,
      "loss": 3.254,
      "step": 1340
    },
    {
      "epoch": 0.036021274309659394,
      "grad_norm": 1.5407532453536987,
      "learning_rate": 9.999999889898587e-05,
      "loss": 3.1018,
      "step": 1341
    },
    {
      "epoch": 0.036048135811754596,
      "grad_norm": 1.6971338987350464,
      "learning_rate": 9.999999889733703e-05,
      "loss": 3.1919,
      "step": 1342
    },
    {
      "epoch": 0.03607499731384979,
      "grad_norm": 1.7509889602661133,
      "learning_rate": 9.999999889568695e-05,
      "loss": 3.4225,
      "step": 1343
    },
    {
      "epoch": 0.036101858815944986,
      "grad_norm": 1.6628385782241821,
      "learning_rate": 9.999999889403564e-05,
      "loss": 3.1497,
      "step": 1344
    },
    {
      "epoch": 0.03612872031804019,
      "grad_norm": 1.778279185295105,
      "learning_rate": 9.999999889238309e-05,
      "loss": 3.6656,
      "step": 1345
    },
    {
      "epoch": 0.03615558182013538,
      "grad_norm": 1.6677271127700806,
      "learning_rate": 9.999999889072933e-05,
      "loss": 3.2719,
      "step": 1346
    },
    {
      "epoch": 0.03618244332223058,
      "grad_norm": 1.6962394714355469,
      "learning_rate": 9.999999888907432e-05,
      "loss": 3.3428,
      "step": 1347
    },
    {
      "epoch": 0.03620930482432578,
      "grad_norm": 1.7792705297470093,
      "learning_rate": 9.999999888741807e-05,
      "loss": 3.5057,
      "step": 1348
    },
    {
      "epoch": 0.036236166326420974,
      "grad_norm": 1.8109219074249268,
      "learning_rate": 9.99999988857606e-05,
      "loss": 3.4714,
      "step": 1349
    },
    {
      "epoch": 0.03626302782851617,
      "grad_norm": 2.049417018890381,
      "learning_rate": 9.999999888410187e-05,
      "loss": 3.4827,
      "step": 1350
    },
    {
      "epoch": 0.03628988933061137,
      "grad_norm": 1.3721641302108765,
      "learning_rate": 9.999999888244193e-05,
      "loss": 3.1125,
      "step": 1351
    },
    {
      "epoch": 0.036316750832706565,
      "grad_norm": 1.4423147439956665,
      "learning_rate": 9.999999888078076e-05,
      "loss": 2.8924,
      "step": 1352
    },
    {
      "epoch": 0.03634361233480176,
      "grad_norm": 1.4216868877410889,
      "learning_rate": 9.999999887911835e-05,
      "loss": 2.8453,
      "step": 1353
    },
    {
      "epoch": 0.03637047383689696,
      "grad_norm": 1.444206953048706,
      "learning_rate": 9.99999988774547e-05,
      "loss": 2.8776,
      "step": 1354
    },
    {
      "epoch": 0.03639733533899216,
      "grad_norm": 1.450270175933838,
      "learning_rate": 9.999999887578981e-05,
      "loss": 2.997,
      "step": 1355
    },
    {
      "epoch": 0.03642419684108735,
      "grad_norm": 1.467962384223938,
      "learning_rate": 9.99999988741237e-05,
      "loss": 3.2055,
      "step": 1356
    },
    {
      "epoch": 0.03645105834318255,
      "grad_norm": 1.3029931783676147,
      "learning_rate": 9.999999887245636e-05,
      "loss": 2.7665,
      "step": 1357
    },
    {
      "epoch": 0.03647791984527775,
      "grad_norm": 1.3137893676757812,
      "learning_rate": 9.999999887078778e-05,
      "loss": 2.974,
      "step": 1358
    },
    {
      "epoch": 0.03650478134737294,
      "grad_norm": 1.6335506439208984,
      "learning_rate": 9.999999886911797e-05,
      "loss": 3.4446,
      "step": 1359
    },
    {
      "epoch": 0.036531642849468145,
      "grad_norm": 1.386962652206421,
      "learning_rate": 9.999999886744691e-05,
      "loss": 3.1126,
      "step": 1360
    },
    {
      "epoch": 0.03655850435156334,
      "grad_norm": 1.363294005393982,
      "learning_rate": 9.999999886577464e-05,
      "loss": 2.7373,
      "step": 1361
    },
    {
      "epoch": 0.036585365853658534,
      "grad_norm": 1.4247697591781616,
      "learning_rate": 9.999999886410112e-05,
      "loss": 2.8468,
      "step": 1362
    },
    {
      "epoch": 0.036612227355753736,
      "grad_norm": 1.3875802755355835,
      "learning_rate": 9.999999886242637e-05,
      "loss": 2.8448,
      "step": 1363
    },
    {
      "epoch": 0.03663908885784893,
      "grad_norm": 1.4311139583587646,
      "learning_rate": 9.99999988607504e-05,
      "loss": 2.8377,
      "step": 1364
    },
    {
      "epoch": 0.036665950359944126,
      "grad_norm": 1.4337817430496216,
      "learning_rate": 9.999999885907316e-05,
      "loss": 2.791,
      "step": 1365
    },
    {
      "epoch": 0.03669281186203933,
      "grad_norm": 1.4047051668167114,
      "learning_rate": 9.999999885739473e-05,
      "loss": 3.131,
      "step": 1366
    },
    {
      "epoch": 0.03671967336413452,
      "grad_norm": 1.412001132965088,
      "learning_rate": 9.999999885571503e-05,
      "loss": 3.0659,
      "step": 1367
    },
    {
      "epoch": 0.03674653486622972,
      "grad_norm": 1.4448264837265015,
      "learning_rate": 9.999999885403411e-05,
      "loss": 2.7289,
      "step": 1368
    },
    {
      "epoch": 0.03677339636832492,
      "grad_norm": 1.4868664741516113,
      "learning_rate": 9.999999885235196e-05,
      "loss": 3.0849,
      "step": 1369
    },
    {
      "epoch": 0.036800257870420114,
      "grad_norm": 1.4030438661575317,
      "learning_rate": 9.999999885066857e-05,
      "loss": 3.0095,
      "step": 1370
    },
    {
      "epoch": 0.03682711937251531,
      "grad_norm": 1.3381463289260864,
      "learning_rate": 9.999999884898397e-05,
      "loss": 2.9221,
      "step": 1371
    },
    {
      "epoch": 0.03685398087461051,
      "grad_norm": 1.3347033262252808,
      "learning_rate": 9.999999884729812e-05,
      "loss": 3.023,
      "step": 1372
    },
    {
      "epoch": 0.036880842376705705,
      "grad_norm": 1.4255144596099854,
      "learning_rate": 9.999999884561102e-05,
      "loss": 3.0732,
      "step": 1373
    },
    {
      "epoch": 0.0369077038788009,
      "grad_norm": 1.49003267288208,
      "learning_rate": 9.999999884392271e-05,
      "loss": 2.9899,
      "step": 1374
    },
    {
      "epoch": 0.0369345653808961,
      "grad_norm": 1.465358853340149,
      "learning_rate": 9.999999884223314e-05,
      "loss": 3.0834,
      "step": 1375
    },
    {
      "epoch": 0.0369614268829913,
      "grad_norm": 1.5213181972503662,
      "learning_rate": 9.999999884054237e-05,
      "loss": 2.9631,
      "step": 1376
    },
    {
      "epoch": 0.03698828838508649,
      "grad_norm": 1.55282723903656,
      "learning_rate": 9.999999883885034e-05,
      "loss": 3.1508,
      "step": 1377
    },
    {
      "epoch": 0.03701514988718169,
      "grad_norm": 1.5790045261383057,
      "learning_rate": 9.99999988371571e-05,
      "loss": 3.2373,
      "step": 1378
    },
    {
      "epoch": 0.03704201138927689,
      "grad_norm": 1.5668582916259766,
      "learning_rate": 9.99999988354626e-05,
      "loss": 3.0675,
      "step": 1379
    },
    {
      "epoch": 0.03706887289137208,
      "grad_norm": 1.5427089929580688,
      "learning_rate": 9.999999883376688e-05,
      "loss": 3.102,
      "step": 1380
    },
    {
      "epoch": 0.037095734393467285,
      "grad_norm": 1.5573800802230835,
      "learning_rate": 9.999999883206991e-05,
      "loss": 3.1938,
      "step": 1381
    },
    {
      "epoch": 0.03712259589556248,
      "grad_norm": 1.4574451446533203,
      "learning_rate": 9.999999883037173e-05,
      "loss": 2.9663,
      "step": 1382
    },
    {
      "epoch": 0.037149457397657674,
      "grad_norm": 1.497205138206482,
      "learning_rate": 9.999999882867231e-05,
      "loss": 3.0097,
      "step": 1383
    },
    {
      "epoch": 0.037176318899752876,
      "grad_norm": 1.5007003545761108,
      "learning_rate": 9.999999882697164e-05,
      "loss": 3.1039,
      "step": 1384
    },
    {
      "epoch": 0.03720318040184807,
      "grad_norm": 1.5595407485961914,
      "learning_rate": 9.999999882526976e-05,
      "loss": 3.3815,
      "step": 1385
    },
    {
      "epoch": 0.037230041903943266,
      "grad_norm": 1.456282615661621,
      "learning_rate": 9.999999882356664e-05,
      "loss": 2.9351,
      "step": 1386
    },
    {
      "epoch": 0.03725690340603847,
      "grad_norm": 1.6644130945205688,
      "learning_rate": 9.999999882186227e-05,
      "loss": 3.3192,
      "step": 1387
    },
    {
      "epoch": 0.03728376490813366,
      "grad_norm": 1.5125242471694946,
      "learning_rate": 9.999999882015669e-05,
      "loss": 3.0214,
      "step": 1388
    },
    {
      "epoch": 0.03731062641022886,
      "grad_norm": 1.468849539756775,
      "learning_rate": 9.999999881844985e-05,
      "loss": 3.1451,
      "step": 1389
    },
    {
      "epoch": 0.03733748791232406,
      "grad_norm": 1.6174612045288086,
      "learning_rate": 9.99999988167418e-05,
      "loss": 3.2808,
      "step": 1390
    },
    {
      "epoch": 0.037364349414419254,
      "grad_norm": 1.575363039970398,
      "learning_rate": 9.99999988150325e-05,
      "loss": 3.225,
      "step": 1391
    },
    {
      "epoch": 0.03739121091651445,
      "grad_norm": 1.5213159322738647,
      "learning_rate": 9.999999881332198e-05,
      "loss": 3.1011,
      "step": 1392
    },
    {
      "epoch": 0.03741807241860965,
      "grad_norm": 1.6819603443145752,
      "learning_rate": 9.999999881161022e-05,
      "loss": 3.1736,
      "step": 1393
    },
    {
      "epoch": 0.037444933920704845,
      "grad_norm": 1.672792673110962,
      "learning_rate": 9.999999880989723e-05,
      "loss": 3.4121,
      "step": 1394
    },
    {
      "epoch": 0.03747179542280004,
      "grad_norm": 1.7510223388671875,
      "learning_rate": 9.9999998808183e-05,
      "loss": 2.8893,
      "step": 1395
    },
    {
      "epoch": 0.03749865692489524,
      "grad_norm": 1.7332818508148193,
      "learning_rate": 9.999999880646754e-05,
      "loss": 3.2137,
      "step": 1396
    },
    {
      "epoch": 0.03752551842699044,
      "grad_norm": 1.836311936378479,
      "learning_rate": 9.999999880475084e-05,
      "loss": 3.605,
      "step": 1397
    },
    {
      "epoch": 0.03755237992908563,
      "grad_norm": 1.770053505897522,
      "learning_rate": 9.999999880303292e-05,
      "loss": 2.998,
      "step": 1398
    },
    {
      "epoch": 0.03757924143118083,
      "grad_norm": 1.8517436981201172,
      "learning_rate": 9.999999880131374e-05,
      "loss": 3.3581,
      "step": 1399
    },
    {
      "epoch": 0.03760610293327603,
      "grad_norm": 1.7875633239746094,
      "learning_rate": 9.999999879959336e-05,
      "loss": 3.0633,
      "step": 1400
    },
    {
      "epoch": 0.03763296443537122,
      "grad_norm": 1.2558006048202515,
      "learning_rate": 9.999999879787172e-05,
      "loss": 3.1465,
      "step": 1401
    },
    {
      "epoch": 0.037659825937466425,
      "grad_norm": 1.3543556928634644,
      "learning_rate": 9.999999879614886e-05,
      "loss": 2.6867,
      "step": 1402
    },
    {
      "epoch": 0.03768668743956162,
      "grad_norm": 1.4992352724075317,
      "learning_rate": 9.999999879442476e-05,
      "loss": 3.1387,
      "step": 1403
    },
    {
      "epoch": 0.037713548941656815,
      "grad_norm": 1.4522989988327026,
      "learning_rate": 9.999999879269943e-05,
      "loss": 2.9011,
      "step": 1404
    },
    {
      "epoch": 0.037740410443752016,
      "grad_norm": 1.5051820278167725,
      "learning_rate": 9.999999879097286e-05,
      "loss": 3.0651,
      "step": 1405
    },
    {
      "epoch": 0.03776727194584721,
      "grad_norm": 1.3663911819458008,
      "learning_rate": 9.999999878924507e-05,
      "loss": 2.7669,
      "step": 1406
    },
    {
      "epoch": 0.037794133447942406,
      "grad_norm": 1.3655157089233398,
      "learning_rate": 9.999999878751603e-05,
      "loss": 3.2849,
      "step": 1407
    },
    {
      "epoch": 0.03782099495003761,
      "grad_norm": 1.4261425733566284,
      "learning_rate": 9.999999878578577e-05,
      "loss": 2.9059,
      "step": 1408
    },
    {
      "epoch": 0.0378478564521328,
      "grad_norm": 1.3434581756591797,
      "learning_rate": 9.999999878405427e-05,
      "loss": 2.6944,
      "step": 1409
    },
    {
      "epoch": 0.037874717954228,
      "grad_norm": 1.4687180519104004,
      "learning_rate": 9.999999878232154e-05,
      "loss": 3.0075,
      "step": 1410
    },
    {
      "epoch": 0.0379015794563232,
      "grad_norm": 1.4739652872085571,
      "learning_rate": 9.999999878058756e-05,
      "loss": 2.8742,
      "step": 1411
    },
    {
      "epoch": 0.037928440958418394,
      "grad_norm": 1.3373268842697144,
      "learning_rate": 9.999999877885236e-05,
      "loss": 2.8139,
      "step": 1412
    },
    {
      "epoch": 0.03795530246051359,
      "grad_norm": 1.6320712566375732,
      "learning_rate": 9.999999877711593e-05,
      "loss": 2.9475,
      "step": 1413
    },
    {
      "epoch": 0.03798216396260879,
      "grad_norm": 1.3583111763000488,
      "learning_rate": 9.999999877537827e-05,
      "loss": 2.838,
      "step": 1414
    },
    {
      "epoch": 0.038009025464703985,
      "grad_norm": 1.3993921279907227,
      "learning_rate": 9.999999877363937e-05,
      "loss": 2.8855,
      "step": 1415
    },
    {
      "epoch": 0.03803588696679918,
      "grad_norm": 1.529668927192688,
      "learning_rate": 9.999999877189923e-05,
      "loss": 3.0322,
      "step": 1416
    },
    {
      "epoch": 0.03806274846889438,
      "grad_norm": 1.4245907068252563,
      "learning_rate": 9.999999877015786e-05,
      "loss": 2.981,
      "step": 1417
    },
    {
      "epoch": 0.03808960997098958,
      "grad_norm": 1.3823530673980713,
      "learning_rate": 9.999999876841527e-05,
      "loss": 3.0361,
      "step": 1418
    },
    {
      "epoch": 0.03811647147308477,
      "grad_norm": 1.3851081132888794,
      "learning_rate": 9.999999876667142e-05,
      "loss": 2.6808,
      "step": 1419
    },
    {
      "epoch": 0.038143332975179974,
      "grad_norm": 1.4888771772384644,
      "learning_rate": 9.999999876492636e-05,
      "loss": 3.1491,
      "step": 1420
    },
    {
      "epoch": 0.03817019447727517,
      "grad_norm": 1.4121801853179932,
      "learning_rate": 9.999999876318005e-05,
      "loss": 2.6652,
      "step": 1421
    },
    {
      "epoch": 0.03819705597937036,
      "grad_norm": 1.4982889890670776,
      "learning_rate": 9.999999876143252e-05,
      "loss": 2.9528,
      "step": 1422
    },
    {
      "epoch": 0.038223917481465565,
      "grad_norm": 1.4127792119979858,
      "learning_rate": 9.999999875968374e-05,
      "loss": 2.9695,
      "step": 1423
    },
    {
      "epoch": 0.03825077898356076,
      "grad_norm": 1.475388765335083,
      "learning_rate": 9.999999875793373e-05,
      "loss": 2.7763,
      "step": 1424
    },
    {
      "epoch": 0.038277640485655955,
      "grad_norm": 1.503794550895691,
      "learning_rate": 9.999999875618248e-05,
      "loss": 3.0564,
      "step": 1425
    },
    {
      "epoch": 0.038304501987751156,
      "grad_norm": 1.58021080493927,
      "learning_rate": 9.999999875443002e-05,
      "loss": 2.9922,
      "step": 1426
    },
    {
      "epoch": 0.03833136348984635,
      "grad_norm": 1.6000361442565918,
      "learning_rate": 9.999999875267632e-05,
      "loss": 3.1504,
      "step": 1427
    },
    {
      "epoch": 0.038358224991941546,
      "grad_norm": 1.5512723922729492,
      "learning_rate": 9.999999875092138e-05,
      "loss": 3.1249,
      "step": 1428
    },
    {
      "epoch": 0.03838508649403675,
      "grad_norm": 1.6090893745422363,
      "learning_rate": 9.99999987491652e-05,
      "loss": 3.1778,
      "step": 1429
    },
    {
      "epoch": 0.03841194799613194,
      "grad_norm": 1.527324914932251,
      "learning_rate": 9.999999874740779e-05,
      "loss": 3.1531,
      "step": 1430
    },
    {
      "epoch": 0.03843880949822714,
      "grad_norm": 1.4883356094360352,
      "learning_rate": 9.999999874564916e-05,
      "loss": 2.9801,
      "step": 1431
    },
    {
      "epoch": 0.03846567100032234,
      "grad_norm": 1.5239859819412231,
      "learning_rate": 9.999999874388928e-05,
      "loss": 3.1369,
      "step": 1432
    },
    {
      "epoch": 0.038492532502417534,
      "grad_norm": 1.6074607372283936,
      "learning_rate": 9.999999874212817e-05,
      "loss": 3.0369,
      "step": 1433
    },
    {
      "epoch": 0.03851939400451273,
      "grad_norm": 1.3886668682098389,
      "learning_rate": 9.999999874036583e-05,
      "loss": 3.0488,
      "step": 1434
    },
    {
      "epoch": 0.03854625550660793,
      "grad_norm": 1.579745888710022,
      "learning_rate": 9.999999873860225e-05,
      "loss": 3.2854,
      "step": 1435
    },
    {
      "epoch": 0.038573117008703126,
      "grad_norm": 1.5234684944152832,
      "learning_rate": 9.999999873683744e-05,
      "loss": 3.0604,
      "step": 1436
    },
    {
      "epoch": 0.03859997851079832,
      "grad_norm": 1.6631017923355103,
      "learning_rate": 9.99999987350714e-05,
      "loss": 3.286,
      "step": 1437
    },
    {
      "epoch": 0.03862684001289352,
      "grad_norm": 1.71566641330719,
      "learning_rate": 9.999999873330412e-05,
      "loss": 3.2429,
      "step": 1438
    },
    {
      "epoch": 0.03865370151498872,
      "grad_norm": 1.5708800554275513,
      "learning_rate": 9.99999987315356e-05,
      "loss": 3.2407,
      "step": 1439
    },
    {
      "epoch": 0.03868056301708392,
      "grad_norm": 1.631901741027832,
      "learning_rate": 9.999999872976587e-05,
      "loss": 3.1672,
      "step": 1440
    },
    {
      "epoch": 0.038707424519179114,
      "grad_norm": 1.5446592569351196,
      "learning_rate": 9.999999872799488e-05,
      "loss": 3.2726,
      "step": 1441
    },
    {
      "epoch": 0.03873428602127431,
      "grad_norm": 1.555907130241394,
      "learning_rate": 9.999999872622267e-05,
      "loss": 3.2899,
      "step": 1442
    },
    {
      "epoch": 0.03876114752336951,
      "grad_norm": 1.5833879709243774,
      "learning_rate": 9.999999872444923e-05,
      "loss": 3.207,
      "step": 1443
    },
    {
      "epoch": 0.038788009025464705,
      "grad_norm": 1.6748602390289307,
      "learning_rate": 9.999999872267456e-05,
      "loss": 3.1702,
      "step": 1444
    },
    {
      "epoch": 0.0388148705275599,
      "grad_norm": 1.6045279502868652,
      "learning_rate": 9.999999872089864e-05,
      "loss": 2.9434,
      "step": 1445
    },
    {
      "epoch": 0.0388417320296551,
      "grad_norm": 1.7981611490249634,
      "learning_rate": 9.999999871912148e-05,
      "loss": 3.1452,
      "step": 1446
    },
    {
      "epoch": 0.038868593531750296,
      "grad_norm": 1.9610273838043213,
      "learning_rate": 9.999999871734311e-05,
      "loss": 3.3071,
      "step": 1447
    },
    {
      "epoch": 0.03889545503384549,
      "grad_norm": 1.8365384340286255,
      "learning_rate": 9.999999871556349e-05,
      "loss": 3.3733,
      "step": 1448
    },
    {
      "epoch": 0.03892231653594069,
      "grad_norm": 1.834254503250122,
      "learning_rate": 9.999999871378266e-05,
      "loss": 3.5926,
      "step": 1449
    },
    {
      "epoch": 0.03894917803803589,
      "grad_norm": 1.7254598140716553,
      "learning_rate": 9.999999871200057e-05,
      "loss": 3.3867,
      "step": 1450
    },
    {
      "epoch": 0.03897603954013108,
      "grad_norm": 1.444594144821167,
      "learning_rate": 9.999999871021726e-05,
      "loss": 3.006,
      "step": 1451
    },
    {
      "epoch": 0.039002901042226285,
      "grad_norm": 1.5244362354278564,
      "learning_rate": 9.99999987084327e-05,
      "loss": 3.0045,
      "step": 1452
    },
    {
      "epoch": 0.03902976254432148,
      "grad_norm": 1.3606665134429932,
      "learning_rate": 9.999999870664693e-05,
      "loss": 2.9012,
      "step": 1453
    },
    {
      "epoch": 0.039056624046416674,
      "grad_norm": 1.3562020063400269,
      "learning_rate": 9.99999987048599e-05,
      "loss": 2.7546,
      "step": 1454
    },
    {
      "epoch": 0.039083485548511876,
      "grad_norm": 1.3499842882156372,
      "learning_rate": 9.999999870307166e-05,
      "loss": 2.7078,
      "step": 1455
    },
    {
      "epoch": 0.03911034705060707,
      "grad_norm": 1.4267544746398926,
      "learning_rate": 9.999999870128218e-05,
      "loss": 2.9684,
      "step": 1456
    },
    {
      "epoch": 0.039137208552702266,
      "grad_norm": 1.4771356582641602,
      "learning_rate": 9.999999869949146e-05,
      "loss": 3.0608,
      "step": 1457
    },
    {
      "epoch": 0.03916407005479747,
      "grad_norm": 1.292306661605835,
      "learning_rate": 9.999999869769952e-05,
      "loss": 2.6641,
      "step": 1458
    },
    {
      "epoch": 0.03919093155689266,
      "grad_norm": 1.3434932231903076,
      "learning_rate": 9.999999869590633e-05,
      "loss": 2.9786,
      "step": 1459
    },
    {
      "epoch": 0.03921779305898786,
      "grad_norm": 1.3802857398986816,
      "learning_rate": 9.99999986941119e-05,
      "loss": 3.0925,
      "step": 1460
    },
    {
      "epoch": 0.03924465456108306,
      "grad_norm": 1.3967256546020508,
      "learning_rate": 9.999999869231625e-05,
      "loss": 2.964,
      "step": 1461
    },
    {
      "epoch": 0.039271516063178254,
      "grad_norm": 1.4296356439590454,
      "learning_rate": 9.999999869051936e-05,
      "loss": 2.7799,
      "step": 1462
    },
    {
      "epoch": 0.03929837756527345,
      "grad_norm": 1.3817452192306519,
      "learning_rate": 9.999999868872126e-05,
      "loss": 2.8511,
      "step": 1463
    },
    {
      "epoch": 0.03932523906736865,
      "grad_norm": 1.372567892074585,
      "learning_rate": 9.999999868692189e-05,
      "loss": 3.0817,
      "step": 1464
    },
    {
      "epoch": 0.039352100569463845,
      "grad_norm": 1.3290953636169434,
      "learning_rate": 9.999999868512131e-05,
      "loss": 2.8261,
      "step": 1465
    },
    {
      "epoch": 0.03937896207155904,
      "grad_norm": 1.4655951261520386,
      "learning_rate": 9.999999868331949e-05,
      "loss": 2.7318,
      "step": 1466
    },
    {
      "epoch": 0.03940582357365424,
      "grad_norm": 1.3358664512634277,
      "learning_rate": 9.999999868151643e-05,
      "loss": 2.5184,
      "step": 1467
    },
    {
      "epoch": 0.03943268507574944,
      "grad_norm": 1.4155128002166748,
      "learning_rate": 9.999999867971215e-05,
      "loss": 2.8334,
      "step": 1468
    },
    {
      "epoch": 0.03945954657784463,
      "grad_norm": 1.5112532377243042,
      "learning_rate": 9.999999867790661e-05,
      "loss": 3.0017,
      "step": 1469
    },
    {
      "epoch": 0.03948640807993983,
      "grad_norm": 1.4111354351043701,
      "learning_rate": 9.999999867609985e-05,
      "loss": 2.99,
      "step": 1470
    },
    {
      "epoch": 0.03951326958203503,
      "grad_norm": 1.3667641878128052,
      "learning_rate": 9.999999867429188e-05,
      "loss": 2.8692,
      "step": 1471
    },
    {
      "epoch": 0.03954013108413022,
      "grad_norm": 1.4051331281661987,
      "learning_rate": 9.999999867248266e-05,
      "loss": 2.9006,
      "step": 1472
    },
    {
      "epoch": 0.039566992586225425,
      "grad_norm": 1.4388536214828491,
      "learning_rate": 9.99999986706722e-05,
      "loss": 2.7888,
      "step": 1473
    },
    {
      "epoch": 0.03959385408832062,
      "grad_norm": 1.400656819343567,
      "learning_rate": 9.99999986688605e-05,
      "loss": 2.7677,
      "step": 1474
    },
    {
      "epoch": 0.039620715590415814,
      "grad_norm": 1.5121233463287354,
      "learning_rate": 9.999999866704759e-05,
      "loss": 3.0488,
      "step": 1475
    },
    {
      "epoch": 0.039647577092511016,
      "grad_norm": 1.4835797548294067,
      "learning_rate": 9.999999866523342e-05,
      "loss": 2.8857,
      "step": 1476
    },
    {
      "epoch": 0.03967443859460621,
      "grad_norm": 1.390968680381775,
      "learning_rate": 9.999999866341804e-05,
      "loss": 3.0377,
      "step": 1477
    },
    {
      "epoch": 0.039701300096701406,
      "grad_norm": 1.4880067110061646,
      "learning_rate": 9.999999866160142e-05,
      "loss": 3.2275,
      "step": 1478
    },
    {
      "epoch": 0.03972816159879661,
      "grad_norm": 1.5887243747711182,
      "learning_rate": 9.999999865978356e-05,
      "loss": 3.3484,
      "step": 1479
    },
    {
      "epoch": 0.0397550231008918,
      "grad_norm": 1.509476900100708,
      "learning_rate": 9.999999865796446e-05,
      "loss": 3.1582,
      "step": 1480
    },
    {
      "epoch": 0.039781884602987,
      "grad_norm": 1.5989946126937866,
      "learning_rate": 9.999999865614414e-05,
      "loss": 3.1388,
      "step": 1481
    },
    {
      "epoch": 0.0398087461050822,
      "grad_norm": 1.590970516204834,
      "learning_rate": 9.999999865432257e-05,
      "loss": 3.1798,
      "step": 1482
    },
    {
      "epoch": 0.039835607607177394,
      "grad_norm": 1.383935809135437,
      "learning_rate": 9.999999865249978e-05,
      "loss": 2.8013,
      "step": 1483
    },
    {
      "epoch": 0.03986246910927259,
      "grad_norm": 1.5269430875778198,
      "learning_rate": 9.999999865067575e-05,
      "loss": 2.7182,
      "step": 1484
    },
    {
      "epoch": 0.03988933061136779,
      "grad_norm": 1.5870354175567627,
      "learning_rate": 9.999999864885049e-05,
      "loss": 3.0641,
      "step": 1485
    },
    {
      "epoch": 0.039916192113462985,
      "grad_norm": 1.4706482887268066,
      "learning_rate": 9.999999864702399e-05,
      "loss": 2.7107,
      "step": 1486
    },
    {
      "epoch": 0.03994305361555818,
      "grad_norm": 1.4456549882888794,
      "learning_rate": 9.999999864519627e-05,
      "loss": 2.8107,
      "step": 1487
    },
    {
      "epoch": 0.03996991511765338,
      "grad_norm": 1.6855895519256592,
      "learning_rate": 9.99999986433673e-05,
      "loss": 3.1461,
      "step": 1488
    },
    {
      "epoch": 0.03999677661974858,
      "grad_norm": 1.5434343814849854,
      "learning_rate": 9.99999986415371e-05,
      "loss": 3.4355,
      "step": 1489
    },
    {
      "epoch": 0.04002363812184377,
      "grad_norm": 1.6378633975982666,
      "learning_rate": 9.999999863970568e-05,
      "loss": 3.1938,
      "step": 1490
    },
    {
      "epoch": 0.04005049962393897,
      "grad_norm": 1.5038949251174927,
      "learning_rate": 9.999999863787301e-05,
      "loss": 3.134,
      "step": 1491
    },
    {
      "epoch": 0.04007736112603417,
      "grad_norm": 1.6363450288772583,
      "learning_rate": 9.999999863603913e-05,
      "loss": 3.3421,
      "step": 1492
    },
    {
      "epoch": 0.04010422262812936,
      "grad_norm": 1.6093322038650513,
      "learning_rate": 9.999999863420399e-05,
      "loss": 3.0592,
      "step": 1493
    },
    {
      "epoch": 0.040131084130224565,
      "grad_norm": 1.5667686462402344,
      "learning_rate": 9.999999863236762e-05,
      "loss": 3.1278,
      "step": 1494
    },
    {
      "epoch": 0.04015794563231976,
      "grad_norm": 1.4725762605667114,
      "learning_rate": 9.999999863053002e-05,
      "loss": 3.1241,
      "step": 1495
    },
    {
      "epoch": 0.040184807134414954,
      "grad_norm": 1.6674635410308838,
      "learning_rate": 9.99999986286912e-05,
      "loss": 3.1671,
      "step": 1496
    },
    {
      "epoch": 0.040211668636510156,
      "grad_norm": 1.7384889125823975,
      "learning_rate": 9.999999862685114e-05,
      "loss": 3.2027,
      "step": 1497
    },
    {
      "epoch": 0.04023853013860535,
      "grad_norm": 1.6738016605377197,
      "learning_rate": 9.999999862500984e-05,
      "loss": 3.3651,
      "step": 1498
    },
    {
      "epoch": 0.040265391640700546,
      "grad_norm": 1.8409940004348755,
      "learning_rate": 9.99999986231673e-05,
      "loss": 3.4753,
      "step": 1499
    },
    {
      "epoch": 0.04029225314279575,
      "grad_norm": 2.1469180583953857,
      "learning_rate": 9.999999862132355e-05,
      "loss": 3.5186,
      "step": 1500
    },
    {
      "epoch": 0.04031911464489094,
      "grad_norm": 1.4282715320587158,
      "learning_rate": 9.999999861947853e-05,
      "loss": 3.2386,
      "step": 1501
    },
    {
      "epoch": 0.04034597614698614,
      "grad_norm": 1.378355860710144,
      "learning_rate": 9.99999986176323e-05,
      "loss": 2.8773,
      "step": 1502
    },
    {
      "epoch": 0.04037283764908134,
      "grad_norm": 1.383313536643982,
      "learning_rate": 9.999999861578483e-05,
      "loss": 2.8719,
      "step": 1503
    },
    {
      "epoch": 0.040399699151176534,
      "grad_norm": 1.3885672092437744,
      "learning_rate": 9.999999861393615e-05,
      "loss": 3.0667,
      "step": 1504
    },
    {
      "epoch": 0.04042656065327173,
      "grad_norm": 1.460296869277954,
      "learning_rate": 9.99999986120862e-05,
      "loss": 3.2173,
      "step": 1505
    },
    {
      "epoch": 0.04045342215536693,
      "grad_norm": 1.3763436079025269,
      "learning_rate": 9.999999861023503e-05,
      "loss": 2.9576,
      "step": 1506
    },
    {
      "epoch": 0.040480283657462125,
      "grad_norm": 1.3316450119018555,
      "learning_rate": 9.999999860838263e-05,
      "loss": 2.8878,
      "step": 1507
    },
    {
      "epoch": 0.04050714515955732,
      "grad_norm": 1.3709266185760498,
      "learning_rate": 9.9999998606529e-05,
      "loss": 2.7754,
      "step": 1508
    },
    {
      "epoch": 0.04053400666165252,
      "grad_norm": 1.2321733236312866,
      "learning_rate": 9.999999860467413e-05,
      "loss": 2.7415,
      "step": 1509
    },
    {
      "epoch": 0.04056086816374772,
      "grad_norm": 1.3904420137405396,
      "learning_rate": 9.9999998602818e-05,
      "loss": 2.9703,
      "step": 1510
    },
    {
      "epoch": 0.04058772966584291,
      "grad_norm": 1.352800965309143,
      "learning_rate": 9.999999860096067e-05,
      "loss": 2.6487,
      "step": 1511
    },
    {
      "epoch": 0.04061459116793811,
      "grad_norm": 1.2946110963821411,
      "learning_rate": 9.999999859910212e-05,
      "loss": 2.8044,
      "step": 1512
    },
    {
      "epoch": 0.04064145267003331,
      "grad_norm": 1.3256781101226807,
      "learning_rate": 9.999999859724231e-05,
      "loss": 2.7241,
      "step": 1513
    },
    {
      "epoch": 0.0406683141721285,
      "grad_norm": 1.4185431003570557,
      "learning_rate": 9.999999859538127e-05,
      "loss": 3.0663,
      "step": 1514
    },
    {
      "epoch": 0.040695175674223705,
      "grad_norm": 1.3325541019439697,
      "learning_rate": 9.9999998593519e-05,
      "loss": 2.8112,
      "step": 1515
    },
    {
      "epoch": 0.0407220371763189,
      "grad_norm": 1.5682042837142944,
      "learning_rate": 9.99999985916555e-05,
      "loss": 2.8277,
      "step": 1516
    },
    {
      "epoch": 0.040748898678414094,
      "grad_norm": 1.333392858505249,
      "learning_rate": 9.999999858979075e-05,
      "loss": 2.6806,
      "step": 1517
    },
    {
      "epoch": 0.040775760180509296,
      "grad_norm": 1.3029935359954834,
      "learning_rate": 9.999999858792477e-05,
      "loss": 2.556,
      "step": 1518
    },
    {
      "epoch": 0.04080262168260449,
      "grad_norm": 1.3836957216262817,
      "learning_rate": 9.999999858605757e-05,
      "loss": 3.1112,
      "step": 1519
    },
    {
      "epoch": 0.040829483184699686,
      "grad_norm": 1.5574411153793335,
      "learning_rate": 9.999999858418914e-05,
      "loss": 2.9932,
      "step": 1520
    },
    {
      "epoch": 0.04085634468679489,
      "grad_norm": 1.3623356819152832,
      "learning_rate": 9.999999858231946e-05,
      "loss": 2.7071,
      "step": 1521
    },
    {
      "epoch": 0.04088320618889008,
      "grad_norm": 1.3509150743484497,
      "learning_rate": 9.999999858044856e-05,
      "loss": 2.7619,
      "step": 1522
    },
    {
      "epoch": 0.04091006769098528,
      "grad_norm": 1.4030135869979858,
      "learning_rate": 9.99999985785764e-05,
      "loss": 2.9066,
      "step": 1523
    },
    {
      "epoch": 0.04093692919308048,
      "grad_norm": 1.4573057889938354,
      "learning_rate": 9.999999857670303e-05,
      "loss": 3.0447,
      "step": 1524
    },
    {
      "epoch": 0.040963790695175674,
      "grad_norm": 1.4426836967468262,
      "learning_rate": 9.999999857482842e-05,
      "loss": 2.8589,
      "step": 1525
    },
    {
      "epoch": 0.04099065219727087,
      "grad_norm": 1.4468977451324463,
      "learning_rate": 9.999999857295259e-05,
      "loss": 2.8131,
      "step": 1526
    },
    {
      "epoch": 0.04101751369936607,
      "grad_norm": 1.6241178512573242,
      "learning_rate": 9.999999857107551e-05,
      "loss": 3.0325,
      "step": 1527
    },
    {
      "epoch": 0.041044375201461265,
      "grad_norm": 1.6806358098983765,
      "learning_rate": 9.99999985691972e-05,
      "loss": 3.2774,
      "step": 1528
    },
    {
      "epoch": 0.04107123670355646,
      "grad_norm": 1.4917420148849487,
      "learning_rate": 9.999999856731765e-05,
      "loss": 2.9552,
      "step": 1529
    },
    {
      "epoch": 0.04109809820565166,
      "grad_norm": 1.6034181118011475,
      "learning_rate": 9.999999856543687e-05,
      "loss": 2.9498,
      "step": 1530
    },
    {
      "epoch": 0.04112495970774686,
      "grad_norm": 1.4520667791366577,
      "learning_rate": 9.999999856355486e-05,
      "loss": 2.9688,
      "step": 1531
    },
    {
      "epoch": 0.04115182120984205,
      "grad_norm": 1.5000613927841187,
      "learning_rate": 9.999999856167161e-05,
      "loss": 3.3126,
      "step": 1532
    },
    {
      "epoch": 0.04117868271193725,
      "grad_norm": 1.5931119918823242,
      "learning_rate": 9.999999855978715e-05,
      "loss": 3.2781,
      "step": 1533
    },
    {
      "epoch": 0.04120554421403245,
      "grad_norm": 1.6332576274871826,
      "learning_rate": 9.999999855790143e-05,
      "loss": 3.2623,
      "step": 1534
    },
    {
      "epoch": 0.04123240571612764,
      "grad_norm": 1.444083333015442,
      "learning_rate": 9.999999855601448e-05,
      "loss": 2.9727,
      "step": 1535
    },
    {
      "epoch": 0.041259267218222845,
      "grad_norm": 1.5641528367996216,
      "learning_rate": 9.99999985541263e-05,
      "loss": 3.105,
      "step": 1536
    },
    {
      "epoch": 0.04128612872031804,
      "grad_norm": 1.5843452215194702,
      "learning_rate": 9.99999985522369e-05,
      "loss": 2.935,
      "step": 1537
    },
    {
      "epoch": 0.041312990222413234,
      "grad_norm": 1.5150201320648193,
      "learning_rate": 9.999999855034624e-05,
      "loss": 2.8343,
      "step": 1538
    },
    {
      "epoch": 0.041339851724508436,
      "grad_norm": 1.5015759468078613,
      "learning_rate": 9.999999854845436e-05,
      "loss": 3.0577,
      "step": 1539
    },
    {
      "epoch": 0.04136671322660363,
      "grad_norm": 1.5382952690124512,
      "learning_rate": 9.999999854656125e-05,
      "loss": 3.1988,
      "step": 1540
    },
    {
      "epoch": 0.041393574728698826,
      "grad_norm": 1.6064475774765015,
      "learning_rate": 9.999999854466689e-05,
      "loss": 3.3073,
      "step": 1541
    },
    {
      "epoch": 0.04142043623079403,
      "grad_norm": 1.4302036762237549,
      "learning_rate": 9.999999854277131e-05,
      "loss": 2.9684,
      "step": 1542
    },
    {
      "epoch": 0.04144729773288922,
      "grad_norm": 1.4662116765975952,
      "learning_rate": 9.99999985408745e-05,
      "loss": 3.0528,
      "step": 1543
    },
    {
      "epoch": 0.04147415923498442,
      "grad_norm": 1.5845084190368652,
      "learning_rate": 9.999999853897646e-05,
      "loss": 3.0684,
      "step": 1544
    },
    {
      "epoch": 0.04150102073707962,
      "grad_norm": 1.603434681892395,
      "learning_rate": 9.999999853707716e-05,
      "loss": 2.9611,
      "step": 1545
    },
    {
      "epoch": 0.041527882239174814,
      "grad_norm": 1.66338312625885,
      "learning_rate": 9.999999853517665e-05,
      "loss": 3.2688,
      "step": 1546
    },
    {
      "epoch": 0.04155474374127001,
      "grad_norm": 1.900890827178955,
      "learning_rate": 9.999999853327491e-05,
      "loss": 3.4207,
      "step": 1547
    },
    {
      "epoch": 0.04158160524336521,
      "grad_norm": 1.735031247138977,
      "learning_rate": 9.999999853137192e-05,
      "loss": 3.5208,
      "step": 1548
    },
    {
      "epoch": 0.041608466745460405,
      "grad_norm": 1.8877718448638916,
      "learning_rate": 9.99999985294677e-05,
      "loss": 3.4457,
      "step": 1549
    },
    {
      "epoch": 0.0416353282475556,
      "grad_norm": 1.8484257459640503,
      "learning_rate": 9.999999852756226e-05,
      "loss": 3.4321,
      "step": 1550
    },
    {
      "epoch": 0.0416621897496508,
      "grad_norm": 1.1651768684387207,
      "learning_rate": 9.999999852565555e-05,
      "loss": 2.7929,
      "step": 1551
    },
    {
      "epoch": 0.041689051251746,
      "grad_norm": 1.3331292867660522,
      "learning_rate": 9.999999852374764e-05,
      "loss": 2.808,
      "step": 1552
    },
    {
      "epoch": 0.04171591275384119,
      "grad_norm": 1.459187388420105,
      "learning_rate": 9.999999852183849e-05,
      "loss": 3.018,
      "step": 1553
    },
    {
      "epoch": 0.04174277425593639,
      "grad_norm": 1.4368375539779663,
      "learning_rate": 9.999999851992811e-05,
      "loss": 3.0412,
      "step": 1554
    },
    {
      "epoch": 0.04176963575803159,
      "grad_norm": 1.465675711631775,
      "learning_rate": 9.99999985180165e-05,
      "loss": 2.9956,
      "step": 1555
    },
    {
      "epoch": 0.04179649726012678,
      "grad_norm": 1.3604998588562012,
      "learning_rate": 9.999999851610364e-05,
      "loss": 2.7662,
      "step": 1556
    },
    {
      "epoch": 0.041823358762221985,
      "grad_norm": 1.3840458393096924,
      "learning_rate": 9.999999851418955e-05,
      "loss": 2.6917,
      "step": 1557
    },
    {
      "epoch": 0.04185022026431718,
      "grad_norm": 1.2864190340042114,
      "learning_rate": 9.999999851227422e-05,
      "loss": 2.8664,
      "step": 1558
    },
    {
      "epoch": 0.041877081766412375,
      "grad_norm": 1.3945748805999756,
      "learning_rate": 9.999999851035767e-05,
      "loss": 2.7021,
      "step": 1559
    },
    {
      "epoch": 0.041903943268507576,
      "grad_norm": 1.3237626552581787,
      "learning_rate": 9.999999850843989e-05,
      "loss": 2.5358,
      "step": 1560
    },
    {
      "epoch": 0.04193080477060277,
      "grad_norm": 1.3199323415756226,
      "learning_rate": 9.999999850652086e-05,
      "loss": 2.83,
      "step": 1561
    },
    {
      "epoch": 0.041957666272697966,
      "grad_norm": 1.2470163106918335,
      "learning_rate": 9.99999985046006e-05,
      "loss": 2.7546,
      "step": 1562
    },
    {
      "epoch": 0.04198452777479317,
      "grad_norm": 1.3408187627792358,
      "learning_rate": 9.999999850267912e-05,
      "loss": 2.7943,
      "step": 1563
    },
    {
      "epoch": 0.04201138927688836,
      "grad_norm": 1.4136046171188354,
      "learning_rate": 9.999999850075639e-05,
      "loss": 3.1962,
      "step": 1564
    },
    {
      "epoch": 0.04203825077898356,
      "grad_norm": 1.4965014457702637,
      "learning_rate": 9.999999849883243e-05,
      "loss": 2.9666,
      "step": 1565
    },
    {
      "epoch": 0.04206511228107876,
      "grad_norm": 1.4411835670471191,
      "learning_rate": 9.999999849690724e-05,
      "loss": 2.8392,
      "step": 1566
    },
    {
      "epoch": 0.042091973783173954,
      "grad_norm": 1.4846420288085938,
      "learning_rate": 9.999999849498082e-05,
      "loss": 2.7586,
      "step": 1567
    },
    {
      "epoch": 0.04211883528526915,
      "grad_norm": 1.3521579504013062,
      "learning_rate": 9.999999849305316e-05,
      "loss": 3.1934,
      "step": 1568
    },
    {
      "epoch": 0.04214569678736435,
      "grad_norm": 1.4549918174743652,
      "learning_rate": 9.999999849112427e-05,
      "loss": 2.9655,
      "step": 1569
    },
    {
      "epoch": 0.042172558289459545,
      "grad_norm": 1.4649477005004883,
      "learning_rate": 9.999999848919415e-05,
      "loss": 2.8873,
      "step": 1570
    },
    {
      "epoch": 0.04219941979155474,
      "grad_norm": 1.4793885946273804,
      "learning_rate": 9.999999848726278e-05,
      "loss": 3.0737,
      "step": 1571
    },
    {
      "epoch": 0.04222628129364994,
      "grad_norm": 1.5236968994140625,
      "learning_rate": 9.99999984853302e-05,
      "loss": 3.1176,
      "step": 1572
    },
    {
      "epoch": 0.04225314279574514,
      "grad_norm": 1.5447379350662231,
      "learning_rate": 9.999999848339636e-05,
      "loss": 3.3341,
      "step": 1573
    },
    {
      "epoch": 0.04228000429784034,
      "grad_norm": 1.3562862873077393,
      "learning_rate": 9.99999984814613e-05,
      "loss": 2.6696,
      "step": 1574
    },
    {
      "epoch": 0.042306865799935534,
      "grad_norm": 1.4716746807098389,
      "learning_rate": 9.999999847952502e-05,
      "loss": 2.9011,
      "step": 1575
    },
    {
      "epoch": 0.04233372730203073,
      "grad_norm": 1.501473307609558,
      "learning_rate": 9.999999847758748e-05,
      "loss": 3.0257,
      "step": 1576
    },
    {
      "epoch": 0.04236058880412593,
      "grad_norm": 1.5300209522247314,
      "learning_rate": 9.999999847564872e-05,
      "loss": 3.2147,
      "step": 1577
    },
    {
      "epoch": 0.042387450306221125,
      "grad_norm": 1.4848058223724365,
      "learning_rate": 9.999999847370872e-05,
      "loss": 3.0189,
      "step": 1578
    },
    {
      "epoch": 0.04241431180831632,
      "grad_norm": 1.4604220390319824,
      "learning_rate": 9.99999984717675e-05,
      "loss": 3.0825,
      "step": 1579
    },
    {
      "epoch": 0.04244117331041152,
      "grad_norm": 1.4743536710739136,
      "learning_rate": 9.999999846982504e-05,
      "loss": 2.968,
      "step": 1580
    },
    {
      "epoch": 0.042468034812506716,
      "grad_norm": 1.4050565958023071,
      "learning_rate": 9.999999846788133e-05,
      "loss": 3.1072,
      "step": 1581
    },
    {
      "epoch": 0.04249489631460191,
      "grad_norm": 1.6336671113967896,
      "learning_rate": 9.999999846593641e-05,
      "loss": 3.4397,
      "step": 1582
    },
    {
      "epoch": 0.04252175781669711,
      "grad_norm": 1.4869568347930908,
      "learning_rate": 9.999999846399024e-05,
      "loss": 3.0334,
      "step": 1583
    },
    {
      "epoch": 0.04254861931879231,
      "grad_norm": 1.423067331314087,
      "learning_rate": 9.999999846204284e-05,
      "loss": 2.8322,
      "step": 1584
    },
    {
      "epoch": 0.0425754808208875,
      "grad_norm": 1.4375420808792114,
      "learning_rate": 9.999999846009423e-05,
      "loss": 3.0971,
      "step": 1585
    },
    {
      "epoch": 0.042602342322982704,
      "grad_norm": 1.5238673686981201,
      "learning_rate": 9.999999845814435e-05,
      "loss": 3.0143,
      "step": 1586
    },
    {
      "epoch": 0.0426292038250779,
      "grad_norm": 1.4856241941452026,
      "learning_rate": 9.999999845619326e-05,
      "loss": 2.8321,
      "step": 1587
    },
    {
      "epoch": 0.042656065327173094,
      "grad_norm": 1.6609350442886353,
      "learning_rate": 9.999999845424093e-05,
      "loss": 3.2155,
      "step": 1588
    },
    {
      "epoch": 0.042682926829268296,
      "grad_norm": 1.5115586519241333,
      "learning_rate": 9.999999845228736e-05,
      "loss": 3.2471,
      "step": 1589
    },
    {
      "epoch": 0.04270978833136349,
      "grad_norm": 1.7532490491867065,
      "learning_rate": 9.999999845033256e-05,
      "loss": 3.1795,
      "step": 1590
    },
    {
      "epoch": 0.042736649833458686,
      "grad_norm": 1.5894776582717896,
      "learning_rate": 9.999999844837652e-05,
      "loss": 3.123,
      "step": 1591
    },
    {
      "epoch": 0.04276351133555389,
      "grad_norm": 1.6659984588623047,
      "learning_rate": 9.999999844641925e-05,
      "loss": 3.3585,
      "step": 1592
    },
    {
      "epoch": 0.04279037283764908,
      "grad_norm": 2.417973756790161,
      "learning_rate": 9.999999844446075e-05,
      "loss": 3.098,
      "step": 1593
    },
    {
      "epoch": 0.04281723433974428,
      "grad_norm": 1.5774519443511963,
      "learning_rate": 9.999999844250102e-05,
      "loss": 3.1257,
      "step": 1594
    },
    {
      "epoch": 0.04284409584183948,
      "grad_norm": 1.5874313116073608,
      "learning_rate": 9.999999844054006e-05,
      "loss": 3.0155,
      "step": 1595
    },
    {
      "epoch": 0.042870957343934674,
      "grad_norm": 1.5999934673309326,
      "learning_rate": 9.999999843857786e-05,
      "loss": 3.1506,
      "step": 1596
    },
    {
      "epoch": 0.04289781884602987,
      "grad_norm": 1.7312874794006348,
      "learning_rate": 9.999999843661442e-05,
      "loss": 3.351,
      "step": 1597
    },
    {
      "epoch": 0.04292468034812507,
      "grad_norm": 1.7281516790390015,
      "learning_rate": 9.999999843464974e-05,
      "loss": 3.5093,
      "step": 1598
    },
    {
      "epoch": 0.042951541850220265,
      "grad_norm": 1.9014663696289062,
      "learning_rate": 9.999999843268384e-05,
      "loss": 3.5891,
      "step": 1599
    },
    {
      "epoch": 0.04297840335231546,
      "grad_norm": 2.007746458053589,
      "learning_rate": 9.99999984307167e-05,
      "loss": 3.5878,
      "step": 1600
    },
    {
      "epoch": 0.04300526485441066,
      "grad_norm": 1.3168871402740479,
      "learning_rate": 9.999999842874834e-05,
      "loss": 2.8627,
      "step": 1601
    },
    {
      "epoch": 0.043032126356505856,
      "grad_norm": 1.3844327926635742,
      "learning_rate": 9.999999842677874e-05,
      "loss": 2.8701,
      "step": 1602
    },
    {
      "epoch": 0.04305898785860105,
      "grad_norm": 1.3635506629943848,
      "learning_rate": 9.999999842480789e-05,
      "loss": 2.6733,
      "step": 1603
    },
    {
      "epoch": 0.04308584936069625,
      "grad_norm": 1.3588570356369019,
      "learning_rate": 9.999999842283582e-05,
      "loss": 2.8806,
      "step": 1604
    },
    {
      "epoch": 0.04311271086279145,
      "grad_norm": 1.3737713098526,
      "learning_rate": 9.999999842086252e-05,
      "loss": 2.9025,
      "step": 1605
    },
    {
      "epoch": 0.04313957236488664,
      "grad_norm": 1.40369713306427,
      "learning_rate": 9.999999841888799e-05,
      "loss": 2.8484,
      "step": 1606
    },
    {
      "epoch": 0.043166433866981845,
      "grad_norm": 1.2884548902511597,
      "learning_rate": 9.99999984169122e-05,
      "loss": 2.7603,
      "step": 1607
    },
    {
      "epoch": 0.04319329536907704,
      "grad_norm": 1.351904273033142,
      "learning_rate": 9.99999984149352e-05,
      "loss": 3.0526,
      "step": 1608
    },
    {
      "epoch": 0.043220156871172234,
      "grad_norm": 1.4556187391281128,
      "learning_rate": 9.999999841295697e-05,
      "loss": 3.0934,
      "step": 1609
    },
    {
      "epoch": 0.043247018373267436,
      "grad_norm": 1.3308874368667603,
      "learning_rate": 9.999999841097748e-05,
      "loss": 2.7632,
      "step": 1610
    },
    {
      "epoch": 0.04327387987536263,
      "grad_norm": 1.3838894367218018,
      "learning_rate": 9.999999840899678e-05,
      "loss": 3.0159,
      "step": 1611
    },
    {
      "epoch": 0.043300741377457826,
      "grad_norm": 1.504431962966919,
      "learning_rate": 9.999999840701483e-05,
      "loss": 3.0511,
      "step": 1612
    },
    {
      "epoch": 0.04332760287955303,
      "grad_norm": 1.2878671884536743,
      "learning_rate": 9.999999840503167e-05,
      "loss": 2.8596,
      "step": 1613
    },
    {
      "epoch": 0.04335446438164822,
      "grad_norm": 1.3833988904953003,
      "learning_rate": 9.999999840304725e-05,
      "loss": 2.7807,
      "step": 1614
    },
    {
      "epoch": 0.04338132588374342,
      "grad_norm": 1.3519421815872192,
      "learning_rate": 9.999999840106163e-05,
      "loss": 2.7777,
      "step": 1615
    },
    {
      "epoch": 0.04340818738583862,
      "grad_norm": 1.4081404209136963,
      "learning_rate": 9.999999839907475e-05,
      "loss": 2.7198,
      "step": 1616
    },
    {
      "epoch": 0.043435048887933814,
      "grad_norm": 1.3711954355239868,
      "learning_rate": 9.999999839708662e-05,
      "loss": 2.717,
      "step": 1617
    },
    {
      "epoch": 0.04346191039002901,
      "grad_norm": 1.3935779333114624,
      "learning_rate": 9.999999839509728e-05,
      "loss": 3.0255,
      "step": 1618
    },
    {
      "epoch": 0.04348877189212421,
      "grad_norm": 1.4566105604171753,
      "learning_rate": 9.99999983931067e-05,
      "loss": 3.1218,
      "step": 1619
    },
    {
      "epoch": 0.043515633394219405,
      "grad_norm": 1.4367201328277588,
      "learning_rate": 9.99999983911149e-05,
      "loss": 3.0931,
      "step": 1620
    },
    {
      "epoch": 0.0435424948963146,
      "grad_norm": 1.390453815460205,
      "learning_rate": 9.999999838912187e-05,
      "loss": 2.903,
      "step": 1621
    },
    {
      "epoch": 0.0435693563984098,
      "grad_norm": 1.2955039739608765,
      "learning_rate": 9.999999838712759e-05,
      "loss": 2.9835,
      "step": 1622
    },
    {
      "epoch": 0.043596217900505,
      "grad_norm": 1.5156861543655396,
      "learning_rate": 9.999999838513207e-05,
      "loss": 3.1388,
      "step": 1623
    },
    {
      "epoch": 0.04362307940260019,
      "grad_norm": 1.4825620651245117,
      "learning_rate": 9.999999838313532e-05,
      "loss": 3.1583,
      "step": 1624
    },
    {
      "epoch": 0.04364994090469539,
      "grad_norm": 1.369698166847229,
      "learning_rate": 9.999999838113734e-05,
      "loss": 2.9362,
      "step": 1625
    },
    {
      "epoch": 0.04367680240679059,
      "grad_norm": 1.3391295671463013,
      "learning_rate": 9.999999837913813e-05,
      "loss": 2.8266,
      "step": 1626
    },
    {
      "epoch": 0.04370366390888578,
      "grad_norm": 1.4127203226089478,
      "learning_rate": 9.999999837713768e-05,
      "loss": 2.827,
      "step": 1627
    },
    {
      "epoch": 0.043730525410980985,
      "grad_norm": 1.542043685913086,
      "learning_rate": 9.999999837513601e-05,
      "loss": 3.0887,
      "step": 1628
    },
    {
      "epoch": 0.04375738691307618,
      "grad_norm": 1.513839602470398,
      "learning_rate": 9.99999983731331e-05,
      "loss": 3.0853,
      "step": 1629
    },
    {
      "epoch": 0.043784248415171374,
      "grad_norm": 1.4729801416397095,
      "learning_rate": 9.999999837112895e-05,
      "loss": 2.98,
      "step": 1630
    },
    {
      "epoch": 0.043811109917266576,
      "grad_norm": 1.509283185005188,
      "learning_rate": 9.999999836912355e-05,
      "loss": 3.2404,
      "step": 1631
    },
    {
      "epoch": 0.04383797141936177,
      "grad_norm": 1.543927550315857,
      "learning_rate": 9.999999836711694e-05,
      "loss": 3.119,
      "step": 1632
    },
    {
      "epoch": 0.043864832921456966,
      "grad_norm": 1.5025025606155396,
      "learning_rate": 9.999999836510909e-05,
      "loss": 3.1039,
      "step": 1633
    },
    {
      "epoch": 0.04389169442355217,
      "grad_norm": 1.3757987022399902,
      "learning_rate": 9.999999836310001e-05,
      "loss": 2.8637,
      "step": 1634
    },
    {
      "epoch": 0.04391855592564736,
      "grad_norm": 1.5544795989990234,
      "learning_rate": 9.99999983610897e-05,
      "loss": 3.0113,
      "step": 1635
    },
    {
      "epoch": 0.04394541742774256,
      "grad_norm": 1.4375749826431274,
      "learning_rate": 9.999999835907815e-05,
      "loss": 2.9539,
      "step": 1636
    },
    {
      "epoch": 0.04397227892983776,
      "grad_norm": 1.557188868522644,
      "learning_rate": 9.999999835706537e-05,
      "loss": 3.1977,
      "step": 1637
    },
    {
      "epoch": 0.043999140431932954,
      "grad_norm": 1.6400185823440552,
      "learning_rate": 9.999999835505136e-05,
      "loss": 3.1995,
      "step": 1638
    },
    {
      "epoch": 0.04402600193402815,
      "grad_norm": 1.4655009508132935,
      "learning_rate": 9.99999983530361e-05,
      "loss": 2.9985,
      "step": 1639
    },
    {
      "epoch": 0.04405286343612335,
      "grad_norm": 1.4965097904205322,
      "learning_rate": 9.999999835101961e-05,
      "loss": 3.1393,
      "step": 1640
    },
    {
      "epoch": 0.044079724938218545,
      "grad_norm": 1.5797890424728394,
      "learning_rate": 9.999999834900189e-05,
      "loss": 3.0014,
      "step": 1641
    },
    {
      "epoch": 0.04410658644031374,
      "grad_norm": 1.567730188369751,
      "learning_rate": 9.999999834698295e-05,
      "loss": 2.9215,
      "step": 1642
    },
    {
      "epoch": 0.04413344794240894,
      "grad_norm": 1.6687631607055664,
      "learning_rate": 9.999999834496276e-05,
      "loss": 3.2531,
      "step": 1643
    },
    {
      "epoch": 0.04416030944450414,
      "grad_norm": 1.6272687911987305,
      "learning_rate": 9.999999834294133e-05,
      "loss": 3.3399,
      "step": 1644
    },
    {
      "epoch": 0.04418717094659933,
      "grad_norm": 1.6043704748153687,
      "learning_rate": 9.999999834091869e-05,
      "loss": 3.0772,
      "step": 1645
    },
    {
      "epoch": 0.04421403244869453,
      "grad_norm": 1.6957578659057617,
      "learning_rate": 9.999999833889479e-05,
      "loss": 3.3066,
      "step": 1646
    },
    {
      "epoch": 0.04424089395078973,
      "grad_norm": 1.7268688678741455,
      "learning_rate": 9.999999833686968e-05,
      "loss": 3.3125,
      "step": 1647
    },
    {
      "epoch": 0.04426775545288492,
      "grad_norm": 1.7406071424484253,
      "learning_rate": 9.999999833484333e-05,
      "loss": 3.2264,
      "step": 1648
    },
    {
      "epoch": 0.044294616954980125,
      "grad_norm": 1.8623542785644531,
      "learning_rate": 9.999999833281574e-05,
      "loss": 3.3522,
      "step": 1649
    },
    {
      "epoch": 0.04432147845707532,
      "grad_norm": 1.8407955169677734,
      "learning_rate": 9.999999833078691e-05,
      "loss": 3.3308,
      "step": 1650
    },
    {
      "epoch": 0.044348339959170514,
      "grad_norm": 1.4079015254974365,
      "learning_rate": 9.999999832875686e-05,
      "loss": 2.9442,
      "step": 1651
    },
    {
      "epoch": 0.044375201461265716,
      "grad_norm": 1.4412444829940796,
      "learning_rate": 9.999999832672557e-05,
      "loss": 3.038,
      "step": 1652
    },
    {
      "epoch": 0.04440206296336091,
      "grad_norm": 1.4149682521820068,
      "learning_rate": 9.999999832469304e-05,
      "loss": 2.7261,
      "step": 1653
    },
    {
      "epoch": 0.044428924465456106,
      "grad_norm": 1.411676287651062,
      "learning_rate": 9.99999983226593e-05,
      "loss": 2.9101,
      "step": 1654
    },
    {
      "epoch": 0.04445578596755131,
      "grad_norm": 1.470566987991333,
      "learning_rate": 9.99999983206243e-05,
      "loss": 2.77,
      "step": 1655
    },
    {
      "epoch": 0.0444826474696465,
      "grad_norm": 1.397863507270813,
      "learning_rate": 9.999999831858808e-05,
      "loss": 2.9582,
      "step": 1656
    },
    {
      "epoch": 0.0445095089717417,
      "grad_norm": 1.3953698873519897,
      "learning_rate": 9.999999831655063e-05,
      "loss": 3.0912,
      "step": 1657
    },
    {
      "epoch": 0.0445363704738369,
      "grad_norm": 1.3832037448883057,
      "learning_rate": 9.999999831451192e-05,
      "loss": 2.8913,
      "step": 1658
    },
    {
      "epoch": 0.044563231975932094,
      "grad_norm": 1.3912192583084106,
      "learning_rate": 9.9999998312472e-05,
      "loss": 2.6594,
      "step": 1659
    },
    {
      "epoch": 0.04459009347802729,
      "grad_norm": 1.3151278495788574,
      "learning_rate": 9.999999831043084e-05,
      "loss": 2.6635,
      "step": 1660
    },
    {
      "epoch": 0.04461695498012249,
      "grad_norm": 1.2868192195892334,
      "learning_rate": 9.999999830838846e-05,
      "loss": 2.7065,
      "step": 1661
    },
    {
      "epoch": 0.044643816482217685,
      "grad_norm": 1.387132167816162,
      "learning_rate": 9.999999830634483e-05,
      "loss": 2.6766,
      "step": 1662
    },
    {
      "epoch": 0.04467067798431288,
      "grad_norm": 1.4079433679580688,
      "learning_rate": 9.999999830429997e-05,
      "loss": 2.8444,
      "step": 1663
    },
    {
      "epoch": 0.04469753948640808,
      "grad_norm": 1.4303300380706787,
      "learning_rate": 9.999999830225387e-05,
      "loss": 2.9663,
      "step": 1664
    },
    {
      "epoch": 0.04472440098850328,
      "grad_norm": 1.39901864528656,
      "learning_rate": 9.999999830020654e-05,
      "loss": 2.9923,
      "step": 1665
    },
    {
      "epoch": 0.04475126249059847,
      "grad_norm": 1.4272572994232178,
      "learning_rate": 9.999999829815798e-05,
      "loss": 2.899,
      "step": 1666
    },
    {
      "epoch": 0.04477812399269367,
      "grad_norm": 1.492218017578125,
      "learning_rate": 9.99999982961082e-05,
      "loss": 3.0391,
      "step": 1667
    },
    {
      "epoch": 0.04480498549478887,
      "grad_norm": 1.450843095779419,
      "learning_rate": 9.999999829405716e-05,
      "loss": 2.9127,
      "step": 1668
    },
    {
      "epoch": 0.04483184699688406,
      "grad_norm": 1.3440508842468262,
      "learning_rate": 9.99999982920049e-05,
      "loss": 3.0001,
      "step": 1669
    },
    {
      "epoch": 0.044858708498979265,
      "grad_norm": 1.4402127265930176,
      "learning_rate": 9.999999828995141e-05,
      "loss": 2.819,
      "step": 1670
    },
    {
      "epoch": 0.04488557000107446,
      "grad_norm": 1.4429413080215454,
      "learning_rate": 9.999999828789667e-05,
      "loss": 3.0246,
      "step": 1671
    },
    {
      "epoch": 0.044912431503169654,
      "grad_norm": 2.6085400581359863,
      "learning_rate": 9.999999828584071e-05,
      "loss": 2.9273,
      "step": 1672
    },
    {
      "epoch": 0.044939293005264856,
      "grad_norm": 1.5993056297302246,
      "learning_rate": 9.999999828378352e-05,
      "loss": 3.2987,
      "step": 1673
    },
    {
      "epoch": 0.04496615450736005,
      "grad_norm": 1.5255417823791504,
      "learning_rate": 9.999999828172508e-05,
      "loss": 2.9552,
      "step": 1674
    },
    {
      "epoch": 0.044993016009455246,
      "grad_norm": 1.557896614074707,
      "learning_rate": 9.999999827966542e-05,
      "loss": 3.123,
      "step": 1675
    },
    {
      "epoch": 0.04501987751155045,
      "grad_norm": 1.558122992515564,
      "learning_rate": 9.999999827760452e-05,
      "loss": 2.9187,
      "step": 1676
    },
    {
      "epoch": 0.04504673901364564,
      "grad_norm": 1.425349473953247,
      "learning_rate": 9.999999827554239e-05,
      "loss": 2.899,
      "step": 1677
    },
    {
      "epoch": 0.04507360051574084,
      "grad_norm": 1.5926076173782349,
      "learning_rate": 9.999999827347903e-05,
      "loss": 3.2055,
      "step": 1678
    },
    {
      "epoch": 0.04510046201783604,
      "grad_norm": 1.6083916425704956,
      "learning_rate": 9.999999827141443e-05,
      "loss": 2.8889,
      "step": 1679
    },
    {
      "epoch": 0.045127323519931234,
      "grad_norm": 1.4613057374954224,
      "learning_rate": 9.999999826934859e-05,
      "loss": 2.968,
      "step": 1680
    },
    {
      "epoch": 0.04515418502202643,
      "grad_norm": 1.591672420501709,
      "learning_rate": 9.999999826728153e-05,
      "loss": 3.0252,
      "step": 1681
    },
    {
      "epoch": 0.04518104652412163,
      "grad_norm": 1.539560317993164,
      "learning_rate": 9.999999826521322e-05,
      "loss": 2.8637,
      "step": 1682
    },
    {
      "epoch": 0.045207908026216825,
      "grad_norm": 1.4949159622192383,
      "learning_rate": 9.99999982631437e-05,
      "loss": 3.05,
      "step": 1683
    },
    {
      "epoch": 0.04523476952831202,
      "grad_norm": 1.5287278890609741,
      "learning_rate": 9.999999826107294e-05,
      "loss": 3.0176,
      "step": 1684
    },
    {
      "epoch": 0.04526163103040722,
      "grad_norm": 1.4513976573944092,
      "learning_rate": 9.999999825900092e-05,
      "loss": 3.0043,
      "step": 1685
    },
    {
      "epoch": 0.04528849253250242,
      "grad_norm": 1.4938799142837524,
      "learning_rate": 9.99999982569277e-05,
      "loss": 2.998,
      "step": 1686
    },
    {
      "epoch": 0.04531535403459761,
      "grad_norm": 1.6789308786392212,
      "learning_rate": 9.999999825485323e-05,
      "loss": 3.108,
      "step": 1687
    },
    {
      "epoch": 0.04534221553669281,
      "grad_norm": 1.4213138818740845,
      "learning_rate": 9.999999825277752e-05,
      "loss": 3.0397,
      "step": 1688
    },
    {
      "epoch": 0.04536907703878801,
      "grad_norm": 1.5696649551391602,
      "learning_rate": 9.999999825070058e-05,
      "loss": 3.1145,
      "step": 1689
    },
    {
      "epoch": 0.0453959385408832,
      "grad_norm": 1.6566909551620483,
      "learning_rate": 9.999999824862241e-05,
      "loss": 3.3197,
      "step": 1690
    },
    {
      "epoch": 0.045422800042978405,
      "grad_norm": 1.591908574104309,
      "learning_rate": 9.999999824654302e-05,
      "loss": 2.9155,
      "step": 1691
    },
    {
      "epoch": 0.0454496615450736,
      "grad_norm": 1.6618692874908447,
      "learning_rate": 9.999999824446238e-05,
      "loss": 3.3152,
      "step": 1692
    },
    {
      "epoch": 0.045476523047168795,
      "grad_norm": 1.660921573638916,
      "learning_rate": 9.999999824238051e-05,
      "loss": 3.1987,
      "step": 1693
    },
    {
      "epoch": 0.045503384549263996,
      "grad_norm": 1.680649757385254,
      "learning_rate": 9.99999982402974e-05,
      "loss": 3.2139,
      "step": 1694
    },
    {
      "epoch": 0.04553024605135919,
      "grad_norm": 1.7884151935577393,
      "learning_rate": 9.999999823821307e-05,
      "loss": 3.2991,
      "step": 1695
    },
    {
      "epoch": 0.045557107553454386,
      "grad_norm": 1.6756354570388794,
      "learning_rate": 9.99999982361275e-05,
      "loss": 3.1816,
      "step": 1696
    },
    {
      "epoch": 0.04558396905554959,
      "grad_norm": 1.6229280233383179,
      "learning_rate": 9.999999823404069e-05,
      "loss": 3.2563,
      "step": 1697
    },
    {
      "epoch": 0.04561083055764478,
      "grad_norm": 1.625819444656372,
      "learning_rate": 9.999999823195265e-05,
      "loss": 3.1226,
      "step": 1698
    },
    {
      "epoch": 0.04563769205973998,
      "grad_norm": 1.7141685485839844,
      "learning_rate": 9.999999822986337e-05,
      "loss": 3.3355,
      "step": 1699
    },
    {
      "epoch": 0.04566455356183518,
      "grad_norm": 1.7747595310211182,
      "learning_rate": 9.999999822777286e-05,
      "loss": 3.2331,
      "step": 1700
    }
  ],
  "logging_steps": 1,
  "max_steps": 20000000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 538,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.320207374503117e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}