{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9766839378238341,
  "eval_steps": 97,
  "global_step": 772,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0025906735751295338,
      "grad_norm": 0.37030652165412903,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.4535,
      "step": 1
    },
    {
      "epoch": 0.0025906735751295338,
      "eval_loss": 1.3420950174331665,
      "eval_runtime": 98.9896,
      "eval_samples_per_second": 2.526,
      "eval_steps_per_second": 0.323,
      "step": 1
    },
    {
      "epoch": 0.0051813471502590676,
      "grad_norm": 0.37870243191719055,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.3744,
      "step": 2
    },
    {
      "epoch": 0.007772020725388601,
      "grad_norm": 0.3809736669063568,
      "learning_rate": 3e-06,
      "loss": 1.413,
      "step": 3
    },
    {
      "epoch": 0.010362694300518135,
      "grad_norm": 0.3608606457710266,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.4169,
      "step": 4
    },
    {
      "epoch": 0.012953367875647668,
      "grad_norm": 0.3465673625469208,
      "learning_rate": 5e-06,
      "loss": 1.3166,
      "step": 5
    },
    {
      "epoch": 0.015544041450777202,
      "grad_norm": 0.36601942777633667,
      "learning_rate": 6e-06,
      "loss": 1.4029,
      "step": 6
    },
    {
      "epoch": 0.018134715025906734,
      "grad_norm": 0.374077171087265,
      "learning_rate": 7e-06,
      "loss": 1.4351,
      "step": 7
    },
    {
      "epoch": 0.02072538860103627,
      "grad_norm": 0.3928413391113281,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.3023,
      "step": 8
    },
    {
      "epoch": 0.023316062176165803,
      "grad_norm": 0.35695162415504456,
      "learning_rate": 9e-06,
      "loss": 1.4251,
      "step": 9
    },
    {
      "epoch": 0.025906735751295335,
      "grad_norm": 0.38057687878608704,
      "learning_rate": 1e-05,
      "loss": 1.4864,
      "step": 10
    },
    {
      "epoch": 0.02849740932642487,
      "grad_norm": 0.3801506757736206,
      "learning_rate": 9.999957505845144e-06,
      "loss": 1.4257,
      "step": 11
    },
    {
      "epoch": 0.031088082901554404,
      "grad_norm": 0.3910161852836609,
      "learning_rate": 9.999830024102874e-06,
      "loss": 1.456,
      "step": 12
    },
    {
      "epoch": 0.03367875647668394,
      "grad_norm": 0.37710297107696533,
      "learning_rate": 9.999617556940085e-06,
      "loss": 1.3441,
      "step": 13
    },
    {
      "epoch": 0.03626943005181347,
      "grad_norm": 0.4258323311805725,
      "learning_rate": 9.99932010796822e-06,
      "loss": 1.3589,
      "step": 14
    },
    {
      "epoch": 0.038860103626943004,
      "grad_norm": 0.42138487100601196,
      "learning_rate": 9.998937682243216e-06,
      "loss": 1.4402,
      "step": 15
    },
    {
      "epoch": 0.04145077720207254,
      "grad_norm": 0.45730850100517273,
      "learning_rate": 9.998470286265415e-06,
      "loss": 1.3916,
      "step": 16
    },
    {
      "epoch": 0.04404145077720207,
      "grad_norm": 0.4608759582042694,
      "learning_rate": 9.99791792797946e-06,
      "loss": 1.4034,
      "step": 17
    },
    {
      "epoch": 0.046632124352331605,
      "grad_norm": 0.45245274901390076,
      "learning_rate": 9.997280616774147e-06,
      "loss": 1.3311,
      "step": 18
    },
    {
      "epoch": 0.04922279792746114,
      "grad_norm": 0.462851881980896,
      "learning_rate": 9.996558363482277e-06,
      "loss": 1.4076,
      "step": 19
    },
    {
      "epoch": 0.05181347150259067,
      "grad_norm": 0.5080270767211914,
      "learning_rate": 9.995751180380468e-06,
      "loss": 1.4116,
      "step": 20
    },
    {
      "epoch": 0.054404145077720206,
      "grad_norm": 0.44085511565208435,
      "learning_rate": 9.994859081188944e-06,
      "loss": 1.399,
      "step": 21
    },
    {
      "epoch": 0.05699481865284974,
      "grad_norm": 0.40889567136764526,
      "learning_rate": 9.993882081071307e-06,
      "loss": 1.3666,
      "step": 22
    },
    {
      "epoch": 0.05958549222797927,
      "grad_norm": 0.42577946186065674,
      "learning_rate": 9.992820196634274e-06,
      "loss": 1.3906,
      "step": 23
    },
    {
      "epoch": 0.06217616580310881,
      "grad_norm": 0.46674349904060364,
      "learning_rate": 9.991673445927399e-06,
      "loss": 1.3446,
      "step": 24
    },
    {
      "epoch": 0.06476683937823834,
      "grad_norm": 0.48815953731536865,
      "learning_rate": 9.99044184844276e-06,
      "loss": 1.3156,
      "step": 25
    },
    {
      "epoch": 0.06735751295336788,
      "grad_norm": 0.4857616424560547,
      "learning_rate": 9.989125425114639e-06,
      "loss": 1.3713,
      "step": 26
    },
    {
      "epoch": 0.06994818652849741,
      "grad_norm": 0.4946143925189972,
      "learning_rate": 9.98772419831915e-06,
      "loss": 1.4438,
      "step": 27
    },
    {
      "epoch": 0.07253886010362694,
      "grad_norm": 0.47251901030540466,
      "learning_rate": 9.986238191873874e-06,
      "loss": 1.3626,
      "step": 28
    },
    {
      "epoch": 0.07512953367875648,
      "grad_norm": 0.5448423027992249,
      "learning_rate": 9.984667431037448e-06,
      "loss": 1.3201,
      "step": 29
    },
    {
      "epoch": 0.07772020725388601,
      "grad_norm": 0.4412786364555359,
      "learning_rate": 9.983011942509131e-06,
      "loss": 1.3581,
      "step": 30
    },
    {
      "epoch": 0.08031088082901554,
      "grad_norm": 0.44406625628471375,
      "learning_rate": 9.981271754428361e-06,
      "loss": 1.2458,
      "step": 31
    },
    {
      "epoch": 0.08290155440414508,
      "grad_norm": 0.5035629272460938,
      "learning_rate": 9.979446896374264e-06,
      "loss": 1.3699,
      "step": 32
    },
    {
      "epoch": 0.08549222797927461,
      "grad_norm": 0.47425153851509094,
      "learning_rate": 9.977537399365159e-06,
      "loss": 1.3558,
      "step": 33
    },
    {
      "epoch": 0.08808290155440414,
      "grad_norm": 0.5448233485221863,
      "learning_rate": 9.975543295858035e-06,
      "loss": 1.3342,
      "step": 34
    },
    {
      "epoch": 0.09067357512953368,
      "grad_norm": 0.49746477603912354,
      "learning_rate": 9.973464619747983e-06,
      "loss": 1.3138,
      "step": 35
    },
    {
      "epoch": 0.09326424870466321,
      "grad_norm": 0.48222672939300537,
      "learning_rate": 9.971301406367644e-06,
      "loss": 1.3991,
      "step": 36
    },
    {
      "epoch": 0.09585492227979274,
      "grad_norm": 0.5209602117538452,
      "learning_rate": 9.969053692486582e-06,
      "loss": 1.2775,
      "step": 37
    },
    {
      "epoch": 0.09844559585492228,
      "grad_norm": 0.4517224133014679,
      "learning_rate": 9.966721516310683e-06,
      "loss": 1.3207,
      "step": 38
    },
    {
      "epoch": 0.10103626943005181,
      "grad_norm": 0.5346600413322449,
      "learning_rate": 9.964304917481482e-06,
      "loss": 1.2108,
      "step": 39
    },
    {
      "epoch": 0.10362694300518134,
      "grad_norm": 0.5563952326774597,
      "learning_rate": 9.961803937075516e-06,
      "loss": 1.2934,
      "step": 40
    },
    {
      "epoch": 0.10621761658031088,
      "grad_norm": 0.5306499004364014,
      "learning_rate": 9.959218617603601e-06,
      "loss": 1.2202,
      "step": 41
    },
    {
      "epoch": 0.10880829015544041,
      "grad_norm": 0.4907849133014679,
      "learning_rate": 9.956549003010122e-06,
      "loss": 1.2261,
      "step": 42
    },
    {
      "epoch": 0.11139896373056994,
      "grad_norm": 0.5063201189041138,
      "learning_rate": 9.953795138672291e-06,
      "loss": 1.2863,
      "step": 43
    },
    {
      "epoch": 0.11398963730569948,
      "grad_norm": 0.5338860750198364,
      "learning_rate": 9.950957071399357e-06,
      "loss": 1.2388,
      "step": 44
    },
    {
      "epoch": 0.11658031088082901,
      "grad_norm": 0.5024104714393616,
      "learning_rate": 9.948034849431831e-06,
      "loss": 1.2573,
      "step": 45
    },
    {
      "epoch": 0.11917098445595854,
      "grad_norm": 0.5430374145507812,
      "learning_rate": 9.945028522440654e-06,
      "loss": 1.1556,
      "step": 46
    },
    {
      "epoch": 0.12176165803108809,
      "grad_norm": 0.4689851701259613,
      "learning_rate": 9.941938141526355e-06,
      "loss": 1.2576,
      "step": 47
    },
    {
      "epoch": 0.12435233160621761,
      "grad_norm": 0.4983338415622711,
      "learning_rate": 9.938763759218186e-06,
      "loss": 1.2247,
      "step": 48
    },
    {
      "epoch": 0.12694300518134716,
      "grad_norm": 0.47061145305633545,
      "learning_rate": 9.935505429473221e-06,
      "loss": 1.2173,
      "step": 49
    },
    {
      "epoch": 0.12953367875647667,
      "grad_norm": 0.5208876729011536,
      "learning_rate": 9.93216320767545e-06,
      "loss": 1.1284,
      "step": 50
    },
    {
      "epoch": 0.13212435233160622,
      "grad_norm": 0.5343345999717712,
      "learning_rate": 9.92873715063483e-06,
      "loss": 1.2766,
      "step": 51
    },
    {
      "epoch": 0.13471502590673576,
      "grad_norm": 0.5376512408256531,
      "learning_rate": 9.925227316586316e-06,
      "loss": 1.3082,
      "step": 52
    },
    {
      "epoch": 0.13730569948186527,
      "grad_norm": 0.49544206261634827,
      "learning_rate": 9.921633765188887e-06,
      "loss": 1.2488,
      "step": 53
    },
    {
      "epoch": 0.13989637305699482,
      "grad_norm": 0.5503423810005188,
      "learning_rate": 9.917956557524511e-06,
      "loss": 1.2837,
      "step": 54
    },
    {
      "epoch": 0.14248704663212436,
      "grad_norm": 0.6678252816200256,
      "learning_rate": 9.91419575609712e-06,
      "loss": 1.2592,
      "step": 55
    },
    {
      "epoch": 0.14507772020725387,
      "grad_norm": 0.5491769313812256,
      "learning_rate": 9.910351424831545e-06,
      "loss": 1.2528,
      "step": 56
    },
    {
      "epoch": 0.14766839378238342,
      "grad_norm": 0.5200887322425842,
      "learning_rate": 9.906423629072435e-06,
      "loss": 1.1433,
      "step": 57
    },
    {
      "epoch": 0.15025906735751296,
      "grad_norm": 0.5037204623222351,
      "learning_rate": 9.902412435583127e-06,
      "loss": 1.204,
      "step": 58
    },
    {
      "epoch": 0.15284974093264247,
      "grad_norm": 0.62909996509552,
      "learning_rate": 9.898317912544537e-06,
      "loss": 1.1012,
      "step": 59
    },
    {
      "epoch": 0.15544041450777202,
      "grad_norm": 0.5567560791969299,
      "learning_rate": 9.89414012955398e-06,
      "loss": 1.2056,
      "step": 60
    },
    {
      "epoch": 0.15803108808290156,
      "grad_norm": 0.4932163655757904,
      "learning_rate": 9.889879157624003e-06,
      "loss": 1.1295,
      "step": 61
    },
    {
      "epoch": 0.16062176165803108,
      "grad_norm": 0.49116799235343933,
      "learning_rate": 9.885535069181163e-06,
      "loss": 1.1767,
      "step": 62
    },
    {
      "epoch": 0.16321243523316062,
      "grad_norm": 0.5316625833511353,
      "learning_rate": 9.881107938064806e-06,
      "loss": 1.2038,
      "step": 63
    },
    {
      "epoch": 0.16580310880829016,
      "grad_norm": 0.5318853259086609,
      "learning_rate": 9.876597839525814e-06,
      "loss": 1.1865,
      "step": 64
    },
    {
      "epoch": 0.16839378238341968,
      "grad_norm": 0.535534143447876,
      "learning_rate": 9.872004850225313e-06,
      "loss": 1.2335,
      "step": 65
    },
    {
      "epoch": 0.17098445595854922,
      "grad_norm": 0.6383189558982849,
      "learning_rate": 9.867329048233387e-06,
      "loss": 1.1985,
      "step": 66
    },
    {
      "epoch": 0.17357512953367876,
      "grad_norm": 0.5795570611953735,
      "learning_rate": 9.862570513027736e-06,
      "loss": 1.1598,
      "step": 67
    },
    {
      "epoch": 0.17616580310880828,
      "grad_norm": 0.5034987330436707,
      "learning_rate": 9.857729325492329e-06,
      "loss": 1.1798,
      "step": 68
    },
    {
      "epoch": 0.17875647668393782,
      "grad_norm": 0.5522433519363403,
      "learning_rate": 9.85280556791604e-06,
      "loss": 1.1093,
      "step": 69
    },
    {
      "epoch": 0.18134715025906736,
      "grad_norm": 0.5660341382026672,
      "learning_rate": 9.847799323991234e-06,
      "loss": 1.1477,
      "step": 70
    },
    {
      "epoch": 0.18393782383419688,
      "grad_norm": 0.529484748840332,
      "learning_rate": 9.842710678812352e-06,
      "loss": 1.1028,
      "step": 71
    },
    {
      "epoch": 0.18652849740932642,
      "grad_norm": 0.4907700717449188,
      "learning_rate": 9.837539718874466e-06,
      "loss": 1.2247,
      "step": 72
    },
    {
      "epoch": 0.18911917098445596,
      "grad_norm": 0.5624697804450989,
      "learning_rate": 9.832286532071802e-06,
      "loss": 1.1997,
      "step": 73
    },
    {
      "epoch": 0.19170984455958548,
      "grad_norm": 0.5344734787940979,
      "learning_rate": 9.826951207696258e-06,
      "loss": 1.1389,
      "step": 74
    },
    {
      "epoch": 0.19430051813471502,
      "grad_norm": 0.5770263671875,
      "learning_rate": 9.82153383643587e-06,
      "loss": 1.2063,
      "step": 75
    },
    {
      "epoch": 0.19689119170984457,
      "grad_norm": 0.5622824430465698,
      "learning_rate": 9.816034510373287e-06,
      "loss": 1.1715,
      "step": 76
    },
    {
      "epoch": 0.19948186528497408,
      "grad_norm": 0.5856156945228577,
      "learning_rate": 9.81045332298419e-06,
      "loss": 1.155,
      "step": 77
    },
    {
      "epoch": 0.20207253886010362,
      "grad_norm": 0.5373365879058838,
      "learning_rate": 9.804790369135719e-06,
      "loss": 1.1258,
      "step": 78
    },
    {
      "epoch": 0.20466321243523317,
      "grad_norm": 0.5527685284614563,
      "learning_rate": 9.799045745084848e-06,
      "loss": 1.1948,
      "step": 79
    },
    {
      "epoch": 0.20725388601036268,
      "grad_norm": 0.5555474162101746,
      "learning_rate": 9.793219548476754e-06,
      "loss": 1.1871,
      "step": 80
    },
    {
      "epoch": 0.20984455958549222,
      "grad_norm": 0.5365428328514099,
      "learning_rate": 9.787311878343158e-06,
      "loss": 1.1591,
      "step": 81
    },
    {
      "epoch": 0.21243523316062177,
      "grad_norm": 0.5377835631370544,
      "learning_rate": 9.781322835100639e-06,
      "loss": 1.1238,
      "step": 82
    },
    {
      "epoch": 0.21502590673575128,
      "grad_norm": 0.5349002480506897,
      "learning_rate": 9.77525252054893e-06,
      "loss": 1.181,
      "step": 83
    },
    {
      "epoch": 0.21761658031088082,
      "grad_norm": 0.6802361607551575,
      "learning_rate": 9.769101037869187e-06,
      "loss": 1.2205,
      "step": 84
    },
    {
      "epoch": 0.22020725388601037,
      "grad_norm": 0.5526275038719177,
      "learning_rate": 9.762868491622229e-06,
      "loss": 1.1173,
      "step": 85
    },
    {
      "epoch": 0.22279792746113988,
      "grad_norm": 0.6218620538711548,
      "learning_rate": 9.756554987746777e-06,
      "loss": 1.1103,
      "step": 86
    },
    {
      "epoch": 0.22538860103626943,
      "grad_norm": 0.6305224299430847,
      "learning_rate": 9.750160633557626e-06,
      "loss": 1.132,
      "step": 87
    },
    {
      "epoch": 0.22797927461139897,
      "grad_norm": 0.6259211301803589,
      "learning_rate": 9.743685537743856e-06,
      "loss": 1.1183,
      "step": 88
    },
    {
      "epoch": 0.23056994818652848,
      "grad_norm": 0.866215169429779,
      "learning_rate": 9.737129810366952e-06,
      "loss": 1.1117,
      "step": 89
    },
    {
      "epoch": 0.23316062176165803,
      "grad_norm": 0.700805127620697,
      "learning_rate": 9.730493562858954e-06,
      "loss": 1.0444,
      "step": 90
    },
    {
      "epoch": 0.23575129533678757,
      "grad_norm": 0.630706250667572,
      "learning_rate": 9.72377690802055e-06,
      "loss": 1.2306,
      "step": 91
    },
    {
      "epoch": 0.23834196891191708,
      "grad_norm": 0.5979788899421692,
      "learning_rate": 9.716979960019173e-06,
      "loss": 1.1418,
      "step": 92
    },
    {
      "epoch": 0.24093264248704663,
      "grad_norm": 0.496480792760849,
      "learning_rate": 9.710102834387043e-06,
      "loss": 1.0891,
      "step": 93
    },
    {
      "epoch": 0.24352331606217617,
      "grad_norm": 0.5865656137466431,
      "learning_rate": 9.70314564801922e-06,
      "loss": 1.0926,
      "step": 94
    },
    {
      "epoch": 0.24611398963730569,
      "grad_norm": 0.6677915453910828,
      "learning_rate": 9.696108519171605e-06,
      "loss": 1.0742,
      "step": 95
    },
    {
      "epoch": 0.24870466321243523,
      "grad_norm": 0.6176425218582153,
      "learning_rate": 9.688991567458934e-06,
      "loss": 1.114,
      "step": 96
    },
    {
      "epoch": 0.25129533678756477,
      "grad_norm": 0.6433624625205994,
      "learning_rate": 9.681794913852747e-06,
      "loss": 1.1304,
      "step": 97
    },
    {
      "epoch": 0.25129533678756477,
      "eval_loss": 1.0442166328430176,
      "eval_runtime": 99.046,
      "eval_samples_per_second": 2.524,
      "eval_steps_per_second": 0.323,
      "step": 97
    },
    {
      "epoch": 0.2538860103626943,
      "grad_norm": 0.634186863899231,
      "learning_rate": 9.67451868067933e-06,
      "loss": 1.0791,
      "step": 98
    },
    {
      "epoch": 0.25647668393782386,
      "grad_norm": 0.6801970601081848,
      "learning_rate": 9.667162991617633e-06,
      "loss": 1.1209,
      "step": 99
    },
    {
      "epoch": 0.25906735751295334,
      "grad_norm": 0.5541910529136658,
      "learning_rate": 9.659727971697173e-06,
      "loss": 1.1145,
      "step": 100
    },
    {
      "epoch": 0.2616580310880829,
      "grad_norm": 0.600925624370575,
      "learning_rate": 9.652213747295906e-06,
      "loss": 1.1005,
      "step": 101
    },
    {
      "epoch": 0.26424870466321243,
      "grad_norm": 0.6611819863319397,
      "learning_rate": 9.644620446138078e-06,
      "loss": 1.1321,
      "step": 102
    },
    {
      "epoch": 0.266839378238342,
      "grad_norm": 0.6193287372589111,
      "learning_rate": 9.636948197292051e-06,
      "loss": 1.069,
      "step": 103
    },
    {
      "epoch": 0.2694300518134715,
      "grad_norm": 0.5518503189086914,
      "learning_rate": 9.629197131168125e-06,
      "loss": 1.092,
      "step": 104
    },
    {
      "epoch": 0.27202072538860106,
      "grad_norm": 0.6933501362800598,
      "learning_rate": 9.621367379516294e-06,
      "loss": 1.0589,
      "step": 105
    },
    {
      "epoch": 0.27461139896373055,
      "grad_norm": 0.7061232924461365,
      "learning_rate": 9.613459075424033e-06,
      "loss": 1.094,
      "step": 106
    },
    {
      "epoch": 0.2772020725388601,
      "grad_norm": 0.7254324555397034,
      "learning_rate": 9.605472353314024e-06,
      "loss": 1.0386,
      "step": 107
    },
    {
      "epoch": 0.27979274611398963,
      "grad_norm": 0.5736768841743469,
      "learning_rate": 9.597407348941865e-06,
      "loss": 1.0804,
      "step": 108
    },
    {
      "epoch": 0.2823834196891192,
      "grad_norm": 0.5374739170074463,
      "learning_rate": 9.589264199393776e-06,
      "loss": 1.0699,
      "step": 109
    },
    {
      "epoch": 0.2849740932642487,
      "grad_norm": 0.7706698775291443,
      "learning_rate": 9.58104304308426e-06,
      "loss": 1.0498,
      "step": 110
    },
    {
      "epoch": 0.28756476683937826,
      "grad_norm": 0.7128419876098633,
      "learning_rate": 9.572744019753753e-06,
      "loss": 1.1212,
      "step": 111
    },
    {
      "epoch": 0.29015544041450775,
      "grad_norm": 0.57721346616745,
      "learning_rate": 9.564367270466247e-06,
      "loss": 1.0665,
      "step": 112
    },
    {
      "epoch": 0.2927461139896373,
      "grad_norm": 0.6171419024467468,
      "learning_rate": 9.555912937606896e-06,
      "loss": 1.1146,
      "step": 113
    },
    {
      "epoch": 0.29533678756476683,
      "grad_norm": 0.575703501701355,
      "learning_rate": 9.54738116487959e-06,
      "loss": 1.1132,
      "step": 114
    },
    {
      "epoch": 0.2979274611398964,
      "grad_norm": 0.6700958609580994,
      "learning_rate": 9.53877209730452e-06,
      "loss": 1.1184,
      "step": 115
    },
    {
      "epoch": 0.3005181347150259,
      "grad_norm": 0.7100483179092407,
      "learning_rate": 9.530085881215705e-06,
      "loss": 1.1258,
      "step": 116
    },
    {
      "epoch": 0.30310880829015546,
      "grad_norm": 0.6098718047142029,
      "learning_rate": 9.52132266425851e-06,
      "loss": 1.1547,
      "step": 117
    },
    {
      "epoch": 0.30569948186528495,
      "grad_norm": 0.6828567981719971,
      "learning_rate": 9.512482595387131e-06,
      "loss": 1.0376,
      "step": 118
    },
    {
      "epoch": 0.3082901554404145,
      "grad_norm": 0.6913807988166809,
      "learning_rate": 9.503565824862076e-06,
      "loss": 1.1741,
      "step": 119
    },
    {
      "epoch": 0.31088082901554404,
      "grad_norm": 0.6839683651924133,
      "learning_rate": 9.494572504247593e-06,
      "loss": 1.1419,
      "step": 120
    },
    {
      "epoch": 0.3134715025906736,
      "grad_norm": 0.6861181855201721,
      "learning_rate": 9.485502786409107e-06,
      "loss": 1.1121,
      "step": 121
    },
    {
      "epoch": 0.3160621761658031,
      "grad_norm": 0.7473586201667786,
      "learning_rate": 9.476356825510613e-06,
      "loss": 1.1247,
      "step": 122
    },
    {
      "epoch": 0.31865284974093266,
      "grad_norm": 0.7107284665107727,
      "learning_rate": 9.467134777012063e-06,
      "loss": 1.0763,
      "step": 123
    },
    {
      "epoch": 0.32124352331606215,
      "grad_norm": 0.7920987010002136,
      "learning_rate": 9.457836797666722e-06,
      "loss": 1.0304,
      "step": 124
    },
    {
      "epoch": 0.3238341968911917,
      "grad_norm": 0.9145889282226562,
      "learning_rate": 9.448463045518498e-06,
      "loss": 1.2063,
      "step": 125
    },
    {
      "epoch": 0.32642487046632124,
      "grad_norm": 0.5906431674957275,
      "learning_rate": 9.439013679899263e-06,
      "loss": 1.0248,
      "step": 126
    },
    {
      "epoch": 0.3290155440414508,
      "grad_norm": 0.7043807506561279,
      "learning_rate": 9.429488861426137e-06,
      "loss": 1.1032,
      "step": 127
    },
    {
      "epoch": 0.3316062176165803,
      "grad_norm": 0.6444696187973022,
      "learning_rate": 9.419888751998768e-06,
      "loss": 1.0548,
      "step": 128
    },
    {
      "epoch": 0.33419689119170987,
      "grad_norm": 0.7440561652183533,
      "learning_rate": 9.410213514796565e-06,
      "loss": 1.0761,
      "step": 129
    },
    {
      "epoch": 0.33678756476683935,
      "grad_norm": 0.6268182992935181,
      "learning_rate": 9.400463314275942e-06,
      "loss": 1.0308,
      "step": 130
    },
    {
      "epoch": 0.3393782383419689,
      "grad_norm": 0.7046973705291748,
      "learning_rate": 9.390638316167513e-06,
      "loss": 1.061,
      "step": 131
    },
    {
      "epoch": 0.34196891191709844,
      "grad_norm": 0.6304768323898315,
      "learning_rate": 9.380738687473274e-06,
      "loss": 1.0291,
      "step": 132
    },
    {
      "epoch": 0.344559585492228,
      "grad_norm": 0.7118530869483948,
      "learning_rate": 9.370764596463764e-06,
      "loss": 1.0132,
      "step": 133
    },
    {
      "epoch": 0.3471502590673575,
      "grad_norm": 0.6919021010398865,
      "learning_rate": 9.360716212675213e-06,
      "loss": 1.0753,
      "step": 134
    },
    {
      "epoch": 0.34974093264248707,
      "grad_norm": 0.6322518587112427,
      "learning_rate": 9.350593706906653e-06,
      "loss": 1.0591,
      "step": 135
    },
    {
      "epoch": 0.35233160621761656,
      "grad_norm": 0.9555734395980835,
      "learning_rate": 9.340397251217009e-06,
      "loss": 1.0681,
      "step": 136
    },
    {
      "epoch": 0.3549222797927461,
      "grad_norm": 0.7190225720405579,
      "learning_rate": 9.330127018922195e-06,
      "loss": 1.1177,
      "step": 137
    },
    {
      "epoch": 0.35751295336787564,
      "grad_norm": 0.6554561853408813,
      "learning_rate": 9.319783184592142e-06,
      "loss": 1.0129,
      "step": 138
    },
    {
      "epoch": 0.3601036269430052,
      "grad_norm": 0.6203082203865051,
      "learning_rate": 9.309365924047853e-06,
      "loss": 1.0762,
      "step": 139
    },
    {
      "epoch": 0.3626943005181347,
      "grad_norm": 0.8245846033096313,
      "learning_rate": 9.298875414358399e-06,
      "loss": 1.0528,
      "step": 140
    },
    {
      "epoch": 0.36528497409326427,
      "grad_norm": 0.7231378555297852,
      "learning_rate": 9.288311833837918e-06,
      "loss": 1.0407,
      "step": 141
    },
    {
      "epoch": 0.36787564766839376,
      "grad_norm": 0.7020997405052185,
      "learning_rate": 9.27767536204258e-06,
      "loss": 1.1054,
      "step": 142
    },
    {
      "epoch": 0.3704663212435233,
      "grad_norm": 0.6527167558670044,
      "learning_rate": 9.266966179767539e-06,
      "loss": 1.0096,
      "step": 143
    },
    {
      "epoch": 0.37305699481865284,
      "grad_norm": 0.6411340236663818,
      "learning_rate": 9.256184469043852e-06,
      "loss": 1.0547,
      "step": 144
    },
    {
      "epoch": 0.3756476683937824,
      "grad_norm": 1.0810225009918213,
      "learning_rate": 9.245330413135395e-06,
      "loss": 1.0752,
      "step": 145
    },
    {
      "epoch": 0.37823834196891193,
      "grad_norm": 0.9295507669448853,
      "learning_rate": 9.23440419653574e-06,
      "loss": 1.0341,
      "step": 146
    },
    {
      "epoch": 0.38082901554404147,
      "grad_norm": 0.7355513572692871,
      "learning_rate": 9.223406004965023e-06,
      "loss": 1.0998,
      "step": 147
    },
    {
      "epoch": 0.38341968911917096,
      "grad_norm": 0.8607478141784668,
      "learning_rate": 9.212336025366789e-06,
      "loss": 1.026,
      "step": 148
    },
    {
      "epoch": 0.3860103626943005,
      "grad_norm": 0.7150827646255493,
      "learning_rate": 9.201194445904804e-06,
      "loss": 1.071,
      "step": 149
    },
    {
      "epoch": 0.38860103626943004,
      "grad_norm": 0.688491940498352,
      "learning_rate": 9.189981455959873e-06,
      "loss": 0.994,
      "step": 150
    },
    {
      "epoch": 0.3911917098445596,
      "grad_norm": 0.8984132409095764,
      "learning_rate": 9.178697246126608e-06,
      "loss": 1.0318,
      "step": 151
    },
    {
      "epoch": 0.39378238341968913,
      "grad_norm": 0.6616289019584656,
      "learning_rate": 9.167342008210191e-06,
      "loss": 1.0407,
      "step": 152
    },
    {
      "epoch": 0.3963730569948187,
      "grad_norm": 0.7567489147186279,
      "learning_rate": 9.15591593522312e-06,
      "loss": 1.0323,
      "step": 153
    },
    {
      "epoch": 0.39896373056994816,
      "grad_norm": 0.6091617345809937,
      "learning_rate": 9.144419221381919e-06,
      "loss": 1.0361,
      "step": 154
    },
    {
      "epoch": 0.4015544041450777,
      "grad_norm": 0.721120297908783,
      "learning_rate": 9.132852062103845e-06,
      "loss": 1.101,
      "step": 155
    },
    {
      "epoch": 0.40414507772020725,
      "grad_norm": 0.7634301781654358,
      "learning_rate": 9.121214654003561e-06,
      "loss": 1.0285,
      "step": 156
    },
    {
      "epoch": 0.4067357512953368,
      "grad_norm": 0.6629643440246582,
      "learning_rate": 9.109507194889793e-06,
      "loss": 1.0858,
      "step": 157
    },
    {
      "epoch": 0.40932642487046633,
      "grad_norm": 0.6693721413612366,
      "learning_rate": 9.097729883761977e-06,
      "loss": 1.0461,
      "step": 158
    },
    {
      "epoch": 0.4119170984455959,
      "grad_norm": 0.6852269768714905,
      "learning_rate": 9.085882920806862e-06,
      "loss": 1.0298,
      "step": 159
    },
    {
      "epoch": 0.41450777202072536,
      "grad_norm": 0.6539198756217957,
      "learning_rate": 9.073966507395123e-06,
      "loss": 1.0298,
      "step": 160
    },
    {
      "epoch": 0.4170984455958549,
      "grad_norm": 0.9533804655075073,
      "learning_rate": 9.061980846077925e-06,
      "loss": 1.062,
      "step": 161
    },
    {
      "epoch": 0.41968911917098445,
      "grad_norm": 0.6733927130699158,
      "learning_rate": 9.049926140583487e-06,
      "loss": 1.0268,
      "step": 162
    },
    {
      "epoch": 0.422279792746114,
      "grad_norm": 0.7857198715209961,
      "learning_rate": 9.037802595813621e-06,
      "loss": 1.052,
      "step": 163
    },
    {
      "epoch": 0.42487046632124353,
      "grad_norm": 0.7269684076309204,
      "learning_rate": 9.025610417840238e-06,
      "loss": 1.0026,
      "step": 164
    },
    {
      "epoch": 0.4274611398963731,
      "grad_norm": 0.8214792013168335,
      "learning_rate": 9.01334981390186e-06,
      "loss": 1.0533,
      "step": 165
    },
    {
      "epoch": 0.43005181347150256,
      "grad_norm": 0.6824995875358582,
      "learning_rate": 9.001020992400086e-06,
      "loss": 1.0437,
      "step": 166
    },
    {
      "epoch": 0.4326424870466321,
      "grad_norm": 0.6578338742256165,
      "learning_rate": 8.988624162896058e-06,
      "loss": 1.097,
      "step": 167
    },
    {
      "epoch": 0.43523316062176165,
      "grad_norm": 0.7836469411849976,
      "learning_rate": 8.976159536106895e-06,
      "loss": 0.977,
      "step": 168
    },
    {
      "epoch": 0.4378238341968912,
      "grad_norm": 0.7745511531829834,
      "learning_rate": 8.963627323902105e-06,
      "loss": 0.9812,
      "step": 169
    },
    {
      "epoch": 0.44041450777202074,
      "grad_norm": 0.712700366973877,
      "learning_rate": 8.951027739299996e-06,
      "loss": 1.0043,
      "step": 170
    },
    {
      "epoch": 0.4430051813471503,
      "grad_norm": 0.6669776439666748,
      "learning_rate": 8.938360996464048e-06,
      "loss": 1.0269,
      "step": 171
    },
    {
      "epoch": 0.44559585492227977,
      "grad_norm": 0.693252444267273,
      "learning_rate": 8.925627310699275e-06,
      "loss": 0.9963,
      "step": 172
    },
    {
      "epoch": 0.4481865284974093,
      "grad_norm": 0.741417407989502,
      "learning_rate": 8.91282689844856e-06,
      "loss": 1.0292,
      "step": 173
    },
    {
      "epoch": 0.45077720207253885,
      "grad_norm": 0.8754829168319702,
      "learning_rate": 8.899959977288988e-06,
      "loss": 1.0089,
      "step": 174
    },
    {
      "epoch": 0.4533678756476684,
      "grad_norm": 0.6871668100357056,
      "learning_rate": 8.887026765928129e-06,
      "loss": 0.9792,
      "step": 175
    },
    {
      "epoch": 0.45595854922279794,
      "grad_norm": 0.7657473683357239,
      "learning_rate": 8.874027484200342e-06,
      "loss": 1.0297,
      "step": 176
    },
    {
      "epoch": 0.4585492227979275,
      "grad_norm": 0.7051308751106262,
      "learning_rate": 8.860962353063022e-06,
      "loss": 1.0033,
      "step": 177
    },
    {
      "epoch": 0.46113989637305697,
      "grad_norm": 0.7424786686897278,
      "learning_rate": 8.847831594592851e-06,
      "loss": 1.0776,
      "step": 178
    },
    {
      "epoch": 0.4637305699481865,
      "grad_norm": 0.9562065601348877,
      "learning_rate": 8.834635431982022e-06,
      "loss": 0.893,
      "step": 179
    },
    {
      "epoch": 0.46632124352331605,
      "grad_norm": 0.7324496507644653,
      "learning_rate": 8.821374089534446e-06,
      "loss": 0.9691,
      "step": 180
    },
    {
      "epoch": 0.4689119170984456,
      "grad_norm": 0.6958107948303223,
      "learning_rate": 8.808047792661941e-06,
      "loss": 1.0113,
      "step": 181
    },
    {
      "epoch": 0.47150259067357514,
      "grad_norm": 0.7926040291786194,
      "learning_rate": 8.794656767880394e-06,
      "loss": 1.054,
      "step": 182
    },
    {
      "epoch": 0.4740932642487047,
      "grad_norm": 0.7949718236923218,
      "learning_rate": 8.781201242805917e-06,
      "loss": 1.0385,
      "step": 183
    },
    {
      "epoch": 0.47668393782383417,
      "grad_norm": 0.7993385195732117,
      "learning_rate": 8.767681446150977e-06,
      "loss": 0.987,
      "step": 184
    },
    {
      "epoch": 0.4792746113989637,
      "grad_norm": 0.8635035157203674,
      "learning_rate": 8.754097607720512e-06,
      "loss": 1.1012,
      "step": 185
    },
    {
      "epoch": 0.48186528497409326,
      "grad_norm": 0.7778316140174866,
      "learning_rate": 8.740449958408006e-06,
      "loss": 1.0547,
      "step": 186
    },
    {
      "epoch": 0.4844559585492228,
      "grad_norm": 0.7622816562652588,
      "learning_rate": 8.726738730191596e-06,
      "loss": 1.057,
      "step": 187
    },
    {
      "epoch": 0.48704663212435234,
      "grad_norm": 0.8010302782058716,
      "learning_rate": 8.7129641561301e-06,
      "loss": 1.06,
      "step": 188
    },
    {
      "epoch": 0.4896373056994819,
      "grad_norm": 0.8265953063964844,
      "learning_rate": 8.699126470359073e-06,
      "loss": 0.9859,
      "step": 189
    },
    {
      "epoch": 0.49222797927461137,
      "grad_norm": 0.7856041193008423,
      "learning_rate": 8.68522590808682e-06,
      "loss": 1.05,
      "step": 190
    },
    {
      "epoch": 0.4948186528497409,
      "grad_norm": 0.7639081478118896,
      "learning_rate": 8.671262705590399e-06,
      "loss": 1.0018,
      "step": 191
    },
    {
      "epoch": 0.49740932642487046,
      "grad_norm": 0.8716084361076355,
      "learning_rate": 8.657237100211604e-06,
      "loss": 1.0073,
      "step": 192
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7365914583206177,
      "learning_rate": 8.643149330352939e-06,
      "loss": 1.0448,
      "step": 193
    },
    {
      "epoch": 0.5025906735751295,
      "grad_norm": 0.7660321593284607,
      "learning_rate": 8.628999635473547e-06,
      "loss": 1.0415,
      "step": 194
    },
    {
      "epoch": 0.5025906735751295,
      "eval_loss": 0.9808682203292847,
      "eval_runtime": 99.1803,
      "eval_samples_per_second": 2.521,
      "eval_steps_per_second": 0.323,
      "step": 194
    },
    {
      "epoch": 0.5051813471502591,
      "grad_norm": 0.8740248680114746,
      "learning_rate": 8.61478825608516e-06,
      "loss": 1.08,
      "step": 195
    },
    {
      "epoch": 0.5077720207253886,
      "grad_norm": 0.8985514640808105,
      "learning_rate": 8.600515433748003e-06,
      "loss": 1.0248,
      "step": 196
    },
    {
      "epoch": 0.5103626943005182,
      "grad_norm": 0.6852197647094727,
      "learning_rate": 8.586181411066684e-06,
      "loss": 1.0513,
      "step": 197
    },
    {
      "epoch": 0.5129533678756477,
      "grad_norm": 0.8582538366317749,
      "learning_rate": 8.571786431686074e-06,
      "loss": 1.0453,
      "step": 198
    },
    {
      "epoch": 0.5155440414507773,
      "grad_norm": 0.7389397621154785,
      "learning_rate": 8.557330740287166e-06,
      "loss": 1.0516,
      "step": 199
    },
    {
      "epoch": 0.5181347150259067,
      "grad_norm": 0.7529143691062927,
      "learning_rate": 8.542814582582917e-06,
      "loss": 1.0294,
      "step": 200
    },
    {
      "epoch": 0.5207253886010362,
      "grad_norm": 0.828758180141449,
      "learning_rate": 8.528238205314067e-06,
      "loss": 1.0357,
      "step": 201
    },
    {
      "epoch": 0.5233160621761658,
      "grad_norm": 0.841716468334198,
      "learning_rate": 8.513601856244951e-06,
      "loss": 1.0827,
      "step": 202
    },
    {
      "epoch": 0.5259067357512953,
      "grad_norm": 0.807770848274231,
      "learning_rate": 8.498905784159282e-06,
      "loss": 1.0939,
      "step": 203
    },
    {
      "epoch": 0.5284974093264249,
      "grad_norm": 0.7827705144882202,
      "learning_rate": 8.484150238855921e-06,
      "loss": 1.0735,
      "step": 204
    },
    {
      "epoch": 0.5310880829015544,
      "grad_norm": 0.7675580382347107,
      "learning_rate": 8.469335471144646e-06,
      "loss": 1.0268,
      "step": 205
    },
    {
      "epoch": 0.533678756476684,
      "grad_norm": 0.767455518245697,
      "learning_rate": 8.454461732841864e-06,
      "loss": 1.0624,
      "step": 206
    },
    {
      "epoch": 0.5362694300518135,
      "grad_norm": 0.6222347617149353,
      "learning_rate": 8.439529276766354e-06,
      "loss": 1.0263,
      "step": 207
    },
    {
      "epoch": 0.538860103626943,
      "grad_norm": 0.7709267735481262,
      "learning_rate": 8.424538356734957e-06,
      "loss": 1.0703,
      "step": 208
    },
    {
      "epoch": 0.5414507772020726,
      "grad_norm": 0.9357248544692993,
      "learning_rate": 8.40948922755826e-06,
      "loss": 1.0299,
      "step": 209
    },
    {
      "epoch": 0.5440414507772021,
      "grad_norm": 0.7862836718559265,
      "learning_rate": 8.394382145036277e-06,
      "loss": 1.017,
      "step": 210
    },
    {
      "epoch": 0.5466321243523317,
      "grad_norm": 0.8551638722419739,
      "learning_rate": 8.379217365954089e-06,
      "loss": 0.9445,
      "step": 211
    },
    {
      "epoch": 0.5492227979274611,
      "grad_norm": 0.7895774841308594,
      "learning_rate": 8.363995148077481e-06,
      "loss": 1.1145,
      "step": 212
    },
    {
      "epoch": 0.5518134715025906,
      "grad_norm": 0.684273362159729,
      "learning_rate": 8.348715750148571e-06,
      "loss": 1.0439,
      "step": 213
    },
    {
      "epoch": 0.5544041450777202,
      "grad_norm": 0.748660147190094,
      "learning_rate": 8.333379431881398e-06,
      "loss": 1.034,
      "step": 214
    },
    {
      "epoch": 0.5569948186528497,
      "grad_norm": 0.8681594133377075,
      "learning_rate": 8.317986453957514e-06,
      "loss": 1.0523,
      "step": 215
    },
    {
      "epoch": 0.5595854922279793,
      "grad_norm": 0.8759576678276062,
      "learning_rate": 8.302537078021555e-06,
      "loss": 1.0167,
      "step": 216
    },
    {
      "epoch": 0.5621761658031088,
      "grad_norm": 0.8709188103675842,
      "learning_rate": 8.28703156667679e-06,
      "loss": 1.0947,
      "step": 217
    },
    {
      "epoch": 0.5647668393782384,
      "grad_norm": 0.7970831990242004,
      "learning_rate": 8.271470183480664e-06,
      "loss": 1.0404,
      "step": 218
    },
    {
      "epoch": 0.5673575129533679,
      "grad_norm": 0.8916209936141968,
      "learning_rate": 8.2558531929403e-06,
      "loss": 1.0681,
      "step": 219
    },
    {
      "epoch": 0.5699481865284974,
      "grad_norm": 0.7815295457839966,
      "learning_rate": 8.240180860508027e-06,
      "loss": 1.0273,
      "step": 220
    },
    {
      "epoch": 0.572538860103627,
      "grad_norm": 0.8419135808944702,
      "learning_rate": 8.224453452576855e-06,
      "loss": 0.974,
      "step": 221
    },
    {
      "epoch": 0.5751295336787565,
      "grad_norm": 1.0326911211013794,
      "learning_rate": 8.208671236475945e-06,
      "loss": 1.0307,
      "step": 222
    },
    {
      "epoch": 0.5777202072538861,
      "grad_norm": 0.8720448017120361,
      "learning_rate": 8.192834480466072e-06,
      "loss": 1.0061,
      "step": 223
    },
    {
      "epoch": 0.5803108808290155,
      "grad_norm": 0.7192279100418091,
      "learning_rate": 8.176943453735062e-06,
      "loss": 0.9383,
      "step": 224
    },
    {
      "epoch": 0.582901554404145,
      "grad_norm": 0.7928858995437622,
      "learning_rate": 8.160998426393214e-06,
      "loss": 1.0538,
      "step": 225
    },
    {
      "epoch": 0.5854922279792746,
      "grad_norm": 0.9114733934402466,
      "learning_rate": 8.144999669468714e-06,
      "loss": 1.0704,
      "step": 226
    },
    {
      "epoch": 0.5880829015544041,
      "grad_norm": 0.8045042753219604,
      "learning_rate": 8.12894745490302e-06,
      "loss": 1.0213,
      "step": 227
    },
    {
      "epoch": 0.5906735751295337,
      "grad_norm": 0.8414791226387024,
      "learning_rate": 8.112842055546254e-06,
      "loss": 1.1058,
      "step": 228
    },
    {
      "epoch": 0.5932642487046632,
      "grad_norm": 0.8457399606704712,
      "learning_rate": 8.096683745152544e-06,
      "loss": 1.0407,
      "step": 229
    },
    {
      "epoch": 0.5958549222797928,
      "grad_norm": 0.7906298637390137,
      "learning_rate": 8.080472798375392e-06,
      "loss": 1.0335,
      "step": 230
    },
    {
      "epoch": 0.5984455958549223,
      "grad_norm": 0.8662751913070679,
      "learning_rate": 8.064209490762988e-06,
      "loss": 1.0809,
      "step": 231
    },
    {
      "epoch": 0.6010362694300518,
      "grad_norm": 0.8119012117385864,
      "learning_rate": 8.04789409875354e-06,
      "loss": 1.0132,
      "step": 232
    },
    {
      "epoch": 0.6036269430051814,
      "grad_norm": 0.7348366975784302,
      "learning_rate": 8.031526899670563e-06,
      "loss": 1.0383,
      "step": 233
    },
    {
      "epoch": 0.6062176165803109,
      "grad_norm": 0.7896949648857117,
      "learning_rate": 8.015108171718177e-06,
      "loss": 1.0656,
      "step": 234
    },
    {
      "epoch": 0.6088082901554405,
      "grad_norm": 0.8619694113731384,
      "learning_rate": 7.998638193976366e-06,
      "loss": 1.0881,
      "step": 235
    },
    {
      "epoch": 0.6113989637305699,
      "grad_norm": 0.7404602766036987,
      "learning_rate": 7.982117246396246e-06,
      "loss": 1.0613,
      "step": 236
    },
    {
      "epoch": 0.6139896373056994,
      "grad_norm": 0.6466916799545288,
      "learning_rate": 7.965545609795297e-06,
      "loss": 1.0333,
      "step": 237
    },
    {
      "epoch": 0.616580310880829,
      "grad_norm": 0.6092397570610046,
      "learning_rate": 7.948923565852597e-06,
      "loss": 1.0161,
      "step": 238
    },
    {
      "epoch": 0.6191709844559585,
      "grad_norm": 0.7040543556213379,
      "learning_rate": 7.932251397104031e-06,
      "loss": 0.9819,
      "step": 239
    },
    {
      "epoch": 0.6217616580310881,
      "grad_norm": 1.0263625383377075,
      "learning_rate": 7.915529386937486e-06,
      "loss": 0.9477,
      "step": 240
    },
    {
      "epoch": 0.6243523316062176,
      "grad_norm": 0.6552597284317017,
      "learning_rate": 7.898757819588038e-06,
      "loss": 0.989,
      "step": 241
    },
    {
      "epoch": 0.6269430051813472,
      "grad_norm": 0.7315415143966675,
      "learning_rate": 7.881936980133118e-06,
      "loss": 1.0219,
      "step": 242
    },
    {
      "epoch": 0.6295336787564767,
      "grad_norm": 0.8863290548324585,
      "learning_rate": 7.86506715448767e-06,
      "loss": 1.0347,
      "step": 243
    },
    {
      "epoch": 0.6321243523316062,
      "grad_norm": 0.6123449206352234,
      "learning_rate": 7.848148629399287e-06,
      "loss": 1.0001,
      "step": 244
    },
    {
      "epoch": 0.6347150259067358,
      "grad_norm": 0.9395535588264465,
      "learning_rate": 7.831181692443338e-06,
      "loss": 1.008,
      "step": 245
    },
    {
      "epoch": 0.6373056994818653,
      "grad_norm": 0.6824455261230469,
      "learning_rate": 7.814166632018083e-06,
      "loss": 1.0588,
      "step": 246
    },
    {
      "epoch": 0.6398963730569949,
      "grad_norm": 0.79184889793396,
      "learning_rate": 7.797103737339767e-06,
      "loss": 1.0284,
      "step": 247
    },
    {
      "epoch": 0.6424870466321243,
      "grad_norm": 0.8167986273765564,
      "learning_rate": 7.779993298437704e-06,
      "loss": 0.9941,
      "step": 248
    },
    {
      "epoch": 0.6450777202072538,
      "grad_norm": 0.8785682916641235,
      "learning_rate": 7.762835606149352e-06,
      "loss": 1.0335,
      "step": 249
    },
    {
      "epoch": 0.6476683937823834,
      "grad_norm": 0.6932119727134705,
      "learning_rate": 7.745630952115365e-06,
      "loss": 1.002,
      "step": 250
    },
    {
      "epoch": 0.6502590673575129,
      "grad_norm": 0.8167779445648193,
      "learning_rate": 7.728379628774632e-06,
      "loss": 0.9496,
      "step": 251
    },
    {
      "epoch": 0.6528497409326425,
      "grad_norm": 0.819342315196991,
      "learning_rate": 7.711081929359316e-06,
      "loss": 1.0228,
      "step": 252
    },
    {
      "epoch": 0.655440414507772,
      "grad_norm": 0.8480210900306702,
      "learning_rate": 7.693738147889868e-06,
      "loss": 1.0217,
      "step": 253
    },
    {
      "epoch": 0.6580310880829016,
      "grad_norm": 0.8472158908843994,
      "learning_rate": 7.67634857917002e-06,
      "loss": 1.0264,
      "step": 254
    },
    {
      "epoch": 0.6606217616580311,
      "grad_norm": 0.7778662443161011,
      "learning_rate": 7.658913518781782e-06,
      "loss": 0.9766,
      "step": 255
    },
    {
      "epoch": 0.6632124352331606,
      "grad_norm": 0.8037154078483582,
      "learning_rate": 7.641433263080418e-06,
      "loss": 1.0129,
      "step": 256
    },
    {
      "epoch": 0.6658031088082902,
      "grad_norm": 0.778190016746521,
      "learning_rate": 7.623908109189404e-06,
      "loss": 1.0401,
      "step": 257
    },
    {
      "epoch": 0.6683937823834197,
      "grad_norm": 1.0345908403396606,
      "learning_rate": 7.606338354995381e-06,
      "loss": 1.0601,
      "step": 258
    },
    {
      "epoch": 0.6709844559585493,
      "grad_norm": 0.7248339653015137,
      "learning_rate": 7.588724299143091e-06,
      "loss": 1.0098,
      "step": 259
    },
    {
      "epoch": 0.6735751295336787,
      "grad_norm": 0.6873196959495544,
      "learning_rate": 7.571066241030302e-06,
      "loss": 1.0574,
      "step": 260
    },
    {
      "epoch": 0.6761658031088082,
      "grad_norm": 0.7845073938369751,
      "learning_rate": 7.553364480802715e-06,
      "loss": 1.0041,
      "step": 261
    },
    {
      "epoch": 0.6787564766839378,
      "grad_norm": 0.7601717114448547,
      "learning_rate": 7.5356193193488655e-06,
      "loss": 0.9021,
      "step": 262
    },
    {
      "epoch": 0.6813471502590673,
      "grad_norm": 0.8694692850112915,
      "learning_rate": 7.517831058295013e-06,
      "loss": 1.0321,
      "step": 263
    },
    {
      "epoch": 0.6839378238341969,
      "grad_norm": 0.8670944571495056,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9876,
      "step": 264
    },
    {
      "epoch": 0.6865284974093264,
      "grad_norm": 0.9318599700927734,
      "learning_rate": 7.4821264475501325e-06,
      "loss": 1.0528,
      "step": 265
    },
    {
      "epoch": 0.689119170984456,
      "grad_norm": 0.9383565187454224,
      "learning_rate": 7.464210704754009e-06,
      "loss": 0.9914,
      "step": 266
    },
    {
      "epoch": 0.6917098445595855,
      "grad_norm": 0.8553853034973145,
      "learning_rate": 7.446253076137372e-06,
      "loss": 1.0909,
      "step": 267
    },
    {
      "epoch": 0.694300518134715,
      "grad_norm": 0.8563231229782104,
      "learning_rate": 7.4282538669379186e-06,
      "loss": 1.0639,
      "step": 268
    },
    {
      "epoch": 0.6968911917098446,
      "grad_norm": 0.8212496042251587,
      "learning_rate": 7.410213383100126e-06,
      "loss": 0.9786,
      "step": 269
    },
    {
      "epoch": 0.6994818652849741,
      "grad_norm": 0.9418748617172241,
      "learning_rate": 7.3921319312700365e-06,
      "loss": 1.0663,
      "step": 270
    },
    {
      "epoch": 0.7020725388601037,
      "grad_norm": 0.7147770524024963,
      "learning_rate": 7.374009818790058e-06,
      "loss": 1.0187,
      "step": 271
    },
    {
      "epoch": 0.7046632124352331,
      "grad_norm": 0.8993127346038818,
      "learning_rate": 7.355847353693729e-06,
      "loss": 0.9813,
      "step": 272
    },
    {
      "epoch": 0.7072538860103627,
      "grad_norm": 0.8212840557098389,
      "learning_rate": 7.337644844700494e-06,
      "loss": 0.9602,
      "step": 273
    },
    {
      "epoch": 0.7098445595854922,
      "grad_norm": 0.8086667656898499,
      "learning_rate": 7.319402601210448e-06,
      "loss": 1.1018,
      "step": 274
    },
    {
      "epoch": 0.7124352331606217,
      "grad_norm": 0.8487491607666016,
      "learning_rate": 7.301120933299076e-06,
      "loss": 0.9453,
      "step": 275
    },
    {
      "epoch": 0.7150259067357513,
      "grad_norm": 1.0112347602844238,
      "learning_rate": 7.282800151711991e-06,
      "loss": 0.9753,
      "step": 276
    },
    {
      "epoch": 0.7176165803108808,
      "grad_norm": 0.7847304940223694,
      "learning_rate": 7.264440567859645e-06,
      "loss": 1.0121,
      "step": 277
    },
    {
      "epoch": 0.7202072538860104,
      "grad_norm": 0.7814239263534546,
      "learning_rate": 7.246042493812036e-06,
      "loss": 1.0152,
      "step": 278
    },
    {
      "epoch": 0.7227979274611399,
      "grad_norm": 0.9602167010307312,
      "learning_rate": 7.227606242293409e-06,
      "loss": 1.057,
      "step": 279
    },
    {
      "epoch": 0.7253886010362695,
      "grad_norm": 0.8330225944519043,
      "learning_rate": 7.209132126676934e-06,
      "loss": 1.0834,
      "step": 280
    },
    {
      "epoch": 0.727979274611399,
      "grad_norm": 0.8452211618423462,
      "learning_rate": 7.190620460979384e-06,
      "loss": 1.0382,
      "step": 281
    },
    {
      "epoch": 0.7305699481865285,
      "grad_norm": 0.7918412089347839,
      "learning_rate": 7.172071559855792e-06,
      "loss": 0.9387,
      "step": 282
    },
    {
      "epoch": 0.7331606217616581,
      "grad_norm": 0.8554849624633789,
      "learning_rate": 7.153485738594111e-06,
      "loss": 1.0589,
      "step": 283
    },
    {
      "epoch": 0.7357512953367875,
      "grad_norm": 0.906253457069397,
      "learning_rate": 7.134863313109847e-06,
      "loss": 1.001,
      "step": 284
    },
    {
      "epoch": 0.7383419689119171,
      "grad_norm": 0.9388474225997925,
      "learning_rate": 7.116204599940693e-06,
      "loss": 1.0046,
      "step": 285
    },
    {
      "epoch": 0.7409326424870466,
      "grad_norm": 0.8858338594436646,
      "learning_rate": 7.097509916241145e-06,
      "loss": 1.0284,
      "step": 286
    },
    {
      "epoch": 0.7435233160621761,
      "grad_norm": 0.8751007914543152,
      "learning_rate": 7.078779579777122e-06,
      "loss": 1.0375,
      "step": 287
    },
    {
      "epoch": 0.7461139896373057,
      "grad_norm": 0.9136319160461426,
      "learning_rate": 7.060013908920549e-06,
      "loss": 1.0254,
      "step": 288
    },
    {
      "epoch": 0.7487046632124352,
      "grad_norm": 0.8126935958862305,
      "learning_rate": 7.041213222643952e-06,
      "loss": 1.0091,
      "step": 289
    },
    {
      "epoch": 0.7512953367875648,
      "grad_norm": 0.8925020098686218,
      "learning_rate": 7.022377840515047e-06,
      "loss": 0.9436,
      "step": 290
    },
    {
      "epoch": 0.7538860103626943,
      "grad_norm": 0.7228291630744934,
      "learning_rate": 7.003508082691286e-06,
      "loss": 0.9353,
      "step": 291
    },
    {
      "epoch": 0.7538860103626943,
      "eval_loss": 0.9649054408073425,
      "eval_runtime": 99.1248,
      "eval_samples_per_second": 2.522,
      "eval_steps_per_second": 0.323,
      "step": 291
    },
    {
      "epoch": 0.7564766839378239,
      "grad_norm": 0.7875744104385376,
      "learning_rate": 6.984604269914437e-06,
      "loss": 1.0115,
      "step": 292
    },
    {
      "epoch": 0.7590673575129534,
      "grad_norm": 0.8201349377632141,
      "learning_rate": 6.965666723505118e-06,
      "loss": 1.0683,
      "step": 293
    },
    {
      "epoch": 0.7616580310880829,
      "grad_norm": 0.9053636193275452,
      "learning_rate": 6.94669576535734e-06,
      "loss": 0.9764,
      "step": 294
    },
    {
      "epoch": 0.7642487046632125,
      "grad_norm": 0.7280160784721375,
      "learning_rate": 6.927691717933038e-06,
      "loss": 1.0246,
      "step": 295
    },
    {
      "epoch": 0.7668393782383419,
      "grad_norm": 0.756608784198761,
      "learning_rate": 6.908654904256584e-06,
      "loss": 1.0057,
      "step": 296
    },
    {
      "epoch": 0.7694300518134715,
      "grad_norm": 0.8075028657913208,
      "learning_rate": 6.889585647909303e-06,
      "loss": 1.0235,
      "step": 297
    },
    {
      "epoch": 0.772020725388601,
      "grad_norm": 0.7520319223403931,
      "learning_rate": 6.870484273023967e-06,
      "loss": 1.007,
      "step": 298
    },
    {
      "epoch": 0.7746113989637305,
      "grad_norm": 0.8978586792945862,
      "learning_rate": 6.8513511042792895e-06,
      "loss": 0.9499,
      "step": 299
    },
    {
      "epoch": 0.7772020725388601,
      "grad_norm": 0.8076562881469727,
      "learning_rate": 6.832186466894402e-06,
      "loss": 0.9844,
      "step": 300
    },
    {
      "epoch": 0.7797927461139896,
      "grad_norm": 0.9009307026863098,
      "learning_rate": 6.812990686623335e-06,
      "loss": 1.0019,
      "step": 301
    },
    {
      "epoch": 0.7823834196891192,
      "grad_norm": 0.7848314046859741,
      "learning_rate": 6.793764089749473e-06,
      "loss": 1.0229,
      "step": 302
    },
    {
      "epoch": 0.7849740932642487,
      "grad_norm": 0.8111161589622498,
      "learning_rate": 6.7745070030800075e-06,
      "loss": 1.0046,
      "step": 303
    },
    {
      "epoch": 0.7875647668393783,
      "grad_norm": 0.8512140512466431,
      "learning_rate": 6.755219753940389e-06,
      "loss": 0.9784,
      "step": 304
    },
    {
      "epoch": 0.7901554404145078,
      "grad_norm": 0.7778370380401611,
      "learning_rate": 6.735902670168758e-06,
      "loss": 1.0444,
      "step": 305
    },
    {
      "epoch": 0.7927461139896373,
      "grad_norm": 0.8520141839981079,
      "learning_rate": 6.716556080110374e-06,
      "loss": 1.036,
      "step": 306
    },
    {
      "epoch": 0.7953367875647669,
      "grad_norm": 0.9404623508453369,
      "learning_rate": 6.6971803126120336e-06,
      "loss": 1.1203,
      "step": 307
    },
    {
      "epoch": 0.7979274611398963,
      "grad_norm": 0.9082056283950806,
      "learning_rate": 6.677775697016484e-06,
      "loss": 1.0798,
      "step": 308
    },
    {
      "epoch": 0.8005181347150259,
      "grad_norm": 0.7969542145729065,
      "learning_rate": 6.658342563156821e-06,
      "loss": 0.9632,
      "step": 309
    },
    {
      "epoch": 0.8031088082901554,
      "grad_norm": 0.7916974425315857,
      "learning_rate": 6.638881241350884e-06,
      "loss": 1.0198,
      "step": 310
    },
    {
      "epoch": 0.805699481865285,
      "grad_norm": 0.9639979600906372,
      "learning_rate": 6.619392062395643e-06,
      "loss": 1.0056,
      "step": 311
    },
    {
      "epoch": 0.8082901554404145,
      "grad_norm": 0.9673221707344055,
      "learning_rate": 6.599875357561572e-06,
      "loss": 0.9875,
      "step": 312
    },
    {
      "epoch": 0.810880829015544,
      "grad_norm": 1.0038220882415771,
      "learning_rate": 6.5803314585870225e-06,
      "loss": 1.1161,
      "step": 313
    },
    {
      "epoch": 0.8134715025906736,
      "grad_norm": 0.9063149094581604,
      "learning_rate": 6.560760697672583e-06,
      "loss": 0.9699,
      "step": 314
    },
    {
      "epoch": 0.8160621761658031,
      "grad_norm": 0.9444339871406555,
      "learning_rate": 6.541163407475433e-06,
      "loss": 1.0099,
      "step": 315
    },
    {
      "epoch": 0.8186528497409327,
      "grad_norm": 0.869704008102417,
      "learning_rate": 6.5215399211036815e-06,
      "loss": 1.0076,
      "step": 316
    },
    {
      "epoch": 0.8212435233160622,
      "grad_norm": 0.7657658457756042,
      "learning_rate": 6.50189057211072e-06,
      "loss": 1.0471,
      "step": 317
    },
    {
      "epoch": 0.8238341968911918,
      "grad_norm": 0.8859752416610718,
      "learning_rate": 6.4822156944895375e-06,
      "loss": 0.9323,
      "step": 318
    },
    {
      "epoch": 0.8264248704663213,
      "grad_norm": 0.7193804979324341,
      "learning_rate": 6.462515622667056e-06,
      "loss": 1.042,
      "step": 319
    },
    {
      "epoch": 0.8290155440414507,
      "grad_norm": 0.7999783754348755,
      "learning_rate": 6.442790691498433e-06,
      "loss": 0.967,
      "step": 320
    },
    {
      "epoch": 0.8316062176165803,
      "grad_norm": 0.8761709928512573,
      "learning_rate": 6.423041236261381e-06,
      "loss": 1.0732,
      "step": 321
    },
    {
      "epoch": 0.8341968911917098,
      "grad_norm": 1.0846220254898071,
      "learning_rate": 6.403267592650466e-06,
      "loss": 0.9741,
      "step": 322
    },
    {
      "epoch": 0.8367875647668394,
      "grad_norm": 0.6978609561920166,
      "learning_rate": 6.383470096771396e-06,
      "loss": 1.0009,
      "step": 323
    },
    {
      "epoch": 0.8393782383419689,
      "grad_norm": 0.8032934069633484,
      "learning_rate": 6.363649085135311e-06,
      "loss": 0.9546,
      "step": 324
    },
    {
      "epoch": 0.8419689119170984,
      "grad_norm": 1.0093727111816406,
      "learning_rate": 6.343804894653072e-06,
      "loss": 1.0592,
      "step": 325
    },
    {
      "epoch": 0.844559585492228,
      "grad_norm": 0.8561227917671204,
      "learning_rate": 6.323937862629513e-06,
      "loss": 1.0551,
      "step": 326
    },
    {
      "epoch": 0.8471502590673575,
      "grad_norm": 0.7876784801483154,
      "learning_rate": 6.304048326757735e-06,
      "loss": 1.0564,
      "step": 327
    },
    {
      "epoch": 0.8497409326424871,
      "grad_norm": 0.7641052603721619,
      "learning_rate": 6.2841366251133405e-06,
      "loss": 0.9427,
      "step": 328
    },
    {
      "epoch": 0.8523316062176166,
      "grad_norm": 0.8268167972564697,
      "learning_rate": 6.2642030961487046e-06,
      "loss": 1.0171,
      "step": 329
    },
    {
      "epoch": 0.8549222797927462,
      "grad_norm": 0.7793816328048706,
      "learning_rate": 6.244248078687213e-06,
      "loss": 0.9475,
      "step": 330
    },
    {
      "epoch": 0.8575129533678757,
      "grad_norm": 0.7105771899223328,
      "learning_rate": 6.224271911917508e-06,
      "loss": 1.0326,
      "step": 331
    },
    {
      "epoch": 0.8601036269430051,
      "grad_norm": 0.8429663777351379,
      "learning_rate": 6.204274935387716e-06,
      "loss": 1.0257,
      "step": 332
    },
    {
      "epoch": 0.8626943005181347,
      "grad_norm": 0.8630242347717285,
      "learning_rate": 6.184257488999688e-06,
      "loss": 1.0037,
      "step": 333
    },
    {
      "epoch": 0.8652849740932642,
      "grad_norm": 1.0064729452133179,
      "learning_rate": 6.164219913003208e-06,
      "loss": 1.0468,
      "step": 334
    },
    {
      "epoch": 0.8678756476683938,
      "grad_norm": 0.9719647765159607,
      "learning_rate": 6.14416254799022e-06,
      "loss": 1.0582,
      "step": 335
    },
    {
      "epoch": 0.8704663212435233,
      "grad_norm": 0.8877604007720947,
      "learning_rate": 6.124085734889034e-06,
      "loss": 0.9288,
      "step": 336
    },
    {
      "epoch": 0.8730569948186528,
      "grad_norm": 0.8161412477493286,
      "learning_rate": 6.1039898149585305e-06,
      "loss": 0.9929,
      "step": 337
    },
    {
      "epoch": 0.8756476683937824,
      "grad_norm": 0.7888549566268921,
      "learning_rate": 6.083875129782366e-06,
      "loss": 1.022,
      "step": 338
    },
    {
      "epoch": 0.8782383419689119,
      "grad_norm": 0.8845784664154053,
      "learning_rate": 6.063742021263157e-06,
      "loss": 0.9448,
      "step": 339
    },
    {
      "epoch": 0.8808290155440415,
      "grad_norm": 0.791270911693573,
      "learning_rate": 6.043590831616677e-06,
      "loss": 0.9783,
      "step": 340
    },
    {
      "epoch": 0.883419689119171,
      "grad_norm": 0.8304092288017273,
      "learning_rate": 6.023421903366034e-06,
      "loss": 1.0105,
      "step": 341
    },
    {
      "epoch": 0.8860103626943006,
      "grad_norm": 0.7290758490562439,
      "learning_rate": 6.003235579335851e-06,
      "loss": 1.011,
      "step": 342
    },
    {
      "epoch": 0.8886010362694301,
      "grad_norm": 0.9048756957054138,
      "learning_rate": 5.9830322026464435e-06,
      "loss": 1.0133,
      "step": 343
    },
    {
      "epoch": 0.8911917098445595,
      "grad_norm": 0.8403552770614624,
      "learning_rate": 5.962812116707977e-06,
      "loss": 1.0031,
      "step": 344
    },
    {
      "epoch": 0.8937823834196891,
      "grad_norm": 0.8173761963844299,
      "learning_rate": 5.942575665214634e-06,
      "loss": 1.007,
      "step": 345
    },
    {
      "epoch": 0.8963730569948186,
      "grad_norm": 0.8470476865768433,
      "learning_rate": 5.92232319213878e-06,
      "loss": 0.9703,
      "step": 346
    },
    {
      "epoch": 0.8989637305699482,
      "grad_norm": 0.816389262676239,
      "learning_rate": 5.902055041725105e-06,
      "loss": 0.9642,
      "step": 347
    },
    {
      "epoch": 0.9015544041450777,
      "grad_norm": 0.7632498741149902,
      "learning_rate": 5.8817715584847744e-06,
      "loss": 1.0025,
      "step": 348
    },
    {
      "epoch": 0.9041450777202072,
      "grad_norm": 0.8325628638267517,
      "learning_rate": 5.861473087189584e-06,
      "loss": 1.0194,
      "step": 349
    },
    {
      "epoch": 0.9067357512953368,
      "grad_norm": 0.921424150466919,
      "learning_rate": 5.841159972866085e-06,
      "loss": 0.9659,
      "step": 350
    },
    {
      "epoch": 0.9093264248704663,
      "grad_norm": 0.8112065196037292,
      "learning_rate": 5.820832560789727e-06,
      "loss": 1.0551,
      "step": 351
    },
    {
      "epoch": 0.9119170984455959,
      "grad_norm": 0.9384793043136597,
      "learning_rate": 5.800491196478989e-06,
      "loss": 0.9915,
      "step": 352
    },
    {
      "epoch": 0.9145077720207254,
      "grad_norm": 0.6935743689537048,
      "learning_rate": 5.780136225689505e-06,
      "loss": 1.0039,
      "step": 353
    },
    {
      "epoch": 0.917098445595855,
      "grad_norm": 0.7410438060760498,
      "learning_rate": 5.759767994408188e-06,
      "loss": 0.8865,
      "step": 354
    },
    {
      "epoch": 0.9196891191709845,
      "grad_norm": 0.6849777102470398,
      "learning_rate": 5.739386848847346e-06,
      "loss": 0.9646,
      "step": 355
    },
    {
      "epoch": 0.9222797927461139,
      "grad_norm": 0.8872363567352295,
      "learning_rate": 5.718993135438803e-06,
      "loss": 1.0218,
      "step": 356
    },
    {
      "epoch": 0.9248704663212435,
      "grad_norm": 0.8674810528755188,
      "learning_rate": 5.6985872008280045e-06,
      "loss": 1.0209,
      "step": 357
    },
    {
      "epoch": 0.927461139896373,
      "grad_norm": 0.9334598779678345,
      "learning_rate": 5.678169391868128e-06,
      "loss": 0.9841,
      "step": 358
    },
    {
      "epoch": 0.9300518134715026,
      "grad_norm": 0.7986560463905334,
      "learning_rate": 5.6577400556141906e-06,
      "loss": 0.9821,
      "step": 359
    },
    {
      "epoch": 0.9326424870466321,
      "grad_norm": 0.7242634296417236,
      "learning_rate": 5.637299539317141e-06,
      "loss": 1.0017,
      "step": 360
    },
    {
      "epoch": 0.9352331606217616,
      "grad_norm": 0.9153129458427429,
      "learning_rate": 5.616848190417965e-06,
      "loss": 0.9975,
      "step": 361
    },
    {
      "epoch": 0.9378238341968912,
      "grad_norm": 1.0346280336380005,
      "learning_rate": 5.596386356541779e-06,
      "loss": 1.0298,
      "step": 362
    },
    {
      "epoch": 0.9404145077720207,
      "grad_norm": 0.7707669734954834,
      "learning_rate": 5.575914385491917e-06,
      "loss": 0.9551,
      "step": 363
    },
    {
      "epoch": 0.9430051813471503,
      "grad_norm": 0.7947671413421631,
      "learning_rate": 5.555432625244024e-06,
      "loss": 0.9803,
      "step": 364
    },
    {
      "epoch": 0.9455958549222798,
      "grad_norm": 0.8810315132141113,
      "learning_rate": 5.534941423940135e-06,
      "loss": 0.9941,
      "step": 365
    },
    {
      "epoch": 0.9481865284974094,
      "grad_norm": 0.848483681678772,
      "learning_rate": 5.51444112988276e-06,
      "loss": 1.0864,
      "step": 366
    },
    {
      "epoch": 0.9507772020725389,
      "grad_norm": 0.816373884677887,
      "learning_rate": 5.493932091528972e-06,
      "loss": 0.9821,
      "step": 367
    },
    {
      "epoch": 0.9533678756476683,
      "grad_norm": 0.8828781843185425,
      "learning_rate": 5.473414657484468e-06,
      "loss": 0.9929,
      "step": 368
    },
    {
      "epoch": 0.9559585492227979,
      "grad_norm": 0.8907682299613953,
      "learning_rate": 5.452889176497659e-06,
      "loss": 0.9888,
      "step": 369
    },
    {
      "epoch": 0.9585492227979274,
      "grad_norm": 0.9210544228553772,
      "learning_rate": 5.432355997453729e-06,
      "loss": 0.9757,
      "step": 370
    },
    {
      "epoch": 0.961139896373057,
      "grad_norm": 0.9528911709785461,
      "learning_rate": 5.4118154693687165e-06,
      "loss": 0.9014,
      "step": 371
    },
    {
      "epoch": 0.9637305699481865,
      "grad_norm": 0.7835102677345276,
      "learning_rate": 5.391267941383572e-06,
      "loss": 0.9736,
      "step": 372
    },
    {
      "epoch": 0.966321243523316,
      "grad_norm": 0.9772538542747498,
      "learning_rate": 5.3707137627582315e-06,
      "loss": 1.0071,
      "step": 373
    },
    {
      "epoch": 0.9689119170984456,
      "grad_norm": 0.8421518802642822,
      "learning_rate": 5.350153282865674e-06,
      "loss": 1.0501,
      "step": 374
    },
    {
      "epoch": 0.9715025906735751,
      "grad_norm": 0.7764169573783875,
      "learning_rate": 5.329586851185987e-06,
      "loss": 1.0491,
      "step": 375
    },
    {
      "epoch": 0.9740932642487047,
      "grad_norm": 0.7631876468658447,
      "learning_rate": 5.309014817300422e-06,
      "loss": 1.0018,
      "step": 376
    },
    {
      "epoch": 0.9766839378238342,
      "grad_norm": 0.816758394241333,
      "learning_rate": 5.2884375308854565e-06,
      "loss": 1.0032,
      "step": 377
    },
    {
      "epoch": 0.9792746113989638,
      "grad_norm": 0.7835342288017273,
      "learning_rate": 5.26785534170685e-06,
      "loss": 0.9686,
      "step": 378
    },
    {
      "epoch": 0.9818652849740933,
      "grad_norm": 0.8044567704200745,
      "learning_rate": 5.247268599613696e-06,
      "loss": 0.9695,
      "step": 379
    },
    {
      "epoch": 0.9844559585492227,
      "grad_norm": 0.883346438407898,
      "learning_rate": 5.226677654532476e-06,
      "loss": 0.9709,
      "step": 380
    },
    {
      "epoch": 0.9870466321243523,
      "grad_norm": 0.9801338911056519,
      "learning_rate": 5.206082856461115e-06,
      "loss": 0.9887,
      "step": 381
    },
    {
      "epoch": 0.9896373056994818,
      "grad_norm": 0.8825283646583557,
      "learning_rate": 5.185484555463026e-06,
      "loss": 0.9649,
      "step": 382
    },
    {
      "epoch": 0.9922279792746114,
      "grad_norm": 1.0225670337677002,
      "learning_rate": 5.16488310166117e-06,
      "loss": 1.0526,
      "step": 383
    },
    {
      "epoch": 0.9948186528497409,
      "grad_norm": 1.0191495418548584,
      "learning_rate": 5.1442788452320915e-06,
      "loss": 0.9138,
      "step": 384
    },
    {
      "epoch": 0.9974093264248705,
      "grad_norm": 0.8653509616851807,
      "learning_rate": 5.123672136399975e-06,
      "loss": 0.9498,
      "step": 385
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9232073426246643,
      "learning_rate": 5.1030633254306935e-06,
      "loss": 0.9828,
      "step": 386
    },
    {
      "epoch": 1.0025906735751295,
      "grad_norm": 0.8782436847686768,
      "learning_rate": 5.082452762625848e-06,
      "loss": 1.0139,
      "step": 387
    },
    {
      "epoch": 1.005181347150259,
      "grad_norm": 0.9029590487480164,
      "learning_rate": 5.061840798316815e-06,
      "loss": 1.0483,
      "step": 388
    },
    {
      "epoch": 1.005181347150259,
      "eval_loss": 0.9548383951187134,
      "eval_runtime": 99.0137,
      "eval_samples_per_second": 2.525,
      "eval_steps_per_second": 0.323,
      "step": 388
    },
    {
      "epoch": 1.0077720207253886,
      "grad_norm": 0.9202731847763062,
      "learning_rate": 5.041227782858799e-06,
      "loss": 1.0369,
      "step": 389
    },
    {
      "epoch": 1.0103626943005182,
      "grad_norm": 0.8768073320388794,
      "learning_rate": 5.020614066624868e-06,
      "loss": 1.0371,
      "step": 390
    },
    {
      "epoch": 1.0129533678756477,
      "grad_norm": 1.0436216592788696,
      "learning_rate": 5e-06,
      "loss": 1.0636,
      "step": 391
    },
    {
      "epoch": 1.0155440414507773,
      "grad_norm": 0.8364196419715881,
      "learning_rate": 4.979385933375133e-06,
      "loss": 0.9556,
      "step": 392
    },
    {
      "epoch": 1.0181347150259068,
      "grad_norm": 0.8371309041976929,
      "learning_rate": 4.958772217141203e-06,
      "loss": 0.9272,
      "step": 393
    },
    {
      "epoch": 1.0207253886010363,
      "grad_norm": 0.7747754454612732,
      "learning_rate": 4.9381592016831856e-06,
      "loss": 1.0231,
      "step": 394
    },
    {
      "epoch": 1.0233160621761659,
      "grad_norm": 0.7783029079437256,
      "learning_rate": 4.917547237374153e-06,
      "loss": 0.9855,
      "step": 395
    },
    {
      "epoch": 1.0025906735751295,
      "grad_norm": 0.9176076650619507,
      "learning_rate": 4.896936674569309e-06,
      "loss": 1.0037,
      "step": 396
    },
    {
      "epoch": 1.005181347150259,
      "grad_norm": 0.9758632183074951,
      "learning_rate": 4.876327863600026e-06,
      "loss": 0.987,
      "step": 397
    },
    {
      "epoch": 1.0077720207253886,
      "grad_norm": 0.8266388177871704,
      "learning_rate": 4.85572115476791e-06,
      "loss": 1.0135,
      "step": 398
    },
    {
      "epoch": 1.0103626943005182,
      "grad_norm": 0.8124467730522156,
      "learning_rate": 4.83511689833883e-06,
      "loss": 0.9964,
      "step": 399
    },
    {
      "epoch": 1.0129533678756477,
      "grad_norm": 1.061855673789978,
      "learning_rate": 4.814515444536975e-06,
      "loss": 1.0406,
      "step": 400
    },
    {
      "epoch": 1.0155440414507773,
      "grad_norm": 0.89823317527771,
      "learning_rate": 4.793917143538887e-06,
      "loss": 1.0009,
      "step": 401
    },
    {
      "epoch": 1.0181347150259068,
      "grad_norm": 0.7810267806053162,
      "learning_rate": 4.773322345467525e-06,
      "loss": 1.0269,
      "step": 402
    },
    {
      "epoch": 1.0207253886010363,
      "grad_norm": 1.0242594480514526,
      "learning_rate": 4.752731400386306e-06,
      "loss": 0.9579,
      "step": 403
    },
    {
      "epoch": 1.0233160621761659,
      "grad_norm": 0.8448682427406311,
      "learning_rate": 4.732144658293151e-06,
      "loss": 1.0156,
      "step": 404
    },
    {
      "epoch": 1.0259067357512954,
      "grad_norm": 1.009443998336792,
      "learning_rate": 4.711562469114544e-06,
      "loss": 1.0358,
      "step": 405
    },
    {
      "epoch": 1.028497409326425,
      "grad_norm": 0.9292247891426086,
      "learning_rate": 4.690985182699581e-06,
      "loss": 0.9328,
      "step": 406
    },
    {
      "epoch": 1.0310880829015545,
      "grad_norm": 0.7662960886955261,
      "learning_rate": 4.670413148814015e-06,
      "loss": 0.9747,
      "step": 407
    },
    {
      "epoch": 1.0336787564766838,
      "grad_norm": 0.8195393681526184,
      "learning_rate": 4.649846717134327e-06,
      "loss": 0.9467,
      "step": 408
    },
    {
      "epoch": 1.0362694300518134,
      "grad_norm": 0.8635261058807373,
      "learning_rate": 4.62928623724177e-06,
      "loss": 0.9653,
      "step": 409
    },
    {
      "epoch": 1.038860103626943,
      "grad_norm": 0.929542064666748,
      "learning_rate": 4.6087320586164296e-06,
      "loss": 0.9214,
      "step": 410
    },
    {
      "epoch": 1.0414507772020725,
      "grad_norm": 0.8770561814308167,
      "learning_rate": 4.588184530631284e-06,
      "loss": 0.9743,
      "step": 411
    },
    {
      "epoch": 1.044041450777202,
      "grad_norm": 0.7577627301216125,
      "learning_rate": 4.567644002546273e-06,
      "loss": 1.016,
      "step": 412
    },
    {
      "epoch": 1.0466321243523315,
      "grad_norm": 0.888802707195282,
      "learning_rate": 4.547110823502343e-06,
      "loss": 0.9667,
      "step": 413
    },
    {
      "epoch": 1.049222797927461,
      "grad_norm": 0.7431952357292175,
      "learning_rate": 4.526585342515533e-06,
      "loss": 1.0157,
      "step": 414
    },
    {
      "epoch": 1.0518134715025906,
      "grad_norm": 0.9357666969299316,
      "learning_rate": 4.506067908471029e-06,
      "loss": 1.0433,
      "step": 415
    },
    {
      "epoch": 1.0544041450777202,
      "grad_norm": 0.9243563413619995,
      "learning_rate": 4.485558870117241e-06,
      "loss": 0.9251,
      "step": 416
    },
    {
      "epoch": 1.0569948186528497,
      "grad_norm": 0.9315341711044312,
      "learning_rate": 4.465058576059868e-06,
      "loss": 0.9785,
      "step": 417
    },
    {
      "epoch": 1.0595854922279793,
      "grad_norm": 1.0324779748916626,
      "learning_rate": 4.444567374755978e-06,
      "loss": 0.994,
      "step": 418
    },
    {
      "epoch": 1.0621761658031088,
      "grad_norm": 0.784785270690918,
      "learning_rate": 4.424085614508084e-06,
      "loss": 1.0312,
      "step": 419
    },
    {
      "epoch": 1.0647668393782384,
      "grad_norm": 0.9518358111381531,
      "learning_rate": 4.403613643458222e-06,
      "loss": 1.0396,
      "step": 420
    },
    {
      "epoch": 1.067357512953368,
      "grad_norm": 1.0345162153244019,
      "learning_rate": 4.383151809582035e-06,
      "loss": 1.0877,
      "step": 421
    },
    {
      "epoch": 1.0699481865284974,
      "grad_norm": 0.8149455189704895,
      "learning_rate": 4.362700460682861e-06,
      "loss": 0.9827,
      "step": 422
    },
    {
      "epoch": 1.072538860103627,
      "grad_norm": 0.8805851936340332,
      "learning_rate": 4.342259944385811e-06,
      "loss": 0.9838,
      "step": 423
    },
    {
      "epoch": 1.0751295336787565,
      "grad_norm": 0.8283563852310181,
      "learning_rate": 4.321830608131872e-06,
      "loss": 1.0128,
      "step": 424
    },
    {
      "epoch": 1.077720207253886,
      "grad_norm": 1.1321141719818115,
      "learning_rate": 4.301412799171998e-06,
      "loss": 0.8987,
      "step": 425
    },
    {
      "epoch": 1.0803108808290156,
      "grad_norm": 0.9323210120201111,
      "learning_rate": 4.281006864561199e-06,
      "loss": 0.9618,
      "step": 426
    },
    {
      "epoch": 1.0829015544041452,
      "grad_norm": 0.7723137140274048,
      "learning_rate": 4.260613151152655e-06,
      "loss": 0.9636,
      "step": 427
    },
    {
      "epoch": 1.0854922279792747,
      "grad_norm": 0.9840835928916931,
      "learning_rate": 4.240232005591816e-06,
      "loss": 0.9963,
      "step": 428
    },
    {
      "epoch": 1.0880829015544042,
      "grad_norm": 0.8220058679580688,
      "learning_rate": 4.219863774310497e-06,
      "loss": 0.9901,
      "step": 429
    },
    {
      "epoch": 1.0906735751295338,
      "grad_norm": 0.9598246216773987,
      "learning_rate": 4.1995088035210126e-06,
      "loss": 0.9561,
      "step": 430
    },
    {
      "epoch": 1.093264248704663,
      "grad_norm": 0.7539969682693481,
      "learning_rate": 4.179167439210275e-06,
      "loss": 0.9904,
      "step": 431
    },
    {
      "epoch": 1.0958549222797926,
      "grad_norm": 0.8430355191230774,
      "learning_rate": 4.158840027133917e-06,
      "loss": 1.0678,
      "step": 432
    },
    {
      "epoch": 1.0984455958549222,
      "grad_norm": 0.9641113877296448,
      "learning_rate": 4.138526912810418e-06,
      "loss": 0.9488,
      "step": 433
    },
    {
      "epoch": 1.1010362694300517,
      "grad_norm": 0.8497964143753052,
      "learning_rate": 4.1182284415152255e-06,
      "loss": 1.0023,
      "step": 434
    },
    {
      "epoch": 1.1036269430051813,
      "grad_norm": 0.8309699892997742,
      "learning_rate": 4.097944958274898e-06,
      "loss": 0.9936,
      "step": 435
    },
    {
      "epoch": 1.1062176165803108,
      "grad_norm": 1.0693992376327515,
      "learning_rate": 4.077676807861221e-06,
      "loss": 0.9811,
      "step": 436
    },
    {
      "epoch": 1.1088082901554404,
      "grad_norm": 0.8500522375106812,
      "learning_rate": 4.057424334785366e-06,
      "loss": 0.9753,
      "step": 437
    },
    {
      "epoch": 1.11139896373057,
      "grad_norm": 0.7945443391799927,
      "learning_rate": 4.037187883292027e-06,
      "loss": 0.9833,
      "step": 438
    },
    {
      "epoch": 1.1139896373056994,
      "grad_norm": 0.7603841423988342,
      "learning_rate": 4.016967797353558e-06,
      "loss": 0.9531,
      "step": 439
    },
    {
      "epoch": 1.116580310880829,
      "grad_norm": 0.9688854217529297,
      "learning_rate": 3.996764420664149e-06,
      "loss": 0.9785,
      "step": 440
    },
    {
      "epoch": 1.1191709844559585,
      "grad_norm": 1.072771668434143,
      "learning_rate": 3.976578096633969e-06,
      "loss": 0.9511,
      "step": 441
    },
    {
      "epoch": 1.121761658031088,
      "grad_norm": 0.8664917349815369,
      "learning_rate": 3.956409168383325e-06,
      "loss": 0.9434,
      "step": 442
    },
    {
      "epoch": 1.1243523316062176,
      "grad_norm": 0.8430099487304688,
      "learning_rate": 3.936257978736845e-06,
      "loss": 0.9397,
      "step": 443
    },
    {
      "epoch": 1.1269430051813472,
      "grad_norm": 0.9500722885131836,
      "learning_rate": 3.916124870217635e-06,
      "loss": 1.0198,
      "step": 444
    },
    {
      "epoch": 1.1295336787564767,
      "grad_norm": 1.0132358074188232,
      "learning_rate": 3.89601018504147e-06,
      "loss": 1.0048,
      "step": 445
    },
    {
      "epoch": 1.1321243523316062,
      "grad_norm": 0.8455902338027954,
      "learning_rate": 3.875914265110967e-06,
      "loss": 0.9248,
      "step": 446
    },
    {
      "epoch": 1.1347150259067358,
      "grad_norm": 0.8625523447990417,
      "learning_rate": 3.85583745200978e-06,
      "loss": 1.0251,
      "step": 447
    },
    {
      "epoch": 1.1373056994818653,
      "grad_norm": 0.9898774027824402,
      "learning_rate": 3.835780086996794e-06,
      "loss": 0.998,
      "step": 448
    },
    {
      "epoch": 1.1398963730569949,
      "grad_norm": 1.021031141281128,
      "learning_rate": 3.815742511000313e-06,
      "loss": 1.0321,
      "step": 449
    },
    {
      "epoch": 1.1424870466321244,
      "grad_norm": 0.8724047541618347,
      "learning_rate": 3.7957250646122843e-06,
      "loss": 0.9535,
      "step": 450
    },
    {
      "epoch": 1.145077720207254,
      "grad_norm": 0.820254921913147,
      "learning_rate": 3.7757280880824946e-06,
      "loss": 0.9489,
      "step": 451
    },
    {
      "epoch": 1.1476683937823835,
      "grad_norm": 1.034761667251587,
      "learning_rate": 3.755751921312788e-06,
      "loss": 0.9903,
      "step": 452
    },
    {
      "epoch": 1.150259067357513,
      "grad_norm": 0.8738400936126709,
      "learning_rate": 3.735796903851297e-06,
      "loss": 0.9504,
      "step": 453
    },
    {
      "epoch": 1.1528497409326426,
      "grad_norm": 0.8705572485923767,
      "learning_rate": 3.715863374886661e-06,
      "loss": 0.9571,
      "step": 454
    },
    {
      "epoch": 1.1554404145077721,
      "grad_norm": 0.8942792415618896,
      "learning_rate": 3.695951673242267e-06,
      "loss": 0.932,
      "step": 455
    },
    {
      "epoch": 1.1580310880829017,
      "grad_norm": 0.9116060733795166,
      "learning_rate": 3.6760621373704867e-06,
      "loss": 0.9602,
      "step": 456
    },
    {
      "epoch": 1.160621761658031,
      "grad_norm": 0.9232982993125916,
      "learning_rate": 3.6561951053469313e-06,
      "loss": 0.9944,
      "step": 457
    },
    {
      "epoch": 1.1632124352331605,
      "grad_norm": 0.6849638223648071,
      "learning_rate": 3.636350914864689e-06,
      "loss": 1.014,
      "step": 458
    },
    {
      "epoch": 1.16580310880829,
      "grad_norm": 1.0841094255447388,
      "learning_rate": 3.6165299032286055e-06,
      "loss": 0.8914,
      "step": 459
    },
    {
      "epoch": 1.1683937823834196,
      "grad_norm": 0.8563059568405151,
      "learning_rate": 3.5967324073495363e-06,
      "loss": 0.972,
      "step": 460
    },
    {
      "epoch": 1.1709844559585492,
      "grad_norm": 1.0777771472930908,
      "learning_rate": 3.5769587637386206e-06,
      "loss": 1.0476,
      "step": 461
    },
    {
      "epoch": 1.1735751295336787,
      "grad_norm": 0.790111780166626,
      "learning_rate": 3.5572093085015683e-06,
      "loss": 0.959,
      "step": 462
    },
    {
      "epoch": 1.1761658031088082,
      "grad_norm": 0.8875803351402283,
      "learning_rate": 3.537484377332945e-06,
      "loss": 1.0109,
      "step": 463
    },
    {
      "epoch": 1.1787564766839378,
      "grad_norm": 0.932518720626831,
      "learning_rate": 3.5177843055104633e-06,
      "loss": 1.0012,
      "step": 464
    },
    {
      "epoch": 1.1813471502590673,
      "grad_norm": 0.776736855506897,
      "learning_rate": 3.4981094278892813e-06,
      "loss": 0.9852,
      "step": 465
    },
    {
      "epoch": 1.1839378238341969,
      "grad_norm": 0.8646918535232544,
      "learning_rate": 3.4784600788963197e-06,
      "loss": 1.0376,
      "step": 466
    },
    {
      "epoch": 1.1865284974093264,
      "grad_norm": 0.9904392957687378,
      "learning_rate": 3.458836592524569e-06,
      "loss": 0.9692,
      "step": 467
    },
    {
      "epoch": 1.189119170984456,
      "grad_norm": 0.8158794641494751,
      "learning_rate": 3.4392393023274173e-06,
      "loss": 1.0176,
      "step": 468
    },
    {
      "epoch": 1.1917098445595855,
      "grad_norm": 0.8880725502967834,
      "learning_rate": 3.419668541412977e-06,
      "loss": 0.9946,
      "step": 469
    },
    {
      "epoch": 1.194300518134715,
      "grad_norm": 1.0934456586837769,
      "learning_rate": 3.4001246424384294e-06,
      "loss": 0.9071,
      "step": 470
    },
    {
      "epoch": 1.1968911917098446,
      "grad_norm": 0.9213566780090332,
      "learning_rate": 3.380607937604358e-06,
      "loss": 0.9489,
      "step": 471
    },
    {
      "epoch": 1.1994818652849741,
      "grad_norm": 0.8582881689071655,
      "learning_rate": 3.361118758649116e-06,
      "loss": 0.9756,
      "step": 472
    },
    {
      "epoch": 1.2020725388601037,
      "grad_norm": 0.9744678735733032,
      "learning_rate": 3.341657436843181e-06,
      "loss": 1.0436,
      "step": 473
    },
    {
      "epoch": 1.2046632124352332,
      "grad_norm": 0.9584829807281494,
      "learning_rate": 3.322224302983517e-06,
      "loss": 1.0492,
      "step": 474
    },
    {
      "epoch": 1.2072538860103628,
      "grad_norm": 0.8227071166038513,
      "learning_rate": 3.302819687387967e-06,
      "loss": 0.9419,
      "step": 475
    },
    {
      "epoch": 1.2098445595854923,
      "grad_norm": 1.1721601486206055,
      "learning_rate": 3.2834439198896285e-06,
      "loss": 1.0105,
      "step": 476
    },
    {
      "epoch": 1.2124352331606219,
      "grad_norm": 0.96018385887146,
      "learning_rate": 3.264097329831244e-06,
      "loss": 1.0335,
      "step": 477
    },
    {
      "epoch": 1.2150259067357512,
      "grad_norm": 0.8224446177482605,
      "learning_rate": 3.2447802460596124e-06,
      "loss": 0.9914,
      "step": 478
    },
    {
      "epoch": 1.2176165803108807,
      "grad_norm": 0.9198357462882996,
      "learning_rate": 3.2254929969199933e-06,
      "loss": 0.957,
      "step": 479
    },
    {
      "epoch": 1.2202072538860103,
      "grad_norm": 1.0417511463165283,
      "learning_rate": 3.206235910250529e-06,
      "loss": 1.1192,
      "step": 480
    },
    {
      "epoch": 1.2227979274611398,
      "grad_norm": 0.9127503037452698,
      "learning_rate": 3.1870093133766653e-06,
      "loss": 0.9585,
      "step": 481
    },
    {
      "epoch": 1.2253886010362693,
      "grad_norm": 0.8656659722328186,
      "learning_rate": 3.167813533105598e-06,
      "loss": 0.9106,
      "step": 482
    },
    {
      "epoch": 1.2279792746113989,
      "grad_norm": 0.9434940218925476,
      "learning_rate": 3.148648895720714e-06,
      "loss": 0.9955,
      "step": 483
    },
    {
      "epoch": 1.2305699481865284,
      "grad_norm": 0.8094484806060791,
      "learning_rate": 3.1295157269760347e-06,
      "loss": 1.007,
      "step": 484
    },
    {
      "epoch": 1.233160621761658,
      "grad_norm": 1.175736904144287,
      "learning_rate": 3.1104143520906976e-06,
      "loss": 0.9983,
      "step": 485
    },
    {
      "epoch": 1.233160621761658,
      "eval_loss": 0.9479498267173767,
      "eval_runtime": 99.1167,
      "eval_samples_per_second": 2.522,
      "eval_steps_per_second": 0.323,
      "step": 485
    },
    {
      "epoch": 1.2357512953367875,
      "grad_norm": 0.872238278388977,
      "learning_rate": 3.0913450957434177e-06,
      "loss": 1.0085,
      "step": 486
    },
    {
      "epoch": 1.238341968911917,
      "grad_norm": 0.8121694326400757,
      "learning_rate": 3.0723082820669634e-06,
      "loss": 0.9683,
      "step": 487
    },
    {
      "epoch": 1.2409326424870466,
      "grad_norm": 0.8866750597953796,
      "learning_rate": 3.0533042346426612e-06,
      "loss": 0.9324,
      "step": 488
    },
    {
      "epoch": 1.2435233160621761,
      "grad_norm": 0.8889084458351135,
      "learning_rate": 3.034333276494884e-06,
      "loss": 0.9622,
      "step": 489
    },
    {
      "epoch": 1.2461139896373057,
      "grad_norm": 0.9424455761909485,
      "learning_rate": 3.015395730085565e-06,
      "loss": 0.9475,
      "step": 490
    },
    {
      "epoch": 1.2487046632124352,
      "grad_norm": 0.8178656101226807,
      "learning_rate": 2.9964919173087154e-06,
      "loss": 0.9323,
      "step": 491
    },
    {
      "epoch": 1.2512953367875648,
      "grad_norm": 0.8948029279708862,
      "learning_rate": 2.9776221594849565e-06,
      "loss": 0.9929,
      "step": 492
    },
    {
      "epoch": 1.2538860103626943,
      "grad_norm": 0.8116942048072815,
      "learning_rate": 2.9587867773560488e-06,
      "loss": 1.0207,
      "step": 493
    },
    {
      "epoch": 1.2564766839378239,
      "grad_norm": 0.8729721903800964,
      "learning_rate": 2.9399860910794532e-06,
      "loss": 1.0275,
      "step": 494
    },
    {
      "epoch": 1.2590673575129534,
      "grad_norm": 0.977170467376709,
      "learning_rate": 2.921220420222878e-06,
      "loss": 1.0432,
      "step": 495
    },
    {
      "epoch": 1.261658031088083,
      "grad_norm": 0.8280233144760132,
      "learning_rate": 2.902490083758856e-06,
      "loss": 0.9642,
      "step": 496
    },
    {
      "epoch": 1.2642487046632125,
      "grad_norm": 0.9376110434532166,
      "learning_rate": 2.8837954000593106e-06,
      "loss": 0.9649,
      "step": 497
    },
    {
      "epoch": 1.266839378238342,
      "grad_norm": 0.7053350210189819,
      "learning_rate": 2.8651366868901543e-06,
      "loss": 0.9851,
      "step": 498
    },
    {
      "epoch": 1.2694300518134716,
      "grad_norm": 0.8334206938743591,
      "learning_rate": 2.8465142614058916e-06,
      "loss": 0.9808,
      "step": 499
    },
    {
      "epoch": 1.2720207253886011,
      "grad_norm": 0.8525857329368591,
      "learning_rate": 2.8279284401442085e-06,
      "loss": 0.9964,
      "step": 500
    },
    {
      "epoch": 1.2746113989637307,
      "grad_norm": 0.8159149289131165,
      "learning_rate": 2.809379539020618e-06,
      "loss": 1.0053,
      "step": 501
    },
    {
      "epoch": 1.2772020725388602,
      "grad_norm": 0.9699737429618835,
      "learning_rate": 2.790867873323067e-06,
      "loss": 0.9218,
      "step": 502
    },
    {
      "epoch": 1.2797927461139897,
      "grad_norm": 0.8307886123657227,
      "learning_rate": 2.7723937577065924e-06,
      "loss": 0.9667,
      "step": 503
    },
    {
      "epoch": 1.2823834196891193,
      "grad_norm": 0.9945038557052612,
      "learning_rate": 2.753957506187964e-06,
      "loss": 1.0314,
      "step": 504
    },
    {
      "epoch": 1.2849740932642488,
      "grad_norm": 1.1403707265853882,
      "learning_rate": 2.735559432140358e-06,
      "loss": 0.9839,
      "step": 505
    },
    {
      "epoch": 1.2875647668393784,
      "grad_norm": 1.1950939893722534,
      "learning_rate": 2.7171998482880093e-06,
      "loss": 1.0106,
      "step": 506
    },
    {
      "epoch": 1.2901554404145077,
      "grad_norm": 0.9658251404762268,
      "learning_rate": 2.6988790667009246e-06,
      "loss": 0.9977,
      "step": 507
    },
    {
      "epoch": 1.2927461139896372,
      "grad_norm": 0.8657079935073853,
      "learning_rate": 2.680597398789554e-06,
      "loss": 0.9296,
      "step": 508
    },
    {
      "epoch": 1.2953367875647668,
      "grad_norm": 0.9752938151359558,
      "learning_rate": 2.6623551552995076e-06,
      "loss": 1.0082,
      "step": 509
    },
    {
      "epoch": 1.2979274611398963,
      "grad_norm": 0.9197008609771729,
      "learning_rate": 2.6441526463062727e-06,
      "loss": 0.9655,
      "step": 510
    },
    {
      "epoch": 1.3005181347150259,
      "grad_norm": 1.0991517305374146,
      "learning_rate": 2.6259901812099432e-06,
      "loss": 0.8592,
      "step": 511
    },
    {
      "epoch": 1.3031088082901554,
      "grad_norm": 0.9919476509094238,
      "learning_rate": 2.607868068729966e-06,
      "loss": 1.014,
      "step": 512
    },
    {
      "epoch": 1.305699481865285,
      "grad_norm": 0.8860031366348267,
      "learning_rate": 2.5897866168998754e-06,
      "loss": 0.9493,
      "step": 513
    },
    {
      "epoch": 1.3082901554404145,
      "grad_norm": 0.9551827907562256,
      "learning_rate": 2.571746133062082e-06,
      "loss": 0.9784,
      "step": 514
    },
    {
      "epoch": 1.310880829015544,
      "grad_norm": 0.9395473003387451,
      "learning_rate": 2.5537469238626296e-06,
      "loss": 0.9847,
      "step": 515
    },
    {
      "epoch": 1.3134715025906736,
      "grad_norm": 0.9098738431930542,
      "learning_rate": 2.5357892952459917e-06,
      "loss": 0.9757,
      "step": 516
    },
    {
      "epoch": 1.3160621761658031,
      "grad_norm": 1.0057542324066162,
      "learning_rate": 2.517873552449869e-06,
      "loss": 1.0406,
      "step": 517
    },
    {
      "epoch": 1.3186528497409327,
      "grad_norm": 0.9436793327331543,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.9948,
      "step": 518
    },
    {
      "epoch": 1.3212435233160622,
      "grad_norm": 0.9765656590461731,
      "learning_rate": 2.4821689417049898e-06,
      "loss": 0.9242,
      "step": 519
    },
    {
      "epoch": 1.3238341968911918,
      "grad_norm": 0.9807612895965576,
      "learning_rate": 2.4643806806511344e-06,
      "loss": 0.979,
      "step": 520
    },
    {
      "epoch": 1.3264248704663213,
      "grad_norm": 1.051147699356079,
      "learning_rate": 2.4466355191972886e-06,
      "loss": 0.905,
      "step": 521
    },
    {
      "epoch": 1.3290155440414508,
      "grad_norm": 0.975245475769043,
      "learning_rate": 2.4289337589697e-06,
      "loss": 1.0436,
      "step": 522
    },
    {
      "epoch": 1.3316062176165804,
      "grad_norm": 1.053765892982483,
      "learning_rate": 2.4112757008569102e-06,
      "loss": 1.0155,
      "step": 523
    },
    {
      "epoch": 1.33419689119171,
      "grad_norm": 0.7797523736953735,
      "learning_rate": 2.3936616450046207e-06,
      "loss": 1.0052,
      "step": 524
    },
    {
      "epoch": 1.3367875647668392,
      "grad_norm": 0.8835865259170532,
      "learning_rate": 2.376091890810598e-06,
      "loss": 0.9177,
      "step": 525
    },
    {
      "epoch": 1.3393782383419688,
      "grad_norm": 0.8106747269630432,
      "learning_rate": 2.3585667369195815e-06,
      "loss": 0.9525,
      "step": 526
    },
    {
      "epoch": 1.3419689119170983,
      "grad_norm": 1.0181241035461426,
      "learning_rate": 2.341086481218217e-06,
      "loss": 0.9024,
      "step": 527
    },
    {
      "epoch": 1.3445595854922279,
      "grad_norm": 0.9752720594406128,
      "learning_rate": 2.32365142082998e-06,
      "loss": 1.0024,
      "step": 528
    },
    {
      "epoch": 1.3471502590673574,
      "grad_norm": 0.8057308793067932,
      "learning_rate": 2.306261852110132e-06,
      "loss": 0.9294,
      "step": 529
    },
    {
      "epoch": 1.349740932642487,
      "grad_norm": 1.02779221534729,
      "learning_rate": 2.288918070640684e-06,
      "loss": 0.9539,
      "step": 530
    },
    {
      "epoch": 1.3523316062176165,
      "grad_norm": 0.8755698800086975,
      "learning_rate": 2.2716203712253708e-06,
      "loss": 1.0245,
      "step": 531
    },
    {
      "epoch": 1.354922279792746,
      "grad_norm": 0.8415625691413879,
      "learning_rate": 2.254369047884639e-06,
      "loss": 0.9994,
      "step": 532
    },
    {
      "epoch": 1.3575129533678756,
      "grad_norm": 1.01762056350708,
      "learning_rate": 2.2371643938506488e-06,
      "loss": 0.9763,
      "step": 533
    },
    {
      "epoch": 1.3601036269430051,
      "grad_norm": 1.0196937322616577,
      "learning_rate": 2.2200067015622986e-06,
      "loss": 0.9745,
      "step": 534
    },
    {
      "epoch": 1.3626943005181347,
      "grad_norm": 1.0955042839050293,
      "learning_rate": 2.2028962626602346e-06,
      "loss": 0.9457,
      "step": 535
    },
    {
      "epoch": 1.3652849740932642,
      "grad_norm": 0.8075233697891235,
      "learning_rate": 2.185833367981918e-06,
      "loss": 0.9042,
      "step": 536
    },
    {
      "epoch": 1.3678756476683938,
      "grad_norm": 0.8643060922622681,
      "learning_rate": 2.168818307556663e-06,
      "loss": 0.9541,
      "step": 537
    },
    {
      "epoch": 1.3704663212435233,
      "grad_norm": 0.7847241759300232,
      "learning_rate": 2.1518513706007154e-06,
      "loss": 0.9874,
      "step": 538
    },
    {
      "epoch": 1.3730569948186528,
      "grad_norm": 0.8477650880813599,
      "learning_rate": 2.13493284551233e-06,
      "loss": 1.0541,
      "step": 539
    },
    {
      "epoch": 1.3756476683937824,
      "grad_norm": 0.993405818939209,
      "learning_rate": 2.118063019866884e-06,
      "loss": 1.0024,
      "step": 540
    },
    {
      "epoch": 1.378238341968912,
      "grad_norm": 0.9656562805175781,
      "learning_rate": 2.101242180411963e-06,
      "loss": 0.9929,
      "step": 541
    },
    {
      "epoch": 1.3808290155440415,
      "grad_norm": 0.88680100440979,
      "learning_rate": 2.0844706130625146e-06,
      "loss": 1.0225,
      "step": 542
    },
    {
      "epoch": 1.383419689119171,
      "grad_norm": 0.9172679781913757,
      "learning_rate": 2.067748602895969e-06,
      "loss": 0.9504,
      "step": 543
    },
    {
      "epoch": 1.3860103626943006,
      "grad_norm": 0.9063828587532043,
      "learning_rate": 2.0510764341474032e-06,
      "loss": 0.9287,
      "step": 544
    },
    {
      "epoch": 1.38860103626943,
      "grad_norm": 0.9309386014938354,
      "learning_rate": 2.0344543902047043e-06,
      "loss": 1.015,
      "step": 545
    },
    {
      "epoch": 1.3911917098445596,
      "grad_norm": 0.9050936102867126,
      "learning_rate": 2.0178827536037547e-06,
      "loss": 0.9952,
      "step": 546
    },
    {
      "epoch": 1.3937823834196892,
      "grad_norm": 1.3205598592758179,
      "learning_rate": 2.001361806023636e-06,
      "loss": 0.911,
      "step": 547
    },
    {
      "epoch": 1.3963730569948187,
      "grad_norm": 0.8794770240783691,
      "learning_rate": 1.9848918282818242e-06,
      "loss": 0.9863,
      "step": 548
    },
    {
      "epoch": 1.3989637305699483,
      "grad_norm": 0.9719308614730835,
      "learning_rate": 1.968473100329437e-06,
      "loss": 0.9862,
      "step": 549
    },
    {
      "epoch": 1.4015544041450778,
      "grad_norm": 0.8657988905906677,
      "learning_rate": 1.952105901246461e-06,
      "loss": 1.0129,
      "step": 550
    },
    {
      "epoch": 1.4041450777202074,
      "grad_norm": 1.1228150129318237,
      "learning_rate": 1.935790509237013e-06,
      "loss": 0.9855,
      "step": 551
    },
    {
      "epoch": 1.406735751295337,
      "grad_norm": 0.838809072971344,
      "learning_rate": 1.9195272016246105e-06,
      "loss": 0.9796,
      "step": 552
    },
    {
      "epoch": 1.4093264248704664,
      "grad_norm": 0.9168699383735657,
      "learning_rate": 1.9033162548474577e-06,
      "loss": 0.9694,
      "step": 553
    },
    {
      "epoch": 1.411917098445596,
      "grad_norm": 0.9007121324539185,
      "learning_rate": 1.887157944453749e-06,
      "loss": 0.9737,
      "step": 554
    },
    {
      "epoch": 1.4145077720207253,
      "grad_norm": 0.8796411156654358,
      "learning_rate": 1.8710525450969803e-06,
      "loss": 0.9293,
      "step": 555
    },
    {
      "epoch": 1.4170984455958548,
      "grad_norm": 0.9988073110580444,
      "learning_rate": 1.855000330531289e-06,
      "loss": 0.9632,
      "step": 556
    },
    {
      "epoch": 1.4196891191709844,
      "grad_norm": 0.9262337684631348,
      "learning_rate": 1.8390015736067869e-06,
      "loss": 0.9679,
      "step": 557
    },
    {
      "epoch": 1.422279792746114,
      "grad_norm": 0.9995079040527344,
      "learning_rate": 1.823056546264939e-06,
      "loss": 0.9558,
      "step": 558
    },
    {
      "epoch": 1.4248704663212435,
      "grad_norm": 0.855469286441803,
      "learning_rate": 1.8071655195339272e-06,
      "loss": 1.0535,
      "step": 559
    },
    {
      "epoch": 1.427461139896373,
      "grad_norm": 0.760111927986145,
      "learning_rate": 1.7913287635240573e-06,
      "loss": 0.9714,
      "step": 560
    },
    {
      "epoch": 1.4300518134715026,
      "grad_norm": 0.9862337708473206,
      "learning_rate": 1.7755465474231465e-06,
      "loss": 1.0004,
      "step": 561
    },
    {
      "epoch": 1.432642487046632,
      "grad_norm": 0.7655313611030579,
      "learning_rate": 1.7598191394919738e-06,
      "loss": 0.9812,
      "step": 562
    },
    {
      "epoch": 1.4352331606217616,
      "grad_norm": 1.0747240781784058,
      "learning_rate": 1.7441468070597017e-06,
      "loss": 1.0208,
      "step": 563
    },
    {
      "epoch": 1.4378238341968912,
      "grad_norm": 0.9578352570533752,
      "learning_rate": 1.7285298165193388e-06,
      "loss": 0.9637,
      "step": 564
    },
    {
      "epoch": 1.4404145077720207,
      "grad_norm": 0.9221131801605225,
      "learning_rate": 1.7129684333232095e-06,
      "loss": 0.952,
      "step": 565
    },
    {
      "epoch": 1.4430051813471503,
      "grad_norm": 1.1166445016860962,
      "learning_rate": 1.697462921978446e-06,
      "loss": 1.0394,
      "step": 566
    },
    {
      "epoch": 1.4455958549222798,
      "grad_norm": 0.8609073758125305,
      "learning_rate": 1.682013546042488e-06,
      "loss": 1.0053,
      "step": 567
    },
    {
      "epoch": 1.4481865284974094,
      "grad_norm": 0.8035356998443604,
      "learning_rate": 1.6666205681186032e-06,
      "loss": 1.0051,
      "step": 568
    },
    {
      "epoch": 1.450777202072539,
      "grad_norm": 0.8815962076187134,
      "learning_rate": 1.6512842498514315e-06,
      "loss": 0.9821,
      "step": 569
    },
    {
      "epoch": 1.4533678756476685,
      "grad_norm": 0.9989993572235107,
      "learning_rate": 1.6360048519225197e-06,
      "loss": 0.9804,
      "step": 570
    },
    {
      "epoch": 1.455958549222798,
      "grad_norm": 0.9338482618331909,
      "learning_rate": 1.6207826340459131e-06,
      "loss": 0.9514,
      "step": 571
    },
    {
      "epoch": 1.4585492227979275,
      "grad_norm": 0.7817990183830261,
      "learning_rate": 1.6056178549637248e-06,
      "loss": 1.0398,
      "step": 572
    },
    {
      "epoch": 1.4611398963730569,
      "grad_norm": 0.9715204834938049,
      "learning_rate": 1.5905107724417412e-06,
      "loss": 0.9859,
      "step": 573
    },
    {
      "epoch": 1.4637305699481864,
      "grad_norm": 0.8073526620864868,
      "learning_rate": 1.5754616432650443e-06,
      "loss": 0.9364,
      "step": 574
    },
    {
      "epoch": 1.466321243523316,
      "grad_norm": 0.908467710018158,
      "learning_rate": 1.5604707232336457e-06,
      "loss": 0.97,
      "step": 575
    },
    {
      "epoch": 1.4689119170984455,
      "grad_norm": 0.7307873964309692,
      "learning_rate": 1.5455382671581365e-06,
      "loss": 0.9081,
      "step": 576
    },
    {
      "epoch": 1.471502590673575,
      "grad_norm": 0.9267300367355347,
      "learning_rate": 1.5306645288553556e-06,
      "loss": 0.9719,
      "step": 577
    },
    {
      "epoch": 1.4740932642487046,
      "grad_norm": 1.200758695602417,
      "learning_rate": 1.5158497611440792e-06,
      "loss": 1.0408,
      "step": 578
    },
    {
      "epoch": 1.4766839378238341,
      "grad_norm": 0.8580562472343445,
      "learning_rate": 1.5010942158407204e-06,
      "loss": 0.8879,
      "step": 579
    },
    {
      "epoch": 1.4792746113989637,
      "grad_norm": 0.7956061959266663,
      "learning_rate": 1.48639814375505e-06,
      "loss": 1.0052,
      "step": 580
    },
    {
      "epoch": 1.4818652849740932,
      "grad_norm": 1.047351360321045,
      "learning_rate": 1.4717617946859319e-06,
      "loss": 1.0048,
      "step": 581
    },
    {
      "epoch": 1.4844559585492227,
      "grad_norm": 0.9025893211364746,
      "learning_rate": 1.4571854174170847e-06,
      "loss": 0.9976,
      "step": 582
    },
    {
      "epoch": 1.4844559585492227,
      "eval_loss": 0.9445139765739441,
      "eval_runtime": 99.1624,
      "eval_samples_per_second": 2.521,
      "eval_steps_per_second": 0.323,
      "step": 582
    },
    {
      "epoch": 1.4870466321243523,
      "grad_norm": 0.8686254620552063,
      "learning_rate": 1.4426692597128339e-06,
      "loss": 0.9141,
      "step": 583
    },
    {
      "epoch": 1.4896373056994818,
      "grad_norm": 0.8144710659980774,
      "learning_rate": 1.428213568313927e-06,
      "loss": 0.9429,
      "step": 584
    },
    {
      "epoch": 1.4922279792746114,
      "grad_norm": 0.8153760433197021,
      "learning_rate": 1.4138185889333172e-06,
      "loss": 0.9564,
      "step": 585
    },
    {
      "epoch": 1.494818652849741,
      "grad_norm": 0.8345690369606018,
      "learning_rate": 1.3994845662519985e-06,
      "loss": 0.9794,
      "step": 586
    },
    {
      "epoch": 1.4974093264248705,
      "grad_norm": 1.0408412218093872,
      "learning_rate": 1.3852117439148416e-06,
      "loss": 0.9961,
      "step": 587
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9831898212432861,
      "learning_rate": 1.3710003645264559e-06,
      "loss": 1.0319,
      "step": 588
    },
    {
      "epoch": 1.5025906735751295,
      "grad_norm": 0.9798739552497864,
      "learning_rate": 1.3568506696470645e-06,
      "loss": 1.0054,
      "step": 589
    },
    {
      "epoch": 1.505181347150259,
      "grad_norm": 0.741041898727417,
      "learning_rate": 1.3427628997883957e-06,
      "loss": 0.9952,
      "step": 590
    },
    {
      "epoch": 1.5077720207253886,
      "grad_norm": 0.7825758457183838,
      "learning_rate": 1.3287372944096021e-06,
      "loss": 0.993,
      "step": 591
    },
    {
      "epoch": 1.5103626943005182,
      "grad_norm": 0.8112067580223083,
      "learning_rate": 1.3147740919131814e-06,
      "loss": 1.0047,
      "step": 592
    },
    {
      "epoch": 1.5129533678756477,
      "grad_norm": 0.8506227135658264,
      "learning_rate": 1.3008735296409281e-06,
      "loss": 0.9148,
      "step": 593
    },
    {
      "epoch": 1.5155440414507773,
      "grad_norm": 0.8730190396308899,
      "learning_rate": 1.2870358438699005e-06,
      "loss": 0.9805,
      "step": 594
    },
    {
      "epoch": 1.5181347150259068,
      "grad_norm": 0.8171835541725159,
      "learning_rate": 1.2732612698084067e-06,
      "loss": 0.9846,
      "step": 595
    },
    {
      "epoch": 1.5207253886010363,
      "grad_norm": 0.8341667652130127,
      "learning_rate": 1.2595500415919948e-06,
      "loss": 0.9912,
      "step": 596
    },
    {
      "epoch": 1.5233160621761659,
      "grad_norm": 0.8373621106147766,
      "learning_rate": 1.245902392279491e-06,
      "loss": 1.0312,
      "step": 597
    },
    {
      "epoch": 1.5259067357512954,
      "grad_norm": 1.0364506244659424,
      "learning_rate": 1.232318553849023e-06,
      "loss": 0.9417,
      "step": 598
    },
    {
      "epoch": 1.528497409326425,
      "grad_norm": 0.8765987157821655,
      "learning_rate": 1.2187987571940846e-06,
      "loss": 1.0039,
      "step": 599
    },
    {
      "epoch": 1.5310880829015545,
      "grad_norm": 0.8896673917770386,
      "learning_rate": 1.2053432321196085e-06,
      "loss": 0.9867,
      "step": 600
    },
    {
      "epoch": 1.533678756476684,
      "grad_norm": 0.7940327525138855,
      "learning_rate": 1.1919522073380614e-06,
      "loss": 1.0547,
      "step": 601
    },
    {
      "epoch": 1.5362694300518136,
      "grad_norm": 0.903992772102356,
      "learning_rate": 1.1786259104655562e-06,
      "loss": 0.9451,
      "step": 602
    },
    {
      "epoch": 1.5388601036269431,
      "grad_norm": 0.8643820285797119,
      "learning_rate": 1.1653645680179792e-06,
      "loss": 0.8786,
      "step": 603
    },
    {
      "epoch": 1.5414507772020727,
      "grad_norm": 0.7279991507530212,
      "learning_rate": 1.1521684054071524e-06,
      "loss": 0.9964,
      "step": 604
    },
    {
      "epoch": 1.5440414507772022,
      "grad_norm": 0.8498103022575378,
      "learning_rate": 1.1390376469369796e-06,
      "loss": 0.9547,
      "step": 605
    },
    {
      "epoch": 1.5466321243523318,
      "grad_norm": 0.800202488899231,
      "learning_rate": 1.1259725157996593e-06,
      "loss": 0.9599,
      "step": 606
    },
    {
      "epoch": 1.549222797927461,
      "grad_norm": 0.9869511127471924,
      "learning_rate": 1.1129732340718702e-06,
      "loss": 1.0126,
      "step": 607
    },
    {
      "epoch": 1.5518134715025906,
      "grad_norm": 0.9484280347824097,
      "learning_rate": 1.1000400227110142e-06,
      "loss": 0.9711,
      "step": 608
    },
    {
      "epoch": 1.5544041450777202,
      "grad_norm": 0.9410507082939148,
      "learning_rate": 1.087173101551439e-06,
      "loss": 0.9845,
      "step": 609
    },
    {
      "epoch": 1.5569948186528497,
      "grad_norm": 0.7972487807273865,
      "learning_rate": 1.0743726893007257e-06,
      "loss": 0.9653,
      "step": 610
    },
    {
      "epoch": 1.5595854922279793,
      "grad_norm": 0.9086237549781799,
      "learning_rate": 1.061639003535952e-06,
      "loss": 1.002,
      "step": 611
    },
    {
      "epoch": 1.5621761658031088,
      "grad_norm": 0.9138615131378174,
      "learning_rate": 1.0489722607000052e-06,
      "loss": 0.9809,
      "step": 612
    },
    {
      "epoch": 1.5647668393782384,
      "grad_norm": 0.9875264167785645,
      "learning_rate": 1.036372676097897e-06,
      "loss": 0.9163,
      "step": 613
    },
    {
      "epoch": 1.567357512953368,
      "grad_norm": 0.9866282939910889,
      "learning_rate": 1.0238404638931077e-06,
      "loss": 1.067,
      "step": 614
    },
    {
      "epoch": 1.5699481865284974,
      "grad_norm": 0.9424428939819336,
      "learning_rate": 1.0113758371039429e-06,
      "loss": 0.9418,
      "step": 615
    },
    {
      "epoch": 1.572538860103627,
      "grad_norm": 0.9368731379508972,
      "learning_rate": 9.989790075999145e-07,
      "loss": 0.9741,
      "step": 616
    },
    {
      "epoch": 1.5751295336787565,
      "grad_norm": 0.8245809674263,
      "learning_rate": 9.866501860981431e-07,
      "loss": 0.9837,
      "step": 617
    },
    {
      "epoch": 1.577720207253886,
      "grad_norm": 0.964534342288971,
      "learning_rate": 9.743895821597638e-07,
      "loss": 0.961,
      "step": 618
    },
    {
      "epoch": 1.5803108808290154,
      "grad_norm": 0.9792007803916931,
      "learning_rate": 9.621974041863813e-07,
      "loss": 0.898,
      "step": 619
    },
    {
      "epoch": 1.582901554404145,
      "grad_norm": 0.8622314929962158,
      "learning_rate": 9.500738594165132e-07,
      "loss": 1.0085,
      "step": 620
    },
    {
      "epoch": 1.5854922279792745,
      "grad_norm": 0.936686098575592,
      "learning_rate": 9.380191539220762e-07,
      "loss": 0.9979,
      "step": 621
    },
    {
      "epoch": 1.588082901554404,
      "grad_norm": 1.0908080339431763,
      "learning_rate": 9.260334926048787e-07,
      "loss": 0.9885,
      "step": 622
    },
    {
      "epoch": 1.5906735751295336,
      "grad_norm": 0.8463078737258911,
      "learning_rate": 9.141170791931386e-07,
      "loss": 0.9314,
      "step": 623
    },
    {
      "epoch": 1.593264248704663,
      "grad_norm": 1.073156714439392,
      "learning_rate": 9.022701162380259e-07,
      "loss": 1.0001,
      "step": 624
    },
    {
      "epoch": 1.5958549222797926,
      "grad_norm": 0.8691221475601196,
      "learning_rate": 8.904928051102074e-07,
      "loss": 0.9589,
      "step": 625
    },
    {
      "epoch": 1.5984455958549222,
      "grad_norm": 0.9988768696784973,
      "learning_rate": 8.787853459964407e-07,
      "loss": 1.0873,
      "step": 626
    },
    {
      "epoch": 1.6010362694300517,
      "grad_norm": 1.0085712671279907,
      "learning_rate": 8.671479378961556e-07,
      "loss": 0.9401,
      "step": 627
    },
    {
      "epoch": 1.6036269430051813,
      "grad_norm": 1.0235176086425781,
      "learning_rate": 8.555807786180814e-07,
      "loss": 1.0291,
      "step": 628
    },
    {
      "epoch": 1.6062176165803108,
      "grad_norm": 0.9803045988082886,
      "learning_rate": 8.4408406477688e-07,
      "loss": 0.9434,
      "step": 629
    },
    {
      "epoch": 1.6088082901554404,
      "grad_norm": 0.9053346514701843,
      "learning_rate": 8.326579917898098e-07,
      "loss": 0.9941,
      "step": 630
    },
    {
      "epoch": 1.61139896373057,
      "grad_norm": 0.9022249579429626,
      "learning_rate": 8.21302753873393e-07,
      "loss": 0.947,
      "step": 631
    },
    {
      "epoch": 1.6139896373056994,
      "grad_norm": 1.0404874086380005,
      "learning_rate": 8.100185440401276e-07,
      "loss": 1.0749,
      "step": 632
    },
    {
      "epoch": 1.616580310880829,
      "grad_norm": 0.9517078399658203,
      "learning_rate": 7.988055540951967e-07,
      "loss": 0.9732,
      "step": 633
    },
    {
      "epoch": 1.6191709844559585,
      "grad_norm": 0.9991124868392944,
      "learning_rate": 7.876639746332132e-07,
      "loss": 1.0463,
      "step": 634
    },
    {
      "epoch": 1.621761658031088,
      "grad_norm": 0.9260480403900146,
      "learning_rate": 7.765939950349776e-07,
      "loss": 0.9353,
      "step": 635
    },
    {
      "epoch": 1.6243523316062176,
      "grad_norm": 0.9367355704307556,
      "learning_rate": 7.655958034642619e-07,
      "loss": 0.9758,
      "step": 636
    },
    {
      "epoch": 1.6269430051813472,
      "grad_norm": 0.777481734752655,
      "learning_rate": 7.54669586864607e-07,
      "loss": 0.927,
      "step": 637
    },
    {
      "epoch": 1.6295336787564767,
      "grad_norm": 0.9329550266265869,
      "learning_rate": 7.43815530956149e-07,
      "loss": 0.9997,
      "step": 638
    },
    {
      "epoch": 1.6321243523316062,
      "grad_norm": 0.9558778405189514,
      "learning_rate": 7.330338202324621e-07,
      "loss": 0.9486,
      "step": 639
    },
    {
      "epoch": 1.6347150259067358,
      "grad_norm": 0.9044818878173828,
      "learning_rate": 7.223246379574206e-07,
      "loss": 0.9678,
      "step": 640
    },
    {
      "epoch": 1.6373056994818653,
      "grad_norm": 1.0184038877487183,
      "learning_rate": 7.116881661620833e-07,
      "loss": 0.9467,
      "step": 641
    },
    {
      "epoch": 1.6398963730569949,
      "grad_norm": 0.82332843542099,
      "learning_rate": 7.011245856416016e-07,
      "loss": 1.0119,
      "step": 642
    },
    {
      "epoch": 1.6424870466321244,
      "grad_norm": 1.0882319211959839,
      "learning_rate": 6.90634075952149e-07,
      "loss": 1.0505,
      "step": 643
    },
    {
      "epoch": 1.645077720207254,
      "grad_norm": 0.8696717619895935,
      "learning_rate": 6.802168154078586e-07,
      "loss": 1.0155,
      "step": 644
    },
    {
      "epoch": 1.6476683937823835,
      "grad_norm": 0.8515833616256714,
      "learning_rate": 6.698729810778065e-07,
      "loss": 1.0446,
      "step": 645
    },
    {
      "epoch": 1.650259067357513,
      "grad_norm": 0.9765702486038208,
      "learning_rate": 6.596027487829915e-07,
      "loss": 0.9914,
      "step": 646
    },
    {
      "epoch": 1.6528497409326426,
      "grad_norm": 0.96961510181427,
      "learning_rate": 6.494062930933497e-07,
      "loss": 0.9454,
      "step": 647
    },
    {
      "epoch": 1.6554404145077721,
      "grad_norm": 0.9544724822044373,
      "learning_rate": 6.392837873247876e-07,
      "loss": 0.953,
      "step": 648
    },
    {
      "epoch": 1.6580310880829017,
      "grad_norm": 1.0494675636291504,
      "learning_rate": 6.292354035362369e-07,
      "loss": 1.0195,
      "step": 649
    },
    {
      "epoch": 1.6606217616580312,
      "grad_norm": 1.032488465309143,
      "learning_rate": 6.192613125267283e-07,
      "loss": 1.0027,
      "step": 650
    },
    {
      "epoch": 1.6632124352331608,
      "grad_norm": 1.010940432548523,
      "learning_rate": 6.093616838324872e-07,
      "loss": 0.9949,
      "step": 651
    },
    {
      "epoch": 1.6658031088082903,
      "grad_norm": 0.7320314645767212,
      "learning_rate": 5.995366857240592e-07,
      "loss": 0.9881,
      "step": 652
    },
    {
      "epoch": 1.6683937823834198,
      "grad_norm": 1.1827306747436523,
      "learning_rate": 5.897864852034368e-07,
      "loss": 0.9126,
      "step": 653
    },
    {
      "epoch": 1.6709844559585494,
      "grad_norm": 0.8718465566635132,
      "learning_rate": 5.801112480012344e-07,
      "loss": 0.8972,
      "step": 654
    },
    {
      "epoch": 1.6735751295336787,
      "grad_norm": 0.8429136276245117,
      "learning_rate": 5.705111385738638e-07,
      "loss": 1.0019,
      "step": 655
    },
    {
      "epoch": 1.6761658031088082,
      "grad_norm": 0.9650006294250488,
      "learning_rate": 5.609863201007382e-07,
      "loss": 0.992,
      "step": 656
    },
    {
      "epoch": 1.6787564766839378,
      "grad_norm": 1.091964840888977,
      "learning_rate": 5.515369544815025e-07,
      "loss": 1.0321,
      "step": 657
    },
    {
      "epoch": 1.6813471502590673,
      "grad_norm": 0.7537882924079895,
      "learning_rate": 5.421632023332779e-07,
      "loss": 0.9601,
      "step": 658
    },
    {
      "epoch": 1.6839378238341969,
      "grad_norm": 0.9340682029724121,
      "learning_rate": 5.328652229879383e-07,
      "loss": 0.9789,
      "step": 659
    },
    {
      "epoch": 1.6865284974093264,
      "grad_norm": 0.7913133502006531,
      "learning_rate": 5.236431744893883e-07,
      "loss": 0.9062,
      "step": 660
    },
    {
      "epoch": 1.689119170984456,
      "grad_norm": 0.9730626344680786,
      "learning_rate": 5.144972135908949e-07,
      "loss": 0.938,
      "step": 661
    },
    {
      "epoch": 1.6917098445595855,
      "grad_norm": 0.8744672536849976,
      "learning_rate": 5.054274957524075e-07,
      "loss": 1.0014,
      "step": 662
    },
    {
      "epoch": 1.694300518134715,
      "grad_norm": 0.9841236472129822,
      "learning_rate": 4.964341751379248e-07,
      "loss": 0.9094,
      "step": 663
    },
    {
      "epoch": 1.6968911917098446,
      "grad_norm": 1.0411112308502197,
      "learning_rate": 4.875174046128684e-07,
      "loss": 1.011,
      "step": 664
    },
    {
      "epoch": 1.6994818652849741,
      "grad_norm": 0.9232543706893921,
      "learning_rate": 4.786773357414926e-07,
      "loss": 0.9188,
      "step": 665
    },
    {
      "epoch": 1.7020725388601037,
      "grad_norm": 0.9863471984863281,
      "learning_rate": 4.6991411878429593e-07,
      "loss": 0.9577,
      "step": 666
    },
    {
      "epoch": 1.704663212435233,
      "grad_norm": 0.8515871167182922,
      "learning_rate": 4.612279026954808e-07,
      "loss": 0.9712,
      "step": 667
    },
    {
      "epoch": 1.7072538860103625,
      "grad_norm": 1.0452429056167603,
      "learning_rate": 4.526188351204103e-07,
      "loss": 0.9525,
      "step": 668
    },
    {
      "epoch": 1.709844559585492,
      "grad_norm": 1.039813756942749,
      "learning_rate": 4.440870623931054e-07,
      "loss": 0.9677,
      "step": 669
    },
    {
      "epoch": 1.7124352331606216,
      "grad_norm": 1.0061904191970825,
      "learning_rate": 4.3563272953375426e-07,
      "loss": 0.9249,
      "step": 670
    },
    {
      "epoch": 1.7150259067357512,
      "grad_norm": 0.9659146666526794,
      "learning_rate": 4.2725598024624783e-07,
      "loss": 0.932,
      "step": 671
    },
    {
      "epoch": 1.7176165803108807,
      "grad_norm": 1.100063443183899,
      "learning_rate": 4.1895695691574146e-07,
      "loss": 0.9303,
      "step": 672
    },
    {
      "epoch": 1.7202072538860103,
      "grad_norm": 0.9187989830970764,
      "learning_rate": 4.1073580060622455e-07,
      "loss": 0.984,
      "step": 673
    },
    {
      "epoch": 1.7227979274611398,
      "grad_norm": 0.8058050870895386,
      "learning_rate": 4.025926510581357e-07,
      "loss": 0.9469,
      "step": 674
    },
    {
      "epoch": 1.7253886010362693,
      "grad_norm": 0.9303697347640991,
      "learning_rate": 3.9452764668597764e-07,
      "loss": 1.0155,
      "step": 675
    },
    {
      "epoch": 1.7279792746113989,
      "grad_norm": 0.875150740146637,
      "learning_rate": 3.8654092457596714e-07,
      "loss": 0.9498,
      "step": 676
    },
    {
      "epoch": 1.7305699481865284,
      "grad_norm": 1.206532597541809,
      "learning_rate": 3.786326204837065e-07,
      "loss": 1.0035,
      "step": 677
    },
    {
      "epoch": 1.733160621761658,
      "grad_norm": 0.9007678627967834,
      "learning_rate": 3.7080286883187713e-07,
      "loss": 0.9713,
      "step": 678
    },
    {
      "epoch": 1.7357512953367875,
      "grad_norm": 1.0224071741104126,
      "learning_rate": 3.6305180270794827e-07,
      "loss": 0.9525,
      "step": 679
    },
    {
      "epoch": 1.7357512953367875,
      "eval_loss": 0.9424797892570496,
      "eval_runtime": 98.9403,
      "eval_samples_per_second": 2.527,
      "eval_steps_per_second": 0.323,
      "step": 679
    },
    {
      "epoch": 1.738341968911917,
      "grad_norm": 0.8955622315406799,
      "learning_rate": 3.553795538619237e-07,
      "loss": 0.9495,
      "step": 680
    },
    {
      "epoch": 1.7409326424870466,
      "grad_norm": 0.9500125050544739,
      "learning_rate": 3.4778625270409484e-07,
      "loss": 0.9822,
      "step": 681
    },
    {
      "epoch": 1.7435233160621761,
      "grad_norm": 0.9673233032226562,
      "learning_rate": 3.402720283028277e-07,
      "loss": 0.9392,
      "step": 682
    },
    {
      "epoch": 1.7461139896373057,
      "grad_norm": 1.0081723928451538,
      "learning_rate": 3.328370083823679e-07,
      "loss": 0.9768,
      "step": 683
    },
    {
      "epoch": 1.7487046632124352,
      "grad_norm": 1.021333932876587,
      "learning_rate": 3.2548131932067184e-07,
      "loss": 0.9438,
      "step": 684
    },
    {
      "epoch": 1.7512953367875648,
      "grad_norm": 0.8333784341812134,
      "learning_rate": 3.182050861472541e-07,
      "loss": 1.0067,
      "step": 685
    },
    {
      "epoch": 1.7538860103626943,
      "grad_norm": 1.0099414587020874,
      "learning_rate": 3.110084325410667e-07,
      "loss": 1.0327,
      "step": 686
    },
    {
      "epoch": 1.7564766839378239,
      "grad_norm": 1.2605355978012085,
      "learning_rate": 3.03891480828396e-07,
      "loss": 0.9497,
      "step": 687
    },
    {
      "epoch": 1.7590673575129534,
      "grad_norm": 0.9966846704483032,
      "learning_rate": 2.9685435198078095e-07,
      "loss": 1.0268,
      "step": 688
    },
    {
      "epoch": 1.761658031088083,
      "grad_norm": 0.920387327671051,
      "learning_rate": 2.898971656129573e-07,
      "loss": 0.9526,
      "step": 689
    },
    {
      "epoch": 1.7642487046632125,
      "grad_norm": 0.9313926100730896,
      "learning_rate": 2.830200399808286e-07,
      "loss": 0.9874,
      "step": 690
    },
    {
      "epoch": 1.766839378238342,
      "grad_norm": 0.9481755495071411,
      "learning_rate": 2.762230919794506e-07,
      "loss": 0.9602,
      "step": 691
    },
    {
      "epoch": 1.7694300518134716,
      "grad_norm": 0.9870350360870361,
      "learning_rate": 2.6950643714104774e-07,
      "loss": 0.9758,
      "step": 692
    },
    {
      "epoch": 1.7720207253886011,
      "grad_norm": 0.8995896577835083,
      "learning_rate": 2.628701896330482e-07,
      "loss": 0.9692,
      "step": 693
    },
    {
      "epoch": 1.7746113989637307,
      "grad_norm": 0.8859801888465881,
      "learning_rate": 2.563144622561453e-07,
      "loss": 0.978,
      "step": 694
    },
    {
      "epoch": 1.7772020725388602,
      "grad_norm": 0.8300227522850037,
      "learning_rate": 2.498393664423737e-07,
      "loss": 0.9199,
      "step": 695
    },
    {
      "epoch": 1.7797927461139897,
      "grad_norm": 0.9782614707946777,
      "learning_rate": 2.4344501225322557e-07,
      "loss": 0.9437,
      "step": 696
    },
    {
      "epoch": 1.7823834196891193,
      "grad_norm": 0.9227774739265442,
      "learning_rate": 2.3713150837777142e-07,
      "loss": 0.9153,
      "step": 697
    },
    {
      "epoch": 1.7849740932642488,
      "grad_norm": 0.854295551776886,
      "learning_rate": 2.3089896213081553e-07,
      "loss": 0.8942,
      "step": 698
    },
    {
      "epoch": 1.7875647668393784,
      "grad_norm": 0.761452853679657,
      "learning_rate": 2.2474747945107068e-07,
      "loss": 1.0242,
      "step": 699
    },
    {
      "epoch": 1.790155440414508,
      "grad_norm": 0.7350905537605286,
      "learning_rate": 2.1867716489936297e-07,
      "loss": 1.0056,
      "step": 700
    },
    {
      "epoch": 1.7927461139896375,
      "grad_norm": 0.9781407117843628,
      "learning_rate": 2.1268812165684382e-07,
      "loss": 0.9554,
      "step": 701
    },
    {
      "epoch": 1.795336787564767,
      "grad_norm": 0.7397287487983704,
      "learning_rate": 2.0678045152324798e-07,
      "loss": 1.0214,
      "step": 702
    },
    {
      "epoch": 1.7979274611398963,
      "grad_norm": 0.9220542311668396,
      "learning_rate": 2.0095425491515386e-07,
      "loss": 0.9359,
      "step": 703
    },
    {
      "epoch": 1.8005181347150259,
      "grad_norm": 0.8612248301506042,
      "learning_rate": 1.9520963086428258e-07,
      "loss": 0.9994,
      "step": 704
    },
    {
      "epoch": 1.8031088082901554,
      "grad_norm": 0.8949776887893677,
      "learning_rate": 1.8954667701581108e-07,
      "loss": 0.9721,
      "step": 705
    },
    {
      "epoch": 1.805699481865285,
      "grad_norm": 1.0004165172576904,
      "learning_rate": 1.8396548962671456e-07,
      "loss": 0.9811,
      "step": 706
    },
    {
      "epoch": 1.8082901554404145,
      "grad_norm": 0.8982537388801575,
      "learning_rate": 1.7846616356413105e-07,
      "loss": 0.9457,
      "step": 707
    },
    {
      "epoch": 1.810880829015544,
      "grad_norm": 0.9139662981033325,
      "learning_rate": 1.7304879230374328e-07,
      "loss": 1.0623,
      "step": 708
    },
    {
      "epoch": 1.8134715025906736,
      "grad_norm": 0.9726191163063049,
      "learning_rate": 1.677134679281983e-07,
      "loss": 1.0008,
      "step": 709
    },
    {
      "epoch": 1.8160621761658031,
      "grad_norm": 0.9855490326881409,
      "learning_rate": 1.6246028112553603e-07,
      "loss": 1.0302,
      "step": 710
    },
    {
      "epoch": 1.8186528497409327,
      "grad_norm": 0.9363789558410645,
      "learning_rate": 1.5728932118764916e-07,
      "loss": 0.9876,
      "step": 711
    },
    {
      "epoch": 1.8212435233160622,
      "grad_norm": 1.0399082899093628,
      "learning_rate": 1.5220067600876686e-07,
      "loss": 0.8963,
      "step": 712
    },
    {
      "epoch": 1.8238341968911918,
      "grad_norm": 0.8987425565719604,
      "learning_rate": 1.4719443208396078e-07,
      "loss": 0.9535,
      "step": 713
    },
    {
      "epoch": 1.8264248704663213,
      "grad_norm": 0.8786448240280151,
      "learning_rate": 1.422706745076713e-07,
      "loss": 1.0302,
      "step": 714
    },
    {
      "epoch": 1.8290155440414506,
      "grad_norm": 1.0369514226913452,
      "learning_rate": 1.3742948697226533e-07,
      "loss": 0.9518,
      "step": 715
    },
    {
      "epoch": 1.8316062176165802,
      "grad_norm": 0.9875585436820984,
      "learning_rate": 1.3267095176661304e-07,
      "loss": 0.9448,
      "step": 716
    },
    {
      "epoch": 1.8341968911917097,
      "grad_norm": 1.0473564863204956,
      "learning_rate": 1.2799514977468618e-07,
      "loss": 0.9618,
      "step": 717
    },
    {
      "epoch": 1.8367875647668392,
      "grad_norm": 1.106512427330017,
      "learning_rate": 1.2340216047418697e-07,
      "loss": 1.02,
      "step": 718
    },
    {
      "epoch": 1.8393782383419688,
      "grad_norm": 0.8493651747703552,
      "learning_rate": 1.188920619351941e-07,
      "loss": 0.9718,
      "step": 719
    },
    {
      "epoch": 1.8419689119170983,
      "grad_norm": 0.8997021317481995,
      "learning_rate": 1.1446493081883891e-07,
      "loss": 0.9508,
      "step": 720
    },
    {
      "epoch": 1.8445595854922279,
      "grad_norm": 0.8545548319816589,
      "learning_rate": 1.1012084237599808e-07,
      "loss": 1.0185,
      "step": 721
    },
    {
      "epoch": 1.8471502590673574,
      "grad_norm": 0.8686257600784302,
      "learning_rate": 1.0585987044602009e-07,
      "loss": 0.9757,
      "step": 722
    },
    {
      "epoch": 1.849740932642487,
      "grad_norm": 0.8913059234619141,
      "learning_rate": 1.0168208745546416e-07,
      "loss": 0.9635,
      "step": 723
    },
    {
      "epoch": 1.8523316062176165,
      "grad_norm": 0.8280147910118103,
      "learning_rate": 9.758756441687333e-08,
      "loss": 1.0558,
      "step": 724
    },
    {
      "epoch": 1.854922279792746,
      "grad_norm": 0.9251644611358643,
      "learning_rate": 9.357637092756667e-08,
      "loss": 0.9797,
      "step": 725
    },
    {
      "epoch": 1.8575129533678756,
      "grad_norm": 0.8681091666221619,
      "learning_rate": 8.964857516845449e-08,
      "loss": 0.98,
      "step": 726
    },
    {
      "epoch": 1.8601036269430051,
      "grad_norm": 0.858220636844635,
      "learning_rate": 8.580424390288167e-08,
      "loss": 0.9679,
      "step": 727
    },
    {
      "epoch": 1.8626943005181347,
      "grad_norm": 1.0083863735198975,
      "learning_rate": 8.204344247549067e-08,
      "loss": 0.9765,
      "step": 728
    },
    {
      "epoch": 1.8652849740932642,
      "grad_norm": 0.7215631604194641,
      "learning_rate": 7.836623481111416e-08,
      "loss": 0.9826,
      "step": 729
    },
    {
      "epoch": 1.8678756476683938,
      "grad_norm": 0.8494441509246826,
      "learning_rate": 7.47726834136836e-08,
      "loss": 0.9878,
      "step": 730
    },
    {
      "epoch": 1.8704663212435233,
      "grad_norm": 0.9821386337280273,
      "learning_rate": 7.126284936517125e-08,
      "loss": 0.98,
      "step": 731
    },
    {
      "epoch": 1.8730569948186528,
      "grad_norm": 0.887163519859314,
      "learning_rate": 6.783679232455043e-08,
      "loss": 0.9275,
      "step": 732
    },
    {
      "epoch": 1.8756476683937824,
      "grad_norm": 0.9071056246757507,
      "learning_rate": 6.449457052677965e-08,
      "loss": 0.9906,
      "step": 733
    },
    {
      "epoch": 1.878238341968912,
      "grad_norm": 0.9382151961326599,
      "learning_rate": 6.123624078181512e-08,
      "loss": 1.0063,
      "step": 734
    },
    {
      "epoch": 1.8808290155440415,
      "grad_norm": 0.8040828704833984,
      "learning_rate": 5.8061858473645315e-08,
      "loss": 0.9858,
      "step": 735
    },
    {
      "epoch": 1.883419689119171,
      "grad_norm": 0.9526659250259399,
      "learning_rate": 5.4971477559346286e-08,
      "loss": 0.9893,
      "step": 736
    },
    {
      "epoch": 1.8860103626943006,
      "grad_norm": 0.8950933218002319,
      "learning_rate": 5.196515056816898e-08,
      "loss": 1.0216,
      "step": 737
    },
    {
      "epoch": 1.88860103626943,
      "grad_norm": 0.8529356718063354,
      "learning_rate": 4.90429286006433e-08,
      "loss": 1.0495,
      "step": 738
    },
    {
      "epoch": 1.8911917098445596,
      "grad_norm": 0.8746633529663086,
      "learning_rate": 4.620486132770996e-08,
      "loss": 0.9701,
      "step": 739
    },
    {
      "epoch": 1.8937823834196892,
      "grad_norm": 0.7848332524299622,
      "learning_rate": 4.34509969898772e-08,
      "loss": 0.9551,
      "step": 740
    },
    {
      "epoch": 1.8963730569948187,
      "grad_norm": 0.844379186630249,
      "learning_rate": 4.078138239639984e-08,
      "loss": 0.9392,
      "step": 741
    },
    {
      "epoch": 1.8989637305699483,
      "grad_norm": 0.9659252762794495,
      "learning_rate": 3.819606292448541e-08,
      "loss": 0.9676,
      "step": 742
    },
    {
      "epoch": 1.9015544041450778,
      "grad_norm": 1.0913461446762085,
      "learning_rate": 3.569508251851816e-08,
      "loss": 1.0477,
      "step": 743
    },
    {
      "epoch": 1.9041450777202074,
      "grad_norm": 1.0511360168457031,
      "learning_rate": 3.327848368931907e-08,
      "loss": 0.9345,
      "step": 744
    },
    {
      "epoch": 1.906735751295337,
      "grad_norm": 0.958997905254364,
      "learning_rate": 3.0946307513418095e-08,
      "loss": 1.0417,
      "step": 745
    },
    {
      "epoch": 1.9093264248704664,
      "grad_norm": 0.8062818050384521,
      "learning_rate": 2.8698593632357496e-08,
      "loss": 0.9683,
      "step": 746
    },
    {
      "epoch": 1.911917098445596,
      "grad_norm": 0.8307934403419495,
      "learning_rate": 2.653538025201685e-08,
      "loss": 0.9277,
      "step": 747
    },
    {
      "epoch": 1.9145077720207255,
      "grad_norm": 0.9197670221328735,
      "learning_rate": 2.4456704141967437e-08,
      "loss": 0.9539,
      "step": 748
    },
    {
      "epoch": 1.917098445595855,
      "grad_norm": 0.885750949382782,
      "learning_rate": 2.2462600634841624e-08,
      "loss": 0.9429,
      "step": 749
    },
    {
      "epoch": 1.9196891191709846,
      "grad_norm": 0.8606031537055969,
      "learning_rate": 2.0553103625737813e-08,
      "loss": 1.0091,
      "step": 750
    },
    {
      "epoch": 1.922279792746114,
      "grad_norm": 0.8774232268333435,
      "learning_rate": 1.8728245571640324e-08,
      "loss": 0.958,
      "step": 751
    },
    {
      "epoch": 1.9248704663212435,
      "grad_norm": 0.9953173398971558,
      "learning_rate": 1.6988057490868736e-08,
      "loss": 1.0015,
      "step": 752
    },
    {
      "epoch": 1.927461139896373,
      "grad_norm": 1.022383213043213,
      "learning_rate": 1.5332568962553306e-08,
      "loss": 0.9427,
      "step": 753
    },
    {
      "epoch": 1.9300518134715026,
      "grad_norm": 1.0359011888504028,
      "learning_rate": 1.3761808126126486e-08,
      "loss": 0.9499,
      "step": 754
    },
    {
      "epoch": 1.932642487046632,
      "grad_norm": 0.8054367899894714,
      "learning_rate": 1.2275801680851629e-08,
      "loss": 1.0467,
      "step": 755
    },
    {
      "epoch": 1.9352331606217616,
      "grad_norm": 0.921036422252655,
      "learning_rate": 1.0874574885362809e-08,
      "loss": 0.9384,
      "step": 756
    },
    {
      "epoch": 1.9378238341968912,
      "grad_norm": 0.8518050909042358,
      "learning_rate": 9.558151557240158e-09,
      "loss": 0.9375,
      "step": 757
    },
    {
      "epoch": 1.9404145077720207,
      "grad_norm": 0.9647573232650757,
      "learning_rate": 8.32655407260241e-09,
      "loss": 1.0186,
      "step": 758
    },
    {
      "epoch": 1.9430051813471503,
      "grad_norm": 0.97606360912323,
      "learning_rate": 7.179803365726656e-09,
      "loss": 0.9963,
      "step": 759
    },
    {
      "epoch": 1.9455958549222798,
      "grad_norm": 1.1201107501983643,
      "learning_rate": 6.117918928693623e-09,
      "loss": 0.983,
      "step": 760
    },
    {
      "epoch": 1.9481865284974094,
      "grad_norm": 0.9616057276725769,
      "learning_rate": 5.140918811056827e-09,
      "loss": 0.9794,
      "step": 761
    },
    {
      "epoch": 1.950777202072539,
      "grad_norm": 0.8792617917060852,
      "learning_rate": 4.248819619533384e-09,
      "loss": 0.9646,
      "step": 762
    },
    {
      "epoch": 1.9533678756476682,
      "grad_norm": 0.8594529032707214,
      "learning_rate": 3.4416365177236675e-09,
      "loss": 1.0342,
      "step": 763
    },
    {
      "epoch": 1.9559585492227978,
      "grad_norm": 0.9270037412643433,
      "learning_rate": 2.7193832258537447e-09,
      "loss": 1.0157,
      "step": 764
    },
    {
      "epoch": 1.9585492227979273,
      "grad_norm": 0.9513561725616455,
      "learning_rate": 2.08207202054056e-09,
      "loss": 1.0843,
      "step": 765
    },
    {
      "epoch": 1.9611398963730569,
      "grad_norm": 0.9862794876098633,
      "learning_rate": 1.5297137345843261e-09,
      "loss": 0.9374,
      "step": 766
    },
    {
      "epoch": 1.9637305699481864,
      "grad_norm": 0.889864444732666,
      "learning_rate": 1.0623177567847809e-09,
      "loss": 1.014,
      "step": 767
    },
    {
      "epoch": 1.966321243523316,
      "grad_norm": 0.850560188293457,
      "learning_rate": 6.798920317807601e-10,
      "loss": 1.0113,
      "step": 768
    },
    {
      "epoch": 1.9689119170984455,
      "grad_norm": 1.0084043741226196,
      "learning_rate": 3.8244305991530504e-10,
      "loss": 0.9921,
      "step": 769
    },
    {
      "epoch": 1.971502590673575,
      "grad_norm": 0.9245330691337585,
      "learning_rate": 1.6997589712575145e-10,
      "loss": 1.0003,
      "step": 770
    },
    {
      "epoch": 1.9740932642487046,
      "grad_norm": 0.9151455163955688,
      "learning_rate": 4.249415485657604e-11,
      "loss": 1.0092,
      "step": 771
    },
    {
      "epoch": 1.9766839378238341,
      "grad_norm": 0.8460503220558167,
      "learning_rate": 0.0,
      "loss": 0.9456,
      "step": 772
    }
  ],
  "logging_steps": 1,
  "max_steps": 772,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 193,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.408535403456889e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}