{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0007405448294102,
  "eval_steps": 296,
  "global_step": 1182,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008463369478973816,
      "grad_norm": 74.46723937988281,
      "learning_rate": 2e-05,
      "loss": 1.686,
      "step": 1
    },
    {
      "epoch": 0.0016926738957947632,
      "grad_norm": 87.14585876464844,
      "learning_rate": 4e-05,
      "loss": 1.9316,
      "step": 2
    },
    {
      "epoch": 0.0025390108436921448,
      "grad_norm": 125.84043884277344,
      "learning_rate": 6e-05,
      "loss": 2.756,
      "step": 3
    },
    {
      "epoch": 0.0033853477915895265,
      "grad_norm": 158.6361541748047,
      "learning_rate": 8e-05,
      "loss": 3.2544,
      "step": 4
    },
    {
      "epoch": 0.004231684739486909,
      "grad_norm": 193.18316650390625,
      "learning_rate": 0.0001,
      "loss": 3.7347,
      "step": 5
    },
    {
      "epoch": 0.0050780216873842895,
      "grad_norm": 168.4342803955078,
      "learning_rate": 0.00012,
      "loss": 3.0451,
      "step": 6
    },
    {
      "epoch": 0.005924358635281671,
      "grad_norm": 135.61651611328125,
      "learning_rate": 0.00014,
      "loss": 2.0432,
      "step": 7
    },
    {
      "epoch": 0.006770695583179053,
      "grad_norm": 133.69590759277344,
      "learning_rate": 0.00016,
      "loss": 1.9678,
      "step": 8
    },
    {
      "epoch": 0.007617032531076435,
      "grad_norm": 138.0289764404297,
      "learning_rate": 0.00018,
      "loss": 1.931,
      "step": 9
    },
    {
      "epoch": 0.008463369478973817,
      "grad_norm": 161.1634063720703,
      "learning_rate": 0.0002,
      "loss": 2.0495,
      "step": 10
    },
    {
      "epoch": 0.009309706426871197,
      "grad_norm": 136.35235595703125,
      "learning_rate": 0.00019999964073553233,
      "loss": 1.733,
      "step": 11
    },
    {
      "epoch": 0.010156043374768579,
      "grad_norm": 103.60092163085938,
      "learning_rate": 0.0001999985629447107,
      "loss": 1.4056,
      "step": 12
    },
    {
      "epoch": 0.01100238032266596,
      "grad_norm": 83.1572494506836,
      "learning_rate": 0.0001999967666352794,
      "loss": 1.2848,
      "step": 13
    },
    {
      "epoch": 0.011848717270563342,
      "grad_norm": 94.966552734375,
      "learning_rate": 0.0001999942518201454,
      "loss": 1.237,
      "step": 14
    },
    {
      "epoch": 0.012695054218460724,
      "grad_norm": 98.76087951660156,
      "learning_rate": 0.0001999910185173784,
      "loss": 1.1532,
      "step": 15
    },
    {
      "epoch": 0.013541391166358106,
      "grad_norm": 80.15461730957031,
      "learning_rate": 0.00019998706675021052,
      "loss": 0.8821,
      "step": 16
    },
    {
      "epoch": 0.014387728114255488,
      "grad_norm": 79.4908218383789,
      "learning_rate": 0.00019998239654703647,
      "loss": 0.8409,
      "step": 17
    },
    {
      "epoch": 0.01523406506215287,
      "grad_norm": 60.96306228637695,
      "learning_rate": 0.000199977007941413,
      "loss": 0.5869,
      "step": 18
    },
    {
      "epoch": 0.016080402010050253,
      "grad_norm": 45.81327819824219,
      "learning_rate": 0.00019997090097205873,
      "loss": 0.4739,
      "step": 19
    },
    {
      "epoch": 0.016926738957947635,
      "grad_norm": 62.55532455444336,
      "learning_rate": 0.000199964075682854,
      "loss": 0.4032,
      "step": 20
    },
    {
      "epoch": 0.017773075905845013,
      "grad_norm": 105.0770034790039,
      "learning_rate": 0.00019995653212284063,
      "loss": 0.378,
      "step": 21
    },
    {
      "epoch": 0.018619412853742395,
      "grad_norm": 99.4818344116211,
      "learning_rate": 0.0001999482703462211,
      "loss": 0.3709,
      "step": 22
    },
    {
      "epoch": 0.019465749801639776,
      "grad_norm": 102.58175659179688,
      "learning_rate": 0.00019993929041235874,
      "loss": 0.3393,
      "step": 23
    },
    {
      "epoch": 0.020312086749537158,
      "grad_norm": 84.75938415527344,
      "learning_rate": 0.00019992959238577705,
      "loss": 0.2595,
      "step": 24
    },
    {
      "epoch": 0.02115842369743454,
      "grad_norm": 35.45841979980469,
      "learning_rate": 0.000199919176336159,
      "loss": 0.2684,
      "step": 25
    },
    {
      "epoch": 0.02200476064533192,
      "grad_norm": 43.51015090942383,
      "learning_rate": 0.00019990804233834705,
      "loss": 0.1653,
      "step": 26
    },
    {
      "epoch": 0.022851097593229303,
      "grad_norm": 57.4715576171875,
      "learning_rate": 0.00019989619047234216,
      "loss": 0.1699,
      "step": 27
    },
    {
      "epoch": 0.023697434541126685,
      "grad_norm": 45.49979782104492,
      "learning_rate": 0.0001998836208233034,
      "loss": 0.1342,
      "step": 28
    },
    {
      "epoch": 0.024543771489024067,
      "grad_norm": 48.7567138671875,
      "learning_rate": 0.0001998703334815473,
      "loss": 0.2004,
      "step": 29
    },
    {
      "epoch": 0.02539010843692145,
      "grad_norm": 22.59009552001953,
      "learning_rate": 0.00019985632854254735,
      "loss": 0.1187,
      "step": 30
    },
    {
      "epoch": 0.02623644538481883,
      "grad_norm": 42.264198303222656,
      "learning_rate": 0.00019984160610693303,
      "loss": 0.1624,
      "step": 31
    },
    {
      "epoch": 0.027082782332716212,
      "grad_norm": 66.26699829101562,
      "learning_rate": 0.0001998261662804893,
      "loss": 0.1597,
      "step": 32
    },
    {
      "epoch": 0.027929119280613594,
      "grad_norm": 51.72114944458008,
      "learning_rate": 0.00019981000917415577,
      "loss": 0.1137,
      "step": 33
    },
    {
      "epoch": 0.028775456228510975,
      "grad_norm": 31.176406860351562,
      "learning_rate": 0.00019979313490402597,
      "loss": 0.0776,
      "step": 34
    },
    {
      "epoch": 0.029621793176408357,
      "grad_norm": 19.646127700805664,
      "learning_rate": 0.00019977554359134638,
      "loss": 0.1053,
      "step": 35
    },
    {
      "epoch": 0.03046813012430574,
      "grad_norm": 32.98014831542969,
      "learning_rate": 0.00019975723536251567,
      "loss": 0.0493,
      "step": 36
    },
    {
      "epoch": 0.031314467072203124,
      "grad_norm": 24.1268310546875,
      "learning_rate": 0.00019973821034908375,
      "loss": 0.0692,
      "step": 37
    },
    {
      "epoch": 0.032160804020100506,
      "grad_norm": 15.55493450164795,
      "learning_rate": 0.0001997184686877509,
      "loss": 0.0543,
      "step": 38
    },
    {
      "epoch": 0.03300714096799789,
      "grad_norm": 18.887033462524414,
      "learning_rate": 0.0001996980105203666,
      "loss": 0.0275,
      "step": 39
    },
    {
      "epoch": 0.03385347791589527,
      "grad_norm": 10.44262981414795,
      "learning_rate": 0.00019967683599392876,
      "loss": 0.0162,
      "step": 40
    },
    {
      "epoch": 0.034699814863792644,
      "grad_norm": 8.52265739440918,
      "learning_rate": 0.00019965494526058246,
      "loss": 0.0245,
      "step": 41
    },
    {
      "epoch": 0.035546151811690026,
      "grad_norm": 11.946599960327148,
      "learning_rate": 0.00019963233847761894,
      "loss": 0.028,
      "step": 42
    },
    {
      "epoch": 0.03639248875958741,
      "grad_norm": 6.178277015686035,
      "learning_rate": 0.00019960901580747448,
      "loss": 0.013,
      "step": 43
    },
    {
      "epoch": 0.03723882570748479,
      "grad_norm": 5.187213897705078,
      "learning_rate": 0.00019958497741772924,
      "loss": 0.0104,
      "step": 44
    },
    {
      "epoch": 0.03808516265538217,
      "grad_norm": 3.2645328044891357,
      "learning_rate": 0.00019956022348110596,
      "loss": 0.0089,
      "step": 45
    },
    {
      "epoch": 0.03893149960327955,
      "grad_norm": 1.7075039148330688,
      "learning_rate": 0.0001995347541754689,
      "loss": 0.0052,
      "step": 46
    },
    {
      "epoch": 0.039777836551176934,
      "grad_norm": 3.5826594829559326,
      "learning_rate": 0.00019950856968382232,
      "loss": 0.0092,
      "step": 47
    },
    {
      "epoch": 0.040624173499074316,
      "grad_norm": 4.57949686050415,
      "learning_rate": 0.00019948167019430941,
      "loss": 0.0084,
      "step": 48
    },
    {
      "epoch": 0.0414705104469717,
      "grad_norm": 13.784544944763184,
      "learning_rate": 0.0001994540559002108,
      "loss": 0.0308,
      "step": 49
    },
    {
      "epoch": 0.04231684739486908,
      "grad_norm": 34.07741165161133,
      "learning_rate": 0.0001994257269999431,
      "loss": 0.0469,
      "step": 50
    },
    {
      "epoch": 0.04316318434276646,
      "grad_norm": 31.998857498168945,
      "learning_rate": 0.00019939668369705778,
      "loss": 0.4136,
      "step": 51
    },
    {
      "epoch": 0.04400952129066384,
      "grad_norm": 44.679222106933594,
      "learning_rate": 0.00019936692620023928,
      "loss": 0.4861,
      "step": 52
    },
    {
      "epoch": 0.044855858238561225,
      "grad_norm": 59.955745697021484,
      "learning_rate": 0.00019933645472330387,
      "loss": 0.7331,
      "step": 53
    },
    {
      "epoch": 0.045702195186458607,
      "grad_norm": 62.51538848876953,
      "learning_rate": 0.00019930526948519793,
      "loss": 0.5997,
      "step": 54
    },
    {
      "epoch": 0.04654853213435599,
      "grad_norm": 71.37804412841797,
      "learning_rate": 0.0001992733707099964,
      "loss": 0.6939,
      "step": 55
    },
    {
      "epoch": 0.04739486908225337,
      "grad_norm": 84.57373046875,
      "learning_rate": 0.00019924075862690118,
      "loss": 0.1067,
      "step": 56
    },
    {
      "epoch": 0.04824120603015075,
      "grad_norm": 84.29151916503906,
      "learning_rate": 0.00019920743347023957,
      "loss": 0.1359,
      "step": 57
    },
    {
      "epoch": 0.04908754297804813,
      "grad_norm": 67.71774291992188,
      "learning_rate": 0.00019917339547946246,
      "loss": 0.143,
      "step": 58
    },
    {
      "epoch": 0.049933879925945515,
      "grad_norm": 48.20841979980469,
      "learning_rate": 0.00019913864489914268,
      "loss": 0.101,
      "step": 59
    },
    {
      "epoch": 0.0507802168738429,
      "grad_norm": 51.68072509765625,
      "learning_rate": 0.00019910318197897319,
      "loss": 0.0844,
      "step": 60
    },
    {
      "epoch": 0.05162655382174028,
      "grad_norm": 46.60186767578125,
      "learning_rate": 0.00019906700697376527,
      "loss": 0.0648,
      "step": 61
    },
    {
      "epoch": 0.05247289076963766,
      "grad_norm": 27.53974723815918,
      "learning_rate": 0.00019903012014344686,
      "loss": 0.0393,
      "step": 62
    },
    {
      "epoch": 0.05331922771753504,
      "grad_norm": 6.218672752380371,
      "learning_rate": 0.0001989925217530605,
      "loss": 0.0125,
      "step": 63
    },
    {
      "epoch": 0.054165564665432424,
      "grad_norm": 6.7888360023498535,
      "learning_rate": 0.00019895421207276155,
      "loss": 0.0133,
      "step": 64
    },
    {
      "epoch": 0.055011901613329806,
      "grad_norm": 7.5940961837768555,
      "learning_rate": 0.00019891519137781608,
      "loss": 0.0106,
      "step": 65
    },
    {
      "epoch": 0.05585823856122719,
      "grad_norm": 5.461939811706543,
      "learning_rate": 0.0001988754599485991,
      "loss": 0.0087,
      "step": 66
    },
    {
      "epoch": 0.05670457550912457,
      "grad_norm": 6.9425740242004395,
      "learning_rate": 0.0001988350180705924,
      "loss": 0.0174,
      "step": 67
    },
    {
      "epoch": 0.05755091245702195,
      "grad_norm": 6.4617533683776855,
      "learning_rate": 0.00019879386603438261,
      "loss": 0.0103,
      "step": 68
    },
    {
      "epoch": 0.05839724940491933,
      "grad_norm": 8.377887725830078,
      "learning_rate": 0.00019875200413565903,
      "loss": 0.0095,
      "step": 69
    },
    {
      "epoch": 0.059243586352816714,
      "grad_norm": 4.740405559539795,
      "learning_rate": 0.00019870943267521145,
      "loss": 0.005,
      "step": 70
    },
    {
      "epoch": 0.060089923300714096,
      "grad_norm": 1.870723843574524,
      "learning_rate": 0.0001986661519589282,
      "loss": 0.0049,
      "step": 71
    },
    {
      "epoch": 0.06093626024861148,
      "grad_norm": 2.44498872756958,
      "learning_rate": 0.0001986221622977937,
      "loss": 0.0043,
      "step": 72
    },
    {
      "epoch": 0.06178259719650886,
      "grad_norm": 6.047572612762451,
      "learning_rate": 0.00019857746400788644,
      "loss": 0.01,
      "step": 73
    },
    {
      "epoch": 0.06262893414440625,
      "grad_norm": 8.071754455566406,
      "learning_rate": 0.00019853205741037652,
      "loss": 0.0172,
      "step": 74
    },
    {
      "epoch": 0.06347527109230362,
      "grad_norm": 8.4547700881958,
      "learning_rate": 0.00019848594283152353,
      "loss": 0.0116,
      "step": 75
    },
    {
      "epoch": 0.06432160804020101,
      "grad_norm": 2.2582836151123047,
      "learning_rate": 0.00019843912060267402,
      "loss": 0.0038,
      "step": 76
    },
    {
      "epoch": 0.06516794498809839,
      "grad_norm": 2.5930304527282715,
      "learning_rate": 0.0001983915910602593,
      "loss": 0.0048,
      "step": 77
    },
    {
      "epoch": 0.06601428193599577,
      "grad_norm": 7.182027339935303,
      "learning_rate": 0.0001983433545457928,
      "loss": 0.0151,
      "step": 78
    },
    {
      "epoch": 0.06686061888389315,
      "grad_norm": 4.962230682373047,
      "learning_rate": 0.00019829441140586793,
      "loss": 0.008,
      "step": 79
    },
    {
      "epoch": 0.06770695583179054,
      "grad_norm": 3.9201576709747314,
      "learning_rate": 0.00019824476199215525,
      "loss": 0.0054,
      "step": 80
    },
    {
      "epoch": 0.06855329277968791,
      "grad_norm": 9.251102447509766,
      "learning_rate": 0.0001981944066614002,
      "loss": 0.0057,
      "step": 81
    },
    {
      "epoch": 0.06939962972758529,
      "grad_norm": 0.8075349926948547,
      "learning_rate": 0.00019814334577542038,
      "loss": 0.0018,
      "step": 82
    },
    {
      "epoch": 0.07024596667548268,
      "grad_norm": 5.284769535064697,
      "learning_rate": 0.00019809157970110302,
      "loss": 0.0031,
      "step": 83
    },
    {
      "epoch": 0.07109230362338005,
      "grad_norm": 7.295186996459961,
      "learning_rate": 0.0001980391088104024,
      "loss": 0.015,
      "step": 84
    },
    {
      "epoch": 0.07193864057127744,
      "grad_norm": 3.265753984451294,
      "learning_rate": 0.00019798593348033697,
      "loss": 0.0034,
      "step": 85
    },
    {
      "epoch": 0.07278497751917481,
      "grad_norm": 0.6690754890441895,
      "learning_rate": 0.00019793205409298693,
      "loss": 0.0017,
      "step": 86
    },
    {
      "epoch": 0.0736313144670722,
      "grad_norm": 2.377958059310913,
      "learning_rate": 0.00019787747103549124,
      "loss": 0.0022,
      "step": 87
    },
    {
      "epoch": 0.07447765141496958,
      "grad_norm": 3.079671621322632,
      "learning_rate": 0.00019782218470004498,
      "loss": 0.0049,
      "step": 88
    },
    {
      "epoch": 0.07532398836286697,
      "grad_norm": 0.35262423753738403,
      "learning_rate": 0.00019776619548389642,
      "loss": 0.001,
      "step": 89
    },
    {
      "epoch": 0.07617032531076434,
      "grad_norm": 0.8201667666435242,
      "learning_rate": 0.00019770950378934435,
      "loss": 0.0019,
      "step": 90
    },
    {
      "epoch": 0.07701666225866173,
      "grad_norm": 0.3472432792186737,
      "learning_rate": 0.00019765211002373493,
      "loss": 0.0012,
      "step": 91
    },
    {
      "epoch": 0.0778629992065591,
      "grad_norm": 1.4362425804138184,
      "learning_rate": 0.00019759401459945898,
      "loss": 0.0019,
      "step": 92
    },
    {
      "epoch": 0.0787093361544565,
      "grad_norm": 1.478244662284851,
      "learning_rate": 0.000197535217933949,
      "loss": 0.0024,
      "step": 93
    },
    {
      "epoch": 0.07955567310235387,
      "grad_norm": 4.533934116363525,
      "learning_rate": 0.000197475720449676,
      "loss": 0.0036,
      "step": 94
    },
    {
      "epoch": 0.08040201005025126,
      "grad_norm": 0.8504026532173157,
      "learning_rate": 0.00019741552257414663,
      "loss": 0.0015,
      "step": 95
    },
    {
      "epoch": 0.08124834699814863,
      "grad_norm": 2.9103963375091553,
      "learning_rate": 0.00019735462473989998,
      "loss": 0.0046,
      "step": 96
    },
    {
      "epoch": 0.08209468394604602,
      "grad_norm": 2.422717571258545,
      "learning_rate": 0.0001972930273845047,
      "loss": 0.0032,
      "step": 97
    },
    {
      "epoch": 0.0829410208939434,
      "grad_norm": 1.085134744644165,
      "learning_rate": 0.00019723073095055557,
      "loss": 0.0018,
      "step": 98
    },
    {
      "epoch": 0.08378735784184078,
      "grad_norm": 2.692281484603882,
      "learning_rate": 0.00019716773588567053,
      "loss": 0.0039,
      "step": 99
    },
    {
      "epoch": 0.08463369478973816,
      "grad_norm": 1.1898086071014404,
      "learning_rate": 0.0001971040426424873,
      "loss": 0.0016,
      "step": 100
    },
    {
      "epoch": 0.08548003173763555,
      "grad_norm": 23.142412185668945,
      "learning_rate": 0.0001970396516786603,
      "loss": 0.3362,
      "step": 101
    },
    {
      "epoch": 0.08632636868553292,
      "grad_norm": 26.89776039123535,
      "learning_rate": 0.0001969745634568572,
      "loss": 0.4391,
      "step": 102
    },
    {
      "epoch": 0.08717270563343031,
      "grad_norm": 27.5157470703125,
      "learning_rate": 0.00019690877844475574,
      "loss": 0.391,
      "step": 103
    },
    {
      "epoch": 0.08801904258132769,
      "grad_norm": 54.70764923095703,
      "learning_rate": 0.00019684229711504025,
      "loss": 0.637,
      "step": 104
    },
    {
      "epoch": 0.08886537952922507,
      "grad_norm": 93.04987335205078,
      "learning_rate": 0.0001967751199453983,
      "loss": 0.988,
      "step": 105
    },
    {
      "epoch": 0.08971171647712245,
      "grad_norm": 55.99843978881836,
      "learning_rate": 0.00019670724741851735,
      "loss": 0.1856,
      "step": 106
    },
    {
      "epoch": 0.09055805342501984,
      "grad_norm": 14.445696830749512,
      "learning_rate": 0.0001966386800220811,
      "loss": 0.0225,
      "step": 107
    },
    {
      "epoch": 0.09140439037291721,
      "grad_norm": 10.762388229370117,
      "learning_rate": 0.00019656941824876617,
      "loss": 0.0169,
      "step": 108
    },
    {
      "epoch": 0.0922507273208146,
      "grad_norm": 10.171407699584961,
      "learning_rate": 0.00019649946259623838,
      "loss": 0.0198,
      "step": 109
    },
    {
      "epoch": 0.09309706426871198,
      "grad_norm": 10.556538581848145,
      "learning_rate": 0.0001964288135671494,
      "loss": 0.0182,
      "step": 110
    },
    {
      "epoch": 0.09394340121660937,
      "grad_norm": 5.872833728790283,
      "learning_rate": 0.0001963574716691329,
      "loss": 0.0142,
      "step": 111
    },
    {
      "epoch": 0.09478973816450674,
      "grad_norm": 3.532670259475708,
      "learning_rate": 0.00019628543741480108,
      "loss": 0.0068,
      "step": 112
    },
    {
      "epoch": 0.09563607511240413,
      "grad_norm": 8.299254417419434,
      "learning_rate": 0.0001962127113217409,
      "loss": 0.0082,
      "step": 113
    },
    {
      "epoch": 0.0964824120603015,
      "grad_norm": 1.9848963022232056,
      "learning_rate": 0.0001961392939125104,
      "loss": 0.0038,
      "step": 114
    },
    {
      "epoch": 0.09732874900819889,
      "grad_norm": 1.5647845268249512,
      "learning_rate": 0.00019606518571463487,
      "loss": 0.0022,
      "step": 115
    },
    {
      "epoch": 0.09817508595609627,
      "grad_norm": 0.4753362238407135,
      "learning_rate": 0.0001959903872606032,
      "loss": 0.0015,
      "step": 116
    },
    {
      "epoch": 0.09902142290399366,
      "grad_norm": 0.9765241146087646,
      "learning_rate": 0.00019591489908786386,
      "loss": 0.0021,
      "step": 117
    },
    {
      "epoch": 0.09986775985189103,
      "grad_norm": 1.2018309831619263,
      "learning_rate": 0.00019583872173882129,
      "loss": 0.0025,
      "step": 118
    },
    {
      "epoch": 0.10071409679978842,
      "grad_norm": 0.5523298382759094,
      "learning_rate": 0.00019576185576083174,
      "loss": 0.0014,
      "step": 119
    },
    {
      "epoch": 0.1015604337476858,
      "grad_norm": 0.46725499629974365,
      "learning_rate": 0.00019568430170619953,
      "loss": 0.0013,
      "step": 120
    },
    {
      "epoch": 0.10240677069558318,
      "grad_norm": 1.9901517629623413,
      "learning_rate": 0.00019560606013217297,
      "loss": 0.0024,
      "step": 121
    },
    {
      "epoch": 0.10325310764348056,
      "grad_norm": 0.9264715313911438,
      "learning_rate": 0.00019552713160094038,
      "loss": 0.0019,
      "step": 122
    },
    {
      "epoch": 0.10409944459137795,
      "grad_norm": 0.8873511552810669,
      "learning_rate": 0.00019544751667962613,
      "loss": 0.0016,
      "step": 123
    },
    {
      "epoch": 0.10494578153927532,
      "grad_norm": 3.455312967300415,
      "learning_rate": 0.0001953672159402865,
      "loss": 0.0025,
      "step": 124
    },
    {
      "epoch": 0.10579211848717271,
      "grad_norm": 0.6087542772293091,
      "learning_rate": 0.00019528622995990545,
      "loss": 0.0015,
      "step": 125
    },
    {
      "epoch": 0.10663845543507008,
      "grad_norm": 9.587512016296387,
      "learning_rate": 0.0001952045593203907,
      "loss": 0.0056,
      "step": 126
    },
    {
      "epoch": 0.10748479238296747,
      "grad_norm": 2.0214486122131348,
      "learning_rate": 0.0001951222046085695,
      "loss": 0.0016,
      "step": 127
    },
    {
      "epoch": 0.10833112933086485,
      "grad_norm": 0.48625630140304565,
      "learning_rate": 0.00019503916641618426,
      "loss": 0.0013,
      "step": 128
    },
    {
      "epoch": 0.10917746627876224,
      "grad_norm": 0.3964328169822693,
      "learning_rate": 0.00019495544533988833,
      "loss": 0.0009,
      "step": 129
    },
    {
      "epoch": 0.11002380322665961,
      "grad_norm": 0.8259016871452332,
      "learning_rate": 0.000194871041981242,
      "loss": 0.0014,
      "step": 130
    },
    {
      "epoch": 0.110870140174557,
      "grad_norm": 1.184308648109436,
      "learning_rate": 0.00019478595694670775,
      "loss": 0.0022,
      "step": 131
    },
    {
      "epoch": 0.11171647712245437,
      "grad_norm": 0.7664253115653992,
      "learning_rate": 0.00019470019084764613,
      "loss": 0.0014,
      "step": 132
    },
    {
      "epoch": 0.11256281407035176,
      "grad_norm": 4.297515869140625,
      "learning_rate": 0.0001946137443003115,
      "loss": 0.0076,
      "step": 133
    },
    {
      "epoch": 0.11340915101824914,
      "grad_norm": 0.5345470309257507,
      "learning_rate": 0.0001945266179258472,
      "loss": 0.0008,
      "step": 134
    },
    {
      "epoch": 0.11425548796614653,
      "grad_norm": 0.8852401971817017,
      "learning_rate": 0.0001944388123502815,
      "loss": 0.0008,
      "step": 135
    },
    {
      "epoch": 0.1151018249140439,
      "grad_norm": 4.631274700164795,
      "learning_rate": 0.0001943503282045229,
      "loss": 0.0152,
      "step": 136
    },
    {
      "epoch": 0.11594816186194129,
      "grad_norm": 0.5090851783752441,
      "learning_rate": 0.0001942611661243555,
      "loss": 0.0011,
      "step": 137
    },
    {
      "epoch": 0.11679449880983866,
      "grad_norm": 0.993028461933136,
      "learning_rate": 0.00019417132675043471,
      "loss": 0.0013,
      "step": 138
    },
    {
      "epoch": 0.11764083575773605,
      "grad_norm": 6.0296101570129395,
      "learning_rate": 0.0001940808107282824,
      "loss": 0.0059,
      "step": 139
    },
    {
      "epoch": 0.11848717270563343,
      "grad_norm": 0.35436925292015076,
      "learning_rate": 0.00019398961870828238,
      "loss": 0.0009,
      "step": 140
    },
    {
      "epoch": 0.11933350965353082,
      "grad_norm": 0.3110350966453552,
      "learning_rate": 0.0001938977513456757,
      "loss": 0.0008,
      "step": 141
    },
    {
      "epoch": 0.12017984660142819,
      "grad_norm": 0.6243657469749451,
      "learning_rate": 0.00019380520930055602,
      "loss": 0.001,
      "step": 142
    },
    {
      "epoch": 0.12102618354932558,
      "grad_norm": 0.31351810693740845,
      "learning_rate": 0.0001937119932378646,
      "loss": 0.0008,
      "step": 143
    },
    {
      "epoch": 0.12187252049722296,
      "grad_norm": 1.0445241928100586,
      "learning_rate": 0.00019361810382738585,
      "loss": 0.0011,
      "step": 144
    },
    {
      "epoch": 0.12271885744512034,
      "grad_norm": 0.5057194232940674,
      "learning_rate": 0.0001935235417437424,
      "loss": 0.001,
      "step": 145
    },
    {
      "epoch": 0.12356519439301772,
      "grad_norm": 0.16824191808700562,
      "learning_rate": 0.00019342830766639013,
      "loss": 0.0006,
      "step": 146
    },
    {
      "epoch": 0.12441153134091511,
      "grad_norm": 0.15476548671722412,
      "learning_rate": 0.00019333240227961345,
      "loss": 0.0005,
      "step": 147
    },
    {
      "epoch": 0.1252578682888125,
      "grad_norm": 0.8760647177696228,
      "learning_rate": 0.00019323582627252036,
      "loss": 0.001,
      "step": 148
    },
    {
      "epoch": 0.12610420523670987,
      "grad_norm": 0.5927007794380188,
      "learning_rate": 0.00019313858033903736,
      "loss": 0.0011,
      "step": 149
    },
    {
      "epoch": 0.12695054218460725,
      "grad_norm": 3.346593141555786,
      "learning_rate": 0.00019304066517790465,
      "loss": 0.0136,
      "step": 150
    },
    {
      "epoch": 0.12779687913250462,
      "grad_norm": 21.881933212280273,
      "learning_rate": 0.00019294208149267102,
      "loss": 0.3441,
      "step": 151
    },
    {
      "epoch": 0.12864321608040202,
      "grad_norm": 27.779556274414062,
      "learning_rate": 0.0001928428299916887,
      "loss": 0.4038,
      "step": 152
    },
    {
      "epoch": 0.1294895530282994,
      "grad_norm": 41.42377471923828,
      "learning_rate": 0.00019274291138810852,
      "loss": 0.4076,
      "step": 153
    },
    {
      "epoch": 0.13033588997619677,
      "grad_norm": 65.22334289550781,
      "learning_rate": 0.0001926423263998745,
      "loss": 0.4623,
      "step": 154
    },
    {
      "epoch": 0.13118222692409415,
      "grad_norm": 92.79143524169922,
      "learning_rate": 0.0001925410757497189,
      "loss": 0.5263,
      "step": 155
    },
    {
      "epoch": 0.13202856387199155,
      "grad_norm": 13.545234680175781,
      "learning_rate": 0.00019243916016515697,
      "loss": 0.0362,
      "step": 156
    },
    {
      "epoch": 0.13287490081988892,
      "grad_norm": 2.329796075820923,
      "learning_rate": 0.00019233658037848167,
      "loss": 0.0055,
      "step": 157
    },
    {
      "epoch": 0.1337212377677863,
      "grad_norm": 2.7054457664489746,
      "learning_rate": 0.00019223333712675838,
      "loss": 0.0072,
      "step": 158
    },
    {
      "epoch": 0.13456757471568367,
      "grad_norm": 5.102527618408203,
      "learning_rate": 0.00019212943115181982,
      "loss": 0.0095,
      "step": 159
    },
    {
      "epoch": 0.13541391166358108,
      "grad_norm": 2.725212812423706,
      "learning_rate": 0.00019202486320026046,
      "loss": 0.0049,
      "step": 160
    },
    {
      "epoch": 0.13626024861147845,
      "grad_norm": 3.888731002807617,
      "learning_rate": 0.00019191963402343128,
      "loss": 0.0064,
      "step": 161
    },
    {
      "epoch": 0.13710658555937583,
      "grad_norm": 0.5881321430206299,
      "learning_rate": 0.00019181374437743438,
      "loss": 0.0018,
      "step": 162
    },
    {
      "epoch": 0.1379529225072732,
      "grad_norm": 0.47292956709861755,
      "learning_rate": 0.00019170719502311747,
      "loss": 0.0013,
      "step": 163
    },
    {
      "epoch": 0.13879925945517058,
      "grad_norm": 0.24725215137004852,
      "learning_rate": 0.00019159998672606853,
      "loss": 0.0006,
      "step": 164
    },
    {
      "epoch": 0.13964559640306798,
      "grad_norm": 0.3073093891143799,
      "learning_rate": 0.00019149212025661017,
      "loss": 0.0008,
      "step": 165
    },
    {
      "epoch": 0.14049193335096535,
      "grad_norm": 0.21486443281173706,
      "learning_rate": 0.0001913835963897942,
      "loss": 0.0006,
      "step": 166
    },
    {
      "epoch": 0.14133827029886273,
      "grad_norm": 0.3804973363876343,
      "learning_rate": 0.00019127441590539602,
      "loss": 0.0007,
      "step": 167
    },
    {
      "epoch": 0.1421846072467601,
      "grad_norm": 0.3088591694831848,
      "learning_rate": 0.00019116457958790893,
      "loss": 0.0008,
      "step": 168
    },
    {
      "epoch": 0.1430309441946575,
      "grad_norm": 0.2884930670261383,
      "learning_rate": 0.00019105408822653875,
      "loss": 0.0007,
      "step": 169
    },
    {
      "epoch": 0.14387728114255488,
      "grad_norm": 2.3264033794403076,
      "learning_rate": 0.00019094294261519785,
      "loss": 0.0016,
      "step": 170
    },
    {
      "epoch": 0.14472361809045226,
      "grad_norm": 0.7139752507209778,
      "learning_rate": 0.00019083114355249956,
      "loss": 0.0012,
      "step": 171
    },
    {
      "epoch": 0.14556995503834963,
      "grad_norm": 4.276805400848389,
      "learning_rate": 0.00019071869184175255,
      "loss": 0.003,
      "step": 172
    },
    {
      "epoch": 0.14641629198624703,
      "grad_norm": 0.36595144867897034,
      "learning_rate": 0.00019060558829095487,
      "loss": 0.0009,
      "step": 173
    },
    {
      "epoch": 0.1472626289341444,
      "grad_norm": 3.2072956562042236,
      "learning_rate": 0.00019049183371278828,
      "loss": 0.0024,
      "step": 174
    },
    {
      "epoch": 0.14810896588204178,
      "grad_norm": 0.3659462332725525,
      "learning_rate": 0.0001903774289246123,
      "loss": 0.0008,
      "step": 175
    },
    {
      "epoch": 0.14895530282993916,
      "grad_norm": 2.719895362854004,
      "learning_rate": 0.00019026237474845852,
      "loss": 0.0015,
      "step": 176
    },
    {
      "epoch": 0.14980163977783656,
      "grad_norm": 2.4224469661712646,
      "learning_rate": 0.0001901466720110244,
      "loss": 0.0032,
      "step": 177
    },
    {
      "epoch": 0.15064797672573393,
      "grad_norm": 0.13008993864059448,
      "learning_rate": 0.0001900303215436676,
      "loss": 0.0004,
      "step": 178
    },
    {
      "epoch": 0.1514943136736313,
      "grad_norm": 4.346362113952637,
      "learning_rate": 0.00018991332418239994,
      "loss": 0.0014,
      "step": 179
    },
    {
      "epoch": 0.15234065062152868,
      "grad_norm": 1.0230494737625122,
      "learning_rate": 0.0001897956807678813,
      "loss": 0.001,
      "step": 180
    },
    {
      "epoch": 0.1531869875694261,
      "grad_norm": 7.090376377105713,
      "learning_rate": 0.00018967739214541363,
      "loss": 0.003,
      "step": 181
    },
    {
      "epoch": 0.15403332451732346,
      "grad_norm": 0.2335856556892395,
      "learning_rate": 0.0001895584591649349,
      "loss": 0.0006,
      "step": 182
    },
    {
      "epoch": 0.15487966146522084,
      "grad_norm": 4.240998268127441,
      "learning_rate": 0.00018943888268101305,
      "loss": 0.0033,
      "step": 183
    },
    {
      "epoch": 0.1557259984131182,
      "grad_norm": 0.1109471321105957,
      "learning_rate": 0.00018931866355283963,
      "loss": 0.0004,
      "step": 184
    },
    {
      "epoch": 0.1565723353610156,
      "grad_norm": 2.551114559173584,
      "learning_rate": 0.00018919780264422395,
      "loss": 0.0018,
      "step": 185
    },
    {
      "epoch": 0.157418672308913,
      "grad_norm": 2.3841819763183594,
      "learning_rate": 0.00018907630082358657,
      "loss": 0.0011,
      "step": 186
    },
    {
      "epoch": 0.15826500925681036,
      "grad_norm": 0.3431253135204315,
      "learning_rate": 0.0001889541589639532,
      "loss": 0.0006,
      "step": 187
    },
    {
      "epoch": 0.15911134620470774,
      "grad_norm": 0.7966493964195251,
      "learning_rate": 0.00018883137794294848,
      "loss": 0.0012,
      "step": 188
    },
    {
      "epoch": 0.15995768315260514,
      "grad_norm": 0.26179561018943787,
      "learning_rate": 0.00018870795864278956,
      "loss": 0.0007,
      "step": 189
    },
    {
      "epoch": 0.16080402010050251,
      "grad_norm": 0.40261805057525635,
      "learning_rate": 0.00018858390195027985,
      "loss": 0.0007,
      "step": 190
    },
    {
      "epoch": 0.1616503570483999,
      "grad_norm": 0.38073089718818665,
      "learning_rate": 0.00018845920875680254,
      "loss": 0.0007,
      "step": 191
    },
    {
      "epoch": 0.16249669399629726,
      "grad_norm": 3.7871763706207275,
      "learning_rate": 0.00018833387995831436,
      "loss": 0.0012,
      "step": 192
    },
    {
      "epoch": 0.16334303094419467,
      "grad_norm": 0.08530396223068237,
      "learning_rate": 0.0001882079164553389,
      "loss": 0.0003,
      "step": 193
    },
    {
      "epoch": 0.16418936789209204,
      "grad_norm": 0.15287600457668304,
      "learning_rate": 0.00018808131915296045,
      "loss": 0.0004,
      "step": 194
    },
    {
      "epoch": 0.16503570483998942,
      "grad_norm": 0.49411728978157043,
      "learning_rate": 0.00018795408896081728,
      "loss": 0.0007,
      "step": 195
    },
    {
      "epoch": 0.1658820417878868,
      "grad_norm": 0.15277883410453796,
      "learning_rate": 0.00018782622679309506,
      "loss": 0.0004,
      "step": 196
    },
    {
      "epoch": 0.1667283787357842,
      "grad_norm": 0.11514627188444138,
      "learning_rate": 0.0001876977335685205,
      "loss": 0.0004,
      "step": 197
    },
    {
      "epoch": 0.16757471568368157,
      "grad_norm": 0.26564082503318787,
      "learning_rate": 0.00018756861021035462,
      "loss": 0.0005,
      "step": 198
    },
    {
      "epoch": 0.16842105263157894,
      "grad_norm": 0.21045762300491333,
      "learning_rate": 0.0001874388576463861,
      "loss": 0.0006,
      "step": 199
    },
    {
      "epoch": 0.16926738957947632,
      "grad_norm": 0.3532085120677948,
      "learning_rate": 0.0001873084768089246,
      "loss": 0.0008,
      "step": 200
    },
    {
      "epoch": 0.17011372652737372,
      "grad_norm": 29.670534133911133,
      "learning_rate": 0.00018717746863479423,
      "loss": 0.2364,
      "step": 201
    },
    {
      "epoch": 0.1709600634752711,
      "grad_norm": 35.507972717285156,
      "learning_rate": 0.00018704583406532662,
      "loss": 0.2491,
      "step": 202
    },
    {
      "epoch": 0.17180640042316847,
      "grad_norm": 39.08292770385742,
      "learning_rate": 0.00018691357404635424,
      "loss": 0.2042,
      "step": 203
    },
    {
      "epoch": 0.17265273737106585,
      "grad_norm": 40.570125579833984,
      "learning_rate": 0.00018678068952820355,
      "loss": 0.1984,
      "step": 204
    },
    {
      "epoch": 0.17349907431896325,
      "grad_norm": 65.52487182617188,
      "learning_rate": 0.0001866471814656883,
      "loss": 0.2995,
      "step": 205
    },
    {
      "epoch": 0.17434541126686062,
      "grad_norm": 28.42360496520996,
      "learning_rate": 0.0001865130508181026,
      "loss": 0.0856,
      "step": 206
    },
    {
      "epoch": 0.175191748214758,
      "grad_norm": 0.7669479846954346,
      "learning_rate": 0.00018637829854921387,
      "loss": 0.0011,
      "step": 207
    },
    {
      "epoch": 0.17603808516265537,
      "grad_norm": 11.103653907775879,
      "learning_rate": 0.0001862429256272562,
      "loss": 0.0107,
      "step": 208
    },
    {
      "epoch": 0.17688442211055277,
      "grad_norm": 12.320133209228516,
      "learning_rate": 0.0001861069330249232,
      "loss": 0.0146,
      "step": 209
    },
    {
      "epoch": 0.17773075905845015,
      "grad_norm": 1.1529252529144287,
      "learning_rate": 0.00018597032171936104,
      "loss": 0.0016,
      "step": 210
    },
    {
      "epoch": 0.17857709600634752,
      "grad_norm": 0.7125049829483032,
      "learning_rate": 0.00018583309269216156,
      "loss": 0.0014,
      "step": 211
    },
    {
      "epoch": 0.1794234329542449,
      "grad_norm": 15.23595905303955,
      "learning_rate": 0.00018569524692935496,
      "loss": 0.0056,
      "step": 212
    },
    {
      "epoch": 0.1802697699021423,
      "grad_norm": 0.48333868384361267,
      "learning_rate": 0.0001855567854214029,
      "loss": 0.0008,
      "step": 213
    },
    {
      "epoch": 0.18111610685003968,
      "grad_norm": 0.19277918338775635,
      "learning_rate": 0.0001854177091631915,
      "loss": 0.0006,
      "step": 214
    },
    {
      "epoch": 0.18196244379793705,
      "grad_norm": 0.9802500009536743,
      "learning_rate": 0.0001852780191540238,
      "loss": 0.0013,
      "step": 215
    },
    {
      "epoch": 0.18280878074583443,
      "grad_norm": 0.16036200523376465,
      "learning_rate": 0.00018513771639761298,
      "loss": 0.0005,
      "step": 216
    },
    {
      "epoch": 0.18365511769373183,
      "grad_norm": 0.32250264286994934,
      "learning_rate": 0.00018499680190207495,
      "loss": 0.0009,
      "step": 217
    },
    {
      "epoch": 0.1845014546416292,
      "grad_norm": 0.777240514755249,
      "learning_rate": 0.00018485527667992115,
      "loss": 0.001,
      "step": 218
    },
    {
      "epoch": 0.18534779158952658,
      "grad_norm": 0.3555632531642914,
      "learning_rate": 0.0001847131417480512,
      "loss": 0.0009,
      "step": 219
    },
    {
      "epoch": 0.18619412853742395,
      "grad_norm": 0.5618864297866821,
      "learning_rate": 0.00018457039812774575,
      "loss": 0.0006,
      "step": 220
    },
    {
      "epoch": 0.18704046548532136,
      "grad_norm": 0.43013933300971985,
      "learning_rate": 0.00018442704684465899,
      "loss": 0.0008,
      "step": 221
    },
    {
      "epoch": 0.18788680243321873,
      "grad_norm": 0.2362668365240097,
      "learning_rate": 0.0001842830889288114,
      "loss": 0.0006,
      "step": 222
    },
    {
      "epoch": 0.1887331393811161,
      "grad_norm": 0.21618768572807312,
      "learning_rate": 0.00018413852541458226,
      "loss": 0.0006,
      "step": 223
    },
    {
      "epoch": 0.18957947632901348,
      "grad_norm": 4.4554901123046875,
      "learning_rate": 0.0001839933573407022,
      "loss": 0.0033,
      "step": 224
    },
    {
      "epoch": 0.19042581327691088,
      "grad_norm": 0.2934055030345917,
      "learning_rate": 0.00018384758575024586,
      "loss": 0.0006,
      "step": 225
    },
    {
      "epoch": 0.19127215022480826,
      "grad_norm": 0.13466045260429382,
      "learning_rate": 0.00018370121169062435,
      "loss": 0.0004,
      "step": 226
    },
    {
      "epoch": 0.19211848717270563,
      "grad_norm": 0.11464786529541016,
      "learning_rate": 0.00018355423621357759,
      "loss": 0.0004,
      "step": 227
    },
    {
      "epoch": 0.192964824120603,
      "grad_norm": 0.6053487062454224,
      "learning_rate": 0.00018340666037516693,
      "loss": 0.0007,
      "step": 228
    },
    {
      "epoch": 0.1938111610685004,
      "grad_norm": 0.09634432196617126,
      "learning_rate": 0.00018325848523576742,
      "loss": 0.0003,
      "step": 229
    },
    {
      "epoch": 0.19465749801639778,
      "grad_norm": 3.126617908477783,
      "learning_rate": 0.00018310971186006038,
      "loss": 0.0012,
      "step": 230
    },
    {
      "epoch": 0.19550383496429516,
      "grad_norm": 0.1292370706796646,
      "learning_rate": 0.00018296034131702553,
      "loss": 0.0004,
      "step": 231
    },
    {
      "epoch": 0.19635017191219253,
      "grad_norm": 0.18858526647090912,
      "learning_rate": 0.00018281037467993345,
      "loss": 0.0004,
      "step": 232
    },
    {
      "epoch": 0.19719650886008994,
      "grad_norm": 3.9159443378448486,
      "learning_rate": 0.00018265981302633782,
      "loss": 0.0101,
      "step": 233
    },
    {
      "epoch": 0.1980428458079873,
      "grad_norm": 0.46279361844062805,
      "learning_rate": 0.00018250865743806767,
      "loss": 0.0006,
      "step": 234
    },
    {
      "epoch": 0.19888918275588469,
      "grad_norm": 0.1234947219491005,
      "learning_rate": 0.00018235690900121965,
      "loss": 0.0004,
      "step": 235
    },
    {
      "epoch": 0.19973551970378206,
      "grad_norm": 0.6481117606163025,
      "learning_rate": 0.0001822045688061502,
      "loss": 0.001,
      "step": 236
    },
    {
      "epoch": 0.20058185665167946,
      "grad_norm": 0.09670120477676392,
      "learning_rate": 0.00018205163794746772,
      "loss": 0.0003,
      "step": 237
    },
    {
      "epoch": 0.20142819359957684,
      "grad_norm": 0.19068674743175507,
      "learning_rate": 0.00018189811752402458,
      "loss": 0.0003,
      "step": 238
    },
    {
      "epoch": 0.2022745305474742,
      "grad_norm": 1.4126213788986206,
      "learning_rate": 0.0001817440086389096,
      "loss": 0.0013,
      "step": 239
    },
    {
      "epoch": 0.2031208674953716,
      "grad_norm": 0.12282940745353699,
      "learning_rate": 0.00018158931239943958,
      "loss": 0.0003,
      "step": 240
    },
    {
      "epoch": 0.203967204443269,
      "grad_norm": 0.08084618300199509,
      "learning_rate": 0.00018143402991715183,
      "loss": 0.0003,
      "step": 241
    },
    {
      "epoch": 0.20481354139116636,
      "grad_norm": 0.13239997625350952,
      "learning_rate": 0.0001812781623077959,
      "loss": 0.0003,
      "step": 242
    },
    {
      "epoch": 0.20565987833906374,
      "grad_norm": 0.6071043014526367,
      "learning_rate": 0.00018112171069132566,
      "loss": 0.0009,
      "step": 243
    },
    {
      "epoch": 0.20650621528696111,
      "grad_norm": 0.39527714252471924,
      "learning_rate": 0.00018096467619189123,
      "loss": 0.0006,
      "step": 244
    },
    {
      "epoch": 0.2073525522348585,
      "grad_norm": 0.17460699379444122,
      "learning_rate": 0.00018080705993783096,
      "loss": 0.0004,
      "step": 245
    },
    {
      "epoch": 0.2081988891827559,
      "grad_norm": 0.20895503461360931,
      "learning_rate": 0.00018064886306166323,
      "loss": 0.0004,
      "step": 246
    },
    {
      "epoch": 0.20904522613065327,
      "grad_norm": 0.5428494811058044,
      "learning_rate": 0.00018049008670007835,
      "loss": 0.0004,
      "step": 247
    },
    {
      "epoch": 0.20989156307855064,
      "grad_norm": 0.139656662940979,
      "learning_rate": 0.00018033073199393048,
      "loss": 0.0003,
      "step": 248
    },
    {
      "epoch": 0.21073790002644802,
      "grad_norm": 0.12410479784011841,
      "learning_rate": 0.0001801708000882292,
      "loss": 0.0003,
      "step": 249
    },
    {
      "epoch": 0.21158423697434542,
      "grad_norm": 1.5398989915847778,
      "learning_rate": 0.00018001029213213162,
      "loss": 0.0008,
      "step": 250
    },
    {
      "epoch": 0.2124305739222428,
      "grad_norm": 83.7887191772461,
      "learning_rate": 0.00017984920927893377,
      "loss": 0.0822,
      "step": 251
    },
    {
      "epoch": 0.21327691087014017,
      "grad_norm": 26.819786071777344,
      "learning_rate": 0.00017968755268606262,
      "loss": 0.0551,
      "step": 252
    },
    {
      "epoch": 0.21412324781803754,
      "grad_norm": 17.227445602416992,
      "learning_rate": 0.00017952532351506753,
      "loss": 0.0476,
      "step": 253
    },
    {
      "epoch": 0.21496958476593495,
      "grad_norm": 12.969980239868164,
      "learning_rate": 0.00017936252293161204,
      "loss": 0.0395,
      "step": 254
    },
    {
      "epoch": 0.21581592171383232,
      "grad_norm": 28.53566551208496,
      "learning_rate": 0.00017919915210546546,
      "loss": 0.0698,
      "step": 255
    },
    {
      "epoch": 0.2166622586617297,
      "grad_norm": 0.27628636360168457,
      "learning_rate": 0.00017903521221049448,
      "loss": 0.0007,
      "step": 256
    },
    {
      "epoch": 0.21750859560962707,
      "grad_norm": 0.11199865490198135,
      "learning_rate": 0.00017887070442465462,
      "loss": 0.0004,
      "step": 257
    },
    {
      "epoch": 0.21835493255752447,
      "grad_norm": 0.18436667323112488,
      "learning_rate": 0.00017870562992998193,
      "loss": 0.0005,
      "step": 258
    },
    {
      "epoch": 0.21920126950542185,
      "grad_norm": 0.3930487632751465,
      "learning_rate": 0.00017853998991258449,
      "loss": 0.0007,
      "step": 259
    },
    {
      "epoch": 0.22004760645331922,
      "grad_norm": 0.37658360600471497,
      "learning_rate": 0.00017837378556263367,
      "loss": 0.0006,
      "step": 260
    },
    {
      "epoch": 0.2208939434012166,
      "grad_norm": 0.11100467294454575,
      "learning_rate": 0.00017820701807435584,
      "loss": 0.0003,
      "step": 261
    },
    {
      "epoch": 0.221740280349114,
      "grad_norm": 0.43430548906326294,
      "learning_rate": 0.0001780396886460237,
      "loss": 0.0005,
      "step": 262
    },
    {
      "epoch": 0.22258661729701137,
      "grad_norm": 0.370032399892807,
      "learning_rate": 0.00017787179847994757,
      "loss": 0.0007,
      "step": 263
    },
    {
      "epoch": 0.22343295424490875,
      "grad_norm": 0.8565439581871033,
      "learning_rate": 0.00017770334878246686,
      "loss": 0.0012,
      "step": 264
    },
    {
      "epoch": 0.22427929119280612,
      "grad_norm": 0.31265321373939514,
      "learning_rate": 0.00017753434076394142,
      "loss": 0.0005,
      "step": 265
    },
    {
      "epoch": 0.22512562814070353,
      "grad_norm": 0.4533150792121887,
      "learning_rate": 0.00017736477563874275,
      "loss": 0.0006,
      "step": 266
    },
    {
      "epoch": 0.2259719650886009,
      "grad_norm": 0.3379235565662384,
      "learning_rate": 0.00017719465462524533,
      "loss": 0.0008,
      "step": 267
    },
    {
      "epoch": 0.22681830203649828,
      "grad_norm": 0.5092006921768188,
      "learning_rate": 0.00017702397894581788,
      "loss": 0.0005,
      "step": 268
    },
    {
      "epoch": 0.22766463898439565,
      "grad_norm": 0.4077492654323578,
      "learning_rate": 0.00017685274982681454,
      "loss": 0.0007,
      "step": 269
    },
    {
      "epoch": 0.22851097593229305,
      "grad_norm": 0.5492392778396606,
      "learning_rate": 0.0001766809684985661,
      "loss": 0.0004,
      "step": 270
    },
    {
      "epoch": 0.22935731288019043,
      "grad_norm": 0.5803564786911011,
      "learning_rate": 0.00017650863619537107,
      "loss": 0.0007,
      "step": 271
    },
    {
      "epoch": 0.2302036498280878,
      "grad_norm": 0.15004323422908783,
      "learning_rate": 0.00017633575415548693,
      "loss": 0.0003,
      "step": 272
    },
    {
      "epoch": 0.23104998677598518,
      "grad_norm": 0.2035101354122162,
      "learning_rate": 0.00017616232362112116,
      "loss": 0.0004,
      "step": 273
    },
    {
      "epoch": 0.23189632372388258,
      "grad_norm": 0.08870142698287964,
      "learning_rate": 0.00017598834583842235,
      "loss": 0.0003,
      "step": 274
    },
    {
      "epoch": 0.23274266067177996,
      "grad_norm": 0.09822133928537369,
      "learning_rate": 0.00017581382205747117,
      "loss": 0.0003,
      "step": 275
    },
    {
      "epoch": 0.23358899761967733,
      "grad_norm": 0.10656624287366867,
      "learning_rate": 0.00017563875353227153,
      "loss": 0.0003,
      "step": 276
    },
    {
      "epoch": 0.2344353345675747,
      "grad_norm": 0.09961220622062683,
      "learning_rate": 0.00017546314152074138,
      "loss": 0.0002,
      "step": 277
    },
    {
      "epoch": 0.2352816715154721,
      "grad_norm": 0.9810962080955505,
      "learning_rate": 0.00017528698728470392,
      "loss": 0.0007,
      "step": 278
    },
    {
      "epoch": 0.23612800846336948,
      "grad_norm": 0.19340015947818756,
      "learning_rate": 0.00017511029208987824,
      "loss": 0.0004,
      "step": 279
    },
    {
      "epoch": 0.23697434541126686,
      "grad_norm": 0.1273375153541565,
      "learning_rate": 0.00017493305720587047,
      "loss": 0.0004,
      "step": 280
    },
    {
      "epoch": 0.23782068235916423,
      "grad_norm": 0.12967373430728912,
      "learning_rate": 0.00017475528390616452,
      "loss": 0.0003,
      "step": 281
    },
    {
      "epoch": 0.23866701930706163,
      "grad_norm": 0.6166614294052124,
      "learning_rate": 0.000174576973468113,
      "loss": 0.0008,
      "step": 282
    },
    {
      "epoch": 0.239513356254959,
      "grad_norm": 0.6157065629959106,
      "learning_rate": 0.00017439812717292798,
      "loss": 0.0008,
      "step": 283
    },
    {
      "epoch": 0.24035969320285638,
      "grad_norm": 0.10683294385671616,
      "learning_rate": 0.00017421874630567188,
      "loss": 0.0003,
      "step": 284
    },
    {
      "epoch": 0.24120603015075376,
      "grad_norm": 1.4939329624176025,
      "learning_rate": 0.00017403883215524811,
      "loss": 0.0008,
      "step": 285
    },
    {
      "epoch": 0.24205236709865116,
      "grad_norm": 1.0055180788040161,
      "learning_rate": 0.0001738583860143919,
      "loss": 0.0009,
      "step": 286
    },
    {
      "epoch": 0.24289870404654854,
      "grad_norm": 0.12554600834846497,
      "learning_rate": 0.000173677409179661,
      "loss": 0.0003,
      "step": 287
    },
    {
      "epoch": 0.2437450409944459,
      "grad_norm": 0.23707199096679688,
      "learning_rate": 0.00017349590295142635,
      "loss": 0.0003,
      "step": 288
    },
    {
      "epoch": 0.24459137794234329,
      "grad_norm": 3.2323226928710938,
      "learning_rate": 0.00017331386863386262,
      "loss": 0.0031,
      "step": 289
    },
    {
      "epoch": 0.2454377148902407,
      "grad_norm": 0.4041217863559723,
      "learning_rate": 0.00017313130753493917,
      "loss": 0.0006,
      "step": 290
    },
    {
      "epoch": 0.24628405183813806,
      "grad_norm": 0.4350222647190094,
      "learning_rate": 0.00017294822096641032,
      "loss": 0.0006,
      "step": 291
    },
    {
      "epoch": 0.24713038878603544,
      "grad_norm": 0.3703482449054718,
      "learning_rate": 0.00017276461024380596,
      "loss": 0.0006,
      "step": 292
    },
    {
      "epoch": 0.2479767257339328,
      "grad_norm": 0.16974994540214539,
      "learning_rate": 0.00017258047668642225,
      "loss": 0.0003,
      "step": 293
    },
    {
      "epoch": 0.24882306268183021,
      "grad_norm": 0.06141228228807449,
      "learning_rate": 0.00017239582161731218,
      "loss": 0.0002,
      "step": 294
    },
    {
      "epoch": 0.2496693996297276,
      "grad_norm": 0.09231980890035629,
      "learning_rate": 0.0001722106463632758,
      "loss": 0.0003,
      "step": 295
    },
    {
      "epoch": 0.250515736577625,
      "grad_norm": 0.1844186633825302,
      "learning_rate": 0.00017202495225485088,
      "loss": 0.0005,
      "step": 296
    },
    {
      "epoch": 0.250515736577625,
      "eval_loss": 0.0016514281742274761,
      "eval_runtime": 22.4693,
      "eval_samples_per_second": 88.565,
      "eval_steps_per_second": 44.283,
      "step": 296
    },
    {
      "epoch": 0.25136207352552237,
      "grad_norm": 0.1142737865447998,
      "learning_rate": 0.00017183874062630335,
      "loss": 0.0003,
      "step": 297
    },
    {
      "epoch": 0.25220841047341974,
      "grad_norm": 0.23479783535003662,
      "learning_rate": 0.0001716520128156176,
      "loss": 0.0004,
      "step": 298
    },
    {
      "epoch": 0.2530547474213171,
      "grad_norm": 0.10498982667922974,
      "learning_rate": 0.000171464770164487,
      "loss": 0.0003,
      "step": 299
    },
    {
      "epoch": 0.2539010843692145,
      "grad_norm": 0.06947942078113556,
      "learning_rate": 0.00017127701401830424,
      "loss": 0.0002,
      "step": 300
    },
    {
      "epoch": 0.25474742131711187,
      "grad_norm": 14.572744369506836,
      "learning_rate": 0.00017108874572615144,
      "loss": 0.0409,
      "step": 301
    },
    {
      "epoch": 0.25559375826500924,
      "grad_norm": 8.615899085998535,
      "learning_rate": 0.00017089996664079084,
      "loss": 0.0149,
      "step": 302
    },
    {
      "epoch": 0.2564400952129066,
      "grad_norm": 5.315328598022461,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.0116,
      "step": 303
    },
    {
      "epoch": 0.25728643216080405,
      "grad_norm": 0.7601366639137268,
      "learning_rate": 0.00017052088151983604,
      "loss": 0.0007,
      "step": 304
    },
    {
      "epoch": 0.2581327691087014,
      "grad_norm": 0.2309204488992691,
      "learning_rate": 0.0001703305782080781,
      "loss": 0.0005,
      "step": 305
    },
    {
      "epoch": 0.2589791060565988,
      "grad_norm": 1.6397736072540283,
      "learning_rate": 0.00017013976955076535,
      "loss": 0.0012,
      "step": 306
    },
    {
      "epoch": 0.25982544300449617,
      "grad_norm": 0.20109611749649048,
      "learning_rate": 0.00016994845691891314,
      "loss": 0.0005,
      "step": 307
    },
    {
      "epoch": 0.26067177995239355,
      "grad_norm": 0.30731967091560364,
      "learning_rate": 0.00016975664168715818,
      "loss": 0.0005,
      "step": 308
    },
    {
      "epoch": 0.2615181169002909,
      "grad_norm": 0.13147425651550293,
      "learning_rate": 0.00016956432523374836,
      "loss": 0.0004,
      "step": 309
    },
    {
      "epoch": 0.2623644538481883,
      "grad_norm": 0.13589197397232056,
      "learning_rate": 0.00016937150894053303,
      "loss": 0.0004,
      "step": 310
    },
    {
      "epoch": 0.26321079079608567,
      "grad_norm": 0.06314504146575928,
      "learning_rate": 0.00016917819419295306,
      "loss": 0.0002,
      "step": 311
    },
    {
      "epoch": 0.2640571277439831,
      "grad_norm": 0.14383579790592194,
      "learning_rate": 0.00016898438238003086,
      "loss": 0.0004,
      "step": 312
    },
    {
      "epoch": 0.2649034646918805,
      "grad_norm": 0.13727350533008575,
      "learning_rate": 0.0001687900748943604,
      "loss": 0.0004,
      "step": 313
    },
    {
      "epoch": 0.26574980163977785,
      "grad_norm": 0.33883053064346313,
      "learning_rate": 0.00016859527313209722,
      "loss": 0.0007,
      "step": 314
    },
    {
      "epoch": 0.2665961385876752,
      "grad_norm": 1.281099796295166,
      "learning_rate": 0.00016839997849294825,
      "loss": 0.0011,
      "step": 315
    },
    {
      "epoch": 0.2674424755355726,
      "grad_norm": 0.15714304149150848,
      "learning_rate": 0.00016820419238016202,
      "loss": 0.0003,
      "step": 316
    },
    {
      "epoch": 0.26828881248347,
      "grad_norm": 0.21545763313770294,
      "learning_rate": 0.0001680079162005184,
      "loss": 0.0004,
      "step": 317
    },
    {
      "epoch": 0.26913514943136735,
      "grad_norm": 0.4178004562854767,
      "learning_rate": 0.00016781115136431856,
      "loss": 0.0006,
      "step": 318
    },
    {
      "epoch": 0.2699814863792647,
      "grad_norm": 0.17474912106990814,
      "learning_rate": 0.00016761389928537481,
      "loss": 0.0004,
      "step": 319
    },
    {
      "epoch": 0.27082782332716215,
      "grad_norm": 2.090693712234497,
      "learning_rate": 0.00016741616138100032,
      "loss": 0.0012,
      "step": 320
    },
    {
      "epoch": 0.27167416027505953,
      "grad_norm": 0.2829761803150177,
      "learning_rate": 0.00016721793907199922,
      "loss": 0.0006,
      "step": 321
    },
    {
      "epoch": 0.2725204972229569,
      "grad_norm": 8.031776428222656,
      "learning_rate": 0.00016701923378265615,
      "loss": 0.008,
      "step": 322
    },
    {
      "epoch": 0.2733668341708543,
      "grad_norm": 0.0836440697312355,
      "learning_rate": 0.00016682004694072608,
      "loss": 0.0002,
      "step": 323
    },
    {
      "epoch": 0.27421317111875165,
      "grad_norm": 0.09077680855989456,
      "learning_rate": 0.00016662037997742415,
      "loss": 0.0002,
      "step": 324
    },
    {
      "epoch": 0.275059508066649,
      "grad_norm": 0.15983553230762482,
      "learning_rate": 0.0001664202343274152,
      "loss": 0.0004,
      "step": 325
    },
    {
      "epoch": 0.2759058450145464,
      "grad_norm": 7.0335001945495605,
      "learning_rate": 0.0001662196114288037,
      "loss": 0.0046,
      "step": 326
    },
    {
      "epoch": 0.2767521819624438,
      "grad_norm": 0.21336182951927185,
      "learning_rate": 0.00016601851272312316,
      "loss": 0.0004,
      "step": 327
    },
    {
      "epoch": 0.27759851891034115,
      "grad_norm": 0.30044668912887573,
      "learning_rate": 0.0001658169396553261,
      "loss": 0.0004,
      "step": 328
    },
    {
      "epoch": 0.2784448558582386,
      "grad_norm": 0.19076892733573914,
      "learning_rate": 0.00016561489367377318,
      "loss": 0.0004,
      "step": 329
    },
    {
      "epoch": 0.27929119280613596,
      "grad_norm": 0.047912344336509705,
      "learning_rate": 0.0001654123762302233,
      "loss": 0.0002,
      "step": 330
    },
    {
      "epoch": 0.28013752975403333,
      "grad_norm": 0.2600057125091553,
      "learning_rate": 0.00016520938877982296,
      "loss": 0.0004,
      "step": 331
    },
    {
      "epoch": 0.2809838667019307,
      "grad_norm": 0.10947916656732559,
      "learning_rate": 0.00016500593278109563,
      "loss": 0.0003,
      "step": 332
    },
    {
      "epoch": 0.2818302036498281,
      "grad_norm": 0.1151491031050682,
      "learning_rate": 0.0001648020096959316,
      "loss": 0.0002,
      "step": 333
    },
    {
      "epoch": 0.28267654059772546,
      "grad_norm": 0.13691584765911102,
      "learning_rate": 0.0001645976209895772,
      "loss": 0.0003,
      "step": 334
    },
    {
      "epoch": 0.28352287754562283,
      "grad_norm": 6.682562828063965,
      "learning_rate": 0.0001643927681306244,
      "loss": 0.0059,
      "step": 335
    },
    {
      "epoch": 0.2843692144935202,
      "grad_norm": 1.8755964040756226,
      "learning_rate": 0.00016418745259100034,
      "loss": 0.0008,
      "step": 336
    },
    {
      "epoch": 0.28521555144141764,
      "grad_norm": 3.117988348007202,
      "learning_rate": 0.00016398167584595657,
      "loss": 0.0013,
      "step": 337
    },
    {
      "epoch": 0.286061888389315,
      "grad_norm": 0.6182642579078674,
      "learning_rate": 0.00016377543937405847,
      "loss": 0.0008,
      "step": 338
    },
    {
      "epoch": 0.2869082253372124,
      "grad_norm": 0.07497651875019073,
      "learning_rate": 0.00016356874465717485,
      "loss": 0.0002,
      "step": 339
    },
    {
      "epoch": 0.28775456228510976,
      "grad_norm": 0.0722075030207634,
      "learning_rate": 0.00016336159318046698,
      "loss": 0.0002,
      "step": 340
    },
    {
      "epoch": 0.28860089923300714,
      "grad_norm": 0.057715825736522675,
      "learning_rate": 0.00016315398643237825,
      "loss": 0.0002,
      "step": 341
    },
    {
      "epoch": 0.2894472361809045,
      "grad_norm": 0.07277728617191315,
      "learning_rate": 0.00016294592590462316,
      "loss": 0.0002,
      "step": 342
    },
    {
      "epoch": 0.2902935731288019,
      "grad_norm": 9.660849571228027,
      "learning_rate": 0.00016273741309217682,
      "loss": 0.0089,
      "step": 343
    },
    {
      "epoch": 0.29113991007669926,
      "grad_norm": 0.14255961775779724,
      "learning_rate": 0.00016252844949326412,
      "loss": 0.0003,
      "step": 344
    },
    {
      "epoch": 0.2919862470245967,
      "grad_norm": 0.08858984708786011,
      "learning_rate": 0.000162319036609349,
      "loss": 0.0003,
      "step": 345
    },
    {
      "epoch": 0.29283258397249406,
      "grad_norm": 4.9274115562438965,
      "learning_rate": 0.00016210917594512356,
      "loss": 0.0031,
      "step": 346
    },
    {
      "epoch": 0.29367892092039144,
      "grad_norm": 0.12270866334438324,
      "learning_rate": 0.0001618988690084975,
      "loss": 0.0003,
      "step": 347
    },
    {
      "epoch": 0.2945252578682888,
      "grad_norm": 0.07490206509828568,
      "learning_rate": 0.00016168811731058692,
      "loss": 0.0002,
      "step": 348
    },
    {
      "epoch": 0.2953715948161862,
      "grad_norm": 0.2741962671279907,
      "learning_rate": 0.00016147692236570378,
      "loss": 0.0003,
      "step": 349
    },
    {
      "epoch": 0.29621793176408356,
      "grad_norm": 1.1869347095489502,
      "learning_rate": 0.00016126528569134488,
      "loss": 0.0008,
      "step": 350
    },
    {
      "epoch": 0.29706426871198094,
      "grad_norm": 7.189521312713623,
      "learning_rate": 0.00016105320880818096,
      "loss": 0.0076,
      "step": 351
    },
    {
      "epoch": 0.2979106056598783,
      "grad_norm": 2.5508530139923096,
      "learning_rate": 0.0001608406932400458,
      "loss": 0.0014,
      "step": 352
    },
    {
      "epoch": 0.29875694260777574,
      "grad_norm": 8.627997398376465,
      "learning_rate": 0.00016062774051392524,
      "loss": 0.007,
      "step": 353
    },
    {
      "epoch": 0.2996032795556731,
      "grad_norm": 4.3146586418151855,
      "learning_rate": 0.00016041435215994622,
      "loss": 0.0037,
      "step": 354
    },
    {
      "epoch": 0.3004496165035705,
      "grad_norm": 7.567290782928467,
      "learning_rate": 0.00016020052971136583,
      "loss": 0.0049,
      "step": 355
    },
    {
      "epoch": 0.30129595345146787,
      "grad_norm": 0.22426241636276245,
      "learning_rate": 0.00015998627470456022,
      "loss": 0.0004,
      "step": 356
    },
    {
      "epoch": 0.30214229039936524,
      "grad_norm": 0.2981072962284088,
      "learning_rate": 0.00015977158867901364,
      "loss": 0.0005,
      "step": 357
    },
    {
      "epoch": 0.3029886273472626,
      "grad_norm": 0.11945097893476486,
      "learning_rate": 0.00015955647317730727,
      "loss": 0.0003,
      "step": 358
    },
    {
      "epoch": 0.30383496429516,
      "grad_norm": 0.08270949125289917,
      "learning_rate": 0.00015934092974510824,
      "loss": 0.0003,
      "step": 359
    },
    {
      "epoch": 0.30468130124305737,
      "grad_norm": 0.1772061437368393,
      "learning_rate": 0.00015912495993115848,
      "loss": 0.0004,
      "step": 360
    },
    {
      "epoch": 0.3055276381909548,
      "grad_norm": 0.7524343729019165,
      "learning_rate": 0.0001589085652872636,
      "loss": 0.0003,
      "step": 361
    },
    {
      "epoch": 0.3063739751388522,
      "grad_norm": 0.32617804408073425,
      "learning_rate": 0.00015869174736828168,
      "loss": 0.0004,
      "step": 362
    },
    {
      "epoch": 0.30722031208674955,
      "grad_norm": 0.5572852492332458,
      "learning_rate": 0.00015847450773211232,
      "loss": 0.0007,
      "step": 363
    },
    {
      "epoch": 0.3080666490346469,
      "grad_norm": 0.07258780300617218,
      "learning_rate": 0.00015825684793968508,
      "loss": 0.0002,
      "step": 364
    },
    {
      "epoch": 0.3089129859825443,
      "grad_norm": 0.31824979186058044,
      "learning_rate": 0.00015803876955494857,
      "loss": 0.0003,
      "step": 365
    },
    {
      "epoch": 0.30975932293044167,
      "grad_norm": 0.1371203064918518,
      "learning_rate": 0.00015782027414485905,
      "loss": 0.0002,
      "step": 366
    },
    {
      "epoch": 0.31060565987833905,
      "grad_norm": 0.09063910692930222,
      "learning_rate": 0.0001576013632793693,
      "loss": 0.0003,
      "step": 367
    },
    {
      "epoch": 0.3114519968262364,
      "grad_norm": 0.09730315208435059,
      "learning_rate": 0.00015738203853141725,
      "loss": 0.0002,
      "step": 368
    },
    {
      "epoch": 0.31229833377413385,
      "grad_norm": 0.08047539740800858,
      "learning_rate": 0.00015716230147691466,
      "loss": 0.0002,
      "step": 369
    },
    {
      "epoch": 0.3131446707220312,
      "grad_norm": 1.4543787240982056,
      "learning_rate": 0.00015694215369473584,
      "loss": 0.0006,
      "step": 370
    },
    {
      "epoch": 0.3139910076699286,
      "grad_norm": 0.1553686261177063,
      "learning_rate": 0.00015672159676670633,
      "loss": 0.0003,
      "step": 371
    },
    {
      "epoch": 0.314837344617826,
      "grad_norm": 0.09920419007539749,
      "learning_rate": 0.00015650063227759143,
      "loss": 0.0002,
      "step": 372
    },
    {
      "epoch": 0.31568368156572335,
      "grad_norm": 0.3528848886489868,
      "learning_rate": 0.00015627926181508494,
      "loss": 0.0005,
      "step": 373
    },
    {
      "epoch": 0.3165300185136207,
      "grad_norm": 0.11126668006181717,
      "learning_rate": 0.0001560574869697977,
      "loss": 0.0002,
      "step": 374
    },
    {
      "epoch": 0.3173763554615181,
      "grad_norm": 0.30505645275115967,
      "learning_rate": 0.00015583530933524617,
      "loss": 0.0004,
      "step": 375
    },
    {
      "epoch": 0.3182226924094155,
      "grad_norm": 0.08159814029932022,
      "learning_rate": 0.00015561273050784092,
      "loss": 0.0001,
      "step": 376
    },
    {
      "epoch": 0.3190690293573129,
      "grad_norm": 6.772775173187256,
      "learning_rate": 0.00015538975208687523,
      "loss": 0.0031,
      "step": 377
    },
    {
      "epoch": 0.3199153663052103,
      "grad_norm": 0.06802649796009064,
      "learning_rate": 0.00015516637567451356,
      "loss": 0.0002,
      "step": 378
    },
    {
      "epoch": 0.32076170325310766,
      "grad_norm": 0.15224559605121613,
      "learning_rate": 0.0001549426028757801,
      "loss": 0.0003,
      "step": 379
    },
    {
      "epoch": 0.32160804020100503,
      "grad_norm": 0.0895279198884964,
      "learning_rate": 0.00015471843529854716,
      "loss": 0.0002,
      "step": 380
    },
    {
      "epoch": 0.3224543771489024,
      "grad_norm": 0.24836435914039612,
      "learning_rate": 0.0001544938745535236,
      "loss": 0.0003,
      "step": 381
    },
    {
      "epoch": 0.3233007140967998,
      "grad_norm": 0.3028913140296936,
      "learning_rate": 0.00015426892225424337,
      "loss": 0.0003,
      "step": 382
    },
    {
      "epoch": 0.32414705104469715,
      "grad_norm": 0.22172853350639343,
      "learning_rate": 0.00015404358001705384,
      "loss": 0.0004,
      "step": 383
    },
    {
      "epoch": 0.32499338799259453,
      "grad_norm": 0.1651873141527176,
      "learning_rate": 0.00015381784946110418,
      "loss": 0.0003,
      "step": 384
    },
    {
      "epoch": 0.32583972494049196,
      "grad_norm": 0.10653572529554367,
      "learning_rate": 0.00015359173220833377,
      "loss": 0.0002,
      "step": 385
    },
    {
      "epoch": 0.32668606188838933,
      "grad_norm": 0.16223099827766418,
      "learning_rate": 0.00015336522988346047,
      "loss": 0.0004,
      "step": 386
    },
    {
      "epoch": 0.3275323988362867,
      "grad_norm": 0.06762652844190598,
      "learning_rate": 0.00015313834411396905,
      "loss": 0.0002,
      "step": 387
    },
    {
      "epoch": 0.3283787357841841,
      "grad_norm": 0.07580212503671646,
      "learning_rate": 0.00015291107653009938,
      "loss": 0.0002,
      "step": 388
    },
    {
      "epoch": 0.32922507273208146,
      "grad_norm": 0.13204818964004517,
      "learning_rate": 0.0001526834287648349,
      "loss": 0.0002,
      "step": 389
    },
    {
      "epoch": 0.33007140967997883,
      "grad_norm": 0.054843708872795105,
      "learning_rate": 0.00015245540245389052,
      "loss": 0.0002,
      "step": 390
    },
    {
      "epoch": 0.3309177466278762,
      "grad_norm": 0.3241541087627411,
      "learning_rate": 0.0001522269992357014,
      "loss": 0.0004,
      "step": 391
    },
    {
      "epoch": 0.3317640835757736,
      "grad_norm": 0.05158082768321037,
      "learning_rate": 0.00015199822075141064,
      "loss": 0.0002,
      "step": 392
    },
    {
      "epoch": 0.332610420523671,
      "grad_norm": 0.08429571986198425,
      "learning_rate": 0.00015176906864485798,
      "loss": 0.0002,
      "step": 393
    },
    {
      "epoch": 0.3334567574715684,
      "grad_norm": 0.12458796799182892,
      "learning_rate": 0.00015153954456256753,
      "loss": 0.0003,
      "step": 394
    },
    {
      "epoch": 0.33430309441946576,
      "grad_norm": 0.06644143909215927,
      "learning_rate": 0.00015130965015373624,
      "loss": 0.0002,
      "step": 395
    },
    {
      "epoch": 0.33514943136736314,
      "grad_norm": 0.341960608959198,
      "learning_rate": 0.00015107938707022197,
      "loss": 0.0005,
      "step": 396
    },
    {
      "epoch": 0.3359957683152605,
      "grad_norm": 0.2635876536369324,
      "learning_rate": 0.0001508487569665316,
      "loss": 0.0004,
      "step": 397
    },
    {
      "epoch": 0.3368421052631579,
      "grad_norm": 0.04443720355629921,
      "learning_rate": 0.00015061776149980914,
      "loss": 0.0002,
      "step": 398
    },
    {
      "epoch": 0.33768844221105526,
      "grad_norm": 0.08289075642824173,
      "learning_rate": 0.0001503864023298239,
      "loss": 0.0002,
      "step": 399
    },
    {
      "epoch": 0.33853477915895264,
      "grad_norm": 1.3839573860168457,
      "learning_rate": 0.0001501546811189584,
      "loss": 0.0012,
      "step": 400
    },
    {
      "epoch": 0.33938111610685007,
      "grad_norm": 3.257554531097412,
      "learning_rate": 0.00014992259953219665,
      "loss": 0.0018,
      "step": 401
    },
    {
      "epoch": 0.34022745305474744,
      "grad_norm": 2.25943660736084,
      "learning_rate": 0.00014969015923711195,
      "loss": 0.0032,
      "step": 402
    },
    {
      "epoch": 0.3410737900026448,
      "grad_norm": 0.6718599200248718,
      "learning_rate": 0.00014945736190385517,
      "loss": 0.0008,
      "step": 403
    },
    {
      "epoch": 0.3419201269505422,
      "grad_norm": 9.728017807006836,
      "learning_rate": 0.0001492242092051424,
      "loss": 0.0077,
      "step": 404
    },
    {
      "epoch": 0.34276646389843957,
      "grad_norm": 0.5138121247291565,
      "learning_rate": 0.00014899070281624327,
      "loss": 0.0005,
      "step": 405
    },
    {
      "epoch": 0.34361280084633694,
      "grad_norm": 0.17196395993232727,
      "learning_rate": 0.00014875684441496883,
      "loss": 0.0004,
      "step": 406
    },
    {
      "epoch": 0.3444591377942343,
      "grad_norm": 0.5897896885871887,
      "learning_rate": 0.00014852263568165927,
      "loss": 0.0007,
      "step": 407
    },
    {
      "epoch": 0.3453054747421317,
      "grad_norm": 0.1742909848690033,
      "learning_rate": 0.00014828807829917213,
      "loss": 0.0003,
      "step": 408
    },
    {
      "epoch": 0.34615181169002907,
      "grad_norm": 0.1630605012178421,
      "learning_rate": 0.00014805317395287008,
      "loss": 0.0003,
      "step": 409
    },
    {
      "epoch": 0.3469981486379265,
      "grad_norm": 0.2970294654369354,
      "learning_rate": 0.00014781792433060884,
      "loss": 0.0004,
      "step": 410
    },
    {
      "epoch": 0.34784448558582387,
      "grad_norm": 0.11012287437915802,
      "learning_rate": 0.00014758233112272502,
      "loss": 0.0003,
      "step": 411
    },
    {
      "epoch": 0.34869082253372125,
      "grad_norm": 0.5044625401496887,
      "learning_rate": 0.00014734639602202393,
      "loss": 0.0003,
      "step": 412
    },
    {
      "epoch": 0.3495371594816186,
      "grad_norm": 0.11342803388834,
      "learning_rate": 0.00014711012072376756,
      "loss": 0.0003,
      "step": 413
    },
    {
      "epoch": 0.350383496429516,
      "grad_norm": 0.06610164046287537,
      "learning_rate": 0.00014687350692566236,
      "loss": 0.0002,
      "step": 414
    },
    {
      "epoch": 0.35122983337741337,
      "grad_norm": 0.11933105438947678,
      "learning_rate": 0.00014663655632784684,
      "loss": 0.0003,
      "step": 415
    },
    {
      "epoch": 0.35207617032531074,
      "grad_norm": 0.1549493819475174,
      "learning_rate": 0.00014639927063287965,
      "loss": 0.0004,
      "step": 416
    },
    {
      "epoch": 0.3529225072732081,
      "grad_norm": 0.43954983353614807,
      "learning_rate": 0.00014616165154572719,
      "loss": 0.0004,
      "step": 417
    },
    {
      "epoch": 0.35376884422110555,
      "grad_norm": 0.07722916454076767,
      "learning_rate": 0.00014592370077375132,
      "loss": 0.0002,
      "step": 418
    },
    {
      "epoch": 0.3546151811690029,
      "grad_norm": 0.12190598994493484,
      "learning_rate": 0.00014568542002669723,
      "loss": 0.0003,
      "step": 419
    },
    {
      "epoch": 0.3554615181169003,
      "grad_norm": 0.0552247054874897,
      "learning_rate": 0.00014544681101668098,
      "loss": 0.0002,
      "step": 420
    },
    {
      "epoch": 0.3563078550647977,
      "grad_norm": 1.4518511295318604,
      "learning_rate": 0.0001452078754581774,
      "loss": 0.0006,
      "step": 421
    },
    {
      "epoch": 0.35715419201269505,
      "grad_norm": 0.12335781008005142,
      "learning_rate": 0.00014496861506800758,
      "loss": 0.0003,
      "step": 422
    },
    {
      "epoch": 0.3580005289605924,
      "grad_norm": 0.09529973566532135,
      "learning_rate": 0.00014472903156532668,
      "loss": 0.0002,
      "step": 423
    },
    {
      "epoch": 0.3588468659084898,
      "grad_norm": 0.1234167143702507,
      "learning_rate": 0.00014448912667161146,
      "loss": 0.0003,
      "step": 424
    },
    {
      "epoch": 0.3596932028563872,
      "grad_norm": 0.05241147801280022,
      "learning_rate": 0.00014424890211064804,
      "loss": 0.0002,
      "step": 425
    },
    {
      "epoch": 0.3605395398042846,
      "grad_norm": 0.433606892824173,
      "learning_rate": 0.00014400835960851936,
      "loss": 0.0005,
      "step": 426
    },
    {
      "epoch": 0.361385876752182,
      "grad_norm": 1.4340794086456299,
      "learning_rate": 0.0001437675008935929,
      "loss": 0.0013,
      "step": 427
    },
    {
      "epoch": 0.36223221370007935,
      "grad_norm": 0.21237996220588684,
      "learning_rate": 0.00014352632769650827,
      "loss": 0.0003,
      "step": 428
    },
    {
      "epoch": 0.3630785506479767,
      "grad_norm": 0.08412837982177734,
      "learning_rate": 0.00014328484175016464,
      "loss": 0.0002,
      "step": 429
    },
    {
      "epoch": 0.3639248875958741,
      "grad_norm": 4.765340328216553,
      "learning_rate": 0.00014304304478970838,
      "loss": 0.0021,
      "step": 430
    },
    {
      "epoch": 0.3647712245437715,
      "grad_norm": 0.15136684477329254,
      "learning_rate": 0.00014280093855252067,
      "loss": 0.0003,
      "step": 431
    },
    {
      "epoch": 0.36561756149166885,
      "grad_norm": 0.0624224916100502,
      "learning_rate": 0.00014255852477820487,
      "loss": 0.0002,
      "step": 432
    },
    {
      "epoch": 0.3664638984395662,
      "grad_norm": 0.16323184967041016,
      "learning_rate": 0.0001423158052085741,
      "loss": 0.0003,
      "step": 433
    },
    {
      "epoch": 0.36731023538746366,
      "grad_norm": 0.0723976343870163,
      "learning_rate": 0.0001420727815876386,
      "loss": 0.0002,
      "step": 434
    },
    {
      "epoch": 0.36815657233536103,
      "grad_norm": 0.14013762772083282,
      "learning_rate": 0.00014182945566159353,
      "loss": 0.0003,
      "step": 435
    },
    {
      "epoch": 0.3690029092832584,
      "grad_norm": 0.1103127971291542,
      "learning_rate": 0.00014158582917880603,
      "loss": 0.0002,
      "step": 436
    },
    {
      "epoch": 0.3698492462311558,
      "grad_norm": 0.0755336657166481,
      "learning_rate": 0.00014134190388980286,
      "loss": 0.0002,
      "step": 437
    },
    {
      "epoch": 0.37069558317905316,
      "grad_norm": 0.1029498502612114,
      "learning_rate": 0.00014109768154725783,
      "loss": 0.0002,
      "step": 438
    },
    {
      "epoch": 0.37154192012695053,
      "grad_norm": 0.10077208280563354,
      "learning_rate": 0.00014085316390597912,
      "loss": 0.0002,
      "step": 439
    },
    {
      "epoch": 0.3723882570748479,
      "grad_norm": 0.06166541576385498,
      "learning_rate": 0.00014060835272289673,
      "loss": 0.0002,
      "step": 440
    },
    {
      "epoch": 0.3732345940227453,
      "grad_norm": 0.1766914576292038,
      "learning_rate": 0.00014036324975704985,
      "loss": 0.0004,
      "step": 441
    },
    {
      "epoch": 0.3740809309706427,
      "grad_norm": 1.185064435005188,
      "learning_rate": 0.00014011785676957422,
      "loss": 0.0011,
      "step": 442
    },
    {
      "epoch": 0.3749272679185401,
      "grad_norm": 0.10013063997030258,
      "learning_rate": 0.00013987217552368945,
      "loss": 0.0002,
      "step": 443
    },
    {
      "epoch": 0.37577360486643746,
      "grad_norm": 0.07872795313596725,
      "learning_rate": 0.0001396262077846864,
      "loss": 0.0002,
      "step": 444
    },
    {
      "epoch": 0.37661994181433484,
      "grad_norm": 0.0872654914855957,
      "learning_rate": 0.00013937995531991442,
      "loss": 0.0002,
      "step": 445
    },
    {
      "epoch": 0.3774662787622322,
      "grad_norm": 9.693881034851074,
      "learning_rate": 0.00013913341989876875,
      "loss": 0.0026,
      "step": 446
    },
    {
      "epoch": 0.3783126157101296,
      "grad_norm": 0.16127417981624603,
      "learning_rate": 0.0001388866032926777,
      "loss": 0.0003,
      "step": 447
    },
    {
      "epoch": 0.37915895265802696,
      "grad_norm": 0.31576570868492126,
      "learning_rate": 0.00013863950727509004,
      "loss": 0.0005,
      "step": 448
    },
    {
      "epoch": 0.38000528960592433,
      "grad_norm": 0.05285055935382843,
      "learning_rate": 0.0001383921336214621,
      "loss": 0.0002,
      "step": 449
    },
    {
      "epoch": 0.38085162655382176,
      "grad_norm": 0.6850122809410095,
      "learning_rate": 0.0001381444841092452,
      "loss": 0.0007,
      "step": 450
    },
    {
      "epoch": 0.38169796350171914,
      "grad_norm": 2.271670341491699,
      "learning_rate": 0.0001378965605178727,
      "loss": 0.0015,
      "step": 451
    },
    {
      "epoch": 0.3825443004496165,
      "grad_norm": 0.08596654236316681,
      "learning_rate": 0.00013764836462874742,
      "loss": 0.0003,
      "step": 452
    },
    {
      "epoch": 0.3833906373975139,
      "grad_norm": 0.13721808791160583,
      "learning_rate": 0.0001373998982252286,
      "loss": 0.0003,
      "step": 453
    },
    {
      "epoch": 0.38423697434541126,
      "grad_norm": 154.0753173828125,
      "learning_rate": 0.0001371511630926192,
      "loss": 0.0816,
      "step": 454
    },
    {
      "epoch": 0.38508331129330864,
      "grad_norm": 0.23801666498184204,
      "learning_rate": 0.00013690216101815318,
      "loss": 0.0004,
      "step": 455
    },
    {
      "epoch": 0.385929648241206,
      "grad_norm": 0.045311469584703445,
      "learning_rate": 0.00013665289379098246,
      "loss": 0.0002,
      "step": 456
    },
    {
      "epoch": 0.3867759851891034,
      "grad_norm": 0.10199946165084839,
      "learning_rate": 0.00013640336320216423,
      "loss": 0.0003,
      "step": 457
    },
    {
      "epoch": 0.3876223221370008,
      "grad_norm": 0.3142765760421753,
      "learning_rate": 0.00013615357104464795,
      "loss": 0.0005,
      "step": 458
    },
    {
      "epoch": 0.3884686590848982,
      "grad_norm": 0.05471539497375488,
      "learning_rate": 0.00013590351911326254,
      "loss": 0.0002,
      "step": 459
    },
    {
      "epoch": 0.38931499603279557,
      "grad_norm": 0.04315619170665741,
      "learning_rate": 0.00013565320920470348,
      "loss": 0.0002,
      "step": 460
    },
    {
      "epoch": 0.39016133298069294,
      "grad_norm": 0.1910562664270401,
      "learning_rate": 0.0001354026431175199,
      "loss": 0.0004,
      "step": 461
    },
    {
      "epoch": 0.3910076699285903,
      "grad_norm": 0.0606267973780632,
      "learning_rate": 0.00013515182265210165,
      "loss": 0.0002,
      "step": 462
    },
    {
      "epoch": 0.3918540068764877,
      "grad_norm": 0.11714379489421844,
      "learning_rate": 0.00013490074961066632,
      "loss": 0.0002,
      "step": 463
    },
    {
      "epoch": 0.39270034382438507,
      "grad_norm": 0.2893092930316925,
      "learning_rate": 0.0001346494257972464,
      "loss": 0.0004,
      "step": 464
    },
    {
      "epoch": 0.39354668077228244,
      "grad_norm": 0.04927147924900055,
      "learning_rate": 0.00013439785301767618,
      "loss": 0.0001,
      "step": 465
    },
    {
      "epoch": 0.3943930177201799,
      "grad_norm": 0.10841896384954453,
      "learning_rate": 0.0001341460330795789,
      "loss": 0.0002,
      "step": 466
    },
    {
      "epoch": 0.39523935466807725,
      "grad_norm": 0.039398279041051865,
      "learning_rate": 0.0001338939677923537,
      "loss": 0.0002,
      "step": 467
    },
    {
      "epoch": 0.3960856916159746,
      "grad_norm": 0.132989764213562,
      "learning_rate": 0.0001336416589671625,
      "loss": 0.0003,
      "step": 468
    },
    {
      "epoch": 0.396932028563872,
      "grad_norm": 0.11296701431274414,
      "learning_rate": 0.00013338910841691732,
      "loss": 0.0002,
      "step": 469
    },
    {
      "epoch": 0.39777836551176937,
      "grad_norm": 0.13828632235527039,
      "learning_rate": 0.0001331363179562669,
      "loss": 0.0002,
      "step": 470
    },
    {
      "epoch": 0.39862470245966675,
      "grad_norm": 0.32025042176246643,
      "learning_rate": 0.00013288328940158384,
      "loss": 0.0003,
      "step": 471
    },
    {
      "epoch": 0.3994710394075641,
      "grad_norm": 0.09038108587265015,
      "learning_rate": 0.0001326300245709516,
      "loss": 0.0002,
      "step": 472
    },
    {
      "epoch": 0.4003173763554615,
      "grad_norm": 0.09762014448642731,
      "learning_rate": 0.00013237652528415113,
      "loss": 0.0003,
      "step": 473
    },
    {
      "epoch": 0.4011637133033589,
      "grad_norm": 0.1327117532491684,
      "learning_rate": 0.00013212279336264823,
      "loss": 0.0002,
      "step": 474
    },
    {
      "epoch": 0.4020100502512563,
      "grad_norm": 0.059437185525894165,
      "learning_rate": 0.00013186883062958022,
      "loss": 0.0001,
      "step": 475
    },
    {
      "epoch": 0.4028563871991537,
      "grad_norm": 0.030865559354424477,
      "learning_rate": 0.00013161463890974277,
      "loss": 0.0001,
      "step": 476
    },
    {
      "epoch": 0.40370272414705105,
      "grad_norm": 0.7980221509933472,
      "learning_rate": 0.00013136022002957693,
      "loss": 0.0008,
      "step": 477
    },
    {
      "epoch": 0.4045490610949484,
      "grad_norm": 0.18203739821910858,
      "learning_rate": 0.000131105575817156,
      "loss": 0.0004,
      "step": 478
    },
    {
      "epoch": 0.4053953980428458,
      "grad_norm": 0.1254805475473404,
      "learning_rate": 0.00013085070810217228,
      "loss": 0.0003,
      "step": 479
    },
    {
      "epoch": 0.4062417349907432,
      "grad_norm": 0.047419462352991104,
      "learning_rate": 0.00013059561871592412,
      "loss": 0.0001,
      "step": 480
    },
    {
      "epoch": 0.40708807193864055,
      "grad_norm": 0.32353323698043823,
      "learning_rate": 0.00013034030949130254,
      "loss": 0.0004,
      "step": 481
    },
    {
      "epoch": 0.407934408886538,
      "grad_norm": 0.5889710783958435,
      "learning_rate": 0.00013008478226277816,
      "loss": 0.0003,
      "step": 482
    },
    {
      "epoch": 0.40878074583443536,
      "grad_norm": 0.28624850511550903,
      "learning_rate": 0.0001298290388663881,
      "loss": 0.0004,
      "step": 483
    },
    {
      "epoch": 0.40962708278233273,
      "grad_norm": 0.5747634768486023,
      "learning_rate": 0.00012957308113972261,
      "loss": 0.0002,
      "step": 484
    },
    {
      "epoch": 0.4104734197302301,
      "grad_norm": 0.12408791482448578,
      "learning_rate": 0.0001293169109219121,
      "loss": 0.0003,
      "step": 485
    },
    {
      "epoch": 0.4113197566781275,
      "grad_norm": 0.49346742033958435,
      "learning_rate": 0.00012906053005361365,
      "loss": 0.0004,
      "step": 486
    },
    {
      "epoch": 0.41216609362602485,
      "grad_norm": 0.04917092248797417,
      "learning_rate": 0.00012880394037699798,
      "loss": 0.0002,
      "step": 487
    },
    {
      "epoch": 0.41301243057392223,
      "grad_norm": 0.12637287378311157,
      "learning_rate": 0.0001285471437357362,
      "loss": 0.0002,
      "step": 488
    },
    {
      "epoch": 0.4138587675218196,
      "grad_norm": 0.07089531421661377,
      "learning_rate": 0.0001282901419749865,
      "loss": 0.0002,
      "step": 489
    },
    {
      "epoch": 0.414705104469717,
      "grad_norm": 0.37326568365097046,
      "learning_rate": 0.00012803293694138077,
      "loss": 0.0004,
      "step": 490
    },
    {
      "epoch": 0.4155514414176144,
      "grad_norm": 0.07765662670135498,
      "learning_rate": 0.00012777553048301171,
      "loss": 0.0002,
      "step": 491
    },
    {
      "epoch": 0.4163977783655118,
      "grad_norm": 0.04970049113035202,
      "learning_rate": 0.0001275179244494192,
      "loss": 0.0001,
      "step": 492
    },
    {
      "epoch": 0.41724411531340916,
      "grad_norm": 0.631595253944397,
      "learning_rate": 0.00012726012069157712,
      "loss": 0.0006,
      "step": 493
    },
    {
      "epoch": 0.41809045226130653,
      "grad_norm": 0.04407462105154991,
      "learning_rate": 0.00012700212106188009,
      "loss": 0.0001,
      "step": 494
    },
    {
      "epoch": 0.4189367892092039,
      "grad_norm": 0.37740379571914673,
      "learning_rate": 0.00012674392741413,
      "loss": 0.0004,
      "step": 495
    },
    {
      "epoch": 0.4197831261571013,
      "grad_norm": 0.06579752266407013,
      "learning_rate": 0.000126485541603523,
      "loss": 0.0002,
      "step": 496
    },
    {
      "epoch": 0.42062946310499866,
      "grad_norm": 0.616669237613678,
      "learning_rate": 0.00012622696548663593,
      "loss": 0.0004,
      "step": 497
    },
    {
      "epoch": 0.42147580005289603,
      "grad_norm": 5.032845973968506,
      "learning_rate": 0.00012596820092141295,
      "loss": 0.0117,
      "step": 498
    },
    {
      "epoch": 0.42232213700079346,
      "grad_norm": 0.10020484775304794,
      "learning_rate": 0.00012570924976715235,
      "loss": 0.0002,
      "step": 499
    },
    {
      "epoch": 0.42316847394869084,
      "grad_norm": 0.23150621354579926,
      "learning_rate": 0.0001254501138844931,
      "loss": 0.0003,
      "step": 500
    },
    {
      "epoch": 0.4240148108965882,
      "grad_norm": 0.8692544102668762,
      "learning_rate": 0.00012519079513540154,
      "loss": 0.0006,
      "step": 501
    },
    {
      "epoch": 0.4248611478444856,
      "grad_norm": 1.7455168962478638,
      "learning_rate": 0.00012493129538315788,
      "loss": 0.0012,
      "step": 502
    },
    {
      "epoch": 0.42570748479238296,
      "grad_norm": 0.291329026222229,
      "learning_rate": 0.00012467161649234293,
      "loss": 0.0004,
      "step": 503
    },
    {
      "epoch": 0.42655382174028034,
      "grad_norm": 0.1808886080980301,
      "learning_rate": 0.00012441176032882467,
      "loss": 0.0003,
      "step": 504
    },
    {
      "epoch": 0.4274001586881777,
      "grad_norm": 0.06194548308849335,
      "learning_rate": 0.00012415172875974482,
      "loss": 0.0002,
      "step": 505
    },
    {
      "epoch": 0.4282464956360751,
      "grad_norm": 0.050222016870975494,
      "learning_rate": 0.0001238915236535054,
      "loss": 0.0002,
      "step": 506
    },
    {
      "epoch": 0.4290928325839725,
      "grad_norm": 0.17352184653282166,
      "learning_rate": 0.0001236311468797555,
      "loss": 0.0004,
      "step": 507
    },
    {
      "epoch": 0.4299391695318699,
      "grad_norm": 7.776711940765381,
      "learning_rate": 0.00012337060030937744,
      "loss": 0.0064,
      "step": 508
    },
    {
      "epoch": 0.43078550647976727,
      "grad_norm": 0.08039094507694244,
      "learning_rate": 0.00012310988581447382,
      "loss": 0.0002,
      "step": 509
    },
    {
      "epoch": 0.43163184342766464,
      "grad_norm": 0.09459773451089859,
      "learning_rate": 0.0001228490052683537,
      "loss": 0.0002,
      "step": 510
    },
    {
      "epoch": 0.432478180375562,
      "grad_norm": 0.068666473031044,
      "learning_rate": 0.0001225879605455193,
      "loss": 0.0002,
      "step": 511
    },
    {
      "epoch": 0.4333245173234594,
      "grad_norm": 0.05744725465774536,
      "learning_rate": 0.00012232675352165246,
      "loss": 0.0002,
      "step": 512
    },
    {
      "epoch": 0.43417085427135677,
      "grad_norm": 0.13892993330955505,
      "learning_rate": 0.00012206538607360125,
      "loss": 0.0003,
      "step": 513
    },
    {
      "epoch": 0.43501719121925414,
      "grad_norm": 0.1475924551486969,
      "learning_rate": 0.00012180386007936637,
      "loss": 0.0002,
      "step": 514
    },
    {
      "epoch": 0.43586352816715157,
      "grad_norm": 0.07325658947229385,
      "learning_rate": 0.00012154217741808783,
      "loss": 0.0002,
      "step": 515
    },
    {
      "epoch": 0.43670986511504895,
      "grad_norm": 0.046727508306503296,
      "learning_rate": 0.00012128033997003126,
      "loss": 0.0002,
      "step": 516
    },
    {
      "epoch": 0.4375562020629463,
      "grad_norm": 0.10566474497318268,
      "learning_rate": 0.0001210183496165744,
      "loss": 0.0002,
      "step": 517
    },
    {
      "epoch": 0.4384025390108437,
      "grad_norm": 2.5309836864471436,
      "learning_rate": 0.00012075620824019384,
      "loss": 0.0008,
      "step": 518
    },
    {
      "epoch": 0.43924887595874107,
      "grad_norm": 0.1438109129667282,
      "learning_rate": 0.0001204939177244512,
      "loss": 0.0003,
      "step": 519
    },
    {
      "epoch": 0.44009521290663844,
      "grad_norm": 0.09494732320308685,
      "learning_rate": 0.0001202314799539797,
      "loss": 0.0002,
      "step": 520
    },
    {
      "epoch": 0.4409415498545358,
      "grad_norm": 0.09209834039211273,
      "learning_rate": 0.00011996889681447068,
      "loss": 0.0002,
      "step": 521
    },
    {
      "epoch": 0.4417878868024332,
      "grad_norm": 0.12126478552818298,
      "learning_rate": 0.00011970617019266,
      "loss": 0.0002,
      "step": 522
    },
    {
      "epoch": 0.4426342237503306,
      "grad_norm": 0.08780597895383835,
      "learning_rate": 0.00011944330197631441,
      "loss": 0.0001,
      "step": 523
    },
    {
      "epoch": 0.443480560698228,
      "grad_norm": 0.06644689291715622,
      "learning_rate": 0.00011918029405421816,
      "loss": 0.0002,
      "step": 524
    },
    {
      "epoch": 0.4443268976461254,
      "grad_norm": 0.14085879921913147,
      "learning_rate": 0.00011891714831615923,
      "loss": 0.0002,
      "step": 525
    },
    {
      "epoch": 0.44517323459402275,
      "grad_norm": 0.11284871399402618,
      "learning_rate": 0.00011865386665291591,
      "loss": 0.0002,
      "step": 526
    },
    {
      "epoch": 0.4460195715419201,
      "grad_norm": 0.5635073781013489,
      "learning_rate": 0.0001183904509562431,
      "loss": 0.0002,
      "step": 527
    },
    {
      "epoch": 0.4468659084898175,
      "grad_norm": 0.09344341605901718,
      "learning_rate": 0.00011812690311885886,
      "loss": 0.0002,
      "step": 528
    },
    {
      "epoch": 0.4477122454377149,
      "grad_norm": 0.08404481410980225,
      "learning_rate": 0.00011786322503443062,
      "loss": 0.0002,
      "step": 529
    },
    {
      "epoch": 0.44855858238561225,
      "grad_norm": 0.06212575361132622,
      "learning_rate": 0.00011759941859756172,
      "loss": 0.0002,
      "step": 530
    },
    {
      "epoch": 0.4494049193335097,
      "grad_norm": 0.037975095212459564,
      "learning_rate": 0.00011733548570377772,
      "loss": 0.0001,
      "step": 531
    },
    {
      "epoch": 0.45025125628140705,
      "grad_norm": 0.12612012028694153,
      "learning_rate": 0.00011707142824951287,
      "loss": 0.0002,
      "step": 532
    },
    {
      "epoch": 0.4510975932293044,
      "grad_norm": 0.13435746729373932,
      "learning_rate": 0.00011680724813209635,
      "loss": 0.0002,
      "step": 533
    },
    {
      "epoch": 0.4519439301772018,
      "grad_norm": 0.042436037212610245,
      "learning_rate": 0.0001165429472497388,
      "loss": 0.0001,
      "step": 534
    },
    {
      "epoch": 0.4527902671250992,
      "grad_norm": 0.10491074621677399,
      "learning_rate": 0.00011627852750151846,
      "loss": 0.0002,
      "step": 535
    },
    {
      "epoch": 0.45363660407299655,
      "grad_norm": 0.03083372488617897,
      "learning_rate": 0.00011601399078736781,
      "loss": 0.0001,
      "step": 536
    },
    {
      "epoch": 0.4544829410208939,
      "grad_norm": 0.10457072407007217,
      "learning_rate": 0.00011574933900805964,
      "loss": 0.0002,
      "step": 537
    },
    {
      "epoch": 0.4553292779687913,
      "grad_norm": 0.7160767912864685,
      "learning_rate": 0.00011548457406519356,
      "loss": 0.0004,
      "step": 538
    },
    {
      "epoch": 0.45617561491668873,
      "grad_norm": 0.17119579017162323,
      "learning_rate": 0.0001152196978611823,
      "loss": 0.0002,
      "step": 539
    },
    {
      "epoch": 0.4570219518645861,
      "grad_norm": 0.4046732485294342,
      "learning_rate": 0.00011495471229923805,
      "loss": 0.0003,
      "step": 540
    },
    {
      "epoch": 0.4578682888124835,
      "grad_norm": 0.17549632489681244,
      "learning_rate": 0.00011468961928335871,
      "loss": 0.0003,
      "step": 541
    },
    {
      "epoch": 0.45871462576038086,
      "grad_norm": 0.07443191111087799,
      "learning_rate": 0.00011442442071831434,
      "loss": 0.0002,
      "step": 542
    },
    {
      "epoch": 0.45956096270827823,
      "grad_norm": 0.05589667335152626,
      "learning_rate": 0.00011415911850963333,
      "loss": 0.0002,
      "step": 543
    },
    {
      "epoch": 0.4604072996561756,
      "grad_norm": 0.1417459100484848,
      "learning_rate": 0.00011389371456358884,
      "loss": 0.0002,
      "step": 544
    },
    {
      "epoch": 0.461253636604073,
      "grad_norm": 0.07161597907543182,
      "learning_rate": 0.00011362821078718502,
      "loss": 0.0002,
      "step": 545
    },
    {
      "epoch": 0.46209997355197036,
      "grad_norm": 0.26568058133125305,
      "learning_rate": 0.00011336260908814336,
      "loss": 0.0003,
      "step": 546
    },
    {
      "epoch": 0.4629463104998678,
      "grad_norm": 0.04489455372095108,
      "learning_rate": 0.00011309691137488885,
      "loss": 0.0001,
      "step": 547
    },
    {
      "epoch": 0.46379264744776516,
      "grad_norm": 0.1145646795630455,
      "learning_rate": 0.00011283111955653647,
      "loss": 0.0002,
      "step": 548
    },
    {
      "epoch": 0.46463898439566254,
      "grad_norm": 0.4978926479816437,
      "learning_rate": 0.00011256523554287735,
      "loss": 0.0005,
      "step": 549
    },
    {
      "epoch": 0.4654853213435599,
      "grad_norm": 6.101826190948486,
      "learning_rate": 0.00011229926124436505,
      "loss": 0.0022,
      "step": 550
    },
    {
      "epoch": 0.4663316582914573,
      "grad_norm": 6.155025482177734,
      "learning_rate": 0.00011203319857210191,
      "loss": 0.0014,
      "step": 551
    },
    {
      "epoch": 0.46717799523935466,
      "grad_norm": 0.09565731883049011,
      "learning_rate": 0.00011176704943782516,
      "loss": 0.0002,
      "step": 552
    },
    {
      "epoch": 0.46802433218725203,
      "grad_norm": 0.06565780192613602,
      "learning_rate": 0.00011150081575389335,
      "loss": 0.0002,
      "step": 553
    },
    {
      "epoch": 0.4688706691351494,
      "grad_norm": 0.18072128295898438,
      "learning_rate": 0.00011123449943327256,
      "loss": 0.0003,
      "step": 554
    },
    {
      "epoch": 0.46971700608304684,
      "grad_norm": 0.05880909785628319,
      "learning_rate": 0.00011096810238952257,
      "loss": 0.0001,
      "step": 555
    },
    {
      "epoch": 0.4705633430309442,
      "grad_norm": 0.3457149267196655,
      "learning_rate": 0.0001107016265367833,
      "loss": 0.0004,
      "step": 556
    },
    {
      "epoch": 0.4714096799788416,
      "grad_norm": 0.055852606892585754,
      "learning_rate": 0.00011043507378976077,
      "loss": 0.0002,
      "step": 557
    },
    {
      "epoch": 0.47225601692673896,
      "grad_norm": 0.03880249708890915,
      "learning_rate": 0.00011016844606371364,
      "loss": 0.0001,
      "step": 558
    },
    {
      "epoch": 0.47310235387463634,
      "grad_norm": 4.957912921905518,
      "learning_rate": 0.0001099017452744393,
      "loss": 0.0114,
      "step": 559
    },
    {
      "epoch": 0.4739486908225337,
      "grad_norm": 0.4188131093978882,
      "learning_rate": 0.00010963497333826002,
      "loss": 0.0005,
      "step": 560
    },
    {
      "epoch": 0.4747950277704311,
      "grad_norm": 0.08571381866931915,
      "learning_rate": 0.00010936813217200941,
      "loss": 0.0002,
      "step": 561
    },
    {
      "epoch": 0.47564136471832846,
      "grad_norm": 0.0913023054599762,
      "learning_rate": 0.00010910122369301842,
      "loss": 0.0002,
      "step": 562
    },
    {
      "epoch": 0.4764877016662259,
      "grad_norm": 0.03830798342823982,
      "learning_rate": 0.00010883424981910175,
      "loss": 0.0001,
      "step": 563
    },
    {
      "epoch": 0.47733403861412327,
      "grad_norm": 0.17392684519290924,
      "learning_rate": 0.0001085672124685439,
      "loss": 0.0003,
      "step": 564
    },
    {
      "epoch": 0.47818037556202064,
      "grad_norm": 0.065266914665699,
      "learning_rate": 0.00010830011356008553,
      "loss": 0.0001,
      "step": 565
    },
    {
      "epoch": 0.479026712509918,
      "grad_norm": 0.05811803787946701,
      "learning_rate": 0.00010803295501290953,
      "loss": 0.0002,
      "step": 566
    },
    {
      "epoch": 0.4798730494578154,
      "grad_norm": 0.08626824617385864,
      "learning_rate": 0.00010776573874662741,
      "loss": 0.0002,
      "step": 567
    },
    {
      "epoch": 0.48071938640571277,
      "grad_norm": 0.11945059150457382,
      "learning_rate": 0.00010749846668126535,
      "loss": 0.0002,
      "step": 568
    },
    {
      "epoch": 0.48156572335361014,
      "grad_norm": 0.0719977393746376,
      "learning_rate": 0.00010723114073725047,
      "loss": 0.0001,
      "step": 569
    },
    {
      "epoch": 0.4824120603015075,
      "grad_norm": 0.08843475580215454,
      "learning_rate": 0.00010696376283539704,
      "loss": 0.0002,
      "step": 570
    },
    {
      "epoch": 0.4832583972494049,
      "grad_norm": 0.049849677830934525,
      "learning_rate": 0.00010669633489689266,
      "loss": 0.0001,
      "step": 571
    },
    {
      "epoch": 0.4841047341973023,
      "grad_norm": 0.14390374720096588,
      "learning_rate": 0.00010642885884328445,
      "loss": 0.0002,
      "step": 572
    },
    {
      "epoch": 0.4849510711451997,
      "grad_norm": 4.275844573974609,
      "learning_rate": 0.00010616133659646521,
      "loss": 0.002,
      "step": 573
    },
    {
      "epoch": 0.48579740809309707,
      "grad_norm": 0.09194169193506241,
      "learning_rate": 0.00010589377007865973,
      "loss": 0.0002,
      "step": 574
    },
    {
      "epoch": 0.48664374504099445,
      "grad_norm": 0.09705609828233719,
      "learning_rate": 0.00010562616121241085,
      "loss": 0.0002,
      "step": 575
    },
    {
      "epoch": 0.4874900819888918,
      "grad_norm": 0.481925904750824,
      "learning_rate": 0.00010535851192056569,
      "loss": 0.0002,
      "step": 576
    },
    {
      "epoch": 0.4883364189367892,
      "grad_norm": 0.07338964194059372,
      "learning_rate": 0.00010509082412626193,
      "loss": 0.0002,
      "step": 577
    },
    {
      "epoch": 0.48918275588468657,
      "grad_norm": 0.24949859082698822,
      "learning_rate": 0.00010482309975291373,
      "loss": 0.0003,
      "step": 578
    },
    {
      "epoch": 0.49002909283258395,
      "grad_norm": 0.06238321214914322,
      "learning_rate": 0.00010455534072419821,
      "loss": 0.0002,
      "step": 579
    },
    {
      "epoch": 0.4908754297804814,
      "grad_norm": 0.03127686679363251,
      "learning_rate": 0.00010428754896404149,
      "loss": 0.0001,
      "step": 580
    },
    {
      "epoch": 0.49172176672837875,
      "grad_norm": 0.039029356092214584,
      "learning_rate": 0.00010401972639660489,
      "loss": 0.0001,
      "step": 581
    },
    {
      "epoch": 0.4925681036762761,
      "grad_norm": 0.08410942554473877,
      "learning_rate": 0.00010375187494627098,
      "loss": 0.0002,
      "step": 582
    },
    {
      "epoch": 0.4934144406241735,
      "grad_norm": 0.0970904678106308,
      "learning_rate": 0.00010348399653762996,
      "loss": 0.0002,
      "step": 583
    },
    {
      "epoch": 0.4942607775720709,
      "grad_norm": 0.04865993186831474,
      "learning_rate": 0.00010321609309546571,
      "loss": 0.0001,
      "step": 584
    },
    {
      "epoch": 0.49510711451996825,
      "grad_norm": 0.2513587176799774,
      "learning_rate": 0.00010294816654474198,
      "loss": 0.0002,
      "step": 585
    },
    {
      "epoch": 0.4959534514678656,
      "grad_norm": 0.04665207117795944,
      "learning_rate": 0.00010268021881058858,
      "loss": 0.0001,
      "step": 586
    },
    {
      "epoch": 0.496799788415763,
      "grad_norm": 0.031066421419382095,
      "learning_rate": 0.00010241225181828749,
      "loss": 0.0001,
      "step": 587
    },
    {
      "epoch": 0.49764612536366043,
      "grad_norm": 1.586833119392395,
      "learning_rate": 0.00010214426749325907,
      "loss": 0.0007,
      "step": 588
    },
    {
      "epoch": 0.4984924623115578,
      "grad_norm": 0.028086965903639793,
      "learning_rate": 0.00010187626776104828,
      "loss": 0.0001,
      "step": 589
    },
    {
      "epoch": 0.4993387992594552,
      "grad_norm": 0.1142040491104126,
      "learning_rate": 0.00010160825454731071,
      "loss": 0.0002,
      "step": 590
    },
    {
      "epoch": 0.5001851362073525,
      "grad_norm": 0.0628841370344162,
      "learning_rate": 0.00010134022977779887,
      "loss": 0.0002,
      "step": 591
    },
    {
      "epoch": 0.50103147315525,
      "grad_norm": 0.09075627475976944,
      "learning_rate": 0.00010107219537834826,
      "loss": 0.0002,
      "step": 592
    },
    {
      "epoch": 0.50103147315525,
      "eval_loss": 0.00019886840891558677,
      "eval_runtime": 22.2577,
      "eval_samples_per_second": 89.407,
      "eval_steps_per_second": 44.704,
      "step": 592
    },
    {
      "epoch": 0.5018778101031474,
      "grad_norm": 0.15841960906982422,
      "learning_rate": 0.00010080415327486363,
      "loss": 0.0002,
      "step": 593
    },
    {
      "epoch": 0.5027241470510447,
      "grad_norm": 0.07168613374233246,
      "learning_rate": 0.00010053610539330507,
      "loss": 0.0002,
      "step": 594
    },
    {
      "epoch": 0.5035704839989421,
      "grad_norm": 0.3676304221153259,
      "learning_rate": 0.00010026805365967409,
      "loss": 0.0002,
      "step": 595
    },
    {
      "epoch": 0.5044168209468395,
      "grad_norm": 0.044699862599372864,
      "learning_rate": 0.0001,
      "loss": 0.0001,
      "step": 596
    },
    {
      "epoch": 0.5052631578947369,
      "grad_norm": 0.24866217374801636,
      "learning_rate": 9.973194634032595e-05,
      "loss": 0.0002,
      "step": 597
    },
    {
      "epoch": 0.5061094948426342,
      "grad_norm": 0.04076084867119789,
      "learning_rate": 9.946389460669496e-05,
      "loss": 0.0001,
      "step": 598
    },
    {
      "epoch": 0.5069558317905316,
      "grad_norm": 0.1308746188879013,
      "learning_rate": 9.919584672513638e-05,
      "loss": 0.0002,
      "step": 599
    },
    {
      "epoch": 0.507802168738429,
      "grad_norm": 0.13659046590328217,
      "learning_rate": 9.892780462165172e-05,
      "loss": 0.0003,
      "step": 600
    },
    {
      "epoch": 0.5086485056863264,
      "grad_norm": 4.073505401611328,
      "learning_rate": 9.865977022220116e-05,
      "loss": 0.0024,
      "step": 601
    },
    {
      "epoch": 0.5094948426342237,
      "grad_norm": 12.728754997253418,
      "learning_rate": 9.839174545268931e-05,
      "loss": 0.0027,
      "step": 602
    },
    {
      "epoch": 0.5103411795821211,
      "grad_norm": 3.483036518096924,
      "learning_rate": 9.812373223895173e-05,
      "loss": 0.0028,
      "step": 603
    },
    {
      "epoch": 0.5111875165300185,
      "grad_norm": 0.052141524851322174,
      "learning_rate": 9.785573250674094e-05,
      "loss": 0.0001,
      "step": 604
    },
    {
      "epoch": 0.5120338534779159,
      "grad_norm": 0.09881985932588577,
      "learning_rate": 9.758774818171256e-05,
      "loss": 0.0002,
      "step": 605
    },
    {
      "epoch": 0.5128801904258132,
      "grad_norm": 0.6806803345680237,
      "learning_rate": 9.731978118941142e-05,
      "loss": 0.0007,
      "step": 606
    },
    {
      "epoch": 0.5137265273737106,
      "grad_norm": 0.03209676593542099,
      "learning_rate": 9.705183345525805e-05,
      "loss": 0.0001,
      "step": 607
    },
    {
      "epoch": 0.5145728643216081,
      "grad_norm": 0.16133540868759155,
      "learning_rate": 9.67839069045343e-05,
      "loss": 0.0003,
      "step": 608
    },
    {
      "epoch": 0.5154192012695055,
      "grad_norm": 3.0394766330718994,
      "learning_rate": 9.651600346237006e-05,
      "loss": 0.0029,
      "step": 609
    },
    {
      "epoch": 0.5162655382174028,
      "grad_norm": 0.07487703114748001,
      "learning_rate": 9.624812505372907e-05,
      "loss": 0.0001,
      "step": 610
    },
    {
      "epoch": 0.5171118751653002,
      "grad_norm": 0.08820661157369614,
      "learning_rate": 9.598027360339513e-05,
      "loss": 0.0002,
      "step": 611
    },
    {
      "epoch": 0.5179582121131976,
      "grad_norm": 0.03363629803061485,
      "learning_rate": 9.571245103595852e-05,
      "loss": 0.0001,
      "step": 612
    },
    {
      "epoch": 0.518804549061095,
      "grad_norm": 0.05476393178105354,
      "learning_rate": 9.54446592758018e-05,
      "loss": 0.0001,
      "step": 613
    },
    {
      "epoch": 0.5196508860089923,
      "grad_norm": 0.028762657195329666,
      "learning_rate": 9.517690024708628e-05,
      "loss": 0.0001,
      "step": 614
    },
    {
      "epoch": 0.5204972229568897,
      "grad_norm": 0.04667352885007858,
      "learning_rate": 9.490917587373812e-05,
      "loss": 0.0001,
      "step": 615
    },
    {
      "epoch": 0.5213435599047871,
      "grad_norm": 0.11764981597661972,
      "learning_rate": 9.464148807943431e-05,
      "loss": 0.0002,
      "step": 616
    },
    {
      "epoch": 0.5221898968526845,
      "grad_norm": 0.3532285690307617,
      "learning_rate": 9.437383878758918e-05,
      "loss": 0.0005,
      "step": 617
    },
    {
      "epoch": 0.5230362338005818,
      "grad_norm": 0.42520013451576233,
      "learning_rate": 9.410622992134032e-05,
      "loss": 0.0006,
      "step": 618
    },
    {
      "epoch": 0.5238825707484792,
      "grad_norm": 0.053856026381254196,
      "learning_rate": 9.383866340353482e-05,
      "loss": 0.0001,
      "step": 619
    },
    {
      "epoch": 0.5247289076963766,
      "grad_norm": 0.11913444846868515,
      "learning_rate": 9.357114115671559e-05,
      "loss": 0.0002,
      "step": 620
    },
    {
      "epoch": 0.525575244644274,
      "grad_norm": 0.08144916594028473,
      "learning_rate": 9.330366510310734e-05,
      "loss": 0.0002,
      "step": 621
    },
    {
      "epoch": 0.5264215815921713,
      "grad_norm": 0.07434901595115662,
      "learning_rate": 9.303623716460297e-05,
      "loss": 0.0002,
      "step": 622
    },
    {
      "epoch": 0.5272679185400687,
      "grad_norm": 0.07391541451215744,
      "learning_rate": 9.276885926274957e-05,
      "loss": 0.0002,
      "step": 623
    },
    {
      "epoch": 0.5281142554879662,
      "grad_norm": 0.08040373027324677,
      "learning_rate": 9.250153331873466e-05,
      "loss": 0.0002,
      "step": 624
    },
    {
      "epoch": 0.5289605924358636,
      "grad_norm": 0.05584714189171791,
      "learning_rate": 9.223426125337261e-05,
      "loss": 0.0001,
      "step": 625
    },
    {
      "epoch": 0.529806929383761,
      "grad_norm": 6.028743743896484,
      "learning_rate": 9.196704498709049e-05,
      "loss": 0.0006,
      "step": 626
    },
    {
      "epoch": 0.5306532663316583,
      "grad_norm": 0.030659710988402367,
      "learning_rate": 9.169988643991449e-05,
      "loss": 0.0001,
      "step": 627
    },
    {
      "epoch": 0.5314996032795557,
      "grad_norm": 0.041605204343795776,
      "learning_rate": 9.143278753145612e-05,
      "loss": 0.0001,
      "step": 628
    },
    {
      "epoch": 0.5323459402274531,
      "grad_norm": 0.029253825545310974,
      "learning_rate": 9.116575018089825e-05,
      "loss": 0.0001,
      "step": 629
    },
    {
      "epoch": 0.5331922771753504,
      "grad_norm": 0.11101424694061279,
      "learning_rate": 9.089877630698158e-05,
      "loss": 0.0002,
      "step": 630
    },
    {
      "epoch": 0.5340386141232478,
      "grad_norm": 1.2533376216888428,
      "learning_rate": 9.063186782799063e-05,
      "loss": 0.0009,
      "step": 631
    },
    {
      "epoch": 0.5348849510711452,
      "grad_norm": 0.11181803047657013,
      "learning_rate": 9.036502666173999e-05,
      "loss": 0.0002,
      "step": 632
    },
    {
      "epoch": 0.5357312880190426,
      "grad_norm": 0.04902556911110878,
      "learning_rate": 9.009825472556074e-05,
      "loss": 0.0001,
      "step": 633
    },
    {
      "epoch": 0.53657762496694,
      "grad_norm": 0.042089883238077164,
      "learning_rate": 8.983155393628635e-05,
      "loss": 0.0001,
      "step": 634
    },
    {
      "epoch": 0.5374239619148373,
      "grad_norm": 0.048657555133104324,
      "learning_rate": 8.956492621023925e-05,
      "loss": 0.0002,
      "step": 635
    },
    {
      "epoch": 0.5382702988627347,
      "grad_norm": 0.07978585362434387,
      "learning_rate": 8.929837346321673e-05,
      "loss": 0.0001,
      "step": 636
    },
    {
      "epoch": 0.5391166358106321,
      "grad_norm": 0.05948410555720329,
      "learning_rate": 8.903189761047744e-05,
      "loss": 0.0002,
      "step": 637
    },
    {
      "epoch": 0.5399629727585294,
      "grad_norm": 0.03661718592047691,
      "learning_rate": 8.876550056672747e-05,
      "loss": 0.0001,
      "step": 638
    },
    {
      "epoch": 0.5408093097064268,
      "grad_norm": 0.03483026102185249,
      "learning_rate": 8.849918424610667e-05,
      "loss": 0.0001,
      "step": 639
    },
    {
      "epoch": 0.5416556466543243,
      "grad_norm": 0.41989028453826904,
      "learning_rate": 8.823295056217486e-05,
      "loss": 0.0004,
      "step": 640
    },
    {
      "epoch": 0.5425019836022217,
      "grad_norm": 0.07068897038698196,
      "learning_rate": 8.796680142789813e-05,
      "loss": 0.0001,
      "step": 641
    },
    {
      "epoch": 0.5433483205501191,
      "grad_norm": 0.20673921704292297,
      "learning_rate": 8.770073875563493e-05,
      "loss": 0.0003,
      "step": 642
    },
    {
      "epoch": 0.5441946574980164,
      "grad_norm": 0.10442963987588882,
      "learning_rate": 8.743476445712268e-05,
      "loss": 0.0002,
      "step": 643
    },
    {
      "epoch": 0.5450409944459138,
      "grad_norm": 0.04070938006043434,
      "learning_rate": 8.716888044346353e-05,
      "loss": 0.0001,
      "step": 644
    },
    {
      "epoch": 0.5458873313938112,
      "grad_norm": 0.0466531440615654,
      "learning_rate": 8.690308862511117e-05,
      "loss": 0.0001,
      "step": 645
    },
    {
      "epoch": 0.5467336683417086,
      "grad_norm": 0.05943473428487778,
      "learning_rate": 8.663739091185668e-05,
      "loss": 0.0002,
      "step": 646
    },
    {
      "epoch": 0.5475800052896059,
      "grad_norm": 0.2231179028749466,
      "learning_rate": 8.637178921281496e-05,
      "loss": 0.0003,
      "step": 647
    },
    {
      "epoch": 0.5484263422375033,
      "grad_norm": 0.034228820353746414,
      "learning_rate": 8.610628543641118e-05,
      "loss": 0.0001,
      "step": 648
    },
    {
      "epoch": 0.5492726791854007,
      "grad_norm": 0.0363926962018013,
      "learning_rate": 8.58408814903667e-05,
      "loss": 0.0001,
      "step": 649
    },
    {
      "epoch": 0.550119016133298,
      "grad_norm": 0.13770565390586853,
      "learning_rate": 8.557557928168568e-05,
      "loss": 0.0003,
      "step": 650
    },
    {
      "epoch": 0.5509653530811954,
      "grad_norm": 6.903998851776123,
      "learning_rate": 8.531038071664133e-05,
      "loss": 0.0124,
      "step": 651
    },
    {
      "epoch": 0.5518116900290928,
      "grad_norm": 2.468278169631958,
      "learning_rate": 8.504528770076196e-05,
      "loss": 0.0011,
      "step": 652
    },
    {
      "epoch": 0.5526580269769902,
      "grad_norm": 5.0674824714660645,
      "learning_rate": 8.478030213881772e-05,
      "loss": 0.0041,
      "step": 653
    },
    {
      "epoch": 0.5535043639248876,
      "grad_norm": 0.20628859102725983,
      "learning_rate": 8.45154259348065e-05,
      "loss": 0.0002,
      "step": 654
    },
    {
      "epoch": 0.5543507008727849,
      "grad_norm": 0.21092236042022705,
      "learning_rate": 8.425066099194038e-05,
      "loss": 0.0003,
      "step": 655
    },
    {
      "epoch": 0.5551970378206823,
      "grad_norm": 0.04090695455670357,
      "learning_rate": 8.398600921263222e-05,
      "loss": 0.0001,
      "step": 656
    },
    {
      "epoch": 0.5560433747685798,
      "grad_norm": 0.21549102663993835,
      "learning_rate": 8.372147249848154e-05,
      "loss": 0.0003,
      "step": 657
    },
    {
      "epoch": 0.5568897117164772,
      "grad_norm": 0.18999320268630981,
      "learning_rate": 8.345705275026123e-05,
      "loss": 0.0003,
      "step": 658
    },
    {
      "epoch": 0.5577360486643745,
      "grad_norm": 0.06413779407739639,
      "learning_rate": 8.319275186790366e-05,
      "loss": 0.0002,
      "step": 659
    },
    {
      "epoch": 0.5585823856122719,
      "grad_norm": 0.05759165436029434,
      "learning_rate": 8.292857175048715e-05,
      "loss": 0.0002,
      "step": 660
    },
    {
      "epoch": 0.5594287225601693,
      "grad_norm": 0.8964185118675232,
      "learning_rate": 8.26645142962223e-05,
      "loss": 0.0006,
      "step": 661
    },
    {
      "epoch": 0.5602750595080667,
      "grad_norm": 0.0584690198302269,
      "learning_rate": 8.240058140243834e-05,
      "loss": 0.0001,
      "step": 662
    },
    {
      "epoch": 0.561121396455964,
      "grad_norm": 0.11676496267318726,
      "learning_rate": 8.213677496556939e-05,
      "loss": 0.0002,
      "step": 663
    },
    {
      "epoch": 0.5619677334038614,
      "grad_norm": 0.1564321517944336,
      "learning_rate": 8.187309688114118e-05,
      "loss": 0.0002,
      "step": 664
    },
    {
      "epoch": 0.5628140703517588,
      "grad_norm": 0.1715901494026184,
      "learning_rate": 8.16095490437569e-05,
      "loss": 0.0002,
      "step": 665
    },
    {
      "epoch": 0.5636604072996562,
      "grad_norm": 0.10471925139427185,
      "learning_rate": 8.134613334708412e-05,
      "loss": 0.0002,
      "step": 666
    },
    {
      "epoch": 0.5645067442475535,
      "grad_norm": 0.05598120763897896,
      "learning_rate": 8.108285168384081e-05,
      "loss": 0.0001,
      "step": 667
    },
    {
      "epoch": 0.5653530811954509,
      "grad_norm": 0.25912487506866455,
      "learning_rate": 8.081970594578185e-05,
      "loss": 0.0002,
      "step": 668
    },
    {
      "epoch": 0.5661994181433483,
      "grad_norm": 1.188771367073059,
      "learning_rate": 8.055669802368561e-05,
      "loss": 0.0005,
      "step": 669
    },
    {
      "epoch": 0.5670457550912457,
      "grad_norm": 0.0743853896856308,
      "learning_rate": 8.029382980734e-05,
      "loss": 0.0001,
      "step": 670
    },
    {
      "epoch": 0.567892092039143,
      "grad_norm": 0.3215715289115906,
      "learning_rate": 8.003110318552933e-05,
      "loss": 0.0003,
      "step": 671
    },
    {
      "epoch": 0.5687384289870404,
      "grad_norm": 0.03677264600992203,
      "learning_rate": 7.976852004602033e-05,
      "loss": 0.0001,
      "step": 672
    },
    {
      "epoch": 0.5695847659349379,
      "grad_norm": 0.17620785534381866,
      "learning_rate": 7.950608227554884e-05,
      "loss": 0.0003,
      "step": 673
    },
    {
      "epoch": 0.5704311028828353,
      "grad_norm": 0.04508533328771591,
      "learning_rate": 7.924379175980618e-05,
      "loss": 0.0002,
      "step": 674
    },
    {
      "epoch": 0.5712774398307326,
      "grad_norm": 0.07219874858856201,
      "learning_rate": 7.898165038342563e-05,
      "loss": 0.0002,
      "step": 675
    },
    {
      "epoch": 0.57212377677863,
      "grad_norm": 0.07334332168102264,
      "learning_rate": 7.871966002996878e-05,
      "loss": 0.0001,
      "step": 676
    },
    {
      "epoch": 0.5729701137265274,
      "grad_norm": 0.0835946574807167,
      "learning_rate": 7.845782258191218e-05,
      "loss": 0.0002,
      "step": 677
    },
    {
      "epoch": 0.5738164506744248,
      "grad_norm": 0.02998744137585163,
      "learning_rate": 7.819613992063361e-05,
      "loss": 0.0001,
      "step": 678
    },
    {
      "epoch": 0.5746627876223221,
      "grad_norm": 0.08256243914365768,
      "learning_rate": 7.793461392639878e-05,
      "loss": 0.0002,
      "step": 679
    },
    {
      "epoch": 0.5755091245702195,
      "grad_norm": 0.05922257527709007,
      "learning_rate": 7.767324647834758e-05,
      "loss": 0.0001,
      "step": 680
    },
    {
      "epoch": 0.5763554615181169,
      "grad_norm": 0.08410824090242386,
      "learning_rate": 7.741203945448071e-05,
      "loss": 0.0002,
      "step": 681
    },
    {
      "epoch": 0.5772017984660143,
      "grad_norm": 0.06941843777894974,
      "learning_rate": 7.715099473164632e-05,
      "loss": 0.0002,
      "step": 682
    },
    {
      "epoch": 0.5780481354139116,
      "grad_norm": 0.0681728646159172,
      "learning_rate": 7.689011418552618e-05,
      "loss": 0.0001,
      "step": 683
    },
    {
      "epoch": 0.578894472361809,
      "grad_norm": 0.035902418196201324,
      "learning_rate": 7.662939969062259e-05,
      "loss": 0.0001,
      "step": 684
    },
    {
      "epoch": 0.5797408093097064,
      "grad_norm": 0.09035813063383102,
      "learning_rate": 7.636885312024456e-05,
      "loss": 0.0002,
      "step": 685
    },
    {
      "epoch": 0.5805871462576038,
      "grad_norm": 0.07556555420160294,
      "learning_rate": 7.610847634649458e-05,
      "loss": 0.0001,
      "step": 686
    },
    {
      "epoch": 0.5814334832055011,
      "grad_norm": 0.1552840620279312,
      "learning_rate": 7.584827124025521e-05,
      "loss": 0.0003,
      "step": 687
    },
    {
      "epoch": 0.5822798201533985,
      "grad_norm": 0.12494683265686035,
      "learning_rate": 7.558823967117537e-05,
      "loss": 0.0002,
      "step": 688
    },
    {
      "epoch": 0.583126157101296,
      "grad_norm": 0.08067876845598221,
      "learning_rate": 7.532838350765708e-05,
      "loss": 0.0002,
      "step": 689
    },
    {
      "epoch": 0.5839724940491934,
      "grad_norm": 0.1234188824892044,
      "learning_rate": 7.506870461684215e-05,
      "loss": 0.0002,
      "step": 690
    },
    {
      "epoch": 0.5848188309970908,
      "grad_norm": 0.06756996363401413,
      "learning_rate": 7.480920486459847e-05,
      "loss": 0.0002,
      "step": 691
    },
    {
      "epoch": 0.5856651679449881,
      "grad_norm": 0.09005871415138245,
      "learning_rate": 7.45498861155069e-05,
      "loss": 0.0002,
      "step": 692
    },
    {
      "epoch": 0.5865115048928855,
      "grad_norm": 0.7519410252571106,
      "learning_rate": 7.42907502328477e-05,
      "loss": 0.0005,
      "step": 693
    },
    {
      "epoch": 0.5873578418407829,
      "grad_norm": 0.05168967694044113,
      "learning_rate": 7.403179907858708e-05,
      "loss": 0.0002,
      "step": 694
    },
    {
      "epoch": 0.5882041787886803,
      "grad_norm": 0.12125031650066376,
      "learning_rate": 7.37730345133641e-05,
      "loss": 0.0002,
      "step": 695
    },
    {
      "epoch": 0.5890505157365776,
      "grad_norm": 0.044862050563097,
      "learning_rate": 7.3514458396477e-05,
      "loss": 0.0001,
      "step": 696
    },
    {
      "epoch": 0.589896852684475,
      "grad_norm": 0.07706248015165329,
      "learning_rate": 7.325607258587002e-05,
      "loss": 0.0001,
      "step": 697
    },
    {
      "epoch": 0.5907431896323724,
      "grad_norm": 0.18789656460285187,
      "learning_rate": 7.299787893811998e-05,
      "loss": 0.0003,
      "step": 698
    },
    {
      "epoch": 0.5915895265802698,
      "grad_norm": 0.07291367650032043,
      "learning_rate": 7.273987930842288e-05,
      "loss": 0.0002,
      "step": 699
    },
    {
      "epoch": 0.5924358635281671,
      "grad_norm": 0.048046525567770004,
      "learning_rate": 7.24820755505808e-05,
      "loss": 0.0002,
      "step": 700
    },
    {
      "epoch": 0.5932822004760645,
      "grad_norm": 0.1294231414794922,
      "learning_rate": 7.222446951698831e-05,
      "loss": 0.0003,
      "step": 701
    },
    {
      "epoch": 0.5941285374239619,
      "grad_norm": 0.052419308573007584,
      "learning_rate": 7.196706305861925e-05,
      "loss": 0.0002,
      "step": 702
    },
    {
      "epoch": 0.5949748743718593,
      "grad_norm": 0.20319993793964386,
      "learning_rate": 7.170985802501358e-05,
      "loss": 0.0003,
      "step": 703
    },
    {
      "epoch": 0.5958212113197566,
      "grad_norm": 3.063783884048462,
      "learning_rate": 7.14528562642638e-05,
      "loss": 0.001,
      "step": 704
    },
    {
      "epoch": 0.5966675482676541,
      "grad_norm": 5.154725074768066,
      "learning_rate": 7.119605962300204e-05,
      "loss": 0.0049,
      "step": 705
    },
    {
      "epoch": 0.5975138852155515,
      "grad_norm": 0.057394057512283325,
      "learning_rate": 7.093946994638638e-05,
      "loss": 0.0001,
      "step": 706
    },
    {
      "epoch": 0.5983602221634489,
      "grad_norm": 0.054040245711803436,
      "learning_rate": 7.068308907808791e-05,
      "loss": 0.0002,
      "step": 707
    },
    {
      "epoch": 0.5992065591113462,
      "grad_norm": 0.046586133539676666,
      "learning_rate": 7.04269188602774e-05,
      "loss": 0.0001,
      "step": 708
    },
    {
      "epoch": 0.6000528960592436,
      "grad_norm": 0.06369137763977051,
      "learning_rate": 7.017096113361192e-05,
      "loss": 0.0002,
      "step": 709
    },
    {
      "epoch": 0.600899233007141,
      "grad_norm": 0.17471066117286682,
      "learning_rate": 6.991521773722186e-05,
      "loss": 0.0002,
      "step": 710
    },
    {
      "epoch": 0.6017455699550384,
      "grad_norm": 0.22996436059474945,
      "learning_rate": 6.965969050869752e-05,
      "loss": 0.0003,
      "step": 711
    },
    {
      "epoch": 0.6025919069029357,
      "grad_norm": 0.05932232365012169,
      "learning_rate": 6.940438128407589e-05,
      "loss": 0.0002,
      "step": 712
    },
    {
      "epoch": 0.6034382438508331,
      "grad_norm": 0.26761987805366516,
      "learning_rate": 6.914929189782773e-05,
      "loss": 0.0003,
      "step": 713
    },
    {
      "epoch": 0.6042845807987305,
      "grad_norm": 0.09831937402486801,
      "learning_rate": 6.889442418284402e-05,
      "loss": 0.0002,
      "step": 714
    },
    {
      "epoch": 0.6051309177466279,
      "grad_norm": 0.11792817711830139,
      "learning_rate": 6.863977997042309e-05,
      "loss": 0.0002,
      "step": 715
    },
    {
      "epoch": 0.6059772546945252,
      "grad_norm": 2.865610361099243,
      "learning_rate": 6.838536109025727e-05,
      "loss": 0.0012,
      "step": 716
    },
    {
      "epoch": 0.6068235916424226,
      "grad_norm": 0.1912592649459839,
      "learning_rate": 6.813116937041979e-05,
      "loss": 0.0002,
      "step": 717
    },
    {
      "epoch": 0.60766992859032,
      "grad_norm": 0.5504553318023682,
      "learning_rate": 6.787720663735178e-05,
      "loss": 0.0005,
      "step": 718
    },
    {
      "epoch": 0.6085162655382174,
      "grad_norm": 1.0158289670944214,
      "learning_rate": 6.76234747158489e-05,
      "loss": 0.0011,
      "step": 719
    },
    {
      "epoch": 0.6093626024861147,
      "grad_norm": 0.04889851436018944,
      "learning_rate": 6.736997542904843e-05,
      "loss": 0.0001,
      "step": 720
    },
    {
      "epoch": 0.6102089394340122,
      "grad_norm": 0.08757238835096359,
      "learning_rate": 6.711671059841615e-05,
      "loss": 0.0002,
      "step": 721
    },
    {
      "epoch": 0.6110552763819096,
      "grad_norm": 0.05490309000015259,
      "learning_rate": 6.68636820437331e-05,
      "loss": 0.0002,
      "step": 722
    },
    {
      "epoch": 0.611901613329807,
      "grad_norm": 0.1514064371585846,
      "learning_rate": 6.66108915830827e-05,
      "loss": 0.0003,
      "step": 723
    },
    {
      "epoch": 0.6127479502777043,
      "grad_norm": 0.06284084916114807,
      "learning_rate": 6.635834103283754e-05,
      "loss": 0.0002,
      "step": 724
    },
    {
      "epoch": 0.6135942872256017,
      "grad_norm": 0.028150109574198723,
      "learning_rate": 6.610603220764635e-05,
      "loss": 0.0001,
      "step": 725
    },
    {
      "epoch": 0.6144406241734991,
      "grad_norm": 0.0833248645067215,
      "learning_rate": 6.585396692042113e-05,
      "loss": 0.0002,
      "step": 726
    },
    {
      "epoch": 0.6152869611213965,
      "grad_norm": 0.06450650095939636,
      "learning_rate": 6.560214698232381e-05,
      "loss": 0.0001,
      "step": 727
    },
    {
      "epoch": 0.6161332980692938,
      "grad_norm": 0.029820643365383148,
      "learning_rate": 6.535057420275364e-05,
      "loss": 0.0001,
      "step": 728
    },
    {
      "epoch": 0.6169796350171912,
      "grad_norm": 0.06921394914388657,
      "learning_rate": 6.509925038933372e-05,
      "loss": 0.0001,
      "step": 729
    },
    {
      "epoch": 0.6178259719650886,
      "grad_norm": 0.083246611058712,
      "learning_rate": 6.484817734789838e-05,
      "loss": 0.0002,
      "step": 730
    },
    {
      "epoch": 0.618672308912986,
      "grad_norm": 0.06405588239431381,
      "learning_rate": 6.459735688248012e-05,
      "loss": 0.0001,
      "step": 731
    },
    {
      "epoch": 0.6195186458608833,
      "grad_norm": 0.0298240277916193,
      "learning_rate": 6.434679079529658e-05,
      "loss": 0.0001,
      "step": 732
    },
    {
      "epoch": 0.6203649828087807,
      "grad_norm": 0.0725536122918129,
      "learning_rate": 6.409648088673749e-05,
      "loss": 0.0002,
      "step": 733
    },
    {
      "epoch": 0.6212113197566781,
      "grad_norm": 0.12698467075824738,
      "learning_rate": 6.384642895535209e-05,
      "loss": 0.0002,
      "step": 734
    },
    {
      "epoch": 0.6220576567045755,
      "grad_norm": 0.10995099693536758,
      "learning_rate": 6.359663679783577e-05,
      "loss": 0.0003,
      "step": 735
    },
    {
      "epoch": 0.6229039936524728,
      "grad_norm": 0.05986730009317398,
      "learning_rate": 6.334710620901753e-05,
      "loss": 0.0001,
      "step": 736
    },
    {
      "epoch": 0.6237503306003702,
      "grad_norm": 0.1268569827079773,
      "learning_rate": 6.309783898184684e-05,
      "loss": 0.0002,
      "step": 737
    },
    {
      "epoch": 0.6245966675482677,
      "grad_norm": 0.03877655416727066,
      "learning_rate": 6.28488369073808e-05,
      "loss": 0.0001,
      "step": 738
    },
    {
      "epoch": 0.6254430044961651,
      "grad_norm": 0.460100382566452,
      "learning_rate": 6.260010177477143e-05,
      "loss": 0.0003,
      "step": 739
    },
    {
      "epoch": 0.6262893414440625,
      "grad_norm": 0.06817807257175446,
      "learning_rate": 6.235163537125257e-05,
      "loss": 0.0001,
      "step": 740
    },
    {
      "epoch": 0.6271356783919598,
      "grad_norm": 0.053915977478027344,
      "learning_rate": 6.21034394821273e-05,
      "loss": 0.0002,
      "step": 741
    },
    {
      "epoch": 0.6279820153398572,
      "grad_norm": 0.05049784854054451,
      "learning_rate": 6.185551589075482e-05,
      "loss": 0.0001,
      "step": 742
    },
    {
      "epoch": 0.6288283522877546,
      "grad_norm": 0.045517224818468094,
      "learning_rate": 6.160786637853793e-05,
      "loss": 0.0001,
      "step": 743
    },
    {
      "epoch": 0.629674689235652,
      "grad_norm": 0.08308801054954529,
      "learning_rate": 6.136049272490998e-05,
      "loss": 0.0002,
      "step": 744
    },
    {
      "epoch": 0.6305210261835493,
      "grad_norm": 0.32736459374427795,
      "learning_rate": 6.111339670732233e-05,
      "loss": 0.0003,
      "step": 745
    },
    {
      "epoch": 0.6313673631314467,
      "grad_norm": 0.04458026960492134,
      "learning_rate": 6.0866580101231255e-05,
      "loss": 0.0002,
      "step": 746
    },
    {
      "epoch": 0.6322137000793441,
      "grad_norm": 0.0794198289513588,
      "learning_rate": 6.06200446800856e-05,
      "loss": 0.0002,
      "step": 747
    },
    {
      "epoch": 0.6330600370272415,
      "grad_norm": 0.04481593891978264,
      "learning_rate": 6.0373792215313606e-05,
      "loss": 0.0001,
      "step": 748
    },
    {
      "epoch": 0.6339063739751388,
      "grad_norm": 0.29709896445274353,
      "learning_rate": 6.0127824476310556e-05,
      "loss": 0.0003,
      "step": 749
    },
    {
      "epoch": 0.6347527109230362,
      "grad_norm": 0.06510290503501892,
      "learning_rate": 5.988214323042581e-05,
      "loss": 0.0002,
      "step": 750
    },
    {
      "epoch": 0.6355990478709336,
      "grad_norm": 0.4633365571498871,
      "learning_rate": 5.9636750242950165e-05,
      "loss": 0.0004,
      "step": 751
    },
    {
      "epoch": 0.636445384818831,
      "grad_norm": 5.964758396148682,
      "learning_rate": 5.9391647277103316e-05,
      "loss": 0.0011,
      "step": 752
    },
    {
      "epoch": 0.6372917217667283,
      "grad_norm": 0.21779651939868927,
      "learning_rate": 5.9146836094020894e-05,
      "loss": 0.0003,
      "step": 753
    },
    {
      "epoch": 0.6381380587146258,
      "grad_norm": 0.09280341118574142,
      "learning_rate": 5.890231845274219e-05,
      "loss": 0.0002,
      "step": 754
    },
    {
      "epoch": 0.6389843956625232,
      "grad_norm": 0.1196620911359787,
      "learning_rate": 5.865809611019716e-05,
      "loss": 0.0002,
      "step": 755
    },
    {
      "epoch": 0.6398307326104206,
      "grad_norm": 0.031216420233249664,
      "learning_rate": 5.841417082119397e-05,
      "loss": 0.0001,
      "step": 756
    },
    {
      "epoch": 0.6406770695583179,
      "grad_norm": 0.07899965345859528,
      "learning_rate": 5.8170544338406497e-05,
      "loss": 0.0002,
      "step": 757
    },
    {
      "epoch": 0.6415234065062153,
      "grad_norm": 0.6815491318702698,
      "learning_rate": 5.792721841236143e-05,
      "loss": 0.0006,
      "step": 758
    },
    {
      "epoch": 0.6423697434541127,
      "grad_norm": 0.08929415047168732,
      "learning_rate": 5.768419479142597e-05,
      "loss": 0.0002,
      "step": 759
    },
    {
      "epoch": 0.6432160804020101,
      "grad_norm": 0.04529585689306259,
      "learning_rate": 5.744147522179515e-05,
      "loss": 0.0001,
      "step": 760
    },
    {
      "epoch": 0.6440624173499074,
      "grad_norm": 0.12001106888055801,
      "learning_rate": 5.719906144747934e-05,
      "loss": 0.0002,
      "step": 761
    },
    {
      "epoch": 0.6449087542978048,
      "grad_norm": 0.032826073467731476,
      "learning_rate": 5.695695521029163e-05,
      "loss": 0.0001,
      "step": 762
    },
    {
      "epoch": 0.6457550912457022,
      "grad_norm": 0.07705069333314896,
      "learning_rate": 5.671515824983539e-05,
      "loss": 0.0002,
      "step": 763
    },
    {
      "epoch": 0.6466014281935996,
      "grad_norm": 0.06072944030165672,
      "learning_rate": 5.647367230349174e-05,
      "loss": 0.0001,
      "step": 764
    },
    {
      "epoch": 0.6474477651414969,
      "grad_norm": 0.037452373653650284,
      "learning_rate": 5.623249910640711e-05,
      "loss": 0.0001,
      "step": 765
    },
    {
      "epoch": 0.6482941020893943,
      "grad_norm": 0.01534605398774147,
      "learning_rate": 5.599164039148066e-05,
      "loss": 0.0001,
      "step": 766
    },
    {
      "epoch": 0.6491404390372917,
      "grad_norm": 0.051084402948617935,
      "learning_rate": 5.5751097889351974e-05,
      "loss": 0.0002,
      "step": 767
    },
    {
      "epoch": 0.6499867759851891,
      "grad_norm": 0.03912018984556198,
      "learning_rate": 5.551087332838858e-05,
      "loss": 0.0001,
      "step": 768
    },
    {
      "epoch": 0.6508331129330864,
      "grad_norm": 0.035196781158447266,
      "learning_rate": 5.527096843467333e-05,
      "loss": 0.0001,
      "step": 769
    },
    {
      "epoch": 0.6516794498809839,
      "grad_norm": 0.1608777642250061,
      "learning_rate": 5.503138493199247e-05,
      "loss": 0.0002,
      "step": 770
    },
    {
      "epoch": 0.6525257868288813,
      "grad_norm": 0.03106590546667576,
      "learning_rate": 5.479212454182264e-05,
      "loss": 0.0001,
      "step": 771
    },
    {
      "epoch": 0.6533721237767787,
      "grad_norm": 0.046204160898923874,
      "learning_rate": 5.455318898331905e-05,
      "loss": 0.0001,
      "step": 772
    },
    {
      "epoch": 0.654218460724676,
      "grad_norm": 12.764196395874023,
      "learning_rate": 5.4314579973302804e-05,
      "loss": 0.0086,
      "step": 773
    },
    {
      "epoch": 0.6550647976725734,
      "grad_norm": 0.07368520647287369,
      "learning_rate": 5.407629922624866e-05,
      "loss": 0.0001,
      "step": 774
    },
    {
      "epoch": 0.6559111346204708,
      "grad_norm": 0.0575093999505043,
      "learning_rate": 5.383834845427284e-05,
      "loss": 0.0002,
      "step": 775
    },
    {
      "epoch": 0.6567574715683682,
      "grad_norm": 0.027164271101355553,
      "learning_rate": 5.360072936712036e-05,
      "loss": 0.0001,
      "step": 776
    },
    {
      "epoch": 0.6576038085162655,
      "grad_norm": 0.08310569077730179,
      "learning_rate": 5.336344367215318e-05,
      "loss": 0.0002,
      "step": 777
    },
    {
      "epoch": 0.6584501454641629,
      "grad_norm": 0.11127579212188721,
      "learning_rate": 5.312649307433768e-05,
      "loss": 0.0002,
      "step": 778
    },
    {
      "epoch": 0.6592964824120603,
      "grad_norm": 0.30127614736557007,
      "learning_rate": 5.2889879276232446e-05,
      "loss": 0.0002,
      "step": 779
    },
    {
      "epoch": 0.6601428193599577,
      "grad_norm": 0.04109237715601921,
      "learning_rate": 5.26536039779761e-05,
      "loss": 0.0001,
      "step": 780
    },
    {
      "epoch": 0.660989156307855,
      "grad_norm": 0.2747434973716736,
      "learning_rate": 5.241766887727505e-05,
      "loss": 0.0004,
      "step": 781
    },
    {
      "epoch": 0.6618354932557524,
      "grad_norm": 0.36636075377464294,
      "learning_rate": 5.218207566939116e-05,
      "loss": 0.0001,
      "step": 782
    },
    {
      "epoch": 0.6626818302036498,
      "grad_norm": 0.088040791451931,
      "learning_rate": 5.19468260471299e-05,
      "loss": 0.0002,
      "step": 783
    },
    {
      "epoch": 0.6635281671515472,
      "grad_norm": 0.1606176346540451,
      "learning_rate": 5.171192170082787e-05,
      "loss": 0.0002,
      "step": 784
    },
    {
      "epoch": 0.6643745040994445,
      "grad_norm": 0.03453480452299118,
      "learning_rate": 5.1477364318340736e-05,
      "loss": 0.0001,
      "step": 785
    },
    {
      "epoch": 0.665220841047342,
      "grad_norm": 0.06563825905323029,
      "learning_rate": 5.124315558503121e-05,
      "loss": 0.0001,
      "step": 786
    },
    {
      "epoch": 0.6660671779952394,
      "grad_norm": 0.06603348255157471,
      "learning_rate": 5.100929718375671e-05,
      "loss": 0.0001,
      "step": 787
    },
    {
      "epoch": 0.6669135149431368,
      "grad_norm": 0.058555878698825836,
      "learning_rate": 5.077579079485764e-05,
      "loss": 0.0001,
      "step": 788
    },
    {
      "epoch": 0.6677598518910342,
      "grad_norm": 0.22287344932556152,
      "learning_rate": 5.054263809614488e-05,
      "loss": 0.0003,
      "step": 789
    },
    {
      "epoch": 0.6686061888389315,
      "grad_norm": 0.05332765355706215,
      "learning_rate": 5.030984076288805e-05,
      "loss": 0.0002,
      "step": 790
    },
    {
      "epoch": 0.6694525257868289,
      "grad_norm": 0.034715961664915085,
      "learning_rate": 5.007740046780337e-05,
      "loss": 0.0001,
      "step": 791
    },
    {
      "epoch": 0.6702988627347263,
      "grad_norm": 0.11441633105278015,
      "learning_rate": 4.9845318881041616e-05,
      "loss": 0.0002,
      "step": 792
    },
    {
      "epoch": 0.6711451996826236,
      "grad_norm": 0.03240574523806572,
      "learning_rate": 4.961359767017613e-05,
      "loss": 0.0001,
      "step": 793
    },
    {
      "epoch": 0.671991536630521,
      "grad_norm": 0.2833597660064697,
      "learning_rate": 4.938223850019087e-05,
      "loss": 0.0003,
      "step": 794
    },
    {
      "epoch": 0.6728378735784184,
      "grad_norm": 0.1415741890668869,
      "learning_rate": 4.915124303346842e-05,
      "loss": 0.0002,
      "step": 795
    },
    {
      "epoch": 0.6736842105263158,
      "grad_norm": 0.028880994766950607,
      "learning_rate": 4.8920612929778044e-05,
      "loss": 0.0001,
      "step": 796
    },
    {
      "epoch": 0.6745305474742131,
      "grad_norm": 0.045264821499586105,
      "learning_rate": 4.869034984626376e-05,
      "loss": 0.0001,
      "step": 797
    },
    {
      "epoch": 0.6753768844221105,
      "grad_norm": 0.043451372534036636,
      "learning_rate": 4.846045543743247e-05,
      "loss": 0.0001,
      "step": 798
    },
    {
      "epoch": 0.6762232213700079,
      "grad_norm": 0.13336284458637238,
      "learning_rate": 4.8230931355142054e-05,
      "loss": 0.0002,
      "step": 799
    },
    {
      "epoch": 0.6770695583179053,
      "grad_norm": 0.07531437277793884,
      "learning_rate": 4.800177924858933e-05,
      "loss": 0.0002,
      "step": 800
    },
    {
      "epoch": 0.6779158952658026,
      "grad_norm": 1.8115055561065674,
      "learning_rate": 4.777300076429866e-05,
      "loss": 0.0008,
      "step": 801
    },
    {
      "epoch": 0.6787622322137001,
      "grad_norm": 0.07725503295660019,
      "learning_rate": 4.7544597546109514e-05,
      "loss": 0.0002,
      "step": 802
    },
    {
      "epoch": 0.6796085691615975,
      "grad_norm": 0.09105274081230164,
      "learning_rate": 4.731657123516517e-05,
      "loss": 0.0002,
      "step": 803
    },
    {
      "epoch": 0.6804549061094949,
      "grad_norm": 0.05316171050071716,
      "learning_rate": 4.7088923469900633e-05,
      "loss": 0.0001,
      "step": 804
    },
    {
      "epoch": 0.6813012430573923,
      "grad_norm": 0.20848077535629272,
      "learning_rate": 4.6861655886030984e-05,
      "loss": 0.0003,
      "step": 805
    },
    {
      "epoch": 0.6821475800052896,
      "grad_norm": 0.0430375374853611,
      "learning_rate": 4.663477011653955e-05,
      "loss": 0.0001,
      "step": 806
    },
    {
      "epoch": 0.682993916953187,
      "grad_norm": 0.17218826711177826,
      "learning_rate": 4.640826779166626e-05,
      "loss": 0.0004,
      "step": 807
    },
    {
      "epoch": 0.6838402539010844,
      "grad_norm": 0.033413782715797424,
      "learning_rate": 4.618215053889584e-05,
      "loss": 0.0001,
      "step": 808
    },
    {
      "epoch": 0.6846865908489818,
      "grad_norm": 0.037513989955186844,
      "learning_rate": 4.595641998294618e-05,
      "loss": 0.0001,
      "step": 809
    },
    {
      "epoch": 0.6855329277968791,
      "grad_norm": 0.026102302595973015,
      "learning_rate": 4.5731077745756644e-05,
      "loss": 0.0001,
      "step": 810
    },
    {
      "epoch": 0.6863792647447765,
      "grad_norm": 0.1992892473936081,
      "learning_rate": 4.550612544647642e-05,
      "loss": 0.0002,
      "step": 811
    },
    {
      "epoch": 0.6872256016926739,
      "grad_norm": 0.05261588096618652,
      "learning_rate": 4.528156470145289e-05,
      "loss": 0.0001,
      "step": 812
    },
    {
      "epoch": 0.6880719386405713,
      "grad_norm": 0.050703030079603195,
      "learning_rate": 4.505739712421988e-05,
      "loss": 0.0001,
      "step": 813
    },
    {
      "epoch": 0.6889182755884686,
      "grad_norm": 0.05695679038763046,
      "learning_rate": 4.4833624325486446e-05,
      "loss": 0.0001,
      "step": 814
    },
    {
      "epoch": 0.689764612536366,
      "grad_norm": 0.06029987335205078,
      "learning_rate": 4.461024791312479e-05,
      "loss": 0.0001,
      "step": 815
    },
    {
      "epoch": 0.6906109494842634,
      "grad_norm": 2.13311767578125,
      "learning_rate": 4.438726949215909e-05,
      "loss": 0.0008,
      "step": 816
    },
    {
      "epoch": 0.6914572864321608,
      "grad_norm": 0.07027161866426468,
      "learning_rate": 4.4164690664753836e-05,
      "loss": 0.0001,
      "step": 817
    },
    {
      "epoch": 0.6923036233800581,
      "grad_norm": 0.07970570772886276,
      "learning_rate": 4.3942513030202305e-05,
      "loss": 0.0002,
      "step": 818
    },
    {
      "epoch": 0.6931499603279556,
      "grad_norm": 0.5385849475860596,
      "learning_rate": 4.372073818491508e-05,
      "loss": 0.0004,
      "step": 819
    },
    {
      "epoch": 0.693996297275853,
      "grad_norm": 0.07817856967449188,
      "learning_rate": 4.34993677224086e-05,
      "loss": 0.0002,
      "step": 820
    },
    {
      "epoch": 0.6948426342237504,
      "grad_norm": 0.03579568490386009,
      "learning_rate": 4.3278403233293694e-05,
      "loss": 0.0001,
      "step": 821
    },
    {
      "epoch": 0.6956889711716477,
      "grad_norm": 0.03255557641386986,
      "learning_rate": 4.305784630526416e-05,
      "loss": 0.0001,
      "step": 822
    },
    {
      "epoch": 0.6965353081195451,
      "grad_norm": 0.041807692497968674,
      "learning_rate": 4.283769852308533e-05,
      "loss": 0.0001,
      "step": 823
    },
    {
      "epoch": 0.6973816450674425,
      "grad_norm": 0.12837781012058258,
      "learning_rate": 4.261796146858274e-05,
      "loss": 0.0002,
      "step": 824
    },
    {
      "epoch": 0.6982279820153399,
      "grad_norm": 0.027660679072141647,
      "learning_rate": 4.239863672063071e-05,
      "loss": 0.0001,
      "step": 825
    },
    {
      "epoch": 0.6990743189632372,
      "grad_norm": 0.04993637651205063,
      "learning_rate": 4.217972585514095e-05,
      "loss": 0.0001,
      "step": 826
    },
    {
      "epoch": 0.6999206559111346,
      "grad_norm": 0.041045546531677246,
      "learning_rate": 4.196123044505148e-05,
      "loss": 0.0001,
      "step": 827
    },
    {
      "epoch": 0.700766992859032,
      "grad_norm": 0.026169411838054657,
      "learning_rate": 4.174315206031496e-05,
      "loss": 0.0001,
      "step": 828
    },
    {
      "epoch": 0.7016133298069294,
      "grad_norm": 0.039118994027376175,
      "learning_rate": 4.1525492267887714e-05,
      "loss": 0.0001,
      "step": 829
    },
    {
      "epoch": 0.7024596667548267,
      "grad_norm": 0.17674730718135834,
      "learning_rate": 4.1308252631718325e-05,
      "loss": 0.0002,
      "step": 830
    },
    {
      "epoch": 0.7033060037027241,
      "grad_norm": 0.1946047693490982,
      "learning_rate": 4.109143471273644e-05,
      "loss": 0.0002,
      "step": 831
    },
    {
      "epoch": 0.7041523406506215,
      "grad_norm": 0.03667005896568298,
      "learning_rate": 4.087504006884155e-05,
      "loss": 0.0001,
      "step": 832
    },
    {
      "epoch": 0.7049986775985189,
      "grad_norm": 0.017610784620046616,
      "learning_rate": 4.0659070254891793e-05,
      "loss": 0.0001,
      "step": 833
    },
    {
      "epoch": 0.7058450145464162,
      "grad_norm": 0.08646859228610992,
      "learning_rate": 4.0443526822692755e-05,
      "loss": 0.0002,
      "step": 834
    },
    {
      "epoch": 0.7066913514943137,
      "grad_norm": 0.06043276563286781,
      "learning_rate": 4.022841132098638e-05,
      "loss": 0.0001,
      "step": 835
    },
    {
      "epoch": 0.7075376884422111,
      "grad_norm": 0.2109387069940567,
      "learning_rate": 4.0013725295439774e-05,
      "loss": 0.0002,
      "step": 836
    },
    {
      "epoch": 0.7083840253901085,
      "grad_norm": 0.02143879421055317,
      "learning_rate": 3.9799470288634177e-05,
      "loss": 0.0001,
      "step": 837
    },
    {
      "epoch": 0.7092303623380058,
      "grad_norm": 0.044113364070653915,
      "learning_rate": 3.958564784005382e-05,
      "loss": 0.0001,
      "step": 838
    },
    {
      "epoch": 0.7100766992859032,
      "grad_norm": 0.034567538648843765,
      "learning_rate": 3.9372259486074766e-05,
      "loss": 0.0001,
      "step": 839
    },
    {
      "epoch": 0.7109230362338006,
      "grad_norm": 0.05990441516041756,
      "learning_rate": 3.915930675995424e-05,
      "loss": 0.0001,
      "step": 840
    },
    {
      "epoch": 0.711769373181698,
      "grad_norm": 0.037960901856422424,
      "learning_rate": 3.894679119181907e-05,
      "loss": 0.0001,
      "step": 841
    },
    {
      "epoch": 0.7126157101295953,
      "grad_norm": 0.07244334369897842,
      "learning_rate": 3.873471430865515e-05,
      "loss": 0.0001,
      "step": 842
    },
    {
      "epoch": 0.7134620470774927,
      "grad_norm": 0.022804200649261475,
      "learning_rate": 3.852307763429626e-05,
      "loss": 0.0001,
      "step": 843
    },
    {
      "epoch": 0.7143083840253901,
      "grad_norm": 0.07347637414932251,
      "learning_rate": 3.831188268941309e-05,
      "loss": 0.0002,
      "step": 844
    },
    {
      "epoch": 0.7151547209732875,
      "grad_norm": 0.11441130191087723,
      "learning_rate": 3.8101130991502543e-05,
      "loss": 0.0002,
      "step": 845
    },
    {
      "epoch": 0.7160010579211848,
      "grad_norm": 0.0762052908539772,
      "learning_rate": 3.789082405487645e-05,
      "loss": 0.0002,
      "step": 846
    },
    {
      "epoch": 0.7168473948690822,
      "grad_norm": 0.03789770230650902,
      "learning_rate": 3.768096339065104e-05,
      "loss": 0.0001,
      "step": 847
    },
    {
      "epoch": 0.7176937318169796,
      "grad_norm": 0.02568461000919342,
      "learning_rate": 3.7471550506735894e-05,
      "loss": 0.0001,
      "step": 848
    },
    {
      "epoch": 0.718540068764877,
      "grad_norm": 0.07044924795627594,
      "learning_rate": 3.7262586907823194e-05,
      "loss": 0.0001,
      "step": 849
    },
    {
      "epoch": 0.7193864057127743,
      "grad_norm": 0.04015934839844704,
      "learning_rate": 3.705407409537684e-05,
      "loss": 0.0001,
      "step": 850
    },
    {
      "epoch": 0.7202327426606718,
      "grad_norm": 0.05022226274013519,
      "learning_rate": 3.6846013567621784e-05,
      "loss": 0.0001,
      "step": 851
    },
    {
      "epoch": 0.7210790796085692,
      "grad_norm": 0.046968862414360046,
      "learning_rate": 3.6638406819533014e-05,
      "loss": 0.0002,
      "step": 852
    },
    {
      "epoch": 0.7219254165564666,
      "grad_norm": 0.04005877673625946,
      "learning_rate": 3.643125534282521e-05,
      "loss": 0.0001,
      "step": 853
    },
    {
      "epoch": 0.722771753504364,
      "grad_norm": 0.029529646039009094,
      "learning_rate": 3.622456062594154e-05,
      "loss": 0.0001,
      "step": 854
    },
    {
      "epoch": 0.7236180904522613,
      "grad_norm": 0.03489358723163605,
      "learning_rate": 3.601832415404346e-05,
      "loss": 0.0001,
      "step": 855
    },
    {
      "epoch": 0.7244644274001587,
      "grad_norm": 0.11569522321224213,
      "learning_rate": 3.581254740899969e-05,
      "loss": 0.0002,
      "step": 856
    },
    {
      "epoch": 0.7253107643480561,
      "grad_norm": 0.057617247104644775,
      "learning_rate": 3.560723186937559e-05,
      "loss": 0.0002,
      "step": 857
    },
    {
      "epoch": 0.7261571012959535,
      "grad_norm": 0.477774977684021,
      "learning_rate": 3.540237901042285e-05,
      "loss": 0.0004,
      "step": 858
    },
    {
      "epoch": 0.7270034382438508,
      "grad_norm": 0.02523122727870941,
      "learning_rate": 3.519799030406844e-05,
      "loss": 0.0001,
      "step": 859
    },
    {
      "epoch": 0.7278497751917482,
      "grad_norm": 0.04599815234541893,
      "learning_rate": 3.4994067218904383e-05,
      "loss": 0.0001,
      "step": 860
    },
    {
      "epoch": 0.7286961121396456,
      "grad_norm": 0.037407319992780685,
      "learning_rate": 3.479061122017706e-05,
      "loss": 0.0001,
      "step": 861
    },
    {
      "epoch": 0.729542449087543,
      "grad_norm": 0.04487038776278496,
      "learning_rate": 3.458762376977669e-05,
      "loss": 0.0001,
      "step": 862
    },
    {
      "epoch": 0.7303887860354403,
      "grad_norm": 0.14074736833572388,
      "learning_rate": 3.438510632622683e-05,
      "loss": 0.0002,
      "step": 863
    },
    {
      "epoch": 0.7312351229833377,
      "grad_norm": 0.04506589099764824,
      "learning_rate": 3.4183060344673934e-05,
      "loss": 0.0001,
      "step": 864
    },
    {
      "epoch": 0.7320814599312351,
      "grad_norm": 0.12804463505744934,
      "learning_rate": 3.398148727687682e-05,
      "loss": 0.0002,
      "step": 865
    },
    {
      "epoch": 0.7329277968791325,
      "grad_norm": 0.043049369007349014,
      "learning_rate": 3.378038857119632e-05,
      "loss": 0.0001,
      "step": 866
    },
    {
      "epoch": 0.7337741338270299,
      "grad_norm": 0.09256184101104736,
      "learning_rate": 3.357976567258481e-05,
      "loss": 0.0002,
      "step": 867
    },
    {
      "epoch": 0.7346204707749273,
      "grad_norm": 0.274862676858902,
      "learning_rate": 3.3379620022575864e-05,
      "loss": 0.0003,
      "step": 868
    },
    {
      "epoch": 0.7354668077228247,
      "grad_norm": 0.030030054971575737,
      "learning_rate": 3.317995305927395e-05,
      "loss": 0.0001,
      "step": 869
    },
    {
      "epoch": 0.7363131446707221,
      "grad_norm": 1.3099826574325562,
      "learning_rate": 3.298076621734385e-05,
      "loss": 0.0008,
      "step": 870
    },
    {
      "epoch": 0.7371594816186194,
      "grad_norm": 0.01840503327548504,
      "learning_rate": 3.27820609280008e-05,
      "loss": 0.0001,
      "step": 871
    },
    {
      "epoch": 0.7380058185665168,
      "grad_norm": 0.06400467455387115,
      "learning_rate": 3.258383861899969e-05,
      "loss": 0.0001,
      "step": 872
    },
    {
      "epoch": 0.7388521555144142,
      "grad_norm": 0.12344048917293549,
      "learning_rate": 3.238610071462522e-05,
      "loss": 0.0002,
      "step": 873
    },
    {
      "epoch": 0.7396984924623116,
      "grad_norm": 0.1406022012233734,
      "learning_rate": 3.2188848635681446e-05,
      "loss": 0.0003,
      "step": 874
    },
    {
      "epoch": 0.7405448294102089,
      "grad_norm": 0.0508735254406929,
      "learning_rate": 3.199208379948161e-05,
      "loss": 0.0001,
      "step": 875
    },
    {
      "epoch": 0.7413911663581063,
      "grad_norm": 0.02087491564452648,
      "learning_rate": 3.179580761983801e-05,
      "loss": 0.0001,
      "step": 876
    },
    {
      "epoch": 0.7422375033060037,
      "grad_norm": 0.1172461062669754,
      "learning_rate": 3.160002150705179e-05,
      "loss": 0.0002,
      "step": 877
    },
    {
      "epoch": 0.7430838402539011,
      "grad_norm": 0.10503999143838882,
      "learning_rate": 3.1404726867902815e-05,
      "loss": 0.0002,
      "step": 878
    },
    {
      "epoch": 0.7439301772017984,
      "grad_norm": 0.037205640226602554,
      "learning_rate": 3.120992510563958e-05,
      "loss": 0.0001,
      "step": 879
    },
    {
      "epoch": 0.7447765141496958,
      "grad_norm": 0.06464976817369461,
      "learning_rate": 3.101561761996912e-05,
      "loss": 0.0002,
      "step": 880
    },
    {
      "epoch": 0.7456228510975932,
      "grad_norm": 0.05341091752052307,
      "learning_rate": 3.082180580704693e-05,
      "loss": 0.0001,
      "step": 881
    },
    {
      "epoch": 0.7464691880454906,
      "grad_norm": 0.043499480932950974,
      "learning_rate": 3.0628491059467014e-05,
      "loss": 0.0001,
      "step": 882
    },
    {
      "epoch": 0.747315524993388,
      "grad_norm": 0.026690788567066193,
      "learning_rate": 3.0435674766251655e-05,
      "loss": 0.0001,
      "step": 883
    },
    {
      "epoch": 0.7481618619412854,
      "grad_norm": 0.144480362534523,
      "learning_rate": 3.0243358312841862e-05,
      "loss": 0.0001,
      "step": 884
    },
    {
      "epoch": 0.7490081988891828,
      "grad_norm": 0.03942825272679329,
      "learning_rate": 3.0051543081086887e-05,
      "loss": 0.0001,
      "step": 885
    },
    {
      "epoch": 0.7498545358370802,
      "grad_norm": 0.0741799920797348,
      "learning_rate": 2.9860230449234706e-05,
      "loss": 0.0002,
      "step": 886
    },
    {
      "epoch": 0.7507008727849775,
      "grad_norm": 0.43694689869880676,
      "learning_rate": 2.9669421791921936e-05,
      "loss": 0.0004,
      "step": 887
    },
    {
      "epoch": 0.7515472097328749,
      "grad_norm": 0.04569239914417267,
      "learning_rate": 2.9479118480164004e-05,
      "loss": 0.0001,
      "step": 888
    },
    {
      "epoch": 0.7515472097328749,
      "eval_loss": 0.00015138309390749782,
      "eval_runtime": 22.3571,
      "eval_samples_per_second": 89.01,
      "eval_steps_per_second": 44.505,
      "step": 888
    },
    {
      "epoch": 0.7523935466807723,
      "grad_norm": 0.051316723227500916,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.0001,
      "step": 889
    },
    {
      "epoch": 0.7532398836286697,
      "grad_norm": 0.1506781429052353,
      "learning_rate": 2.910003335920918e-05,
      "loss": 0.0002,
      "step": 890
    },
    {
      "epoch": 0.754086220576567,
      "grad_norm": 0.17350022494792938,
      "learning_rate": 2.8911254273848564e-05,
      "loss": 0.0002,
      "step": 891
    },
    {
      "epoch": 0.7549325575244644,
      "grad_norm": 0.10590571910142899,
      "learning_rate": 2.872298598169578e-05,
      "loss": 0.0002,
      "step": 892
    },
    {
      "epoch": 0.7557788944723618,
      "grad_norm": 0.347629576921463,
      "learning_rate": 2.8535229835512988e-05,
      "loss": 0.0003,
      "step": 893
    },
    {
      "epoch": 0.7566252314202592,
      "grad_norm": 0.06805393844842911,
      "learning_rate": 2.8347987184382398e-05,
      "loss": 0.0002,
      "step": 894
    },
    {
      "epoch": 0.7574715683681565,
      "grad_norm": 0.053487904369831085,
      "learning_rate": 2.816125937369669e-05,
      "loss": 0.0001,
      "step": 895
    },
    {
      "epoch": 0.7583179053160539,
      "grad_norm": 0.1019781231880188,
      "learning_rate": 2.7975047745149118e-05,
      "loss": 0.0002,
      "step": 896
    },
    {
      "epoch": 0.7591642422639513,
      "grad_norm": 0.03230949491262436,
      "learning_rate": 2.7789353636724224e-05,
      "loss": 0.0001,
      "step": 897
    },
    {
      "epoch": 0.7600105792118487,
      "grad_norm": 0.10666634142398834,
      "learning_rate": 2.760417838268784e-05,
      "loss": 0.0002,
      "step": 898
    },
    {
      "epoch": 0.760856916159746,
      "grad_norm": 0.043601736426353455,
      "learning_rate": 2.7419523313577767e-05,
      "loss": 0.0001,
      "step": 899
    },
    {
      "epoch": 0.7617032531076435,
      "grad_norm": 0.2986220717430115,
      "learning_rate": 2.7235389756194095e-05,
      "loss": 0.0004,
      "step": 900
    },
    {
      "epoch": 0.7625495900555409,
      "grad_norm": 0.0650421679019928,
      "learning_rate": 2.7051779033589732e-05,
      "loss": 0.0001,
      "step": 901
    },
    {
      "epoch": 0.7633959270034383,
      "grad_norm": 0.07894416153430939,
      "learning_rate": 2.6868692465060828e-05,
      "loss": 0.0002,
      "step": 902
    },
    {
      "epoch": 0.7642422639513357,
      "grad_norm": 15.576688766479492,
      "learning_rate": 2.6686131366137378e-05,
      "loss": 0.0099,
      "step": 903
    },
    {
      "epoch": 0.765088600899233,
      "grad_norm": 0.02983132377266884,
      "learning_rate": 2.6504097048573685e-05,
      "loss": 0.0001,
      "step": 904
    },
    {
      "epoch": 0.7659349378471304,
      "grad_norm": 0.07050719857215881,
      "learning_rate": 2.6322590820339e-05,
      "loss": 0.0002,
      "step": 905
    },
    {
      "epoch": 0.7667812747950278,
      "grad_norm": 0.2947550117969513,
      "learning_rate": 2.6141613985608093e-05,
      "loss": 0.0003,
      "step": 906
    },
    {
      "epoch": 0.7676276117429252,
      "grad_norm": 0.10146462917327881,
      "learning_rate": 2.59611678447519e-05,
      "loss": 0.0002,
      "step": 907
    },
    {
      "epoch": 0.7684739486908225,
      "grad_norm": 0.03578224405646324,
      "learning_rate": 2.5781253694328155e-05,
      "loss": 0.0001,
      "step": 908
    },
    {
      "epoch": 0.7693202856387199,
      "grad_norm": 0.11798228323459625,
      "learning_rate": 2.5601872827072016e-05,
      "loss": 0.0002,
      "step": 909
    },
    {
      "epoch": 0.7701666225866173,
      "grad_norm": 0.03726814314723015,
      "learning_rate": 2.542302653188704e-05,
      "loss": 0.0001,
      "step": 910
    },
    {
      "epoch": 0.7710129595345147,
      "grad_norm": 0.05878552049398422,
      "learning_rate": 2.524471609383552e-05,
      "loss": 0.0001,
      "step": 911
    },
    {
      "epoch": 0.771859296482412,
      "grad_norm": 0.212065190076828,
      "learning_rate": 2.5066942794129567e-05,
      "loss": 0.0003,
      "step": 912
    },
    {
      "epoch": 0.7727056334303094,
      "grad_norm": 0.023151680827140808,
      "learning_rate": 2.4889707910121797e-05,
      "loss": 0.0001,
      "step": 913
    },
    {
      "epoch": 0.7735519703782068,
      "grad_norm": 0.04849415272474289,
      "learning_rate": 2.4713012715296113e-05,
      "loss": 0.0001,
      "step": 914
    },
    {
      "epoch": 0.7743983073261042,
      "grad_norm": 0.3643008768558502,
      "learning_rate": 2.4536858479258628e-05,
      "loss": 0.0004,
      "step": 915
    },
    {
      "epoch": 0.7752446442740016,
      "grad_norm": 0.07301183044910431,
      "learning_rate": 2.4361246467728506e-05,
      "loss": 0.0001,
      "step": 916
    },
    {
      "epoch": 0.776090981221899,
      "grad_norm": 0.049862127751111984,
      "learning_rate": 2.4186177942528844e-05,
      "loss": 0.0001,
      "step": 917
    },
    {
      "epoch": 0.7769373181697964,
      "grad_norm": 0.029499627649784088,
      "learning_rate": 2.4011654161577667e-05,
      "loss": 0.0001,
      "step": 918
    },
    {
      "epoch": 0.7777836551176938,
      "grad_norm": 0.0962710827589035,
      "learning_rate": 2.383767637887885e-05,
      "loss": 0.0002,
      "step": 919
    },
    {
      "epoch": 0.7786299920655911,
      "grad_norm": 0.07912716269493103,
      "learning_rate": 2.3664245844513078e-05,
      "loss": 0.0002,
      "step": 920
    },
    {
      "epoch": 0.7794763290134885,
      "grad_norm": 0.0478321835398674,
      "learning_rate": 2.3491363804628962e-05,
      "loss": 0.0001,
      "step": 921
    },
    {
      "epoch": 0.7803226659613859,
      "grad_norm": 0.17505258321762085,
      "learning_rate": 2.331903150143391e-05,
      "loss": 0.0002,
      "step": 922
    },
    {
      "epoch": 0.7811690029092833,
      "grad_norm": 0.15537269413471222,
      "learning_rate": 2.3147250173185486e-05,
      "loss": 0.0003,
      "step": 923
    },
    {
      "epoch": 0.7820153398571806,
      "grad_norm": 0.33590516448020935,
      "learning_rate": 2.2976021054182152e-05,
      "loss": 0.0004,
      "step": 924
    },
    {
      "epoch": 0.782861676805078,
      "grad_norm": 1.057529330253601,
      "learning_rate": 2.2805345374754684e-05,
      "loss": 0.0007,
      "step": 925
    },
    {
      "epoch": 0.7837080137529754,
      "grad_norm": 0.03747454658150673,
      "learning_rate": 2.263522436125729e-05,
      "loss": 0.0001,
      "step": 926
    },
    {
      "epoch": 0.7845543507008728,
      "grad_norm": 0.03142997995018959,
      "learning_rate": 2.246565923605859e-05,
      "loss": 0.0001,
      "step": 927
    },
    {
      "epoch": 0.7854006876487701,
      "grad_norm": 0.02467474155128002,
      "learning_rate": 2.229665121753317e-05,
      "loss": 0.0001,
      "step": 928
    },
    {
      "epoch": 0.7862470245966675,
      "grad_norm": 0.029198601841926575,
      "learning_rate": 2.212820152005246e-05,
      "loss": 0.0001,
      "step": 929
    },
    {
      "epoch": 0.7870933615445649,
      "grad_norm": 0.09220435470342636,
      "learning_rate": 2.1960311353976316e-05,
      "loss": 0.0002,
      "step": 930
    },
    {
      "epoch": 0.7879396984924623,
      "grad_norm": 8.980354309082031,
      "learning_rate": 2.1792981925644162e-05,
      "loss": 0.0034,
      "step": 931
    },
    {
      "epoch": 0.7887860354403597,
      "grad_norm": 0.05904144421219826,
      "learning_rate": 2.1626214437366345e-05,
      "loss": 0.0001,
      "step": 932
    },
    {
      "epoch": 0.7896323723882571,
      "grad_norm": 0.12579579651355743,
      "learning_rate": 2.1460010087415527e-05,
      "loss": 0.0002,
      "step": 933
    },
    {
      "epoch": 0.7904787093361545,
      "grad_norm": 0.03405604511499405,
      "learning_rate": 2.1294370070018076e-05,
      "loss": 0.0001,
      "step": 934
    },
    {
      "epoch": 0.7913250462840519,
      "grad_norm": 0.022147029638290405,
      "learning_rate": 2.1129295575345385e-05,
      "loss": 0.0001,
      "step": 935
    },
    {
      "epoch": 0.7921713832319492,
      "grad_norm": 0.2529233992099762,
      "learning_rate": 2.0964787789505523e-05,
      "loss": 0.0003,
      "step": 936
    },
    {
      "epoch": 0.7930177201798466,
      "grad_norm": 0.30088552832603455,
      "learning_rate": 2.0800847894534525e-05,
      "loss": 0.0004,
      "step": 937
    },
    {
      "epoch": 0.793864057127744,
      "grad_norm": 0.07343331724405289,
      "learning_rate": 2.0637477068387957e-05,
      "loss": 0.0001,
      "step": 938
    },
    {
      "epoch": 0.7947103940756414,
      "grad_norm": 0.03723829239606857,
      "learning_rate": 2.047467648493251e-05,
      "loss": 0.0001,
      "step": 939
    },
    {
      "epoch": 0.7955567310235387,
      "grad_norm": 0.09210750460624695,
      "learning_rate": 2.0312447313937387e-05,
      "loss": 0.0002,
      "step": 940
    },
    {
      "epoch": 0.7964030679714361,
      "grad_norm": 0.018357712775468826,
      "learning_rate": 2.015079072106626e-05,
      "loss": 0.0001,
      "step": 941
    },
    {
      "epoch": 0.7972494049193335,
      "grad_norm": 0.06646177917718887,
      "learning_rate": 1.9989707867868425e-05,
      "loss": 0.0001,
      "step": 942
    },
    {
      "epoch": 0.7980957418672309,
      "grad_norm": 0.017834974452853203,
      "learning_rate": 1.9829199911770812e-05,
      "loss": 0.0001,
      "step": 943
    },
    {
      "epoch": 0.7989420788151282,
      "grad_norm": 0.032671969383955,
      "learning_rate": 1.966926800606955e-05,
      "loss": 0.0001,
      "step": 944
    },
    {
      "epoch": 0.7997884157630256,
      "grad_norm": 0.23378333449363708,
      "learning_rate": 1.9509913299921644e-05,
      "loss": 0.0003,
      "step": 945
    },
    {
      "epoch": 0.800634752710923,
      "grad_norm": 3.638793468475342,
      "learning_rate": 1.9351136938336777e-05,
      "loss": 0.0038,
      "step": 946
    },
    {
      "epoch": 0.8014810896588204,
      "grad_norm": 0.024274148046970367,
      "learning_rate": 1.919294006216904e-05,
      "loss": 0.0001,
      "step": 947
    },
    {
      "epoch": 0.8023274266067179,
      "grad_norm": 0.1327366828918457,
      "learning_rate": 1.9035323808108775e-05,
      "loss": 0.0002,
      "step": 948
    },
    {
      "epoch": 0.8031737635546152,
      "grad_norm": 0.03646604344248772,
      "learning_rate": 1.8878289308674356e-05,
      "loss": 0.0001,
      "step": 949
    },
    {
      "epoch": 0.8040201005025126,
      "grad_norm": 0.055946316570043564,
      "learning_rate": 1.8721837692204115e-05,
      "loss": 0.0001,
      "step": 950
    },
    {
      "epoch": 0.80486643745041,
      "grad_norm": 0.1100902110338211,
      "learning_rate": 1.8565970082848173e-05,
      "loss": 0.0002,
      "step": 951
    },
    {
      "epoch": 0.8057127743983074,
      "grad_norm": 0.04728550836443901,
      "learning_rate": 1.8410687600560438e-05,
      "loss": 0.0001,
      "step": 952
    },
    {
      "epoch": 0.8065591113462047,
      "grad_norm": 0.16050080955028534,
      "learning_rate": 1.8255991361090408e-05,
      "loss": 0.0002,
      "step": 953
    },
    {
      "epoch": 0.8074054482941021,
      "grad_norm": 0.03241283819079399,
      "learning_rate": 1.8101882475975417e-05,
      "loss": 0.0001,
      "step": 954
    },
    {
      "epoch": 0.8082517852419995,
      "grad_norm": 0.04175794497132301,
      "learning_rate": 1.794836205253233e-05,
      "loss": 0.0001,
      "step": 955
    },
    {
      "epoch": 0.8090981221898969,
      "grad_norm": 0.13068246841430664,
      "learning_rate": 1.779543119384982e-05,
      "loss": 0.0002,
      "step": 956
    },
    {
      "epoch": 0.8099444591377942,
      "grad_norm": 0.6095410585403442,
      "learning_rate": 1.7643090998780367e-05,
      "loss": 0.0005,
      "step": 957
    },
    {
      "epoch": 0.8107907960856916,
      "grad_norm": 0.2803589999675751,
      "learning_rate": 1.7491342561932355e-05,
      "loss": 0.0003,
      "step": 958
    },
    {
      "epoch": 0.811637133033589,
      "grad_norm": 0.15300843119621277,
      "learning_rate": 1.734018697366221e-05,
      "loss": 0.0002,
      "step": 959
    },
    {
      "epoch": 0.8124834699814864,
      "grad_norm": 0.16421805322170258,
      "learning_rate": 1.7189625320066572e-05,
      "loss": 0.0002,
      "step": 960
    },
    {
      "epoch": 0.8133298069293837,
      "grad_norm": 0.41529351472854614,
      "learning_rate": 1.703965868297448e-05,
      "loss": 0.0004,
      "step": 961
    },
    {
      "epoch": 0.8141761438772811,
      "grad_norm": 0.07753802090883255,
      "learning_rate": 1.6890288139939625e-05,
      "loss": 0.0002,
      "step": 962
    },
    {
      "epoch": 0.8150224808251785,
      "grad_norm": 0.019801927730441093,
      "learning_rate": 1.674151476423258e-05,
      "loss": 0.0001,
      "step": 963
    },
    {
      "epoch": 0.815868817773076,
      "grad_norm": 0.12162253260612488,
      "learning_rate": 1.6593339624833094e-05,
      "loss": 0.0002,
      "step": 964
    },
    {
      "epoch": 0.8167151547209733,
      "grad_norm": 0.03215118870139122,
      "learning_rate": 1.644576378642244e-05,
      "loss": 0.0001,
      "step": 965
    },
    {
      "epoch": 0.8175614916688707,
      "grad_norm": 0.03967051953077316,
      "learning_rate": 1.6298788309375644e-05,
      "loss": 0.0001,
      "step": 966
    },
    {
      "epoch": 0.8184078286167681,
      "grad_norm": 0.046636469662189484,
      "learning_rate": 1.6152414249754132e-05,
      "loss": 0.0001,
      "step": 967
    },
    {
      "epoch": 0.8192541655646655,
      "grad_norm": 0.029653459787368774,
      "learning_rate": 1.600664265929782e-05,
      "loss": 0.0001,
      "step": 968
    },
    {
      "epoch": 0.8201005025125628,
      "grad_norm": 0.04834069684147835,
      "learning_rate": 1.5861474585417778e-05,
      "loss": 0.0001,
      "step": 969
    },
    {
      "epoch": 0.8209468394604602,
      "grad_norm": 0.03304549679160118,
      "learning_rate": 1.571691107118861e-05,
      "loss": 0.0001,
      "step": 970
    },
    {
      "epoch": 0.8217931764083576,
      "grad_norm": 0.03528258204460144,
      "learning_rate": 1.5572953155341017e-05,
      "loss": 0.0001,
      "step": 971
    },
    {
      "epoch": 0.822639513356255,
      "grad_norm": 0.06700039654970169,
      "learning_rate": 1.5429601872254273e-05,
      "loss": 0.0001,
      "step": 972
    },
    {
      "epoch": 0.8234858503041523,
      "grad_norm": 0.026155125349760056,
      "learning_rate": 1.528685825194882e-05,
      "loss": 0.0001,
      "step": 973
    },
    {
      "epoch": 0.8243321872520497,
      "grad_norm": 0.07379820197820663,
      "learning_rate": 1.5144723320078869e-05,
      "loss": 0.0002,
      "step": 974
    },
    {
      "epoch": 0.8251785241999471,
      "grad_norm": 0.07782360166311264,
      "learning_rate": 1.5003198097925052e-05,
      "loss": 0.0002,
      "step": 975
    },
    {
      "epoch": 0.8260248611478445,
      "grad_norm": 0.1054888591170311,
      "learning_rate": 1.4862283602387017e-05,
      "loss": 0.0002,
      "step": 976
    },
    {
      "epoch": 0.8268711980957418,
      "grad_norm": 0.13800279796123505,
      "learning_rate": 1.4721980845976202e-05,
      "loss": 0.0002,
      "step": 977
    },
    {
      "epoch": 0.8277175350436392,
      "grad_norm": 0.14023710787296295,
      "learning_rate": 1.4582290836808543e-05,
      "loss": 0.0002,
      "step": 978
    },
    {
      "epoch": 0.8285638719915366,
      "grad_norm": 0.039280444383621216,
      "learning_rate": 1.4443214578597075e-05,
      "loss": 0.0001,
      "step": 979
    },
    {
      "epoch": 0.829410208939434,
      "grad_norm": 0.22727468609809875,
      "learning_rate": 1.4304753070645083e-05,
      "loss": 0.0002,
      "step": 980
    },
    {
      "epoch": 0.8302565458873314,
      "grad_norm": 0.18732726573944092,
      "learning_rate": 1.4166907307838472e-05,
      "loss": 0.0002,
      "step": 981
    },
    {
      "epoch": 0.8311028828352288,
      "grad_norm": 0.052902307361364365,
      "learning_rate": 1.402967828063897e-05,
      "loss": 0.0001,
      "step": 982
    },
    {
      "epoch": 0.8319492197831262,
      "grad_norm": 0.0993070900440216,
      "learning_rate": 1.3893066975076829e-05,
      "loss": 0.0002,
      "step": 983
    },
    {
      "epoch": 0.8327955567310236,
      "grad_norm": 0.06284849345684052,
      "learning_rate": 1.3757074372743828e-05,
      "loss": 0.0001,
      "step": 984
    },
    {
      "epoch": 0.8336418936789209,
      "grad_norm": 0.0746096521615982,
      "learning_rate": 1.3621701450786162e-05,
      "loss": 0.0002,
      "step": 985
    },
    {
      "epoch": 0.8344882306268183,
      "grad_norm": 0.15257956087589264,
      "learning_rate": 1.3486949181897434e-05,
      "loss": 0.0002,
      "step": 986
    },
    {
      "epoch": 0.8353345675747157,
      "grad_norm": 0.0508284792304039,
      "learning_rate": 1.3352818534311695e-05,
      "loss": 0.0001,
      "step": 987
    },
    {
      "epoch": 0.8361809045226131,
      "grad_norm": 0.05081188678741455,
      "learning_rate": 1.3219310471796453e-05,
      "loss": 0.0001,
      "step": 988
    },
    {
      "epoch": 0.8370272414705104,
      "grad_norm": 0.05635140463709831,
      "learning_rate": 1.3086425953645787e-05,
      "loss": 0.0001,
      "step": 989
    },
    {
      "epoch": 0.8378735784184078,
      "grad_norm": 0.05115867406129837,
      "learning_rate": 1.295416593467338e-05,
      "loss": 0.0001,
      "step": 990
    },
    {
      "epoch": 0.8387199153663052,
      "grad_norm": 0.08419815450906754,
      "learning_rate": 1.2822531365205781e-05,
      "loss": 0.0002,
      "step": 991
    },
    {
      "epoch": 0.8395662523142026,
      "grad_norm": 0.048064954578876495,
      "learning_rate": 1.2691523191075395e-05,
      "loss": 0.0001,
      "step": 992
    },
    {
      "epoch": 0.8404125892620999,
      "grad_norm": 0.2124854326248169,
      "learning_rate": 1.2561142353613942e-05,
      "loss": 0.0002,
      "step": 993
    },
    {
      "epoch": 0.8412589262099973,
      "grad_norm": 0.032767366617918015,
      "learning_rate": 1.2431389789645399e-05,
      "loss": 0.0001,
      "step": 994
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.03317883610725403,
      "learning_rate": 1.2302266431479503e-05,
      "loss": 0.0001,
      "step": 995
    },
    {
      "epoch": 0.8429516001057921,
      "grad_norm": 0.10786723345518112,
      "learning_rate": 1.217377320690496e-05,
      "loss": 0.0001,
      "step": 996
    },
    {
      "epoch": 0.8437979370536896,
      "grad_norm": 0.03594963252544403,
      "learning_rate": 1.2045911039182734e-05,
      "loss": 0.0001,
      "step": 997
    },
    {
      "epoch": 0.8446442740015869,
      "grad_norm": 0.041848160326480865,
      "learning_rate": 1.1918680847039554e-05,
      "loss": 0.0001,
      "step": 998
    },
    {
      "epoch": 0.8454906109494843,
      "grad_norm": 0.9690784811973572,
      "learning_rate": 1.1792083544661115e-05,
      "loss": 0.0008,
      "step": 999
    },
    {
      "epoch": 0.8463369478973817,
      "grad_norm": 0.21381966769695282,
      "learning_rate": 1.166612004168568e-05,
      "loss": 0.0002,
      "step": 1000
    },
    {
      "epoch": 0.847183284845279,
      "grad_norm": 0.039850324392318726,
      "learning_rate": 1.1540791243197469e-05,
      "loss": 0.0001,
      "step": 1001
    },
    {
      "epoch": 0.8480296217931764,
      "grad_norm": 0.04686881601810455,
      "learning_rate": 1.141609804972017e-05,
      "loss": 0.0001,
      "step": 1002
    },
    {
      "epoch": 0.8488759587410738,
      "grad_norm": 0.1030183807015419,
      "learning_rate": 1.1292041357210436e-05,
      "loss": 0.0002,
      "step": 1003
    },
    {
      "epoch": 0.8497222956889712,
      "grad_norm": 3.8446195125579834,
      "learning_rate": 1.1168622057051547e-05,
      "loss": 0.0014,
      "step": 1004
    },
    {
      "epoch": 0.8505686326368685,
      "grad_norm": 0.0340881273150444,
      "learning_rate": 1.104584103604681e-05,
      "loss": 0.0001,
      "step": 1005
    },
    {
      "epoch": 0.8514149695847659,
      "grad_norm": 0.0387510247528553,
      "learning_rate": 1.0923699176413448e-05,
      "loss": 0.0001,
      "step": 1006
    },
    {
      "epoch": 0.8522613065326633,
      "grad_norm": 0.04323866590857506,
      "learning_rate": 1.0802197355776055e-05,
      "loss": 0.0001,
      "step": 1007
    },
    {
      "epoch": 0.8531076434805607,
      "grad_norm": 0.04572153091430664,
      "learning_rate": 1.0681336447160362e-05,
      "loss": 0.0001,
      "step": 1008
    },
    {
      "epoch": 0.853953980428458,
      "grad_norm": 0.07120076566934586,
      "learning_rate": 1.056111731898699e-05,
      "loss": 0.0002,
      "step": 1009
    },
    {
      "epoch": 0.8548003173763554,
      "grad_norm": 0.028867924585938454,
      "learning_rate": 1.04415408350651e-05,
      "loss": 0.0001,
      "step": 1010
    },
    {
      "epoch": 0.8556466543242528,
      "grad_norm": 0.06583672016859055,
      "learning_rate": 1.0322607854586409e-05,
      "loss": 0.0001,
      "step": 1011
    },
    {
      "epoch": 0.8564929912721502,
      "grad_norm": 0.0212421752512455,
      "learning_rate": 1.020431923211873e-05,
      "loss": 0.0001,
      "step": 1012
    },
    {
      "epoch": 0.8573393282200477,
      "grad_norm": 0.0732201635837555,
      "learning_rate": 1.0086675817600078e-05,
      "loss": 0.0002,
      "step": 1013
    },
    {
      "epoch": 0.858185665167945,
      "grad_norm": 0.050640542060136795,
      "learning_rate": 9.969678456332409e-06,
      "loss": 0.0001,
      "step": 1014
    },
    {
      "epoch": 0.8590320021158424,
      "grad_norm": 0.020134979858994484,
      "learning_rate": 9.853327988975625e-06,
      "loss": 0.0001,
      "step": 1015
    },
    {
      "epoch": 0.8598783390637398,
      "grad_norm": 0.04244556277990341,
      "learning_rate": 9.737625251541504e-06,
      "loss": 0.0001,
      "step": 1016
    },
    {
      "epoch": 0.8607246760116372,
      "grad_norm": 1.2951719760894775,
      "learning_rate": 9.622571075387698e-06,
      "loss": 0.0008,
      "step": 1017
    },
    {
      "epoch": 0.8615710129595345,
      "grad_norm": 0.0249020978808403,
      "learning_rate": 9.508166287211739e-06,
      "loss": 0.0001,
      "step": 1018
    },
    {
      "epoch": 0.8624173499074319,
      "grad_norm": 0.045827802270650864,
      "learning_rate": 9.394411709045137e-06,
      "loss": 0.0001,
      "step": 1019
    },
    {
      "epoch": 0.8632636868553293,
      "grad_norm": 0.02237636409699917,
      "learning_rate": 9.28130815824746e-06,
      "loss": 0.0001,
      "step": 1020
    },
    {
      "epoch": 0.8641100238032267,
      "grad_norm": 0.0756639912724495,
      "learning_rate": 9.16885644750044e-06,
      "loss": 0.0002,
      "step": 1021
    },
    {
      "epoch": 0.864956360751124,
      "grad_norm": 0.05225581303238869,
      "learning_rate": 9.057057384802181e-06,
      "loss": 0.0001,
      "step": 1022
    },
    {
      "epoch": 0.8658026976990214,
      "grad_norm": 0.18432854115962982,
      "learning_rate": 8.945911773461234e-06,
      "loss": 0.0002,
      "step": 1023
    },
    {
      "epoch": 0.8666490346469188,
      "grad_norm": 0.03696737065911293,
      "learning_rate": 8.835420412091066e-06,
      "loss": 0.0001,
      "step": 1024
    },
    {
      "epoch": 0.8674953715948162,
      "grad_norm": 0.26212799549102783,
      "learning_rate": 8.725584094604012e-06,
      "loss": 0.0003,
      "step": 1025
    },
    {
      "epoch": 0.8683417085427135,
      "grad_norm": 0.29439830780029297,
      "learning_rate": 8.616403610205814e-06,
      "loss": 0.0002,
      "step": 1026
    },
    {
      "epoch": 0.8691880454906109,
      "grad_norm": 0.05446379631757736,
      "learning_rate": 8.507879743389846e-06,
      "loss": 0.0001,
      "step": 1027
    },
    {
      "epoch": 0.8700343824385083,
      "grad_norm": 0.04564208164811134,
      "learning_rate": 8.400013273931485e-06,
      "loss": 0.0001,
      "step": 1028
    },
    {
      "epoch": 0.8708807193864058,
      "grad_norm": 0.09591326117515564,
      "learning_rate": 8.29280497688254e-06,
      "loss": 0.0002,
      "step": 1029
    },
    {
      "epoch": 0.8717270563343031,
      "grad_norm": 0.04426273703575134,
      "learning_rate": 8.186255622565642e-06,
      "loss": 0.0001,
      "step": 1030
    },
    {
      "epoch": 0.8725733932822005,
      "grad_norm": 0.03719878941774368,
      "learning_rate": 8.080365976568728e-06,
      "loss": 0.0001,
      "step": 1031
    },
    {
      "epoch": 0.8734197302300979,
      "grad_norm": 0.05031093955039978,
      "learning_rate": 7.975136799739546e-06,
      "loss": 0.0001,
      "step": 1032
    },
    {
      "epoch": 0.8742660671779953,
      "grad_norm": 0.05599454790353775,
      "learning_rate": 7.870568848180182e-06,
      "loss": 0.0001,
      "step": 1033
    },
    {
      "epoch": 0.8751124041258926,
      "grad_norm": 0.06366157531738281,
      "learning_rate": 7.766662873241614e-06,
      "loss": 0.0001,
      "step": 1034
    },
    {
      "epoch": 0.87595874107379,
      "grad_norm": 0.13091598451137543,
      "learning_rate": 7.663419621518376e-06,
      "loss": 0.0002,
      "step": 1035
    },
    {
      "epoch": 0.8768050780216874,
      "grad_norm": 0.3583017587661743,
      "learning_rate": 7.560839834843037e-06,
      "loss": 0.0003,
      "step": 1036
    },
    {
      "epoch": 0.8776514149695848,
      "grad_norm": 0.2321922481060028,
      "learning_rate": 7.458924250281119e-06,
      "loss": 0.0002,
      "step": 1037
    },
    {
      "epoch": 0.8784977519174821,
      "grad_norm": 0.09537674486637115,
      "learning_rate": 7.357673600125525e-06,
      "loss": 0.0002,
      "step": 1038
    },
    {
      "epoch": 0.8793440888653795,
      "grad_norm": 0.0997953936457634,
      "learning_rate": 7.257088611891516e-06,
      "loss": 0.0002,
      "step": 1039
    },
    {
      "epoch": 0.8801904258132769,
      "grad_norm": 0.03599618375301361,
      "learning_rate": 7.1571700083113205e-06,
      "loss": 0.0001,
      "step": 1040
    },
    {
      "epoch": 0.8810367627611743,
      "grad_norm": 0.08133365213871002,
      "learning_rate": 7.057918507329009e-06,
      "loss": 0.0001,
      "step": 1041
    },
    {
      "epoch": 0.8818830997090716,
      "grad_norm": 0.21962670981884003,
      "learning_rate": 6.959334822095354e-06,
      "loss": 0.0001,
      "step": 1042
    },
    {
      "epoch": 0.882729436656969,
      "grad_norm": 0.1007503941655159,
      "learning_rate": 6.861419660962643e-06,
      "loss": 0.0002,
      "step": 1043
    },
    {
      "epoch": 0.8835757736048664,
      "grad_norm": 0.0671425312757492,
      "learning_rate": 6.764173727479662e-06,
      "loss": 0.0001,
      "step": 1044
    },
    {
      "epoch": 0.8844221105527639,
      "grad_norm": 0.08273983746767044,
      "learning_rate": 6.667597720386553e-06,
      "loss": 0.0002,
      "step": 1045
    },
    {
      "epoch": 0.8852684475006612,
      "grad_norm": 0.03656056150794029,
      "learning_rate": 6.571692333609891e-06,
      "loss": 0.0001,
      "step": 1046
    },
    {
      "epoch": 0.8861147844485586,
      "grad_norm": 0.07200013846158981,
      "learning_rate": 6.4764582562576225e-06,
      "loss": 0.0002,
      "step": 1047
    },
    {
      "epoch": 0.886961121396456,
      "grad_norm": 0.1970086544752121,
      "learning_rate": 6.381896172614177e-06,
      "loss": 0.0002,
      "step": 1048
    },
    {
      "epoch": 0.8878074583443534,
      "grad_norm": 0.08710719645023346,
      "learning_rate": 6.288006762135423e-06,
      "loss": 0.0001,
      "step": 1049
    },
    {
      "epoch": 0.8886537952922507,
      "grad_norm": 0.10266397893428802,
      "learning_rate": 6.1947906994440195e-06,
      "loss": 0.0002,
      "step": 1050
    },
    {
      "epoch": 0.8895001322401481,
      "grad_norm": 0.0269624013453722,
      "learning_rate": 6.102248654324294e-06,
      "loss": 0.0001,
      "step": 1051
    },
    {
      "epoch": 0.8903464691880455,
      "grad_norm": 0.06637012213468552,
      "learning_rate": 6.010381291717637e-06,
      "loss": 0.0002,
      "step": 1052
    },
    {
      "epoch": 0.8911928061359429,
      "grad_norm": 0.21953462064266205,
      "learning_rate": 5.919189271717629e-06,
      "loss": 0.0002,
      "step": 1053
    },
    {
      "epoch": 0.8920391430838402,
      "grad_norm": 67.29975891113281,
      "learning_rate": 5.8286732495653196e-06,
      "loss": 0.0368,
      "step": 1054
    },
    {
      "epoch": 0.8928854800317376,
      "grad_norm": 0.04297763481736183,
      "learning_rate": 5.738833875644523e-06,
      "loss": 0.0001,
      "step": 1055
    },
    {
      "epoch": 0.893731816979635,
      "grad_norm": 0.12288635969161987,
      "learning_rate": 5.649671795477129e-06,
      "loss": 0.0002,
      "step": 1056
    },
    {
      "epoch": 0.8945781539275324,
      "grad_norm": 2.0618977546691895,
      "learning_rate": 5.561187649718491e-06,
      "loss": 0.0005,
      "step": 1057
    },
    {
      "epoch": 0.8954244908754297,
      "grad_norm": 0.49036017060279846,
      "learning_rate": 5.4733820741528e-06,
      "loss": 0.0003,
      "step": 1058
    },
    {
      "epoch": 0.8962708278233271,
      "grad_norm": 0.09468761831521988,
      "learning_rate": 5.386255699688525e-06,
      "loss": 0.0002,
      "step": 1059
    },
    {
      "epoch": 0.8971171647712245,
      "grad_norm": 0.07483547180891037,
      "learning_rate": 5.299809152353863e-06,
      "loss": 0.0002,
      "step": 1060
    },
    {
      "epoch": 0.8979635017191219,
      "grad_norm": 0.04199570044875145,
      "learning_rate": 5.214043053292295e-06,
      "loss": 0.0001,
      "step": 1061
    },
    {
      "epoch": 0.8988098386670194,
      "grad_norm": 0.033658865839242935,
      "learning_rate": 5.128958018758012e-06,
      "loss": 0.0001,
      "step": 1062
    },
    {
      "epoch": 0.8996561756149167,
      "grad_norm": 0.10445345938205719,
      "learning_rate": 5.044554660111666e-06,
      "loss": 0.0002,
      "step": 1063
    },
    {
      "epoch": 0.9005025125628141,
      "grad_norm": 0.15214763581752777,
      "learning_rate": 4.960833583815772e-06,
      "loss": 0.0002,
      "step": 1064
    },
    {
      "epoch": 0.9013488495107115,
      "grad_norm": 0.044543132185935974,
      "learning_rate": 4.877795391430506e-06,
      "loss": 0.0001,
      "step": 1065
    },
    {
      "epoch": 0.9021951864586089,
      "grad_norm": 0.06599655747413635,
      "learning_rate": 4.795440679609298e-06,
      "loss": 0.0001,
      "step": 1066
    },
    {
      "epoch": 0.9030415234065062,
      "grad_norm": 0.04786626994609833,
      "learning_rate": 4.713770040094567e-06,
      "loss": 0.0001,
      "step": 1067
    },
    {
      "epoch": 0.9038878603544036,
      "grad_norm": 0.05292282998561859,
      "learning_rate": 4.632784059713535e-06,
      "loss": 0.0001,
      "step": 1068
    },
    {
      "epoch": 0.904734197302301,
      "grad_norm": 0.07457330822944641,
      "learning_rate": 4.552483320373868e-06,
      "loss": 0.0001,
      "step": 1069
    },
    {
      "epoch": 0.9055805342501984,
      "grad_norm": 0.02598341554403305,
      "learning_rate": 4.472868399059626e-06,
      "loss": 0.0001,
      "step": 1070
    },
    {
      "epoch": 0.9064268711980957,
      "grad_norm": 0.04329909756779671,
      "learning_rate": 4.393939867827046e-06,
      "loss": 0.0001,
      "step": 1071
    },
    {
      "epoch": 0.9072732081459931,
      "grad_norm": 0.06844758242368698,
      "learning_rate": 4.315698293800475e-06,
      "loss": 0.0001,
      "step": 1072
    },
    {
      "epoch": 0.9081195450938905,
      "grad_norm": 0.030138617381453514,
      "learning_rate": 4.2381442391682514e-06,
      "loss": 0.0001,
      "step": 1073
    },
    {
      "epoch": 0.9089658820417879,
      "grad_norm": 0.8961422443389893,
      "learning_rate": 4.161278261178714e-06,
      "loss": 0.0006,
      "step": 1074
    },
    {
      "epoch": 0.9098122189896852,
      "grad_norm": 0.07998895645141602,
      "learning_rate": 4.085100912136131e-06,
      "loss": 0.0001,
      "step": 1075
    },
    {
      "epoch": 0.9106585559375826,
      "grad_norm": 0.054003313183784485,
      "learning_rate": 4.009612739396817e-06,
      "loss": 0.0001,
      "step": 1076
    },
    {
      "epoch": 0.91150489288548,
      "grad_norm": 0.017176540568470955,
      "learning_rate": 3.93481428536513e-06,
      "loss": 0.0001,
      "step": 1077
    },
    {
      "epoch": 0.9123512298333775,
      "grad_norm": 0.013484030961990356,
      "learning_rate": 3.860706087489607e-06,
      "loss": 0.0001,
      "step": 1078
    },
    {
      "epoch": 0.9131975667812748,
      "grad_norm": 0.03977302834391594,
      "learning_rate": 3.7872886782591154e-06,
      "loss": 0.0001,
      "step": 1079
    },
    {
      "epoch": 0.9140439037291722,
      "grad_norm": 0.04018683359026909,
      "learning_rate": 3.714562585198922e-06,
      "loss": 0.0001,
      "step": 1080
    },
    {
      "epoch": 0.9148902406770696,
      "grad_norm": 0.05887274816632271,
      "learning_rate": 3.6425283308671232e-06,
      "loss": 0.0001,
      "step": 1081
    },
    {
      "epoch": 0.915736577624967,
      "grad_norm": 0.6434818506240845,
      "learning_rate": 3.571186432850626e-06,
      "loss": 0.0006,
      "step": 1082
    },
    {
      "epoch": 0.9165829145728643,
      "grad_norm": 0.15346260368824005,
      "learning_rate": 3.500537403761628e-06,
      "loss": 0.0002,
      "step": 1083
    },
    {
      "epoch": 0.9174292515207617,
      "grad_norm": 0.033555272966623306,
      "learning_rate": 3.4305817512338525e-06,
      "loss": 0.0001,
      "step": 1084
    },
    {
      "epoch": 0.9182755884686591,
      "grad_norm": 0.039782192558050156,
      "learning_rate": 3.3613199779188976e-06,
      "loss": 0.0001,
      "step": 1085
    },
    {
      "epoch": 0.9191219254165565,
      "grad_norm": 0.15752553939819336,
      "learning_rate": 3.29275258148265e-06,
      "loss": 0.0002,
      "step": 1086
    },
    {
      "epoch": 0.9199682623644538,
      "grad_norm": 0.08154158294200897,
      "learning_rate": 3.2248800546016866e-06,
      "loss": 0.0001,
      "step": 1087
    },
    {
      "epoch": 0.9208145993123512,
      "grad_norm": 0.05125689134001732,
      "learning_rate": 3.1577028849597567e-06,
      "loss": 0.0001,
      "step": 1088
    },
    {
      "epoch": 0.9216609362602486,
      "grad_norm": 0.40674006938934326,
      "learning_rate": 3.0912215552442725e-06,
      "loss": 0.0003,
      "step": 1089
    },
    {
      "epoch": 0.922507273208146,
      "grad_norm": 0.025004560127854347,
      "learning_rate": 3.0254365431428013e-06,
      "loss": 0.0001,
      "step": 1090
    },
    {
      "epoch": 0.9233536101560433,
      "grad_norm": 0.027129800990223885,
      "learning_rate": 2.960348321339712e-06,
      "loss": 0.0001,
      "step": 1091
    },
    {
      "epoch": 0.9241999471039407,
      "grad_norm": 0.10096831619739532,
      "learning_rate": 2.895957357512713e-06,
      "loss": 0.0002,
      "step": 1092
    },
    {
      "epoch": 0.9250462840518381,
      "grad_norm": 0.060557637363672256,
      "learning_rate": 2.8322641143294747e-06,
      "loss": 0.0001,
      "step": 1093
    },
    {
      "epoch": 0.9258926209997356,
      "grad_norm": 0.01941130869090557,
      "learning_rate": 2.7692690494444227e-06,
      "loss": 0.0001,
      "step": 1094
    },
    {
      "epoch": 0.926738957947633,
      "grad_norm": 0.143899604678154,
      "learning_rate": 2.7069726154953067e-06,
      "loss": 0.0002,
      "step": 1095
    },
    {
      "epoch": 0.9275852948955303,
      "grad_norm": 0.05427510291337967,
      "learning_rate": 2.6453752601000246e-06,
      "loss": 0.0001,
      "step": 1096
    },
    {
      "epoch": 0.9284316318434277,
      "grad_norm": 0.03041023574769497,
      "learning_rate": 2.5844774258534155e-06,
      "loss": 0.0001,
      "step": 1097
    },
    {
      "epoch": 0.9292779687913251,
      "grad_norm": 0.09313998371362686,
      "learning_rate": 2.524279550324027e-06,
      "loss": 0.0002,
      "step": 1098
    },
    {
      "epoch": 0.9301243057392224,
      "grad_norm": 0.09529302269220352,
      "learning_rate": 2.464782066051008e-06,
      "loss": 0.0002,
      "step": 1099
    },
    {
      "epoch": 0.9309706426871198,
      "grad_norm": 0.14089873433113098,
      "learning_rate": 2.405985400541022e-06,
      "loss": 0.0002,
      "step": 1100
    },
    {
      "epoch": 0.9318169796350172,
      "grad_norm": 0.14107897877693176,
      "learning_rate": 2.3478899762650943e-06,
      "loss": 0.0002,
      "step": 1101
    },
    {
      "epoch": 0.9326633165829146,
      "grad_norm": 0.07372616231441498,
      "learning_rate": 2.2904962106556793e-06,
      "loss": 0.0001,
      "step": 1102
    },
    {
      "epoch": 0.933509653530812,
      "grad_norm": 0.06261898577213287,
      "learning_rate": 2.2338045161035883e-06,
      "loss": 0.0002,
      "step": 1103
    },
    {
      "epoch": 0.9343559904787093,
      "grad_norm": 0.5789007544517517,
      "learning_rate": 2.177815299955044e-06,
      "loss": 0.0004,
      "step": 1104
    },
    {
      "epoch": 0.9352023274266067,
      "grad_norm": 0.0687805637717247,
      "learning_rate": 2.122528964508774e-06,
      "loss": 0.0001,
      "step": 1105
    },
    {
      "epoch": 0.9360486643745041,
      "grad_norm": 0.0731336921453476,
      "learning_rate": 2.067945907013069e-06,
      "loss": 0.0002,
      "step": 1106
    },
    {
      "epoch": 0.9368950013224014,
      "grad_norm": 0.07129161059856415,
      "learning_rate": 2.014066519663027e-06,
      "loss": 0.0001,
      "step": 1107
    },
    {
      "epoch": 0.9377413382702988,
      "grad_norm": 0.0385940745472908,
      "learning_rate": 1.9608911895976134e-06,
      "loss": 0.0001,
      "step": 1108
    },
    {
      "epoch": 0.9385876752181962,
      "grad_norm": 0.039715465158224106,
      "learning_rate": 1.9084202988969735e-06,
      "loss": 0.0001,
      "step": 1109
    },
    {
      "epoch": 0.9394340121660937,
      "grad_norm": 0.113487608730793,
      "learning_rate": 1.8566542245796347e-06,
      "loss": 0.0002,
      "step": 1110
    },
    {
      "epoch": 0.940280349113991,
      "grad_norm": 0.37704917788505554,
      "learning_rate": 1.8055933385998091e-06,
      "loss": 0.0004,
      "step": 1111
    },
    {
      "epoch": 0.9411266860618884,
      "grad_norm": 0.026006758213043213,
      "learning_rate": 1.7552380078447505e-06,
      "loss": 0.0001,
      "step": 1112
    },
    {
      "epoch": 0.9419730230097858,
      "grad_norm": 0.06526373326778412,
      "learning_rate": 1.7055885941320793e-06,
      "loss": 0.0001,
      "step": 1113
    },
    {
      "epoch": 0.9428193599576832,
      "grad_norm": 0.1013353019952774,
      "learning_rate": 1.6566454542071951e-06,
      "loss": 0.0001,
      "step": 1114
    },
    {
      "epoch": 0.9436656969055806,
      "grad_norm": 0.024974433705210686,
      "learning_rate": 1.6084089397407355e-06,
      "loss": 0.0001,
      "step": 1115
    },
    {
      "epoch": 0.9445120338534779,
      "grad_norm": 0.06891551613807678,
      "learning_rate": 1.5608793973259873e-06,
      "loss": 0.0001,
      "step": 1116
    },
    {
      "epoch": 0.9453583708013753,
      "grad_norm": 0.2101244330406189,
      "learning_rate": 1.5140571684764792e-06,
      "loss": 0.0002,
      "step": 1117
    },
    {
      "epoch": 0.9462047077492727,
      "grad_norm": 0.03794321045279503,
      "learning_rate": 1.4679425896234833e-06,
      "loss": 0.0001,
      "step": 1118
    },
    {
      "epoch": 0.94705104469717,
      "grad_norm": 0.06963124126195908,
      "learning_rate": 1.4225359921135717e-06,
      "loss": 0.0002,
      "step": 1119
    },
    {
      "epoch": 0.9478973816450674,
      "grad_norm": 0.05082397162914276,
      "learning_rate": 1.3778377022063084e-06,
      "loss": 0.0001,
      "step": 1120
    },
    {
      "epoch": 0.9487437185929648,
      "grad_norm": 0.06863516569137573,
      "learning_rate": 1.333848041071828e-06,
      "loss": 0.0001,
      "step": 1121
    },
    {
      "epoch": 0.9495900555408622,
      "grad_norm": 0.10396652668714523,
      "learning_rate": 1.2905673247885718e-06,
      "loss": 0.0002,
      "step": 1122
    },
    {
      "epoch": 0.9504363924887596,
      "grad_norm": 0.04176372289657593,
      "learning_rate": 1.2479958643410006e-06,
      "loss": 0.0001,
      "step": 1123
    },
    {
      "epoch": 0.9512827294366569,
      "grad_norm": 0.050088297575712204,
      "learning_rate": 1.2061339656174064e-06,
      "loss": 0.0001,
      "step": 1124
    },
    {
      "epoch": 0.9521290663845543,
      "grad_norm": 0.03956186771392822,
      "learning_rate": 1.1649819294076158e-06,
      "loss": 0.0001,
      "step": 1125
    },
    {
      "epoch": 0.9529754033324518,
      "grad_norm": 0.05234998092055321,
      "learning_rate": 1.1245400514009351e-06,
      "loss": 0.0001,
      "step": 1126
    },
    {
      "epoch": 0.9538217402803492,
      "grad_norm": 0.05241256207227707,
      "learning_rate": 1.0848086221839416e-06,
      "loss": 0.0001,
      "step": 1127
    },
    {
      "epoch": 0.9546680772282465,
      "grad_norm": 0.22163714468479156,
      "learning_rate": 1.045787927238462e-06,
      "loss": 0.0002,
      "step": 1128
    },
    {
      "epoch": 0.9555144141761439,
      "grad_norm": 0.10826706886291504,
      "learning_rate": 1.0074782469394861e-06,
      "loss": 0.0002,
      "step": 1129
    },
    {
      "epoch": 0.9563607511240413,
      "grad_norm": 0.02859027124941349,
      "learning_rate": 9.698798565531464e-07,
      "loss": 0.0001,
      "step": 1130
    },
    {
      "epoch": 0.9572070880719387,
      "grad_norm": 0.020412441343069077,
      "learning_rate": 9.329930262347519e-07,
      "loss": 0.0001,
      "step": 1131
    },
    {
      "epoch": 0.958053425019836,
      "grad_norm": 0.18546128273010254,
      "learning_rate": 8.968180210268462e-07,
      "loss": 0.0002,
      "step": 1132
    },
    {
      "epoch": 0.9588997619677334,
      "grad_norm": 0.04801074042916298,
      "learning_rate": 8.613551008573306e-07,
      "loss": 0.0001,
      "step": 1133
    },
    {
      "epoch": 0.9597460989156308,
      "grad_norm": 0.045858047902584076,
      "learning_rate": 8.266045205375328e-07,
      "loss": 0.0001,
      "step": 1134
    },
    {
      "epoch": 0.9605924358635282,
      "grad_norm": 0.042931508272886276,
      "learning_rate": 7.925665297604301e-07,
      "loss": 0.0001,
      "step": 1135
    },
    {
      "epoch": 0.9614387728114255,
      "grad_norm": 0.1204577311873436,
      "learning_rate": 7.592413730988291e-07,
      "loss": 0.0001,
      "step": 1136
    },
    {
      "epoch": 0.9622851097593229,
      "grad_norm": 0.13001196086406708,
      "learning_rate": 7.266292900036331e-07,
      "loss": 0.0001,
      "step": 1137
    },
    {
      "epoch": 0.9631314467072203,
      "grad_norm": 0.04208778962492943,
      "learning_rate": 6.947305148020889e-07,
      "loss": 0.0001,
      "step": 1138
    },
    {
      "epoch": 0.9639777836551177,
      "grad_norm": 0.04732511192560196,
      "learning_rate": 6.635452766961314e-07,
      "loss": 0.0001,
      "step": 1139
    },
    {
      "epoch": 0.964824120603015,
      "grad_norm": 0.3013063073158264,
      "learning_rate": 6.330737997607195e-07,
      "loss": 0.0004,
      "step": 1140
    },
    {
      "epoch": 0.9656704575509124,
      "grad_norm": 0.02021421305835247,
      "learning_rate": 6.033163029422361e-07,
      "loss": 0.0001,
      "step": 1141
    },
    {
      "epoch": 0.9665167944988098,
      "grad_norm": 0.08578869700431824,
      "learning_rate": 5.742730000568908e-07,
      "loss": 0.0001,
      "step": 1142
    },
    {
      "epoch": 0.9673631314467073,
      "grad_norm": 0.04271700233221054,
      "learning_rate": 5.459440997892307e-07,
      "loss": 0.0001,
      "step": 1143
    },
    {
      "epoch": 0.9682094683946046,
      "grad_norm": 0.16835899651050568,
      "learning_rate": 5.183298056906094e-07,
      "loss": 0.0002,
      "step": 1144
    },
    {
      "epoch": 0.969055805342502,
      "grad_norm": 0.0224732868373394,
      "learning_rate": 4.91430316177699e-07,
      "loss": 0.0001,
      "step": 1145
    },
    {
      "epoch": 0.9699021422903994,
      "grad_norm": 0.0165510643273592,
      "learning_rate": 4.652458245311242e-07,
      "loss": 0.0001,
      "step": 1146
    },
    {
      "epoch": 0.9707484792382968,
      "grad_norm": 0.04145815595984459,
      "learning_rate": 4.397765188940417e-07,
      "loss": 0.0001,
      "step": 1147
    },
    {
      "epoch": 0.9715948161861941,
      "grad_norm": 0.031001733615994453,
      "learning_rate": 4.1502258227077427e-07,
      "loss": 0.0001,
      "step": 1148
    },
    {
      "epoch": 0.9724411531340915,
      "grad_norm": 0.04481024295091629,
      "learning_rate": 3.909841925255231e-07,
      "loss": 0.0001,
      "step": 1149
    },
    {
      "epoch": 0.9732874900819889,
      "grad_norm": 0.25295722484588623,
      "learning_rate": 3.6766152238106865e-07,
      "loss": 0.0004,
      "step": 1150
    },
    {
      "epoch": 0.9741338270298863,
      "grad_norm": 0.13082540035247803,
      "learning_rate": 3.450547394175496e-07,
      "loss": 0.0002,
      "step": 1151
    },
    {
      "epoch": 0.9749801639777836,
      "grad_norm": 1.5636310577392578,
      "learning_rate": 3.231640060712415e-07,
      "loss": 0.0003,
      "step": 1152
    },
    {
      "epoch": 0.975826500925681,
      "grad_norm": 0.03289796784520149,
      "learning_rate": 3.0198947963340217e-07,
      "loss": 0.0001,
      "step": 1153
    },
    {
      "epoch": 0.9766728378735784,
      "grad_norm": 0.05520953610539436,
      "learning_rate": 2.81531312249117e-07,
      "loss": 0.0001,
      "step": 1154
    },
    {
      "epoch": 0.9775191748214758,
      "grad_norm": 0.04525551199913025,
      "learning_rate": 2.617896509162554e-07,
      "loss": 0.0001,
      "step": 1155
    },
    {
      "epoch": 0.9783655117693731,
      "grad_norm": 0.035422325134277344,
      "learning_rate": 2.427646374843495e-07,
      "loss": 0.0001,
      "step": 1156
    },
    {
      "epoch": 0.9792118487172705,
      "grad_norm": 0.0195600725710392,
      "learning_rate": 2.2445640865363937e-07,
      "loss": 0.0001,
      "step": 1157
    },
    {
      "epoch": 0.9800581856651679,
      "grad_norm": 0.04410228133201599,
      "learning_rate": 2.0686509597404037e-07,
      "loss": 0.0001,
      "step": 1158
    },
    {
      "epoch": 0.9809045226130654,
      "grad_norm": 0.14861729741096497,
      "learning_rate": 1.899908258442329e-07,
      "loss": 0.0002,
      "step": 1159
    },
    {
      "epoch": 0.9817508595609628,
      "grad_norm": 0.31148141622543335,
      "learning_rate": 1.738337195107187e-07,
      "loss": 0.0002,
      "step": 1160
    },
    {
      "epoch": 0.9825971965088601,
      "grad_norm": 0.04191156476736069,
      "learning_rate": 1.5839389306698815e-07,
      "loss": 0.0001,
      "step": 1161
    },
    {
      "epoch": 0.9834435334567575,
      "grad_norm": 0.06189923733472824,
      "learning_rate": 1.436714574526543e-07,
      "loss": 0.0001,
      "step": 1162
    },
    {
      "epoch": 0.9842898704046549,
      "grad_norm": 0.06969114392995834,
      "learning_rate": 1.2966651845268685e-07,
      "loss": 0.0001,
      "step": 1163
    },
    {
      "epoch": 0.9851362073525523,
      "grad_norm": 0.03229764476418495,
      "learning_rate": 1.1637917669662379e-07,
      "loss": 0.0001,
      "step": 1164
    },
    {
      "epoch": 0.9859825443004496,
      "grad_norm": 0.07409042119979858,
      "learning_rate": 1.0380952765786101e-07,
      "loss": 0.0002,
      "step": 1165
    },
    {
      "epoch": 0.986828881248347,
      "grad_norm": 0.11478302627801895,
      "learning_rate": 9.195766165295272e-08,
      "loss": 0.0002,
      "step": 1166
    },
    {
      "epoch": 0.9876752181962444,
      "grad_norm": 0.05197848379611969,
      "learning_rate": 8.082366384098982e-08,
      "loss": 0.0001,
      "step": 1167
    },
    {
      "epoch": 0.9885215551441418,
      "grad_norm": 0.05330824479460716,
      "learning_rate": 7.040761422297815e-08,
      "loss": 0.0001,
      "step": 1168
    },
    {
      "epoch": 0.9893678920920391,
      "grad_norm": 0.02525501511991024,
      "learning_rate": 6.070958764125001e-08,
      "loss": 0.0001,
      "step": 1169
    },
    {
      "epoch": 0.9902142290399365,
      "grad_norm": 0.05971555411815643,
      "learning_rate": 5.172965377890915e-08,
      "loss": 0.0001,
      "step": 1170
    },
    {
      "epoch": 0.9910605659878339,
      "grad_norm": 0.040896084159612656,
      "learning_rate": 4.346787715939771e-08,
      "loss": 0.0001,
      "step": 1171
    },
    {
      "epoch": 0.9919069029357312,
      "grad_norm": 0.2015872299671173,
      "learning_rate": 3.592431714597444e-08,
      "loss": 0.0003,
      "step": 1172
    },
    {
      "epoch": 0.9927532398836286,
      "grad_norm": 0.018765635788440704,
      "learning_rate": 2.9099027941292822e-08,
      "loss": 0.0001,
      "step": 1173
    },
    {
      "epoch": 0.993599576831526,
      "grad_norm": 0.03560919314622879,
      "learning_rate": 2.2992058587023578e-08,
      "loss": 0.0001,
      "step": 1174
    },
    {
      "epoch": 0.9944459137794235,
      "grad_norm": 0.0527128130197525,
      "learning_rate": 1.7603452963532718e-08,
      "loss": 0.0001,
      "step": 1175
    },
    {
      "epoch": 0.9952922507273209,
      "grad_norm": 0.0666029080748558,
      "learning_rate": 1.293324978948185e-08,
      "loss": 0.0001,
      "step": 1176
    },
    {
      "epoch": 0.9961385876752182,
      "grad_norm": 0.31935206055641174,
      "learning_rate": 8.981482621639448e-09,
      "loss": 0.0003,
      "step": 1177
    },
    {
      "epoch": 0.9969849246231156,
      "grad_norm": 0.036114055663347244,
      "learning_rate": 5.7481798546144e-09,
      "loss": 0.0001,
      "step": 1178
    },
    {
      "epoch": 0.997831261571013,
      "grad_norm": 0.03680175915360451,
      "learning_rate": 3.2333647206117534e-09,
      "loss": 0.0001,
      "step": 1179
    },
    {
      "epoch": 0.9986775985189104,
      "grad_norm": 0.17211928963661194,
      "learning_rate": 1.4370552892883915e-09,
      "loss": 0.0003,
      "step": 1180
    },
    {
      "epoch": 0.9995239354668077,
      "grad_norm": 0.06715048849582672,
      "learning_rate": 3.592644676753132e-10,
      "loss": 0.0002,
      "step": 1181
    },
    {
      "epoch": 1.0007405448294102,
      "grad_norm": 0.027510175481438637,
      "learning_rate": 0.0,
      "loss": 0.0001,
      "step": 1182
    }
  ],
  "logging_steps": 1,
  "max_steps": 1182,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 296,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.2191624098349056e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}