{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 11528,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017349063150589867,
      "grad_norm": 4.815335750579834,
      "learning_rate": 1.7346053772766696e-08,
      "loss": 1.6353,
      "step": 1
    },
    {
      "epoch": 0.00034698126301179735,
      "grad_norm": 6.2609639167785645,
      "learning_rate": 3.469210754553339e-08,
      "loss": 1.6396,
      "step": 2
    },
    {
      "epoch": 0.000520471894517696,
      "grad_norm": 8.841588973999023,
      "learning_rate": 5.203816131830009e-08,
      "loss": 1.834,
      "step": 3
    },
    {
      "epoch": 0.0006939625260235947,
      "grad_norm": 4.740471363067627,
      "learning_rate": 6.938421509106678e-08,
      "loss": 1.5742,
      "step": 4
    },
    {
      "epoch": 0.0008674531575294934,
      "grad_norm": 6.70953369140625,
      "learning_rate": 8.673026886383348e-08,
      "loss": 1.5811,
      "step": 5
    },
    {
      "epoch": 0.001040943789035392,
      "grad_norm": 5.489251136779785,
      "learning_rate": 1.0407632263660019e-07,
      "loss": 1.6089,
      "step": 6
    },
    {
      "epoch": 0.0012144344205412907,
      "grad_norm": 6.307178020477295,
      "learning_rate": 1.214223764093669e-07,
      "loss": 1.5957,
      "step": 7
    },
    {
      "epoch": 0.0013879250520471894,
      "grad_norm": 8.668987274169922,
      "learning_rate": 1.3876843018213356e-07,
      "loss": 1.5532,
      "step": 8
    },
    {
      "epoch": 0.001561415683553088,
      "grad_norm": 6.398656368255615,
      "learning_rate": 1.5611448395490029e-07,
      "loss": 1.7119,
      "step": 9
    },
    {
      "epoch": 0.0017349063150589867,
      "grad_norm": 9.915180206298828,
      "learning_rate": 1.7346053772766696e-07,
      "loss": 1.5229,
      "step": 10
    },
    {
      "epoch": 0.0019083969465648854,
      "grad_norm": 7.884058952331543,
      "learning_rate": 1.9080659150043368e-07,
      "loss": 1.6274,
      "step": 11
    },
    {
      "epoch": 0.002081887578070784,
      "grad_norm": 5.94698429107666,
      "learning_rate": 2.0815264527320037e-07,
      "loss": 1.5659,
      "step": 12
    },
    {
      "epoch": 0.002255378209576683,
      "grad_norm": 5.980604648590088,
      "learning_rate": 2.2549869904596704e-07,
      "loss": 1.6509,
      "step": 13
    },
    {
      "epoch": 0.0024288688410825814,
      "grad_norm": 5.658161640167236,
      "learning_rate": 2.428447528187338e-07,
      "loss": 1.5566,
      "step": 14
    },
    {
      "epoch": 0.0026023594725884803,
      "grad_norm": 5.7178192138671875,
      "learning_rate": 2.6019080659150043e-07,
      "loss": 1.5186,
      "step": 15
    },
    {
      "epoch": 0.002775850104094379,
      "grad_norm": 5.625595569610596,
      "learning_rate": 2.7753686036426713e-07,
      "loss": 1.6133,
      "step": 16
    },
    {
      "epoch": 0.0029493407356002777,
      "grad_norm": 5.517483711242676,
      "learning_rate": 2.948829141370339e-07,
      "loss": 1.6411,
      "step": 17
    },
    {
      "epoch": 0.003122831367106176,
      "grad_norm": 5.840810298919678,
      "learning_rate": 3.1222896790980057e-07,
      "loss": 1.5659,
      "step": 18
    },
    {
      "epoch": 0.003296321998612075,
      "grad_norm": 5.345042705535889,
      "learning_rate": 3.295750216825672e-07,
      "loss": 1.7026,
      "step": 19
    },
    {
      "epoch": 0.0034698126301179735,
      "grad_norm": 12.300100326538086,
      "learning_rate": 3.469210754553339e-07,
      "loss": 1.5674,
      "step": 20
    },
    {
      "epoch": 0.0036433032616238724,
      "grad_norm": 5.376595973968506,
      "learning_rate": 3.642671292281006e-07,
      "loss": 1.6851,
      "step": 21
    },
    {
      "epoch": 0.003816793893129771,
      "grad_norm": 6.235954761505127,
      "learning_rate": 3.8161318300086735e-07,
      "loss": 1.5957,
      "step": 22
    },
    {
      "epoch": 0.003990284524635669,
      "grad_norm": 6.174750804901123,
      "learning_rate": 3.9895923677363405e-07,
      "loss": 1.6279,
      "step": 23
    },
    {
      "epoch": 0.004163775156141568,
      "grad_norm": 5.745070934295654,
      "learning_rate": 4.1630529054640075e-07,
      "loss": 1.6548,
      "step": 24
    },
    {
      "epoch": 0.004337265787647467,
      "grad_norm": 5.718823432922363,
      "learning_rate": 4.3365134431916744e-07,
      "loss": 1.5581,
      "step": 25
    },
    {
      "epoch": 0.004510756419153366,
      "grad_norm": 5.081756114959717,
      "learning_rate": 4.509973980919341e-07,
      "loss": 1.5991,
      "step": 26
    },
    {
      "epoch": 0.004684247050659264,
      "grad_norm": 7.115768909454346,
      "learning_rate": 4.683434518647008e-07,
      "loss": 1.8408,
      "step": 27
    },
    {
      "epoch": 0.004857737682165163,
      "grad_norm": 4.278243064880371,
      "learning_rate": 4.856895056374676e-07,
      "loss": 1.5459,
      "step": 28
    },
    {
      "epoch": 0.005031228313671062,
      "grad_norm": 5.30607271194458,
      "learning_rate": 5.030355594102343e-07,
      "loss": 1.5732,
      "step": 29
    },
    {
      "epoch": 0.005204718945176961,
      "grad_norm": 4.440147399902344,
      "learning_rate": 5.203816131830009e-07,
      "loss": 1.5776,
      "step": 30
    },
    {
      "epoch": 0.005378209576682859,
      "grad_norm": 5.735405921936035,
      "learning_rate": 5.377276669557676e-07,
      "loss": 1.6392,
      "step": 31
    },
    {
      "epoch": 0.005551700208188758,
      "grad_norm": 5.50547456741333,
      "learning_rate": 5.550737207285343e-07,
      "loss": 1.541,
      "step": 32
    },
    {
      "epoch": 0.0057251908396946565,
      "grad_norm": 5.245941638946533,
      "learning_rate": 5.72419774501301e-07,
      "loss": 1.564,
      "step": 33
    },
    {
      "epoch": 0.005898681471200555,
      "grad_norm": 5.313120365142822,
      "learning_rate": 5.897658282740678e-07,
      "loss": 1.522,
      "step": 34
    },
    {
      "epoch": 0.006072172102706454,
      "grad_norm": 5.691841125488281,
      "learning_rate": 6.071118820468344e-07,
      "loss": 1.5229,
      "step": 35
    },
    {
      "epoch": 0.006245662734212352,
      "grad_norm": 6.22916316986084,
      "learning_rate": 6.244579358196011e-07,
      "loss": 1.5361,
      "step": 36
    },
    {
      "epoch": 0.006419153365718251,
      "grad_norm": 5.261981964111328,
      "learning_rate": 6.418039895923677e-07,
      "loss": 1.6646,
      "step": 37
    },
    {
      "epoch": 0.00659264399722415,
      "grad_norm": 6.444558620452881,
      "learning_rate": 6.591500433651344e-07,
      "loss": 1.5459,
      "step": 38
    },
    {
      "epoch": 0.006766134628730049,
      "grad_norm": 6.428016662597656,
      "learning_rate": 6.764960971379011e-07,
      "loss": 1.5625,
      "step": 39
    },
    {
      "epoch": 0.006939625260235947,
      "grad_norm": 6.500727653503418,
      "learning_rate": 6.938421509106678e-07,
      "loss": 1.5977,
      "step": 40
    },
    {
      "epoch": 0.007113115891741846,
      "grad_norm": 5.337209701538086,
      "learning_rate": 7.111882046834345e-07,
      "loss": 1.54,
      "step": 41
    },
    {
      "epoch": 0.007286606523247745,
      "grad_norm": 4.817502498626709,
      "learning_rate": 7.285342584562012e-07,
      "loss": 1.5469,
      "step": 42
    },
    {
      "epoch": 0.007460097154753644,
      "grad_norm": 5.187300205230713,
      "learning_rate": 7.45880312228968e-07,
      "loss": 1.5542,
      "step": 43
    },
    {
      "epoch": 0.007633587786259542,
      "grad_norm": 5.807931900024414,
      "learning_rate": 7.632263660017347e-07,
      "loss": 1.4937,
      "step": 44
    },
    {
      "epoch": 0.007807078417765441,
      "grad_norm": 4.923823356628418,
      "learning_rate": 7.805724197745014e-07,
      "loss": 1.5527,
      "step": 45
    },
    {
      "epoch": 0.007980569049271339,
      "grad_norm": 6.542623043060303,
      "learning_rate": 7.979184735472681e-07,
      "loss": 1.584,
      "step": 46
    },
    {
      "epoch": 0.008154059680777238,
      "grad_norm": 5.157179355621338,
      "learning_rate": 8.152645273200348e-07,
      "loss": 1.4624,
      "step": 47
    },
    {
      "epoch": 0.008327550312283136,
      "grad_norm": 5.175705909729004,
      "learning_rate": 8.326105810928015e-07,
      "loss": 1.4834,
      "step": 48
    },
    {
      "epoch": 0.008501040943789036,
      "grad_norm": 4.367042541503906,
      "learning_rate": 8.499566348655682e-07,
      "loss": 1.4966,
      "step": 49
    },
    {
      "epoch": 0.008674531575294934,
      "grad_norm": 6.201111793518066,
      "learning_rate": 8.673026886383349e-07,
      "loss": 1.4912,
      "step": 50
    },
    {
      "epoch": 0.008848022206800832,
      "grad_norm": 4.258865833282471,
      "learning_rate": 8.846487424111015e-07,
      "loss": 1.5898,
      "step": 51
    },
    {
      "epoch": 0.009021512838306732,
      "grad_norm": 5.785416603088379,
      "learning_rate": 9.019947961838682e-07,
      "loss": 1.4507,
      "step": 52
    },
    {
      "epoch": 0.00919500346981263,
      "grad_norm": 5.257880687713623,
      "learning_rate": 9.193408499566349e-07,
      "loss": 1.5249,
      "step": 53
    },
    {
      "epoch": 0.009368494101318528,
      "grad_norm": 4.440859794616699,
      "learning_rate": 9.366869037294016e-07,
      "loss": 1.4727,
      "step": 54
    },
    {
      "epoch": 0.009541984732824428,
      "grad_norm": 5.027373313903809,
      "learning_rate": 9.540329575021685e-07,
      "loss": 1.5361,
      "step": 55
    },
    {
      "epoch": 0.009715475364330326,
      "grad_norm": 5.587261199951172,
      "learning_rate": 9.713790112749352e-07,
      "loss": 1.4995,
      "step": 56
    },
    {
      "epoch": 0.009888965995836226,
      "grad_norm": 5.661511421203613,
      "learning_rate": 9.887250650477019e-07,
      "loss": 1.4868,
      "step": 57
    },
    {
      "epoch": 0.010062456627342124,
      "grad_norm": 5.535928726196289,
      "learning_rate": 1.0060711188204686e-06,
      "loss": 1.47,
      "step": 58
    },
    {
      "epoch": 0.010235947258848022,
      "grad_norm": 7.044625759124756,
      "learning_rate": 1.0234171725932352e-06,
      "loss": 1.5581,
      "step": 59
    },
    {
      "epoch": 0.010409437890353921,
      "grad_norm": 5.890305519104004,
      "learning_rate": 1.0407632263660017e-06,
      "loss": 1.4434,
      "step": 60
    },
    {
      "epoch": 0.01058292852185982,
      "grad_norm": 4.891403675079346,
      "learning_rate": 1.0581092801387684e-06,
      "loss": 1.46,
      "step": 61
    },
    {
      "epoch": 0.010756419153365717,
      "grad_norm": 3.755310297012329,
      "learning_rate": 1.0754553339115351e-06,
      "loss": 1.5347,
      "step": 62
    },
    {
      "epoch": 0.010929909784871617,
      "grad_norm": 4.348376750946045,
      "learning_rate": 1.0928013876843018e-06,
      "loss": 1.4775,
      "step": 63
    },
    {
      "epoch": 0.011103400416377515,
      "grad_norm": 4.379542827606201,
      "learning_rate": 1.1101474414570685e-06,
      "loss": 1.3525,
      "step": 64
    },
    {
      "epoch": 0.011276891047883415,
      "grad_norm": 3.707447052001953,
      "learning_rate": 1.1274934952298352e-06,
      "loss": 1.5146,
      "step": 65
    },
    {
      "epoch": 0.011450381679389313,
      "grad_norm": 3.2854342460632324,
      "learning_rate": 1.144839549002602e-06,
      "loss": 1.4409,
      "step": 66
    },
    {
      "epoch": 0.011623872310895211,
      "grad_norm": 4.887171268463135,
      "learning_rate": 1.1621856027753688e-06,
      "loss": 1.29,
      "step": 67
    },
    {
      "epoch": 0.01179736294240111,
      "grad_norm": 3.731964349746704,
      "learning_rate": 1.1795316565481355e-06,
      "loss": 1.3916,
      "step": 68
    },
    {
      "epoch": 0.011970853573907009,
      "grad_norm": 4.650766372680664,
      "learning_rate": 1.1968777103209022e-06,
      "loss": 1.3745,
      "step": 69
    },
    {
      "epoch": 0.012144344205412909,
      "grad_norm": 4.351012706756592,
      "learning_rate": 1.214223764093669e-06,
      "loss": 1.2866,
      "step": 70
    },
    {
      "epoch": 0.012317834836918807,
      "grad_norm": 3.9581916332244873,
      "learning_rate": 1.2315698178664356e-06,
      "loss": 1.3257,
      "step": 71
    },
    {
      "epoch": 0.012491325468424705,
      "grad_norm": 3.449936628341675,
      "learning_rate": 1.2489158716392023e-06,
      "loss": 1.4653,
      "step": 72
    },
    {
      "epoch": 0.012664816099930604,
      "grad_norm": 3.5131375789642334,
      "learning_rate": 1.266261925411969e-06,
      "loss": 1.3169,
      "step": 73
    },
    {
      "epoch": 0.012838306731436502,
      "grad_norm": 4.998897075653076,
      "learning_rate": 1.2836079791847355e-06,
      "loss": 1.2939,
      "step": 74
    },
    {
      "epoch": 0.0130117973629424,
      "grad_norm": 3.755049705505371,
      "learning_rate": 1.3009540329575024e-06,
      "loss": 1.3618,
      "step": 75
    },
    {
      "epoch": 0.0131852879944483,
      "grad_norm": 4.029270172119141,
      "learning_rate": 1.3183000867302689e-06,
      "loss": 1.2971,
      "step": 76
    },
    {
      "epoch": 0.013358778625954198,
      "grad_norm": 3.4073169231414795,
      "learning_rate": 1.3356461405030358e-06,
      "loss": 1.3311,
      "step": 77
    },
    {
      "epoch": 0.013532269257460098,
      "grad_norm": 4.988164901733398,
      "learning_rate": 1.3529921942758023e-06,
      "loss": 1.2456,
      "step": 78
    },
    {
      "epoch": 0.013705759888965996,
      "grad_norm": 3.605191230773926,
      "learning_rate": 1.3703382480485692e-06,
      "loss": 1.3921,
      "step": 79
    },
    {
      "epoch": 0.013879250520471894,
      "grad_norm": 3.802180051803589,
      "learning_rate": 1.3876843018213356e-06,
      "loss": 1.2939,
      "step": 80
    },
    {
      "epoch": 0.014052741151977794,
      "grad_norm": 4.355827808380127,
      "learning_rate": 1.4050303555941025e-06,
      "loss": 1.1885,
      "step": 81
    },
    {
      "epoch": 0.014226231783483692,
      "grad_norm": 2.91510009765625,
      "learning_rate": 1.422376409366869e-06,
      "loss": 1.2898,
      "step": 82
    },
    {
      "epoch": 0.01439972241498959,
      "grad_norm": 3.5141520500183105,
      "learning_rate": 1.439722463139636e-06,
      "loss": 1.2383,
      "step": 83
    },
    {
      "epoch": 0.01457321304649549,
      "grad_norm": 4.397307872772217,
      "learning_rate": 1.4570685169124024e-06,
      "loss": 1.3105,
      "step": 84
    },
    {
      "epoch": 0.014746703678001388,
      "grad_norm": 5.1851983070373535,
      "learning_rate": 1.4744145706851693e-06,
      "loss": 1.1816,
      "step": 85
    },
    {
      "epoch": 0.014920194309507287,
      "grad_norm": 3.397026777267456,
      "learning_rate": 1.491760624457936e-06,
      "loss": 1.2275,
      "step": 86
    },
    {
      "epoch": 0.015093684941013185,
      "grad_norm": 3.9106953144073486,
      "learning_rate": 1.5091066782307025e-06,
      "loss": 1.1262,
      "step": 87
    },
    {
      "epoch": 0.015267175572519083,
      "grad_norm": 3.204820394515991,
      "learning_rate": 1.5264527320034694e-06,
      "loss": 1.1958,
      "step": 88
    },
    {
      "epoch": 0.015440666204024983,
      "grad_norm": 4.559689998626709,
      "learning_rate": 1.543798785776236e-06,
      "loss": 1.3069,
      "step": 89
    },
    {
      "epoch": 0.015614156835530881,
      "grad_norm": 2.379528284072876,
      "learning_rate": 1.5611448395490028e-06,
      "loss": 1.0981,
      "step": 90
    },
    {
      "epoch": 0.01578764746703678,
      "grad_norm": 6.517797946929932,
      "learning_rate": 1.5784908933217693e-06,
      "loss": 0.998,
      "step": 91
    },
    {
      "epoch": 0.015961138098542677,
      "grad_norm": 2.010162115097046,
      "learning_rate": 1.5958369470945362e-06,
      "loss": 1.0688,
      "step": 92
    },
    {
      "epoch": 0.01613462873004858,
      "grad_norm": 2.0165321826934814,
      "learning_rate": 1.6131830008673027e-06,
      "loss": 1.1516,
      "step": 93
    },
    {
      "epoch": 0.016308119361554477,
      "grad_norm": 1.6602797508239746,
      "learning_rate": 1.6305290546400696e-06,
      "loss": 1.1128,
      "step": 94
    },
    {
      "epoch": 0.016481609993060375,
      "grad_norm": 2.1719162464141846,
      "learning_rate": 1.647875108412836e-06,
      "loss": 1.0203,
      "step": 95
    },
    {
      "epoch": 0.016655100624566273,
      "grad_norm": 1.6706749200820923,
      "learning_rate": 1.665221162185603e-06,
      "loss": 1.0068,
      "step": 96
    },
    {
      "epoch": 0.01682859125607217,
      "grad_norm": 1.6050095558166504,
      "learning_rate": 1.6825672159583695e-06,
      "loss": 1.0884,
      "step": 97
    },
    {
      "epoch": 0.017002081887578072,
      "grad_norm": 1.8640276193618774,
      "learning_rate": 1.6999132697311364e-06,
      "loss": 1.0833,
      "step": 98
    },
    {
      "epoch": 0.01717557251908397,
      "grad_norm": 1.4145679473876953,
      "learning_rate": 1.717259323503903e-06,
      "loss": 0.991,
      "step": 99
    },
    {
      "epoch": 0.01734906315058987,
      "grad_norm": 1.441550850868225,
      "learning_rate": 1.7346053772766698e-06,
      "loss": 1.033,
      "step": 100
    },
    {
      "epoch": 0.017522553782095766,
      "grad_norm": 1.8363375663757324,
      "learning_rate": 1.7519514310494365e-06,
      "loss": 1.1899,
      "step": 101
    },
    {
      "epoch": 0.017696044413601664,
      "grad_norm": 1.7074156999588013,
      "learning_rate": 1.769297484822203e-06,
      "loss": 0.9524,
      "step": 102
    },
    {
      "epoch": 0.017869535045107566,
      "grad_norm": 1.4260603189468384,
      "learning_rate": 1.7866435385949699e-06,
      "loss": 1.0457,
      "step": 103
    },
    {
      "epoch": 0.018043025676613464,
      "grad_norm": 1.432376742362976,
      "learning_rate": 1.8039895923677363e-06,
      "loss": 0.9365,
      "step": 104
    },
    {
      "epoch": 0.018216516308119362,
      "grad_norm": 1.3644927740097046,
      "learning_rate": 1.8213356461405032e-06,
      "loss": 1.0605,
      "step": 105
    },
    {
      "epoch": 0.01839000693962526,
      "grad_norm": 1.4237080812454224,
      "learning_rate": 1.8386816999132697e-06,
      "loss": 1.1375,
      "step": 106
    },
    {
      "epoch": 0.018563497571131158,
      "grad_norm": 1.280415415763855,
      "learning_rate": 1.8560277536860366e-06,
      "loss": 1.0276,
      "step": 107
    },
    {
      "epoch": 0.018736988202637056,
      "grad_norm": 1.3138136863708496,
      "learning_rate": 1.8733738074588031e-06,
      "loss": 0.9885,
      "step": 108
    },
    {
      "epoch": 0.018910478834142957,
      "grad_norm": 1.2917845249176025,
      "learning_rate": 1.89071986123157e-06,
      "loss": 1.1091,
      "step": 109
    },
    {
      "epoch": 0.019083969465648856,
      "grad_norm": 1.70353102684021,
      "learning_rate": 1.908065915004337e-06,
      "loss": 1.1401,
      "step": 110
    },
    {
      "epoch": 0.019257460097154754,
      "grad_norm": 1.7633841037750244,
      "learning_rate": 1.925411968777103e-06,
      "loss": 1.0864,
      "step": 111
    },
    {
      "epoch": 0.01943095072866065,
      "grad_norm": 1.9823323488235474,
      "learning_rate": 1.9427580225498703e-06,
      "loss": 1.2375,
      "step": 112
    },
    {
      "epoch": 0.01960444136016655,
      "grad_norm": 1.3398616313934326,
      "learning_rate": 1.9601040763226366e-06,
      "loss": 1.0801,
      "step": 113
    },
    {
      "epoch": 0.01977793199167245,
      "grad_norm": 1.2672946453094482,
      "learning_rate": 1.9774501300954037e-06,
      "loss": 1.1201,
      "step": 114
    },
    {
      "epoch": 0.01995142262317835,
      "grad_norm": 1.2124594449996948,
      "learning_rate": 1.99479618386817e-06,
      "loss": 1.0732,
      "step": 115
    },
    {
      "epoch": 0.020124913254684247,
      "grad_norm": 1.215266227722168,
      "learning_rate": 2.012142237640937e-06,
      "loss": 1.075,
      "step": 116
    },
    {
      "epoch": 0.020298403886190145,
      "grad_norm": 1.3705044984817505,
      "learning_rate": 2.0294882914137034e-06,
      "loss": 1.1982,
      "step": 117
    },
    {
      "epoch": 0.020471894517696043,
      "grad_norm": 1.158579707145691,
      "learning_rate": 2.0468343451864705e-06,
      "loss": 1.0884,
      "step": 118
    },
    {
      "epoch": 0.020645385149201945,
      "grad_norm": 1.1770604848861694,
      "learning_rate": 2.0641803989592368e-06,
      "loss": 1.0688,
      "step": 119
    },
    {
      "epoch": 0.020818875780707843,
      "grad_norm": 1.2560598850250244,
      "learning_rate": 2.0815264527320035e-06,
      "loss": 0.9971,
      "step": 120
    },
    {
      "epoch": 0.02099236641221374,
      "grad_norm": 1.1297913789749146,
      "learning_rate": 2.09887250650477e-06,
      "loss": 0.9954,
      "step": 121
    },
    {
      "epoch": 0.02116585704371964,
      "grad_norm": 1.1109462976455688,
      "learning_rate": 2.116218560277537e-06,
      "loss": 0.991,
      "step": 122
    },
    {
      "epoch": 0.021339347675225537,
      "grad_norm": 2.530078411102295,
      "learning_rate": 2.133564614050304e-06,
      "loss": 1.1199,
      "step": 123
    },
    {
      "epoch": 0.021512838306731435,
      "grad_norm": 2.565696954727173,
      "learning_rate": 2.1509106678230702e-06,
      "loss": 1.0833,
      "step": 124
    },
    {
      "epoch": 0.021686328938237336,
      "grad_norm": 1.5229138135910034,
      "learning_rate": 2.1682567215958374e-06,
      "loss": 0.9124,
      "step": 125
    },
    {
      "epoch": 0.021859819569743234,
      "grad_norm": 1.0388314723968506,
      "learning_rate": 2.1856027753686036e-06,
      "loss": 0.9087,
      "step": 126
    },
    {
      "epoch": 0.022033310201249132,
      "grad_norm": 1.0649573802947998,
      "learning_rate": 2.2029488291413708e-06,
      "loss": 0.9292,
      "step": 127
    },
    {
      "epoch": 0.02220680083275503,
      "grad_norm": 1.1507086753845215,
      "learning_rate": 2.220294882914137e-06,
      "loss": 0.8608,
      "step": 128
    },
    {
      "epoch": 0.02238029146426093,
      "grad_norm": 0.8832144141197205,
      "learning_rate": 2.237640936686904e-06,
      "loss": 0.8677,
      "step": 129
    },
    {
      "epoch": 0.02255378209576683,
      "grad_norm": 1.2134603261947632,
      "learning_rate": 2.2549869904596704e-06,
      "loss": 1.1113,
      "step": 130
    },
    {
      "epoch": 0.022727272727272728,
      "grad_norm": 0.953169047832489,
      "learning_rate": 2.2723330442324375e-06,
      "loss": 0.9861,
      "step": 131
    },
    {
      "epoch": 0.022900763358778626,
      "grad_norm": 0.9815278649330139,
      "learning_rate": 2.289679098005204e-06,
      "loss": 0.9607,
      "step": 132
    },
    {
      "epoch": 0.023074253990284524,
      "grad_norm": 0.8123059868812561,
      "learning_rate": 2.3070251517779705e-06,
      "loss": 0.8423,
      "step": 133
    },
    {
      "epoch": 0.023247744621790422,
      "grad_norm": 0.8939048647880554,
      "learning_rate": 2.3243712055507376e-06,
      "loss": 0.9629,
      "step": 134
    },
    {
      "epoch": 0.023421235253296323,
      "grad_norm": 0.8204233646392822,
      "learning_rate": 2.341717259323504e-06,
      "loss": 1.0103,
      "step": 135
    },
    {
      "epoch": 0.02359472588480222,
      "grad_norm": 1.450378656387329,
      "learning_rate": 2.359063313096271e-06,
      "loss": 0.9797,
      "step": 136
    },
    {
      "epoch": 0.02376821651630812,
      "grad_norm": 0.69654381275177,
      "learning_rate": 2.3764093668690373e-06,
      "loss": 0.8779,
      "step": 137
    },
    {
      "epoch": 0.023941707147814018,
      "grad_norm": 0.876659095287323,
      "learning_rate": 2.3937554206418044e-06,
      "loss": 0.9592,
      "step": 138
    },
    {
      "epoch": 0.024115197779319916,
      "grad_norm": 0.967021644115448,
      "learning_rate": 2.4111014744145707e-06,
      "loss": 0.9807,
      "step": 139
    },
    {
      "epoch": 0.024288688410825817,
      "grad_norm": 0.9644420742988586,
      "learning_rate": 2.428447528187338e-06,
      "loss": 0.9805,
      "step": 140
    },
    {
      "epoch": 0.024462179042331715,
      "grad_norm": 0.7096670269966125,
      "learning_rate": 2.445793581960104e-06,
      "loss": 0.9441,
      "step": 141
    },
    {
      "epoch": 0.024635669673837613,
      "grad_norm": 0.8356378078460693,
      "learning_rate": 2.463139635732871e-06,
      "loss": 0.937,
      "step": 142
    },
    {
      "epoch": 0.02480916030534351,
      "grad_norm": 0.7720776796340942,
      "learning_rate": 2.4804856895056375e-06,
      "loss": 1.0525,
      "step": 143
    },
    {
      "epoch": 0.02498265093684941,
      "grad_norm": 1.4250718355178833,
      "learning_rate": 2.4978317432784046e-06,
      "loss": 0.8657,
      "step": 144
    },
    {
      "epoch": 0.025156141568355307,
      "grad_norm": 0.8530842661857605,
      "learning_rate": 2.515177797051171e-06,
      "loss": 0.7996,
      "step": 145
    },
    {
      "epoch": 0.02532963219986121,
      "grad_norm": 0.7666248679161072,
      "learning_rate": 2.532523850823938e-06,
      "loss": 1.0222,
      "step": 146
    },
    {
      "epoch": 0.025503122831367107,
      "grad_norm": 1.1478543281555176,
      "learning_rate": 2.5498699045967047e-06,
      "loss": 0.8037,
      "step": 147
    },
    {
      "epoch": 0.025676613462873005,
      "grad_norm": 0.852233350276947,
      "learning_rate": 2.567215958369471e-06,
      "loss": 0.8386,
      "step": 148
    },
    {
      "epoch": 0.025850104094378903,
      "grad_norm": 1.0843455791473389,
      "learning_rate": 2.5845620121422376e-06,
      "loss": 0.7981,
      "step": 149
    },
    {
      "epoch": 0.0260235947258848,
      "grad_norm": 0.7627226710319519,
      "learning_rate": 2.6019080659150048e-06,
      "loss": 0.938,
      "step": 150
    },
    {
      "epoch": 0.026197085357390702,
      "grad_norm": 1.3740365505218506,
      "learning_rate": 2.6192541196877714e-06,
      "loss": 1.0828,
      "step": 151
    },
    {
      "epoch": 0.0263705759888966,
      "grad_norm": 0.8279482126235962,
      "learning_rate": 2.6366001734605377e-06,
      "loss": 0.7927,
      "step": 152
    },
    {
      "epoch": 0.0265440666204025,
      "grad_norm": 0.7392839789390564,
      "learning_rate": 2.6539462272333044e-06,
      "loss": 0.9153,
      "step": 153
    },
    {
      "epoch": 0.026717557251908396,
      "grad_norm": 1.2634882926940918,
      "learning_rate": 2.6712922810060715e-06,
      "loss": 0.8733,
      "step": 154
    },
    {
      "epoch": 0.026891047883414294,
      "grad_norm": 0.8666257858276367,
      "learning_rate": 2.6886383347788382e-06,
      "loss": 0.8186,
      "step": 155
    },
    {
      "epoch": 0.027064538514920196,
      "grad_norm": 0.6611669063568115,
      "learning_rate": 2.7059843885516045e-06,
      "loss": 0.9236,
      "step": 156
    },
    {
      "epoch": 0.027238029146426094,
      "grad_norm": 0.6729313135147095,
      "learning_rate": 2.723330442324371e-06,
      "loss": 1.0305,
      "step": 157
    },
    {
      "epoch": 0.027411519777931992,
      "grad_norm": 0.9559786915779114,
      "learning_rate": 2.7406764960971383e-06,
      "loss": 0.8596,
      "step": 158
    },
    {
      "epoch": 0.02758501040943789,
      "grad_norm": 0.8773760199546814,
      "learning_rate": 2.758022549869905e-06,
      "loss": 0.8076,
      "step": 159
    },
    {
      "epoch": 0.027758501040943788,
      "grad_norm": 0.8217472434043884,
      "learning_rate": 2.7753686036426713e-06,
      "loss": 1.0337,
      "step": 160
    },
    {
      "epoch": 0.02793199167244969,
      "grad_norm": 0.8561400175094604,
      "learning_rate": 2.7927146574154384e-06,
      "loss": 0.8726,
      "step": 161
    },
    {
      "epoch": 0.028105482303955587,
      "grad_norm": 0.7257559299468994,
      "learning_rate": 2.810060711188205e-06,
      "loss": 0.8669,
      "step": 162
    },
    {
      "epoch": 0.028278972935461485,
      "grad_norm": 0.7245901823043823,
      "learning_rate": 2.8274067649609714e-06,
      "loss": 1.0156,
      "step": 163
    },
    {
      "epoch": 0.028452463566967384,
      "grad_norm": 0.8338601589202881,
      "learning_rate": 2.844752818733738e-06,
      "loss": 0.8984,
      "step": 164
    },
    {
      "epoch": 0.02862595419847328,
      "grad_norm": 1.102246880531311,
      "learning_rate": 2.862098872506505e-06,
      "loss": 0.8914,
      "step": 165
    },
    {
      "epoch": 0.02879944482997918,
      "grad_norm": 0.6566537618637085,
      "learning_rate": 2.879444926279272e-06,
      "loss": 1.0024,
      "step": 166
    },
    {
      "epoch": 0.02897293546148508,
      "grad_norm": 0.7692321538925171,
      "learning_rate": 2.896790980052038e-06,
      "loss": 0.8635,
      "step": 167
    },
    {
      "epoch": 0.02914642609299098,
      "grad_norm": 0.7698785662651062,
      "learning_rate": 2.914137033824805e-06,
      "loss": 1.071,
      "step": 168
    },
    {
      "epoch": 0.029319916724496877,
      "grad_norm": 0.6429572701454163,
      "learning_rate": 2.931483087597572e-06,
      "loss": 0.8945,
      "step": 169
    },
    {
      "epoch": 0.029493407356002775,
      "grad_norm": 0.7103275656700134,
      "learning_rate": 2.9488291413703387e-06,
      "loss": 0.884,
      "step": 170
    },
    {
      "epoch": 0.029666897987508673,
      "grad_norm": 0.7187206149101257,
      "learning_rate": 2.966175195143105e-06,
      "loss": 0.9307,
      "step": 171
    },
    {
      "epoch": 0.029840388619014575,
      "grad_norm": 0.5949711799621582,
      "learning_rate": 2.983521248915872e-06,
      "loss": 0.9526,
      "step": 172
    },
    {
      "epoch": 0.030013879250520473,
      "grad_norm": 0.7339666485786438,
      "learning_rate": 3.0008673026886387e-06,
      "loss": 0.8528,
      "step": 173
    },
    {
      "epoch": 0.03018736988202637,
      "grad_norm": 0.7466350197792053,
      "learning_rate": 3.018213356461405e-06,
      "loss": 0.8169,
      "step": 174
    },
    {
      "epoch": 0.03036086051353227,
      "grad_norm": 1.0726069211959839,
      "learning_rate": 3.0355594102341717e-06,
      "loss": 0.8491,
      "step": 175
    },
    {
      "epoch": 0.030534351145038167,
      "grad_norm": 0.8756661415100098,
      "learning_rate": 3.052905464006939e-06,
      "loss": 0.897,
      "step": 176
    },
    {
      "epoch": 0.030707841776544068,
      "grad_norm": 0.6892372369766235,
      "learning_rate": 3.0702515177797055e-06,
      "loss": 0.8511,
      "step": 177
    },
    {
      "epoch": 0.030881332408049966,
      "grad_norm": 1.579180359840393,
      "learning_rate": 3.087597571552472e-06,
      "loss": 0.7449,
      "step": 178
    },
    {
      "epoch": 0.031054823039555864,
      "grad_norm": 0.9959123134613037,
      "learning_rate": 3.1049436253252385e-06,
      "loss": 0.7668,
      "step": 179
    },
    {
      "epoch": 0.031228313671061762,
      "grad_norm": 0.8271302580833435,
      "learning_rate": 3.1222896790980056e-06,
      "loss": 0.8684,
      "step": 180
    },
    {
      "epoch": 0.031401804302567664,
      "grad_norm": 0.6397276520729065,
      "learning_rate": 3.1396357328707723e-06,
      "loss": 0.8875,
      "step": 181
    },
    {
      "epoch": 0.03157529493407356,
      "grad_norm": 0.9459241032600403,
      "learning_rate": 3.1569817866435386e-06,
      "loss": 0.8673,
      "step": 182
    },
    {
      "epoch": 0.03174878556557946,
      "grad_norm": 0.7759047150611877,
      "learning_rate": 3.1743278404163057e-06,
      "loss": 0.8711,
      "step": 183
    },
    {
      "epoch": 0.031922276197085354,
      "grad_norm": 0.7790178656578064,
      "learning_rate": 3.1916738941890724e-06,
      "loss": 0.8477,
      "step": 184
    },
    {
      "epoch": 0.032095766828591256,
      "grad_norm": 1.604477047920227,
      "learning_rate": 3.209019947961839e-06,
      "loss": 0.8777,
      "step": 185
    },
    {
      "epoch": 0.03226925746009716,
      "grad_norm": 0.676693856716156,
      "learning_rate": 3.2263660017346054e-06,
      "loss": 0.8643,
      "step": 186
    },
    {
      "epoch": 0.03244274809160305,
      "grad_norm": 0.644795298576355,
      "learning_rate": 3.2437120555073725e-06,
      "loss": 0.8586,
      "step": 187
    },
    {
      "epoch": 0.03261623872310895,
      "grad_norm": 0.6525770425796509,
      "learning_rate": 3.261058109280139e-06,
      "loss": 0.9644,
      "step": 188
    },
    {
      "epoch": 0.03278972935461485,
      "grad_norm": 0.7150494456291199,
      "learning_rate": 3.2784041630529055e-06,
      "loss": 0.9407,
      "step": 189
    },
    {
      "epoch": 0.03296321998612075,
      "grad_norm": 0.673234760761261,
      "learning_rate": 3.295750216825672e-06,
      "loss": 0.792,
      "step": 190
    },
    {
      "epoch": 0.03313671061762665,
      "grad_norm": 0.6626754403114319,
      "learning_rate": 3.3130962705984393e-06,
      "loss": 0.9353,
      "step": 191
    },
    {
      "epoch": 0.033310201249132546,
      "grad_norm": 0.7096927165985107,
      "learning_rate": 3.330442324371206e-06,
      "loss": 0.8811,
      "step": 192
    },
    {
      "epoch": 0.03348369188063845,
      "grad_norm": 0.7351794242858887,
      "learning_rate": 3.3477883781439722e-06,
      "loss": 0.6892,
      "step": 193
    },
    {
      "epoch": 0.03365718251214434,
      "grad_norm": 0.7929335832595825,
      "learning_rate": 3.365134431916739e-06,
      "loss": 0.7441,
      "step": 194
    },
    {
      "epoch": 0.03383067314365024,
      "grad_norm": 0.7624382376670837,
      "learning_rate": 3.382480485689506e-06,
      "loss": 0.8257,
      "step": 195
    },
    {
      "epoch": 0.034004163775156145,
      "grad_norm": 0.7748669385910034,
      "learning_rate": 3.3998265394622727e-06,
      "loss": 0.8481,
      "step": 196
    },
    {
      "epoch": 0.03417765440666204,
      "grad_norm": 0.814811110496521,
      "learning_rate": 3.417172593235039e-06,
      "loss": 0.8699,
      "step": 197
    },
    {
      "epoch": 0.03435114503816794,
      "grad_norm": 0.7616357803344727,
      "learning_rate": 3.434518647007806e-06,
      "loss": 0.803,
      "step": 198
    },
    {
      "epoch": 0.034524635669673835,
      "grad_norm": 0.6103156208992004,
      "learning_rate": 3.451864700780573e-06,
      "loss": 0.9177,
      "step": 199
    },
    {
      "epoch": 0.03469812630117974,
      "grad_norm": 0.6358505487442017,
      "learning_rate": 3.4692107545533395e-06,
      "loss": 0.9548,
      "step": 200
    },
    {
      "epoch": 0.03487161693268564,
      "grad_norm": 0.8462308645248413,
      "learning_rate": 3.486556808326106e-06,
      "loss": 0.876,
      "step": 201
    },
    {
      "epoch": 0.03504510756419153,
      "grad_norm": 0.6185916662216187,
      "learning_rate": 3.503902862098873e-06,
      "loss": 0.9285,
      "step": 202
    },
    {
      "epoch": 0.035218598195697434,
      "grad_norm": 0.810539722442627,
      "learning_rate": 3.5212489158716396e-06,
      "loss": 0.7559,
      "step": 203
    },
    {
      "epoch": 0.03539208882720333,
      "grad_norm": 0.7707465887069702,
      "learning_rate": 3.538594969644406e-06,
      "loss": 0.7766,
      "step": 204
    },
    {
      "epoch": 0.03556557945870923,
      "grad_norm": 0.8495586514472961,
      "learning_rate": 3.5559410234171726e-06,
      "loss": 0.8794,
      "step": 205
    },
    {
      "epoch": 0.03573907009021513,
      "grad_norm": 0.8893985748291016,
      "learning_rate": 3.5732870771899397e-06,
      "loss": 0.8918,
      "step": 206
    },
    {
      "epoch": 0.035912560721721026,
      "grad_norm": 1.4733084440231323,
      "learning_rate": 3.5906331309627064e-06,
      "loss": 0.6824,
      "step": 207
    },
    {
      "epoch": 0.03608605135322693,
      "grad_norm": 0.7809956073760986,
      "learning_rate": 3.6079791847354727e-06,
      "loss": 0.8,
      "step": 208
    },
    {
      "epoch": 0.03625954198473282,
      "grad_norm": 0.9555914998054504,
      "learning_rate": 3.62532523850824e-06,
      "loss": 0.7207,
      "step": 209
    },
    {
      "epoch": 0.036433032616238724,
      "grad_norm": 0.7463303804397583,
      "learning_rate": 3.6426712922810065e-06,
      "loss": 0.8175,
      "step": 210
    },
    {
      "epoch": 0.03660652324774462,
      "grad_norm": 0.769335150718689,
      "learning_rate": 3.660017346053773e-06,
      "loss": 0.7695,
      "step": 211
    },
    {
      "epoch": 0.03678001387925052,
      "grad_norm": 0.536637544631958,
      "learning_rate": 3.6773633998265395e-06,
      "loss": 0.9326,
      "step": 212
    },
    {
      "epoch": 0.03695350451075642,
      "grad_norm": 0.686669111251831,
      "learning_rate": 3.6947094535993066e-06,
      "loss": 0.8235,
      "step": 213
    },
    {
      "epoch": 0.037126995142262316,
      "grad_norm": 0.908027172088623,
      "learning_rate": 3.7120555073720733e-06,
      "loss": 0.802,
      "step": 214
    },
    {
      "epoch": 0.03730048577376822,
      "grad_norm": 0.89664226770401,
      "learning_rate": 3.72940156114484e-06,
      "loss": 0.7837,
      "step": 215
    },
    {
      "epoch": 0.03747397640527411,
      "grad_norm": 0.8224126100540161,
      "learning_rate": 3.7467476149176062e-06,
      "loss": 0.7007,
      "step": 216
    },
    {
      "epoch": 0.037647467036780013,
      "grad_norm": 0.8188750743865967,
      "learning_rate": 3.7640936686903734e-06,
      "loss": 0.8289,
      "step": 217
    },
    {
      "epoch": 0.037820957668285915,
      "grad_norm": 2.1264846324920654,
      "learning_rate": 3.78143972246314e-06,
      "loss": 0.7163,
      "step": 218
    },
    {
      "epoch": 0.03799444829979181,
      "grad_norm": 0.7793065309524536,
      "learning_rate": 3.7987857762359063e-06,
      "loss": 0.7161,
      "step": 219
    },
    {
      "epoch": 0.03816793893129771,
      "grad_norm": 7.660075664520264,
      "learning_rate": 3.816131830008674e-06,
      "loss": 0.7937,
      "step": 220
    },
    {
      "epoch": 0.038341429562803606,
      "grad_norm": 0.9559584856033325,
      "learning_rate": 3.8334778837814406e-06,
      "loss": 0.7888,
      "step": 221
    },
    {
      "epoch": 0.03851492019430951,
      "grad_norm": 0.8368854522705078,
      "learning_rate": 3.850823937554206e-06,
      "loss": 0.8337,
      "step": 222
    },
    {
      "epoch": 0.03868841082581541,
      "grad_norm": 0.8978548645973206,
      "learning_rate": 3.868169991326973e-06,
      "loss": 0.8235,
      "step": 223
    },
    {
      "epoch": 0.0388619014573213,
      "grad_norm": 0.7543613314628601,
      "learning_rate": 3.885516045099741e-06,
      "loss": 0.7761,
      "step": 224
    },
    {
      "epoch": 0.039035392088827205,
      "grad_norm": 0.7012503147125244,
      "learning_rate": 3.9028620988725065e-06,
      "loss": 0.8542,
      "step": 225
    },
    {
      "epoch": 0.0392088827203331,
      "grad_norm": 0.64545077085495,
      "learning_rate": 3.920208152645273e-06,
      "loss": 0.8572,
      "step": 226
    },
    {
      "epoch": 0.039382373351839,
      "grad_norm": 1.1598503589630127,
      "learning_rate": 3.93755420641804e-06,
      "loss": 0.8157,
      "step": 227
    },
    {
      "epoch": 0.0395558639833449,
      "grad_norm": 0.8046950101852417,
      "learning_rate": 3.9549002601908074e-06,
      "loss": 0.835,
      "step": 228
    },
    {
      "epoch": 0.0397293546148508,
      "grad_norm": 0.9517707824707031,
      "learning_rate": 3.972246313963573e-06,
      "loss": 0.8191,
      "step": 229
    },
    {
      "epoch": 0.0399028452463567,
      "grad_norm": 1.009547472000122,
      "learning_rate": 3.98959236773634e-06,
      "loss": 0.6919,
      "step": 230
    },
    {
      "epoch": 0.04007633587786259,
      "grad_norm": 0.7407523989677429,
      "learning_rate": 4.0069384215091075e-06,
      "loss": 0.7905,
      "step": 231
    },
    {
      "epoch": 0.040249826509368494,
      "grad_norm": 0.6676948666572571,
      "learning_rate": 4.024284475281874e-06,
      "loss": 0.832,
      "step": 232
    },
    {
      "epoch": 0.040423317140874396,
      "grad_norm": 0.9606331586837769,
      "learning_rate": 4.04163052905464e-06,
      "loss": 0.7549,
      "step": 233
    },
    {
      "epoch": 0.04059680777238029,
      "grad_norm": 0.7595999836921692,
      "learning_rate": 4.058976582827407e-06,
      "loss": 0.8694,
      "step": 234
    },
    {
      "epoch": 0.04077029840388619,
      "grad_norm": 2.211470127105713,
      "learning_rate": 4.076322636600174e-06,
      "loss": 0.844,
      "step": 235
    },
    {
      "epoch": 0.040943789035392086,
      "grad_norm": 0.6662604212760925,
      "learning_rate": 4.093668690372941e-06,
      "loss": 0.9075,
      "step": 236
    },
    {
      "epoch": 0.04111727966689799,
      "grad_norm": 0.7361257076263428,
      "learning_rate": 4.111014744145707e-06,
      "loss": 0.7053,
      "step": 237
    },
    {
      "epoch": 0.04129077029840389,
      "grad_norm": 0.8374466896057129,
      "learning_rate": 4.1283607979184735e-06,
      "loss": 0.9587,
      "step": 238
    },
    {
      "epoch": 0.041464260929909784,
      "grad_norm": 0.8268419504165649,
      "learning_rate": 4.145706851691241e-06,
      "loss": 0.7429,
      "step": 239
    },
    {
      "epoch": 0.041637751561415685,
      "grad_norm": 0.7588842511177063,
      "learning_rate": 4.163052905464007e-06,
      "loss": 0.8174,
      "step": 240
    },
    {
      "epoch": 0.04181124219292158,
      "grad_norm": 0.66733717918396,
      "learning_rate": 4.180398959236774e-06,
      "loss": 0.9172,
      "step": 241
    },
    {
      "epoch": 0.04198473282442748,
      "grad_norm": 0.7593885064125061,
      "learning_rate": 4.19774501300954e-06,
      "loss": 0.7698,
      "step": 242
    },
    {
      "epoch": 0.04215822345593338,
      "grad_norm": 1.1506482362747192,
      "learning_rate": 4.215091066782308e-06,
      "loss": 0.8696,
      "step": 243
    },
    {
      "epoch": 0.04233171408743928,
      "grad_norm": 0.8607063293457031,
      "learning_rate": 4.232437120555074e-06,
      "loss": 0.9065,
      "step": 244
    },
    {
      "epoch": 0.04250520471894518,
      "grad_norm": 0.7805898189544678,
      "learning_rate": 4.24978317432784e-06,
      "loss": 0.7156,
      "step": 245
    },
    {
      "epoch": 0.042678695350451074,
      "grad_norm": 0.9701886177062988,
      "learning_rate": 4.267129228100608e-06,
      "loss": 0.6403,
      "step": 246
    },
    {
      "epoch": 0.042852185981956975,
      "grad_norm": 0.8391703963279724,
      "learning_rate": 4.284475281873375e-06,
      "loss": 0.6885,
      "step": 247
    },
    {
      "epoch": 0.04302567661346287,
      "grad_norm": 0.7802613973617554,
      "learning_rate": 4.3018213356461405e-06,
      "loss": 0.8601,
      "step": 248
    },
    {
      "epoch": 0.04319916724496877,
      "grad_norm": 0.7520661354064941,
      "learning_rate": 4.319167389418907e-06,
      "loss": 1.0027,
      "step": 249
    },
    {
      "epoch": 0.04337265787647467,
      "grad_norm": 0.8130558729171753,
      "learning_rate": 4.336513443191675e-06,
      "loss": 0.7485,
      "step": 250
    },
    {
      "epoch": 0.04354614850798057,
      "grad_norm": 0.8154857158660889,
      "learning_rate": 4.353859496964441e-06,
      "loss": 0.8755,
      "step": 251
    },
    {
      "epoch": 0.04371963913948647,
      "grad_norm": 0.757587730884552,
      "learning_rate": 4.371205550737207e-06,
      "loss": 0.8047,
      "step": 252
    },
    {
      "epoch": 0.04389312977099236,
      "grad_norm": 0.6265802979469299,
      "learning_rate": 4.388551604509974e-06,
      "loss": 0.7207,
      "step": 253
    },
    {
      "epoch": 0.044066620402498265,
      "grad_norm": 0.6111220717430115,
      "learning_rate": 4.4058976582827415e-06,
      "loss": 0.9375,
      "step": 254
    },
    {
      "epoch": 0.044240111034004166,
      "grad_norm": 0.9138938188552856,
      "learning_rate": 4.423243712055507e-06,
      "loss": 0.8058,
      "step": 255
    },
    {
      "epoch": 0.04441360166551006,
      "grad_norm": 0.8813202381134033,
      "learning_rate": 4.440589765828274e-06,
      "loss": 0.7773,
      "step": 256
    },
    {
      "epoch": 0.04458709229701596,
      "grad_norm": 1.5841827392578125,
      "learning_rate": 4.457935819601042e-06,
      "loss": 0.9963,
      "step": 257
    },
    {
      "epoch": 0.04476058292852186,
      "grad_norm": 0.766584038734436,
      "learning_rate": 4.475281873373808e-06,
      "loss": 0.7659,
      "step": 258
    },
    {
      "epoch": 0.04493407356002776,
      "grad_norm": 0.6755748987197876,
      "learning_rate": 4.492627927146574e-06,
      "loss": 0.7212,
      "step": 259
    },
    {
      "epoch": 0.04510756419153366,
      "grad_norm": 2.221755266189575,
      "learning_rate": 4.509973980919341e-06,
      "loss": 0.7344,
      "step": 260
    },
    {
      "epoch": 0.045281054823039554,
      "grad_norm": 0.805280327796936,
      "learning_rate": 4.527320034692108e-06,
      "loss": 0.9041,
      "step": 261
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 3.478299617767334,
      "learning_rate": 4.544666088464875e-06,
      "loss": 0.6799,
      "step": 262
    },
    {
      "epoch": 0.04562803608605135,
      "grad_norm": 0.8958662152290344,
      "learning_rate": 4.562012142237641e-06,
      "loss": 0.7192,
      "step": 263
    },
    {
      "epoch": 0.04580152671755725,
      "grad_norm": 0.9214358925819397,
      "learning_rate": 4.579358196010408e-06,
      "loss": 0.8047,
      "step": 264
    },
    {
      "epoch": 0.04597501734906315,
      "grad_norm": 0.6886780261993408,
      "learning_rate": 4.596704249783175e-06,
      "loss": 0.8123,
      "step": 265
    },
    {
      "epoch": 0.04614850798056905,
      "grad_norm": 0.7185443043708801,
      "learning_rate": 4.614050303555941e-06,
      "loss": 0.8203,
      "step": 266
    },
    {
      "epoch": 0.04632199861207495,
      "grad_norm": 0.7269342541694641,
      "learning_rate": 4.631396357328708e-06,
      "loss": 0.7539,
      "step": 267
    },
    {
      "epoch": 0.046495489243580844,
      "grad_norm": 0.7610074877738953,
      "learning_rate": 4.648742411101475e-06,
      "loss": 0.7434,
      "step": 268
    },
    {
      "epoch": 0.046668979875086745,
      "grad_norm": 0.5354163646697998,
      "learning_rate": 4.666088464874242e-06,
      "loss": 0.8367,
      "step": 269
    },
    {
      "epoch": 0.04684247050659265,
      "grad_norm": 0.7159690260887146,
      "learning_rate": 4.683434518647008e-06,
      "loss": 0.7781,
      "step": 270
    },
    {
      "epoch": 0.04701596113809854,
      "grad_norm": 0.639028012752533,
      "learning_rate": 4.7007805724197745e-06,
      "loss": 0.8618,
      "step": 271
    },
    {
      "epoch": 0.04718945176960444,
      "grad_norm": 1.0831190347671509,
      "learning_rate": 4.718126626192542e-06,
      "loss": 0.8411,
      "step": 272
    },
    {
      "epoch": 0.04736294240111034,
      "grad_norm": 1.0538662672042847,
      "learning_rate": 4.735472679965309e-06,
      "loss": 0.7742,
      "step": 273
    },
    {
      "epoch": 0.04753643303261624,
      "grad_norm": 0.7081946134567261,
      "learning_rate": 4.7528187337380746e-06,
      "loss": 0.8293,
      "step": 274
    },
    {
      "epoch": 0.04770992366412214,
      "grad_norm": 0.7760938405990601,
      "learning_rate": 4.770164787510841e-06,
      "loss": 0.7058,
      "step": 275
    },
    {
      "epoch": 0.047883414295628035,
      "grad_norm": 0.885649561882019,
      "learning_rate": 4.787510841283609e-06,
      "loss": 0.7678,
      "step": 276
    },
    {
      "epoch": 0.04805690492713394,
      "grad_norm": 4.859396457672119,
      "learning_rate": 4.8048568950563755e-06,
      "loss": 0.7527,
      "step": 277
    },
    {
      "epoch": 0.04823039555863983,
      "grad_norm": 0.6299145817756653,
      "learning_rate": 4.822202948829141e-06,
      "loss": 0.8083,
      "step": 278
    },
    {
      "epoch": 0.04840388619014573,
      "grad_norm": 0.7858087420463562,
      "learning_rate": 4.839549002601908e-06,
      "loss": 0.8025,
      "step": 279
    },
    {
      "epoch": 0.048577376821651634,
      "grad_norm": 0.7333070039749146,
      "learning_rate": 4.856895056374676e-06,
      "loss": 0.7378,
      "step": 280
    },
    {
      "epoch": 0.04875086745315753,
      "grad_norm": 0.6566837430000305,
      "learning_rate": 4.8742411101474414e-06,
      "loss": 0.8901,
      "step": 281
    },
    {
      "epoch": 0.04892435808466343,
      "grad_norm": 0.9724940061569214,
      "learning_rate": 4.891587163920208e-06,
      "loss": 0.7305,
      "step": 282
    },
    {
      "epoch": 0.049097848716169325,
      "grad_norm": 10.982410430908203,
      "learning_rate": 4.908933217692976e-06,
      "loss": 0.7285,
      "step": 283
    },
    {
      "epoch": 0.049271339347675226,
      "grad_norm": 0.6716983318328857,
      "learning_rate": 4.926279271465742e-06,
      "loss": 0.907,
      "step": 284
    },
    {
      "epoch": 0.04944482997918112,
      "grad_norm": 0.5775556564331055,
      "learning_rate": 4.943625325238508e-06,
      "loss": 0.9473,
      "step": 285
    },
    {
      "epoch": 0.04961832061068702,
      "grad_norm": 0.6030986309051514,
      "learning_rate": 4.960971379011275e-06,
      "loss": 0.7859,
      "step": 286
    },
    {
      "epoch": 0.049791811242192924,
      "grad_norm": 0.6085160374641418,
      "learning_rate": 4.9783174327840425e-06,
      "loss": 0.8865,
      "step": 287
    },
    {
      "epoch": 0.04996530187369882,
      "grad_norm": 0.5863941311836243,
      "learning_rate": 4.995663486556809e-06,
      "loss": 0.8601,
      "step": 288
    },
    {
      "epoch": 0.05013879250520472,
      "grad_norm": 0.6125705242156982,
      "learning_rate": 5.013009540329575e-06,
      "loss": 0.8037,
      "step": 289
    },
    {
      "epoch": 0.050312283136710614,
      "grad_norm": 0.571257472038269,
      "learning_rate": 5.030355594102342e-06,
      "loss": 0.9082,
      "step": 290
    },
    {
      "epoch": 0.050485773768216516,
      "grad_norm": 0.791437566280365,
      "learning_rate": 5.047701647875108e-06,
      "loss": 0.6273,
      "step": 291
    },
    {
      "epoch": 0.05065926439972242,
      "grad_norm": 0.8369519710540771,
      "learning_rate": 5.065047701647876e-06,
      "loss": 0.8008,
      "step": 292
    },
    {
      "epoch": 0.05083275503122831,
      "grad_norm": 0.56186443567276,
      "learning_rate": 5.082393755420643e-06,
      "loss": 0.9321,
      "step": 293
    },
    {
      "epoch": 0.05100624566273421,
      "grad_norm": 0.7457464933395386,
      "learning_rate": 5.099739809193409e-06,
      "loss": 0.8228,
      "step": 294
    },
    {
      "epoch": 0.05117973629424011,
      "grad_norm": 0.528501033782959,
      "learning_rate": 5.117085862966176e-06,
      "loss": 0.8887,
      "step": 295
    },
    {
      "epoch": 0.05135322692574601,
      "grad_norm": 0.7900983691215515,
      "learning_rate": 5.134431916738942e-06,
      "loss": 0.6978,
      "step": 296
    },
    {
      "epoch": 0.05152671755725191,
      "grad_norm": 0.6286918520927429,
      "learning_rate": 5.1517779705117086e-06,
      "loss": 0.9106,
      "step": 297
    },
    {
      "epoch": 0.051700208188757805,
      "grad_norm": 0.7337638139724731,
      "learning_rate": 5.169124024284475e-06,
      "loss": 0.8862,
      "step": 298
    },
    {
      "epoch": 0.05187369882026371,
      "grad_norm": 0.5130578875541687,
      "learning_rate": 5.186470078057242e-06,
      "loss": 0.9341,
      "step": 299
    },
    {
      "epoch": 0.0520471894517696,
      "grad_norm": 0.8467931747436523,
      "learning_rate": 5.2038161318300095e-06,
      "loss": 0.8159,
      "step": 300
    },
    {
      "epoch": 0.0522206800832755,
      "grad_norm": 0.7748866081237793,
      "learning_rate": 5.221162185602776e-06,
      "loss": 0.8328,
      "step": 301
    },
    {
      "epoch": 0.052394170714781405,
      "grad_norm": 1.04865300655365,
      "learning_rate": 5.238508239375543e-06,
      "loss": 0.8247,
      "step": 302
    },
    {
      "epoch": 0.0525676613462873,
      "grad_norm": 0.8614583611488342,
      "learning_rate": 5.25585429314831e-06,
      "loss": 0.771,
      "step": 303
    },
    {
      "epoch": 0.0527411519777932,
      "grad_norm": 0.8058211207389832,
      "learning_rate": 5.2732003469210754e-06,
      "loss": 0.8032,
      "step": 304
    },
    {
      "epoch": 0.052914642609299095,
      "grad_norm": 1.5144559144973755,
      "learning_rate": 5.290546400693842e-06,
      "loss": 0.6587,
      "step": 305
    },
    {
      "epoch": 0.053088133240805,
      "grad_norm": 0.7051360011100769,
      "learning_rate": 5.307892454466609e-06,
      "loss": 0.7852,
      "step": 306
    },
    {
      "epoch": 0.0532616238723109,
      "grad_norm": 1.4300090074539185,
      "learning_rate": 5.325238508239376e-06,
      "loss": 0.6649,
      "step": 307
    },
    {
      "epoch": 0.05343511450381679,
      "grad_norm": 0.8560264110565186,
      "learning_rate": 5.342584562012143e-06,
      "loss": 0.7117,
      "step": 308
    },
    {
      "epoch": 0.053608605135322694,
      "grad_norm": 0.8068621158599854,
      "learning_rate": 5.35993061578491e-06,
      "loss": 0.782,
      "step": 309
    },
    {
      "epoch": 0.05378209576682859,
      "grad_norm": 0.831801176071167,
      "learning_rate": 5.3772766695576765e-06,
      "loss": 0.748,
      "step": 310
    },
    {
      "epoch": 0.05395558639833449,
      "grad_norm": 0.7599542140960693,
      "learning_rate": 5.394622723330442e-06,
      "loss": 0.74,
      "step": 311
    },
    {
      "epoch": 0.05412907702984039,
      "grad_norm": 0.6961326599121094,
      "learning_rate": 5.411968777103209e-06,
      "loss": 0.6924,
      "step": 312
    },
    {
      "epoch": 0.054302567661346286,
      "grad_norm": 0.8084877729415894,
      "learning_rate": 5.429314830875976e-06,
      "loss": 0.6763,
      "step": 313
    },
    {
      "epoch": 0.05447605829285219,
      "grad_norm": 0.7455685138702393,
      "learning_rate": 5.446660884648742e-06,
      "loss": 0.6255,
      "step": 314
    },
    {
      "epoch": 0.05464954892435808,
      "grad_norm": 0.5308278203010559,
      "learning_rate": 5.46400693842151e-06,
      "loss": 0.9922,
      "step": 315
    },
    {
      "epoch": 0.054823039555863984,
      "grad_norm": 0.9737206101417542,
      "learning_rate": 5.481352992194277e-06,
      "loss": 0.7515,
      "step": 316
    },
    {
      "epoch": 0.054996530187369885,
      "grad_norm": 0.7302623987197876,
      "learning_rate": 5.498699045967043e-06,
      "loss": 0.9568,
      "step": 317
    },
    {
      "epoch": 0.05517002081887578,
      "grad_norm": 1.1479965448379517,
      "learning_rate": 5.51604509973981e-06,
      "loss": 0.6812,
      "step": 318
    },
    {
      "epoch": 0.05534351145038168,
      "grad_norm": 0.7652915120124817,
      "learning_rate": 5.533391153512576e-06,
      "loss": 0.7554,
      "step": 319
    },
    {
      "epoch": 0.055517002081887576,
      "grad_norm": 0.6627578139305115,
      "learning_rate": 5.5507372072853426e-06,
      "loss": 0.8398,
      "step": 320
    },
    {
      "epoch": 0.05569049271339348,
      "grad_norm": 0.6127437949180603,
      "learning_rate": 5.568083261058109e-06,
      "loss": 0.8469,
      "step": 321
    },
    {
      "epoch": 0.05586398334489938,
      "grad_norm": 1.0261603593826294,
      "learning_rate": 5.585429314830877e-06,
      "loss": 0.6429,
      "step": 322
    },
    {
      "epoch": 0.05603747397640527,
      "grad_norm": 0.6037549376487732,
      "learning_rate": 5.6027753686036435e-06,
      "loss": 0.8923,
      "step": 323
    },
    {
      "epoch": 0.056210964607911175,
      "grad_norm": 0.7665083408355713,
      "learning_rate": 5.62012142237641e-06,
      "loss": 0.7959,
      "step": 324
    },
    {
      "epoch": 0.05638445523941707,
      "grad_norm": 0.6456735134124756,
      "learning_rate": 5.637467476149177e-06,
      "loss": 0.8284,
      "step": 325
    },
    {
      "epoch": 0.05655794587092297,
      "grad_norm": 0.6334142684936523,
      "learning_rate": 5.654813529921943e-06,
      "loss": 0.8318,
      "step": 326
    },
    {
      "epoch": 0.056731436502428866,
      "grad_norm": 0.695650041103363,
      "learning_rate": 5.6721595836947094e-06,
      "loss": 0.9285,
      "step": 327
    },
    {
      "epoch": 0.05690492713393477,
      "grad_norm": 2.0656869411468506,
      "learning_rate": 5.689505637467476e-06,
      "loss": 0.7832,
      "step": 328
    },
    {
      "epoch": 0.05707841776544067,
      "grad_norm": 0.5046230554580688,
      "learning_rate": 5.706851691240244e-06,
      "loss": 0.8992,
      "step": 329
    },
    {
      "epoch": 0.05725190839694656,
      "grad_norm": 0.5988643169403076,
      "learning_rate": 5.72419774501301e-06,
      "loss": 0.804,
      "step": 330
    },
    {
      "epoch": 0.057425399028452465,
      "grad_norm": 0.692200243473053,
      "learning_rate": 5.741543798785777e-06,
      "loss": 0.7488,
      "step": 331
    },
    {
      "epoch": 0.05759888965995836,
      "grad_norm": 0.6809567213058472,
      "learning_rate": 5.758889852558544e-06,
      "loss": 0.8037,
      "step": 332
    },
    {
      "epoch": 0.05777238029146426,
      "grad_norm": 0.638801097869873,
      "learning_rate": 5.7762359063313105e-06,
      "loss": 0.7747,
      "step": 333
    },
    {
      "epoch": 0.05794587092297016,
      "grad_norm": 0.9376187920570374,
      "learning_rate": 5.793581960104076e-06,
      "loss": 0.6567,
      "step": 334
    },
    {
      "epoch": 0.05811936155447606,
      "grad_norm": 0.89469313621521,
      "learning_rate": 5.810928013876843e-06,
      "loss": 0.9426,
      "step": 335
    },
    {
      "epoch": 0.05829285218598196,
      "grad_norm": 0.8213267922401428,
      "learning_rate": 5.82827406764961e-06,
      "loss": 0.731,
      "step": 336
    },
    {
      "epoch": 0.05846634281748785,
      "grad_norm": 1.1361161470413208,
      "learning_rate": 5.845620121422377e-06,
      "loss": 0.6674,
      "step": 337
    },
    {
      "epoch": 0.058639833448993754,
      "grad_norm": 1.0722969770431519,
      "learning_rate": 5.862966175195144e-06,
      "loss": 0.7927,
      "step": 338
    },
    {
      "epoch": 0.058813324080499656,
      "grad_norm": 0.7767350077629089,
      "learning_rate": 5.880312228967911e-06,
      "loss": 0.8269,
      "step": 339
    },
    {
      "epoch": 0.05898681471200555,
      "grad_norm": 0.7055554986000061,
      "learning_rate": 5.897658282740677e-06,
      "loss": 0.8315,
      "step": 340
    },
    {
      "epoch": 0.05916030534351145,
      "grad_norm": 0.7638046145439148,
      "learning_rate": 5.915004336513443e-06,
      "loss": 0.7021,
      "step": 341
    },
    {
      "epoch": 0.059333795975017346,
      "grad_norm": 0.7253721952438354,
      "learning_rate": 5.93235039028621e-06,
      "loss": 0.8967,
      "step": 342
    },
    {
      "epoch": 0.05950728660652325,
      "grad_norm": 0.5444093346595764,
      "learning_rate": 5.9496964440589766e-06,
      "loss": 0.7219,
      "step": 343
    },
    {
      "epoch": 0.05968077723802915,
      "grad_norm": 0.5293662548065186,
      "learning_rate": 5.967042497831744e-06,
      "loss": 0.7992,
      "step": 344
    },
    {
      "epoch": 0.059854267869535044,
      "grad_norm": 0.567878782749176,
      "learning_rate": 5.984388551604511e-06,
      "loss": 0.7272,
      "step": 345
    },
    {
      "epoch": 0.060027758501040945,
      "grad_norm": 0.7837110757827759,
      "learning_rate": 6.0017346053772775e-06,
      "loss": 0.6542,
      "step": 346
    },
    {
      "epoch": 0.06020124913254684,
      "grad_norm": 1.6302368640899658,
      "learning_rate": 6.019080659150044e-06,
      "loss": 0.8364,
      "step": 347
    },
    {
      "epoch": 0.06037473976405274,
      "grad_norm": 0.7898486852645874,
      "learning_rate": 6.03642671292281e-06,
      "loss": 0.8018,
      "step": 348
    },
    {
      "epoch": 0.06054823039555864,
      "grad_norm": 0.7227967381477356,
      "learning_rate": 6.053772766695577e-06,
      "loss": 0.8589,
      "step": 349
    },
    {
      "epoch": 0.06072172102706454,
      "grad_norm": 0.918053925037384,
      "learning_rate": 6.0711188204683434e-06,
      "loss": 0.855,
      "step": 350
    },
    {
      "epoch": 0.06089521165857044,
      "grad_norm": 0.7263798117637634,
      "learning_rate": 6.08846487424111e-06,
      "loss": 0.7437,
      "step": 351
    },
    {
      "epoch": 0.061068702290076333,
      "grad_norm": 0.6784784197807312,
      "learning_rate": 6.105810928013878e-06,
      "loss": 0.7058,
      "step": 352
    },
    {
      "epoch": 0.061242192921582235,
      "grad_norm": 0.5894147157669067,
      "learning_rate": 6.123156981786644e-06,
      "loss": 0.8665,
      "step": 353
    },
    {
      "epoch": 0.061415683553088136,
      "grad_norm": 0.5161333680152893,
      "learning_rate": 6.140503035559411e-06,
      "loss": 0.8225,
      "step": 354
    },
    {
      "epoch": 0.06158917418459403,
      "grad_norm": 0.8817012310028076,
      "learning_rate": 6.157849089332178e-06,
      "loss": 0.8091,
      "step": 355
    },
    {
      "epoch": 0.06176266481609993,
      "grad_norm": 1.0701371431350708,
      "learning_rate": 6.175195143104944e-06,
      "loss": 0.6826,
      "step": 356
    },
    {
      "epoch": 0.06193615544760583,
      "grad_norm": 0.5561087727546692,
      "learning_rate": 6.19254119687771e-06,
      "loss": 0.7378,
      "step": 357
    },
    {
      "epoch": 0.06210964607911173,
      "grad_norm": 0.6870754957199097,
      "learning_rate": 6.209887250650477e-06,
      "loss": 0.9226,
      "step": 358
    },
    {
      "epoch": 0.06228313671061763,
      "grad_norm": 0.9816715717315674,
      "learning_rate": 6.2272333044232445e-06,
      "loss": 0.7219,
      "step": 359
    },
    {
      "epoch": 0.062456627342123525,
      "grad_norm": 0.7335995435714722,
      "learning_rate": 6.244579358196011e-06,
      "loss": 0.7476,
      "step": 360
    },
    {
      "epoch": 0.06263011797362943,
      "grad_norm": 0.9058141708374023,
      "learning_rate": 6.261925411968778e-06,
      "loss": 0.9377,
      "step": 361
    },
    {
      "epoch": 0.06280360860513533,
      "grad_norm": 0.6555805802345276,
      "learning_rate": 6.279271465741545e-06,
      "loss": 0.908,
      "step": 362
    },
    {
      "epoch": 0.06297709923664122,
      "grad_norm": 0.6734438538551331,
      "learning_rate": 6.2966175195143105e-06,
      "loss": 0.7739,
      "step": 363
    },
    {
      "epoch": 0.06315058986814712,
      "grad_norm": 0.8846603035926819,
      "learning_rate": 6.313963573287077e-06,
      "loss": 0.834,
      "step": 364
    },
    {
      "epoch": 0.06332408049965302,
      "grad_norm": 0.8238341212272644,
      "learning_rate": 6.331309627059844e-06,
      "loss": 0.6595,
      "step": 365
    },
    {
      "epoch": 0.06349757113115892,
      "grad_norm": 0.8995324969291687,
      "learning_rate": 6.348655680832611e-06,
      "loss": 0.8555,
      "step": 366
    },
    {
      "epoch": 0.06367106176266482,
      "grad_norm": 0.8729636073112488,
      "learning_rate": 6.366001734605378e-06,
      "loss": 0.6772,
      "step": 367
    },
    {
      "epoch": 0.06384455239417071,
      "grad_norm": 0.8361624479293823,
      "learning_rate": 6.383347788378145e-06,
      "loss": 0.8401,
      "step": 368
    },
    {
      "epoch": 0.06401804302567661,
      "grad_norm": 0.7549445033073425,
      "learning_rate": 6.4006938421509115e-06,
      "loss": 0.7329,
      "step": 369
    },
    {
      "epoch": 0.06419153365718251,
      "grad_norm": 0.7796569466590881,
      "learning_rate": 6.418039895923678e-06,
      "loss": 0.8467,
      "step": 370
    },
    {
      "epoch": 0.06436502428868841,
      "grad_norm": 0.7562674283981323,
      "learning_rate": 6.435385949696444e-06,
      "loss": 0.6274,
      "step": 371
    },
    {
      "epoch": 0.06453851492019431,
      "grad_norm": 0.8655654191970825,
      "learning_rate": 6.452732003469211e-06,
      "loss": 0.7244,
      "step": 372
    },
    {
      "epoch": 0.0647120055517002,
      "grad_norm": 0.6695210933685303,
      "learning_rate": 6.4700780572419774e-06,
      "loss": 0.7799,
      "step": 373
    },
    {
      "epoch": 0.0648854961832061,
      "grad_norm": 1.7294037342071533,
      "learning_rate": 6.487424111014745e-06,
      "loss": 0.7656,
      "step": 374
    },
    {
      "epoch": 0.065058986814712,
      "grad_norm": 0.8687999248504639,
      "learning_rate": 6.504770164787512e-06,
      "loss": 0.7002,
      "step": 375
    },
    {
      "epoch": 0.0652324774462179,
      "grad_norm": 0.7759955525398254,
      "learning_rate": 6.522116218560278e-06,
      "loss": 0.7661,
      "step": 376
    },
    {
      "epoch": 0.06540596807772381,
      "grad_norm": 0.6977130770683289,
      "learning_rate": 6.539462272333045e-06,
      "loss": 0.8188,
      "step": 377
    },
    {
      "epoch": 0.0655794587092297,
      "grad_norm": 1.5170788764953613,
      "learning_rate": 6.556808326105811e-06,
      "loss": 0.6729,
      "step": 378
    },
    {
      "epoch": 0.0657529493407356,
      "grad_norm": 0.7143145203590393,
      "learning_rate": 6.574154379878578e-06,
      "loss": 0.8057,
      "step": 379
    },
    {
      "epoch": 0.0659264399722415,
      "grad_norm": 0.7431565523147583,
      "learning_rate": 6.591500433651344e-06,
      "loss": 0.76,
      "step": 380
    },
    {
      "epoch": 0.0660999306037474,
      "grad_norm": 0.8921603560447693,
      "learning_rate": 6.608846487424112e-06,
      "loss": 0.7434,
      "step": 381
    },
    {
      "epoch": 0.0662734212352533,
      "grad_norm": 0.791877031326294,
      "learning_rate": 6.6261925411968785e-06,
      "loss": 0.7627,
      "step": 382
    },
    {
      "epoch": 0.06644691186675919,
      "grad_norm": 0.6708717346191406,
      "learning_rate": 6.643538594969645e-06,
      "loss": 0.7662,
      "step": 383
    },
    {
      "epoch": 0.06662040249826509,
      "grad_norm": 0.8915305733680725,
      "learning_rate": 6.660884648742412e-06,
      "loss": 0.6855,
      "step": 384
    },
    {
      "epoch": 0.06679389312977099,
      "grad_norm": 1.2064861059188843,
      "learning_rate": 6.678230702515179e-06,
      "loss": 0.6848,
      "step": 385
    },
    {
      "epoch": 0.0669673837612769,
      "grad_norm": 1.4230811595916748,
      "learning_rate": 6.6955767562879445e-06,
      "loss": 0.631,
      "step": 386
    },
    {
      "epoch": 0.0671408743927828,
      "grad_norm": 0.8477129936218262,
      "learning_rate": 6.712922810060711e-06,
      "loss": 0.7196,
      "step": 387
    },
    {
      "epoch": 0.06731436502428868,
      "grad_norm": 0.794474720954895,
      "learning_rate": 6.730268863833478e-06,
      "loss": 0.7865,
      "step": 388
    },
    {
      "epoch": 0.06748785565579458,
      "grad_norm": 0.7628791928291321,
      "learning_rate": 6.747614917606245e-06,
      "loss": 0.7771,
      "step": 389
    },
    {
      "epoch": 0.06766134628730049,
      "grad_norm": 0.669377863407135,
      "learning_rate": 6.764960971379012e-06,
      "loss": 0.79,
      "step": 390
    },
    {
      "epoch": 0.06783483691880639,
      "grad_norm": 0.930276095867157,
      "learning_rate": 6.782307025151779e-06,
      "loss": 0.751,
      "step": 391
    },
    {
      "epoch": 0.06800832755031229,
      "grad_norm": 1.2251073122024536,
      "learning_rate": 6.7996530789245455e-06,
      "loss": 0.6873,
      "step": 392
    },
    {
      "epoch": 0.06818181818181818,
      "grad_norm": 0.5957617163658142,
      "learning_rate": 6.816999132697311e-06,
      "loss": 0.877,
      "step": 393
    },
    {
      "epoch": 0.06835530881332408,
      "grad_norm": 0.6212051510810852,
      "learning_rate": 6.834345186470078e-06,
      "loss": 0.8433,
      "step": 394
    },
    {
      "epoch": 0.06852879944482998,
      "grad_norm": 0.8099094033241272,
      "learning_rate": 6.851691240242845e-06,
      "loss": 0.7192,
      "step": 395
    },
    {
      "epoch": 0.06870229007633588,
      "grad_norm": 0.7998353838920593,
      "learning_rate": 6.869037294015612e-06,
      "loss": 0.8076,
      "step": 396
    },
    {
      "epoch": 0.06887578070784178,
      "grad_norm": 0.7063225507736206,
      "learning_rate": 6.886383347788379e-06,
      "loss": 0.6953,
      "step": 397
    },
    {
      "epoch": 0.06904927133934767,
      "grad_norm": 0.5355812311172485,
      "learning_rate": 6.903729401561146e-06,
      "loss": 0.8667,
      "step": 398
    },
    {
      "epoch": 0.06922276197085357,
      "grad_norm": 1.0308690071105957,
      "learning_rate": 6.921075455333912e-06,
      "loss": 0.7527,
      "step": 399
    },
    {
      "epoch": 0.06939625260235947,
      "grad_norm": 0.5810636878013611,
      "learning_rate": 6.938421509106679e-06,
      "loss": 0.7284,
      "step": 400
    },
    {
      "epoch": 0.06956974323386537,
      "grad_norm": 0.7493751049041748,
      "learning_rate": 6.955767562879445e-06,
      "loss": 0.7474,
      "step": 401
    },
    {
      "epoch": 0.06974323386537128,
      "grad_norm": 1.0948079824447632,
      "learning_rate": 6.973113616652212e-06,
      "loss": 0.7709,
      "step": 402
    },
    {
      "epoch": 0.06991672449687716,
      "grad_norm": 0.7085433006286621,
      "learning_rate": 6.990459670424979e-06,
      "loss": 0.8494,
      "step": 403
    },
    {
      "epoch": 0.07009021512838307,
      "grad_norm": 0.6970533728599548,
      "learning_rate": 7.007805724197746e-06,
      "loss": 0.7302,
      "step": 404
    },
    {
      "epoch": 0.07026370575988897,
      "grad_norm": 0.7963132858276367,
      "learning_rate": 7.0251517779705125e-06,
      "loss": 0.725,
      "step": 405
    },
    {
      "epoch": 0.07043719639139487,
      "grad_norm": 0.6666185259819031,
      "learning_rate": 7.042497831743279e-06,
      "loss": 0.7947,
      "step": 406
    },
    {
      "epoch": 0.07061068702290077,
      "grad_norm": 0.7390024065971375,
      "learning_rate": 7.059843885516046e-06,
      "loss": 0.6677,
      "step": 407
    },
    {
      "epoch": 0.07078417765440666,
      "grad_norm": 0.9136267304420471,
      "learning_rate": 7.077189939288812e-06,
      "loss": 0.6533,
      "step": 408
    },
    {
      "epoch": 0.07095766828591256,
      "grad_norm": 0.694338858127594,
      "learning_rate": 7.0945359930615785e-06,
      "loss": 0.7598,
      "step": 409
    },
    {
      "epoch": 0.07113115891741846,
      "grad_norm": 0.9569571614265442,
      "learning_rate": 7.111882046834345e-06,
      "loss": 0.7534,
      "step": 410
    },
    {
      "epoch": 0.07130464954892436,
      "grad_norm": 1.0424975156784058,
      "learning_rate": 7.129228100607113e-06,
      "loss": 0.6542,
      "step": 411
    },
    {
      "epoch": 0.07147814018043026,
      "grad_norm": 0.9410294890403748,
      "learning_rate": 7.146574154379879e-06,
      "loss": 0.6138,
      "step": 412
    },
    {
      "epoch": 0.07165163081193615,
      "grad_norm": 0.7803980708122253,
      "learning_rate": 7.163920208152646e-06,
      "loss": 0.7131,
      "step": 413
    },
    {
      "epoch": 0.07182512144344205,
      "grad_norm": 1.045674204826355,
      "learning_rate": 7.181266261925413e-06,
      "loss": 0.7546,
      "step": 414
    },
    {
      "epoch": 0.07199861207494795,
      "grad_norm": 0.8654783964157104,
      "learning_rate": 7.1986123156981795e-06,
      "loss": 0.7412,
      "step": 415
    },
    {
      "epoch": 0.07217210270645386,
      "grad_norm": 0.6371574401855469,
      "learning_rate": 7.215958369470945e-06,
      "loss": 0.8594,
      "step": 416
    },
    {
      "epoch": 0.07234559333795976,
      "grad_norm": 0.8476478457450867,
      "learning_rate": 7.233304423243712e-06,
      "loss": 0.6602,
      "step": 417
    },
    {
      "epoch": 0.07251908396946564,
      "grad_norm": 0.560556948184967,
      "learning_rate": 7.25065047701648e-06,
      "loss": 0.8,
      "step": 418
    },
    {
      "epoch": 0.07269257460097155,
      "grad_norm": 0.6393320560455322,
      "learning_rate": 7.267996530789246e-06,
      "loss": 0.8726,
      "step": 419
    },
    {
      "epoch": 0.07286606523247745,
      "grad_norm": 1.0716561079025269,
      "learning_rate": 7.285342584562013e-06,
      "loss": 0.6736,
      "step": 420
    },
    {
      "epoch": 0.07303955586398335,
      "grad_norm": 0.8586274981498718,
      "learning_rate": 7.30268863833478e-06,
      "loss": 0.7964,
      "step": 421
    },
    {
      "epoch": 0.07321304649548924,
      "grad_norm": 0.6543982028961182,
      "learning_rate": 7.320034692107546e-06,
      "loss": 0.771,
      "step": 422
    },
    {
      "epoch": 0.07338653712699514,
      "grad_norm": 0.9546947479248047,
      "learning_rate": 7.337380745880312e-06,
      "loss": 0.6929,
      "step": 423
    },
    {
      "epoch": 0.07356002775850104,
      "grad_norm": 0.7377446293830872,
      "learning_rate": 7.354726799653079e-06,
      "loss": 0.677,
      "step": 424
    },
    {
      "epoch": 0.07373351839000694,
      "grad_norm": 0.7322527170181274,
      "learning_rate": 7.3720728534258464e-06,
      "loss": 0.7683,
      "step": 425
    },
    {
      "epoch": 0.07390700902151284,
      "grad_norm": 0.7413867115974426,
      "learning_rate": 7.389418907198613e-06,
      "loss": 0.771,
      "step": 426
    },
    {
      "epoch": 0.07408049965301873,
      "grad_norm": 0.7413671612739563,
      "learning_rate": 7.40676496097138e-06,
      "loss": 0.822,
      "step": 427
    },
    {
      "epoch": 0.07425399028452463,
      "grad_norm": 0.8946639895439148,
      "learning_rate": 7.4241110147441465e-06,
      "loss": 0.854,
      "step": 428
    },
    {
      "epoch": 0.07442748091603053,
      "grad_norm": 0.8664313554763794,
      "learning_rate": 7.441457068516913e-06,
      "loss": 0.7322,
      "step": 429
    },
    {
      "epoch": 0.07460097154753643,
      "grad_norm": 0.9482120275497437,
      "learning_rate": 7.45880312228968e-06,
      "loss": 0.6526,
      "step": 430
    },
    {
      "epoch": 0.07477446217904234,
      "grad_norm": 0.5458512902259827,
      "learning_rate": 7.476149176062446e-06,
      "loss": 0.8723,
      "step": 431
    },
    {
      "epoch": 0.07494795281054822,
      "grad_norm": 0.8940516710281372,
      "learning_rate": 7.4934952298352125e-06,
      "loss": 0.7019,
      "step": 432
    },
    {
      "epoch": 0.07512144344205413,
      "grad_norm": 0.6466241478919983,
      "learning_rate": 7.51084128360798e-06,
      "loss": 0.7361,
      "step": 433
    },
    {
      "epoch": 0.07529493407356003,
      "grad_norm": 0.7802069187164307,
      "learning_rate": 7.528187337380747e-06,
      "loss": 0.6299,
      "step": 434
    },
    {
      "epoch": 0.07546842470506593,
      "grad_norm": 0.7344783544540405,
      "learning_rate": 7.545533391153513e-06,
      "loss": 0.7388,
      "step": 435
    },
    {
      "epoch": 0.07564191533657183,
      "grad_norm": 0.6571396589279175,
      "learning_rate": 7.56287944492628e-06,
      "loss": 0.7869,
      "step": 436
    },
    {
      "epoch": 0.07581540596807772,
      "grad_norm": 3.0189146995544434,
      "learning_rate": 7.580225498699047e-06,
      "loss": 0.7114,
      "step": 437
    },
    {
      "epoch": 0.07598889659958362,
      "grad_norm": 0.6549769043922424,
      "learning_rate": 7.597571552471813e-06,
      "loss": 0.7844,
      "step": 438
    },
    {
      "epoch": 0.07616238723108952,
      "grad_norm": 1.1466854810714722,
      "learning_rate": 7.614917606244579e-06,
      "loss": 0.7739,
      "step": 439
    },
    {
      "epoch": 0.07633587786259542,
      "grad_norm": 0.9282358288764954,
      "learning_rate": 7.632263660017348e-06,
      "loss": 0.845,
      "step": 440
    },
    {
      "epoch": 0.07650936849410132,
      "grad_norm": 0.9557682275772095,
      "learning_rate": 7.649609713790114e-06,
      "loss": 0.7102,
      "step": 441
    },
    {
      "epoch": 0.07668285912560721,
      "grad_norm": 0.6585580706596375,
      "learning_rate": 7.666955767562881e-06,
      "loss": 0.811,
      "step": 442
    },
    {
      "epoch": 0.07685634975711311,
      "grad_norm": 0.7832518219947815,
      "learning_rate": 7.684301821335646e-06,
      "loss": 0.6974,
      "step": 443
    },
    {
      "epoch": 0.07702984038861901,
      "grad_norm": 0.6745266318321228,
      "learning_rate": 7.701647875108413e-06,
      "loss": 0.8176,
      "step": 444
    },
    {
      "epoch": 0.07720333102012492,
      "grad_norm": 0.852424144744873,
      "learning_rate": 7.71899392888118e-06,
      "loss": 0.7551,
      "step": 445
    },
    {
      "epoch": 0.07737682165163082,
      "grad_norm": 0.8548586368560791,
      "learning_rate": 7.736339982653946e-06,
      "loss": 0.7634,
      "step": 446
    },
    {
      "epoch": 0.0775503122831367,
      "grad_norm": 1.3087246417999268,
      "learning_rate": 7.753686036426713e-06,
      "loss": 0.7422,
      "step": 447
    },
    {
      "epoch": 0.0777238029146426,
      "grad_norm": 0.9239414930343628,
      "learning_rate": 7.771032090199481e-06,
      "loss": 0.7153,
      "step": 448
    },
    {
      "epoch": 0.07789729354614851,
      "grad_norm": 0.9753161072731018,
      "learning_rate": 7.788378143972248e-06,
      "loss": 0.6571,
      "step": 449
    },
    {
      "epoch": 0.07807078417765441,
      "grad_norm": 0.8641350269317627,
      "learning_rate": 7.805724197745013e-06,
      "loss": 0.6698,
      "step": 450
    },
    {
      "epoch": 0.07824427480916031,
      "grad_norm": 0.773655354976654,
      "learning_rate": 7.82307025151778e-06,
      "loss": 0.7451,
      "step": 451
    },
    {
      "epoch": 0.0784177654406662,
      "grad_norm": 0.8763969540596008,
      "learning_rate": 7.840416305290546e-06,
      "loss": 0.5737,
      "step": 452
    },
    {
      "epoch": 0.0785912560721721,
      "grad_norm": 1.2770614624023438,
      "learning_rate": 7.857762359063313e-06,
      "loss": 0.6241,
      "step": 453
    },
    {
      "epoch": 0.078764746703678,
      "grad_norm": 0.7275820970535278,
      "learning_rate": 7.87510841283608e-06,
      "loss": 0.7694,
      "step": 454
    },
    {
      "epoch": 0.0789382373351839,
      "grad_norm": 0.8261591196060181,
      "learning_rate": 7.892454466608848e-06,
      "loss": 0.7981,
      "step": 455
    },
    {
      "epoch": 0.0791117279666898,
      "grad_norm": 0.703579843044281,
      "learning_rate": 7.909800520381615e-06,
      "loss": 0.6826,
      "step": 456
    },
    {
      "epoch": 0.07928521859819569,
      "grad_norm": 1.005600929260254,
      "learning_rate": 7.927146574154382e-06,
      "loss": 0.7288,
      "step": 457
    },
    {
      "epoch": 0.0794587092297016,
      "grad_norm": 0.7515043616294861,
      "learning_rate": 7.944492627927147e-06,
      "loss": 0.8726,
      "step": 458
    },
    {
      "epoch": 0.0796321998612075,
      "grad_norm": 1.3452553749084473,
      "learning_rate": 7.961838681699913e-06,
      "loss": 0.8608,
      "step": 459
    },
    {
      "epoch": 0.0798056904927134,
      "grad_norm": 0.7214851379394531,
      "learning_rate": 7.97918473547268e-06,
      "loss": 0.6246,
      "step": 460
    },
    {
      "epoch": 0.0799791811242193,
      "grad_norm": 0.7828449606895447,
      "learning_rate": 7.996530789245447e-06,
      "loss": 0.7883,
      "step": 461
    },
    {
      "epoch": 0.08015267175572519,
      "grad_norm": 0.7184955477714539,
      "learning_rate": 8.013876843018215e-06,
      "loss": 0.7754,
      "step": 462
    },
    {
      "epoch": 0.08032616238723109,
      "grad_norm": 1.6390373706817627,
      "learning_rate": 8.031222896790982e-06,
      "loss": 0.8508,
      "step": 463
    },
    {
      "epoch": 0.08049965301873699,
      "grad_norm": 0.7775904536247253,
      "learning_rate": 8.048568950563748e-06,
      "loss": 0.7139,
      "step": 464
    },
    {
      "epoch": 0.08067314365024289,
      "grad_norm": 0.579129695892334,
      "learning_rate": 8.065915004336513e-06,
      "loss": 0.8235,
      "step": 465
    },
    {
      "epoch": 0.08084663428174879,
      "grad_norm": 0.775214433670044,
      "learning_rate": 8.08326105810928e-06,
      "loss": 0.7097,
      "step": 466
    },
    {
      "epoch": 0.08102012491325468,
      "grad_norm": 0.7665948271751404,
      "learning_rate": 8.100607111882047e-06,
      "loss": 0.605,
      "step": 467
    },
    {
      "epoch": 0.08119361554476058,
      "grad_norm": 0.6703997254371643,
      "learning_rate": 8.117953165654814e-06,
      "loss": 0.7177,
      "step": 468
    },
    {
      "epoch": 0.08136710617626648,
      "grad_norm": 0.7012144327163696,
      "learning_rate": 8.13529921942758e-06,
      "loss": 0.6602,
      "step": 469
    },
    {
      "epoch": 0.08154059680777238,
      "grad_norm": 0.7739162445068359,
      "learning_rate": 8.152645273200349e-06,
      "loss": 0.8223,
      "step": 470
    },
    {
      "epoch": 0.08171408743927829,
      "grad_norm": 1.3429744243621826,
      "learning_rate": 8.169991326973115e-06,
      "loss": 0.8232,
      "step": 471
    },
    {
      "epoch": 0.08188757807078417,
      "grad_norm": 0.7664818167686462,
      "learning_rate": 8.187337380745882e-06,
      "loss": 0.7393,
      "step": 472
    },
    {
      "epoch": 0.08206106870229007,
      "grad_norm": 0.709116518497467,
      "learning_rate": 8.204683434518647e-06,
      "loss": 0.8679,
      "step": 473
    },
    {
      "epoch": 0.08223455933379598,
      "grad_norm": 1.02998685836792,
      "learning_rate": 8.222029488291414e-06,
      "loss": 0.728,
      "step": 474
    },
    {
      "epoch": 0.08240804996530188,
      "grad_norm": 0.8173246383666992,
      "learning_rate": 8.23937554206418e-06,
      "loss": 0.8169,
      "step": 475
    },
    {
      "epoch": 0.08258154059680778,
      "grad_norm": 0.8525285124778748,
      "learning_rate": 8.256721595836947e-06,
      "loss": 0.6626,
      "step": 476
    },
    {
      "epoch": 0.08275503122831367,
      "grad_norm": 0.8937381505966187,
      "learning_rate": 8.274067649609715e-06,
      "loss": 0.5542,
      "step": 477
    },
    {
      "epoch": 0.08292852185981957,
      "grad_norm": 0.7560815215110779,
      "learning_rate": 8.291413703382482e-06,
      "loss": 0.6812,
      "step": 478
    },
    {
      "epoch": 0.08310201249132547,
      "grad_norm": 0.8020071983337402,
      "learning_rate": 8.308759757155249e-06,
      "loss": 0.7202,
      "step": 479
    },
    {
      "epoch": 0.08327550312283137,
      "grad_norm": 0.8825773596763611,
      "learning_rate": 8.326105810928014e-06,
      "loss": 0.7566,
      "step": 480
    },
    {
      "epoch": 0.08344899375433727,
      "grad_norm": 0.5909084677696228,
      "learning_rate": 8.34345186470078e-06,
      "loss": 0.8176,
      "step": 481
    },
    {
      "epoch": 0.08362248438584316,
      "grad_norm": 0.8789088129997253,
      "learning_rate": 8.360797918473547e-06,
      "loss": 0.7727,
      "step": 482
    },
    {
      "epoch": 0.08379597501734906,
      "grad_norm": 0.9357159733772278,
      "learning_rate": 8.378143972246314e-06,
      "loss": 0.5996,
      "step": 483
    },
    {
      "epoch": 0.08396946564885496,
      "grad_norm": 1.0077883005142212,
      "learning_rate": 8.39549002601908e-06,
      "loss": 0.5942,
      "step": 484
    },
    {
      "epoch": 0.08414295628036086,
      "grad_norm": 0.6832888126373291,
      "learning_rate": 8.412836079791849e-06,
      "loss": 0.7163,
      "step": 485
    },
    {
      "epoch": 0.08431644691186677,
      "grad_norm": 0.9494338631629944,
      "learning_rate": 8.430182133564616e-06,
      "loss": 0.5977,
      "step": 486
    },
    {
      "epoch": 0.08448993754337265,
      "grad_norm": 1.0955969095230103,
      "learning_rate": 8.44752818733738e-06,
      "loss": 0.6746,
      "step": 487
    },
    {
      "epoch": 0.08466342817487855,
      "grad_norm": 1.162654995918274,
      "learning_rate": 8.464874241110147e-06,
      "loss": 0.7969,
      "step": 488
    },
    {
      "epoch": 0.08483691880638446,
      "grad_norm": 0.5996276140213013,
      "learning_rate": 8.482220294882914e-06,
      "loss": 0.8638,
      "step": 489
    },
    {
      "epoch": 0.08501040943789036,
      "grad_norm": 0.9170894622802734,
      "learning_rate": 8.49956634865568e-06,
      "loss": 0.632,
      "step": 490
    },
    {
      "epoch": 0.08518390006939626,
      "grad_norm": 0.9191092848777771,
      "learning_rate": 8.516912402428448e-06,
      "loss": 0.7981,
      "step": 491
    },
    {
      "epoch": 0.08535739070090215,
      "grad_norm": 0.7416355609893799,
      "learning_rate": 8.534258456201216e-06,
      "loss": 0.5829,
      "step": 492
    },
    {
      "epoch": 0.08553088133240805,
      "grad_norm": 0.779184103012085,
      "learning_rate": 8.551604509973983e-06,
      "loss": 0.6583,
      "step": 493
    },
    {
      "epoch": 0.08570437196391395,
      "grad_norm": 0.8102035522460938,
      "learning_rate": 8.56895056374675e-06,
      "loss": 0.6511,
      "step": 494
    },
    {
      "epoch": 0.08587786259541985,
      "grad_norm": 0.8218130469322205,
      "learning_rate": 8.586296617519514e-06,
      "loss": 0.7849,
      "step": 495
    },
    {
      "epoch": 0.08605135322692574,
      "grad_norm": 0.628180742263794,
      "learning_rate": 8.603642671292281e-06,
      "loss": 0.7087,
      "step": 496
    },
    {
      "epoch": 0.08622484385843164,
      "grad_norm": 1.953553318977356,
      "learning_rate": 8.620988725065048e-06,
      "loss": 0.7119,
      "step": 497
    },
    {
      "epoch": 0.08639833448993754,
      "grad_norm": 0.8177529573440552,
      "learning_rate": 8.638334778837814e-06,
      "loss": 0.623,
      "step": 498
    },
    {
      "epoch": 0.08657182512144344,
      "grad_norm": 0.6821890473365784,
      "learning_rate": 8.655680832610583e-06,
      "loss": 0.7053,
      "step": 499
    },
    {
      "epoch": 0.08674531575294935,
      "grad_norm": 0.885227620601654,
      "learning_rate": 8.67302688638335e-06,
      "loss": 0.6406,
      "step": 500
    },
    {
      "epoch": 0.08691880638445523,
      "grad_norm": 0.7139533758163452,
      "learning_rate": 8.690372940156116e-06,
      "loss": 0.7063,
      "step": 501
    },
    {
      "epoch": 0.08709229701596113,
      "grad_norm": 1.3664354085922241,
      "learning_rate": 8.707718993928881e-06,
      "loss": 0.6804,
      "step": 502
    },
    {
      "epoch": 0.08726578764746704,
      "grad_norm": 0.8646285533905029,
      "learning_rate": 8.725065047701648e-06,
      "loss": 0.7297,
      "step": 503
    },
    {
      "epoch": 0.08743927827897294,
      "grad_norm": 1.329119324684143,
      "learning_rate": 8.742411101474415e-06,
      "loss": 0.7087,
      "step": 504
    },
    {
      "epoch": 0.08761276891047884,
      "grad_norm": 0.7398525476455688,
      "learning_rate": 8.759757155247181e-06,
      "loss": 0.7896,
      "step": 505
    },
    {
      "epoch": 0.08778625954198473,
      "grad_norm": 0.7473292350769043,
      "learning_rate": 8.777103209019948e-06,
      "loss": 0.663,
      "step": 506
    },
    {
      "epoch": 0.08795975017349063,
      "grad_norm": 0.8798092603683472,
      "learning_rate": 8.794449262792716e-06,
      "loss": 0.6587,
      "step": 507
    },
    {
      "epoch": 0.08813324080499653,
      "grad_norm": 0.8144434094429016,
      "learning_rate": 8.811795316565483e-06,
      "loss": 0.6016,
      "step": 508
    },
    {
      "epoch": 0.08830673143650243,
      "grad_norm": 1.0975896120071411,
      "learning_rate": 8.82914137033825e-06,
      "loss": 0.6672,
      "step": 509
    },
    {
      "epoch": 0.08848022206800833,
      "grad_norm": 0.764312744140625,
      "learning_rate": 8.846487424111015e-06,
      "loss": 0.7397,
      "step": 510
    },
    {
      "epoch": 0.08865371269951422,
      "grad_norm": 0.689689040184021,
      "learning_rate": 8.863833477883781e-06,
      "loss": 0.7532,
      "step": 511
    },
    {
      "epoch": 0.08882720333102012,
      "grad_norm": 0.9319776296615601,
      "learning_rate": 8.881179531656548e-06,
      "loss": 0.7139,
      "step": 512
    },
    {
      "epoch": 0.08900069396252602,
      "grad_norm": 1.0532619953155518,
      "learning_rate": 8.898525585429315e-06,
      "loss": 0.6583,
      "step": 513
    },
    {
      "epoch": 0.08917418459403192,
      "grad_norm": 0.6886184811592102,
      "learning_rate": 8.915871639202083e-06,
      "loss": 0.7584,
      "step": 514
    },
    {
      "epoch": 0.08934767522553783,
      "grad_norm": 0.7394210696220398,
      "learning_rate": 8.93321769297485e-06,
      "loss": 0.7539,
      "step": 515
    },
    {
      "epoch": 0.08952116585704371,
      "grad_norm": 0.5901085734367371,
      "learning_rate": 8.950563746747617e-06,
      "loss": 0.7644,
      "step": 516
    },
    {
      "epoch": 0.08969465648854962,
      "grad_norm": 0.8676567077636719,
      "learning_rate": 8.967909800520382e-06,
      "loss": 0.7686,
      "step": 517
    },
    {
      "epoch": 0.08986814712005552,
      "grad_norm": 0.7285106182098389,
      "learning_rate": 8.985255854293148e-06,
      "loss": 0.6838,
      "step": 518
    },
    {
      "epoch": 0.09004163775156142,
      "grad_norm": 0.6929370760917664,
      "learning_rate": 9.002601908065915e-06,
      "loss": 0.7634,
      "step": 519
    },
    {
      "epoch": 0.09021512838306732,
      "grad_norm": 0.8217961192131042,
      "learning_rate": 9.019947961838682e-06,
      "loss": 0.7397,
      "step": 520
    },
    {
      "epoch": 0.09038861901457321,
      "grad_norm": 0.9813176393508911,
      "learning_rate": 9.037294015611448e-06,
      "loss": 0.7473,
      "step": 521
    },
    {
      "epoch": 0.09056210964607911,
      "grad_norm": 0.755930483341217,
      "learning_rate": 9.054640069384217e-06,
      "loss": 0.6104,
      "step": 522
    },
    {
      "epoch": 0.09073560027758501,
      "grad_norm": 0.8116580843925476,
      "learning_rate": 9.071986123156983e-06,
      "loss": 0.5685,
      "step": 523
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.7591921091079712,
      "learning_rate": 9.08933217692975e-06,
      "loss": 0.6926,
      "step": 524
    },
    {
      "epoch": 0.09108258154059681,
      "grad_norm": 0.7296175360679626,
      "learning_rate": 9.106678230702515e-06,
      "loss": 0.696,
      "step": 525
    },
    {
      "epoch": 0.0912560721721027,
      "grad_norm": 0.6052044034004211,
      "learning_rate": 9.124024284475282e-06,
      "loss": 0.813,
      "step": 526
    },
    {
      "epoch": 0.0914295628036086,
      "grad_norm": 0.9392281174659729,
      "learning_rate": 9.141370338248049e-06,
      "loss": 0.6506,
      "step": 527
    },
    {
      "epoch": 0.0916030534351145,
      "grad_norm": 0.7543035745620728,
      "learning_rate": 9.158716392020815e-06,
      "loss": 0.714,
      "step": 528
    },
    {
      "epoch": 0.0917765440666204,
      "grad_norm": 1.0716018676757812,
      "learning_rate": 9.176062445793584e-06,
      "loss": 0.6085,
      "step": 529
    },
    {
      "epoch": 0.0919500346981263,
      "grad_norm": 1.0834380388259888,
      "learning_rate": 9.19340849956635e-06,
      "loss": 0.6614,
      "step": 530
    },
    {
      "epoch": 0.0921235253296322,
      "grad_norm": 0.9322414398193359,
      "learning_rate": 9.210754553339117e-06,
      "loss": 0.585,
      "step": 531
    },
    {
      "epoch": 0.0922970159611381,
      "grad_norm": 1.064762830734253,
      "learning_rate": 9.228100607111882e-06,
      "loss": 0.905,
      "step": 532
    },
    {
      "epoch": 0.092470506592644,
      "grad_norm": 0.7074531316757202,
      "learning_rate": 9.245446660884649e-06,
      "loss": 0.7267,
      "step": 533
    },
    {
      "epoch": 0.0926439972241499,
      "grad_norm": 0.7268208861351013,
      "learning_rate": 9.262792714657415e-06,
      "loss": 0.7759,
      "step": 534
    },
    {
      "epoch": 0.0928174878556558,
      "grad_norm": 1.8754379749298096,
      "learning_rate": 9.280138768430182e-06,
      "loss": 0.5776,
      "step": 535
    },
    {
      "epoch": 0.09299097848716169,
      "grad_norm": 2.2080729007720947,
      "learning_rate": 9.29748482220295e-06,
      "loss": 0.7365,
      "step": 536
    },
    {
      "epoch": 0.09316446911866759,
      "grad_norm": 0.8226796984672546,
      "learning_rate": 9.314830875975717e-06,
      "loss": 0.8542,
      "step": 537
    },
    {
      "epoch": 0.09333795975017349,
      "grad_norm": 1.653834581375122,
      "learning_rate": 9.332176929748484e-06,
      "loss": 0.6907,
      "step": 538
    },
    {
      "epoch": 0.09351145038167939,
      "grad_norm": 0.6772366166114807,
      "learning_rate": 9.34952298352125e-06,
      "loss": 0.7031,
      "step": 539
    },
    {
      "epoch": 0.0936849410131853,
      "grad_norm": 0.5401658415794373,
      "learning_rate": 9.366869037294016e-06,
      "loss": 0.8594,
      "step": 540
    },
    {
      "epoch": 0.09385843164469118,
      "grad_norm": 1.2171071767807007,
      "learning_rate": 9.384215091066782e-06,
      "loss": 0.657,
      "step": 541
    },
    {
      "epoch": 0.09403192227619708,
      "grad_norm": 0.568236231803894,
      "learning_rate": 9.401561144839549e-06,
      "loss": 0.7146,
      "step": 542
    },
    {
      "epoch": 0.09420541290770298,
      "grad_norm": 0.5544914603233337,
      "learning_rate": 9.418907198612316e-06,
      "loss": 0.8723,
      "step": 543
    },
    {
      "epoch": 0.09437890353920889,
      "grad_norm": 0.7629676461219788,
      "learning_rate": 9.436253252385084e-06,
      "loss": 0.7893,
      "step": 544
    },
    {
      "epoch": 0.09455239417071479,
      "grad_norm": 0.8096640110015869,
      "learning_rate": 9.45359930615785e-06,
      "loss": 0.8293,
      "step": 545
    },
    {
      "epoch": 0.09472588480222068,
      "grad_norm": 0.9710381627082825,
      "learning_rate": 9.470945359930617e-06,
      "loss": 0.6384,
      "step": 546
    },
    {
      "epoch": 0.09489937543372658,
      "grad_norm": 0.6319253444671631,
      "learning_rate": 9.488291413703382e-06,
      "loss": 0.8062,
      "step": 547
    },
    {
      "epoch": 0.09507286606523248,
      "grad_norm": 0.7259004712104797,
      "learning_rate": 9.505637467476149e-06,
      "loss": 0.7434,
      "step": 548
    },
    {
      "epoch": 0.09524635669673838,
      "grad_norm": 0.9125682711601257,
      "learning_rate": 9.522983521248916e-06,
      "loss": 0.7483,
      "step": 549
    },
    {
      "epoch": 0.09541984732824428,
      "grad_norm": 0.7326458692550659,
      "learning_rate": 9.540329575021683e-06,
      "loss": 0.749,
      "step": 550
    },
    {
      "epoch": 0.09559333795975017,
      "grad_norm": 0.7823507785797119,
      "learning_rate": 9.557675628794451e-06,
      "loss": 0.689,
      "step": 551
    },
    {
      "epoch": 0.09576682859125607,
      "grad_norm": 0.8902698755264282,
      "learning_rate": 9.575021682567218e-06,
      "loss": 0.7332,
      "step": 552
    },
    {
      "epoch": 0.09594031922276197,
      "grad_norm": 0.7595632076263428,
      "learning_rate": 9.592367736339984e-06,
      "loss": 0.7043,
      "step": 553
    },
    {
      "epoch": 0.09611380985426787,
      "grad_norm": 0.792465090751648,
      "learning_rate": 9.609713790112751e-06,
      "loss": 0.7013,
      "step": 554
    },
    {
      "epoch": 0.09628730048577377,
      "grad_norm": 0.7701244950294495,
      "learning_rate": 9.627059843885516e-06,
      "loss": 0.791,
      "step": 555
    },
    {
      "epoch": 0.09646079111727966,
      "grad_norm": 0.8943719267845154,
      "learning_rate": 9.644405897658283e-06,
      "loss": 0.7175,
      "step": 556
    },
    {
      "epoch": 0.09663428174878556,
      "grad_norm": 1.3731937408447266,
      "learning_rate": 9.66175195143105e-06,
      "loss": 0.7277,
      "step": 557
    },
    {
      "epoch": 0.09680777238029147,
      "grad_norm": 0.8585498332977295,
      "learning_rate": 9.679098005203816e-06,
      "loss": 0.6948,
      "step": 558
    },
    {
      "epoch": 0.09698126301179737,
      "grad_norm": 0.6464988589286804,
      "learning_rate": 9.696444058976584e-06,
      "loss": 0.8181,
      "step": 559
    },
    {
      "epoch": 0.09715475364330327,
      "grad_norm": 0.8568048477172852,
      "learning_rate": 9.713790112749351e-06,
      "loss": 0.7307,
      "step": 560
    },
    {
      "epoch": 0.09732824427480916,
      "grad_norm": 0.8411333560943604,
      "learning_rate": 9.731136166522118e-06,
      "loss": 0.6648,
      "step": 561
    },
    {
      "epoch": 0.09750173490631506,
      "grad_norm": 7.223629474639893,
      "learning_rate": 9.748482220294883e-06,
      "loss": 0.6119,
      "step": 562
    },
    {
      "epoch": 0.09767522553782096,
      "grad_norm": 0.6756853461265564,
      "learning_rate": 9.76582827406765e-06,
      "loss": 0.6816,
      "step": 563
    },
    {
      "epoch": 0.09784871616932686,
      "grad_norm": 0.6368600130081177,
      "learning_rate": 9.783174327840416e-06,
      "loss": 0.7976,
      "step": 564
    },
    {
      "epoch": 0.09802220680083276,
      "grad_norm": 1.1108633279800415,
      "learning_rate": 9.800520381613183e-06,
      "loss": 0.6875,
      "step": 565
    },
    {
      "epoch": 0.09819569743233865,
      "grad_norm": 0.842835009098053,
      "learning_rate": 9.817866435385951e-06,
      "loss": 0.7048,
      "step": 566
    },
    {
      "epoch": 0.09836918806384455,
      "grad_norm": 0.6681727170944214,
      "learning_rate": 9.835212489158718e-06,
      "loss": 0.7976,
      "step": 567
    },
    {
      "epoch": 0.09854267869535045,
      "grad_norm": 0.8945640325546265,
      "learning_rate": 9.852558542931485e-06,
      "loss": 0.8228,
      "step": 568
    },
    {
      "epoch": 0.09871616932685635,
      "grad_norm": 0.6476752161979675,
      "learning_rate": 9.869904596704251e-06,
      "loss": 0.7466,
      "step": 569
    },
    {
      "epoch": 0.09888965995836224,
      "grad_norm": 0.9636639356613159,
      "learning_rate": 9.887250650477016e-06,
      "loss": 0.8408,
      "step": 570
    },
    {
      "epoch": 0.09906315058986814,
      "grad_norm": 0.7499241828918457,
      "learning_rate": 9.904596704249783e-06,
      "loss": 0.7227,
      "step": 571
    },
    {
      "epoch": 0.09923664122137404,
      "grad_norm": 0.8054186105728149,
      "learning_rate": 9.92194275802255e-06,
      "loss": 0.712,
      "step": 572
    },
    {
      "epoch": 0.09941013185287995,
      "grad_norm": 0.5592566132545471,
      "learning_rate": 9.939288811795318e-06,
      "loss": 0.8115,
      "step": 573
    },
    {
      "epoch": 0.09958362248438585,
      "grad_norm": 1.003616213798523,
      "learning_rate": 9.956634865568085e-06,
      "loss": 0.6499,
      "step": 574
    },
    {
      "epoch": 0.09975711311589174,
      "grad_norm": 0.7742766737937927,
      "learning_rate": 9.973980919340852e-06,
      "loss": 0.7188,
      "step": 575
    },
    {
      "epoch": 0.09993060374739764,
      "grad_norm": 1.2237392663955688,
      "learning_rate": 9.991326973113618e-06,
      "loss": 0.6455,
      "step": 576
    },
    {
      "epoch": 0.10010409437890354,
      "grad_norm": 0.8949840068817139,
      "learning_rate": 1.0008673026886383e-05,
      "loss": 0.5955,
      "step": 577
    },
    {
      "epoch": 0.10027758501040944,
      "grad_norm": 0.9954709410667419,
      "learning_rate": 1.002601908065915e-05,
      "loss": 0.6442,
      "step": 578
    },
    {
      "epoch": 0.10045107564191534,
      "grad_norm": 0.939723551273346,
      "learning_rate": 1.0043365134431917e-05,
      "loss": 0.6877,
      "step": 579
    },
    {
      "epoch": 0.10062456627342123,
      "grad_norm": 0.7950365543365479,
      "learning_rate": 1.0060711188204683e-05,
      "loss": 0.7377,
      "step": 580
    },
    {
      "epoch": 0.10079805690492713,
      "grad_norm": 0.8031066656112671,
      "learning_rate": 1.007805724197745e-05,
      "loss": 0.7664,
      "step": 581
    },
    {
      "epoch": 0.10097154753643303,
      "grad_norm": 0.7567896842956543,
      "learning_rate": 1.0095403295750217e-05,
      "loss": 0.7759,
      "step": 582
    },
    {
      "epoch": 0.10114503816793893,
      "grad_norm": 0.6213170886039734,
      "learning_rate": 1.0112749349522983e-05,
      "loss": 0.8481,
      "step": 583
    },
    {
      "epoch": 0.10131852879944483,
      "grad_norm": 0.6094651222229004,
      "learning_rate": 1.0130095403295752e-05,
      "loss": 0.7006,
      "step": 584
    },
    {
      "epoch": 0.10149201943095072,
      "grad_norm": 1.0541772842407227,
      "learning_rate": 1.0147441457068519e-05,
      "loss": 0.6729,
      "step": 585
    },
    {
      "epoch": 0.10166551006245662,
      "grad_norm": 0.7511570453643799,
      "learning_rate": 1.0164787510841285e-05,
      "loss": 0.7402,
      "step": 586
    },
    {
      "epoch": 0.10183900069396253,
      "grad_norm": 0.6804909110069275,
      "learning_rate": 1.0182133564614052e-05,
      "loss": 0.5862,
      "step": 587
    },
    {
      "epoch": 0.10201249132546843,
      "grad_norm": 0.9981895685195923,
      "learning_rate": 1.0199479618386819e-05,
      "loss": 0.8435,
      "step": 588
    },
    {
      "epoch": 0.10218598195697433,
      "grad_norm": 0.9453763365745544,
      "learning_rate": 1.0216825672159585e-05,
      "loss": 0.6151,
      "step": 589
    },
    {
      "epoch": 0.10235947258848022,
      "grad_norm": 0.5910636186599731,
      "learning_rate": 1.0234171725932352e-05,
      "loss": 0.7754,
      "step": 590
    },
    {
      "epoch": 0.10253296321998612,
      "grad_norm": 0.6517502069473267,
      "learning_rate": 1.0251517779705119e-05,
      "loss": 0.7522,
      "step": 591
    },
    {
      "epoch": 0.10270645385149202,
      "grad_norm": 1.0400340557098389,
      "learning_rate": 1.0268863833477884e-05,
      "loss": 0.884,
      "step": 592
    },
    {
      "epoch": 0.10287994448299792,
      "grad_norm": 0.9368115067481995,
      "learning_rate": 1.028620988725065e-05,
      "loss": 0.6853,
      "step": 593
    },
    {
      "epoch": 0.10305343511450382,
      "grad_norm": 0.6270506978034973,
      "learning_rate": 1.0303555941023417e-05,
      "loss": 0.7229,
      "step": 594
    },
    {
      "epoch": 0.10322692574600971,
      "grad_norm": 1.6493935585021973,
      "learning_rate": 1.0320901994796184e-05,
      "loss": 0.6746,
      "step": 595
    },
    {
      "epoch": 0.10340041637751561,
      "grad_norm": 0.6048089861869812,
      "learning_rate": 1.033824804856895e-05,
      "loss": 0.7174,
      "step": 596
    },
    {
      "epoch": 0.10357390700902151,
      "grad_norm": 1.015458345413208,
      "learning_rate": 1.0355594102341717e-05,
      "loss": 0.8799,
      "step": 597
    },
    {
      "epoch": 0.10374739764052741,
      "grad_norm": 0.6777968406677246,
      "learning_rate": 1.0372940156114484e-05,
      "loss": 0.8088,
      "step": 598
    },
    {
      "epoch": 0.10392088827203332,
      "grad_norm": 0.8517946600914001,
      "learning_rate": 1.0390286209887252e-05,
      "loss": 0.6487,
      "step": 599
    },
    {
      "epoch": 0.1040943789035392,
      "grad_norm": 0.880364179611206,
      "learning_rate": 1.0407632263660019e-05,
      "loss": 0.7549,
      "step": 600
    },
    {
      "epoch": 0.1042678695350451,
      "grad_norm": 0.6786611676216125,
      "learning_rate": 1.0424978317432786e-05,
      "loss": 0.7529,
      "step": 601
    },
    {
      "epoch": 0.104441360166551,
      "grad_norm": 1.0398542881011963,
      "learning_rate": 1.0442324371205552e-05,
      "loss": 0.6331,
      "step": 602
    },
    {
      "epoch": 0.10461485079805691,
      "grad_norm": 0.731816828250885,
      "learning_rate": 1.0459670424978319e-05,
      "loss": 0.9141,
      "step": 603
    },
    {
      "epoch": 0.10478834142956281,
      "grad_norm": 0.8709906935691833,
      "learning_rate": 1.0477016478751086e-05,
      "loss": 0.6921,
      "step": 604
    },
    {
      "epoch": 0.1049618320610687,
      "grad_norm": 0.9583622217178345,
      "learning_rate": 1.0494362532523852e-05,
      "loss": 0.636,
      "step": 605
    },
    {
      "epoch": 0.1051353226925746,
      "grad_norm": 1.5152076482772827,
      "learning_rate": 1.051170858629662e-05,
      "loss": 0.676,
      "step": 606
    },
    {
      "epoch": 0.1053088133240805,
      "grad_norm": 1.03229820728302,
      "learning_rate": 1.0529054640069384e-05,
      "loss": 0.5743,
      "step": 607
    },
    {
      "epoch": 0.1054823039555864,
      "grad_norm": 0.5453727841377258,
      "learning_rate": 1.0546400693842151e-05,
      "loss": 0.7979,
      "step": 608
    },
    {
      "epoch": 0.1056557945870923,
      "grad_norm": 0.625810444355011,
      "learning_rate": 1.0563746747614918e-05,
      "loss": 0.7998,
      "step": 609
    },
    {
      "epoch": 0.10582928521859819,
      "grad_norm": 1.0403211116790771,
      "learning_rate": 1.0581092801387684e-05,
      "loss": 0.582,
      "step": 610
    },
    {
      "epoch": 0.10600277585010409,
      "grad_norm": 1.1652127504348755,
      "learning_rate": 1.0598438855160451e-05,
      "loss": 0.6367,
      "step": 611
    },
    {
      "epoch": 0.10617626648161,
      "grad_norm": 0.8623735904693604,
      "learning_rate": 1.0615784908933218e-05,
      "loss": 0.7267,
      "step": 612
    },
    {
      "epoch": 0.1063497571131159,
      "grad_norm": 0.5244912505149841,
      "learning_rate": 1.0633130962705984e-05,
      "loss": 0.8396,
      "step": 613
    },
    {
      "epoch": 0.1065232477446218,
      "grad_norm": 0.7100646495819092,
      "learning_rate": 1.0650477016478753e-05,
      "loss": 0.8169,
      "step": 614
    },
    {
      "epoch": 0.10669673837612768,
      "grad_norm": 0.5875173211097717,
      "learning_rate": 1.066782307025152e-05,
      "loss": 0.6633,
      "step": 615
    },
    {
      "epoch": 0.10687022900763359,
      "grad_norm": 0.7332322597503662,
      "learning_rate": 1.0685169124024286e-05,
      "loss": 0.8625,
      "step": 616
    },
    {
      "epoch": 0.10704371963913949,
      "grad_norm": 0.6259750127792358,
      "learning_rate": 1.0702515177797053e-05,
      "loss": 0.8584,
      "step": 617
    },
    {
      "epoch": 0.10721721027064539,
      "grad_norm": 0.7651997208595276,
      "learning_rate": 1.071986123156982e-05,
      "loss": 0.7227,
      "step": 618
    },
    {
      "epoch": 0.10739070090215129,
      "grad_norm": 0.9559131264686584,
      "learning_rate": 1.0737207285342586e-05,
      "loss": 0.8584,
      "step": 619
    },
    {
      "epoch": 0.10756419153365718,
      "grad_norm": 0.8246591687202454,
      "learning_rate": 1.0754553339115353e-05,
      "loss": 0.6956,
      "step": 620
    },
    {
      "epoch": 0.10773768216516308,
      "grad_norm": 1.5528041124343872,
      "learning_rate": 1.077189939288812e-05,
      "loss": 0.6495,
      "step": 621
    },
    {
      "epoch": 0.10791117279666898,
      "grad_norm": 0.8507909774780273,
      "learning_rate": 1.0789245446660885e-05,
      "loss": 0.6914,
      "step": 622
    },
    {
      "epoch": 0.10808466342817488,
      "grad_norm": 1.2405694723129272,
      "learning_rate": 1.0806591500433651e-05,
      "loss": 0.6771,
      "step": 623
    },
    {
      "epoch": 0.10825815405968078,
      "grad_norm": 0.8569530248641968,
      "learning_rate": 1.0823937554206418e-05,
      "loss": 0.5836,
      "step": 624
    },
    {
      "epoch": 0.10843164469118667,
      "grad_norm": 0.7054854035377502,
      "learning_rate": 1.0841283607979185e-05,
      "loss": 0.7961,
      "step": 625
    },
    {
      "epoch": 0.10860513532269257,
      "grad_norm": 0.6264186501502991,
      "learning_rate": 1.0858629661751951e-05,
      "loss": 0.7217,
      "step": 626
    },
    {
      "epoch": 0.10877862595419847,
      "grad_norm": 0.7073953747749329,
      "learning_rate": 1.0875975715524718e-05,
      "loss": 0.7085,
      "step": 627
    },
    {
      "epoch": 0.10895211658570438,
      "grad_norm": 1.7915912866592407,
      "learning_rate": 1.0893321769297485e-05,
      "loss": 0.6115,
      "step": 628
    },
    {
      "epoch": 0.10912560721721028,
      "grad_norm": 1.0424374341964722,
      "learning_rate": 1.0910667823070253e-05,
      "loss": 0.7844,
      "step": 629
    },
    {
      "epoch": 0.10929909784871616,
      "grad_norm": 1.5371145009994507,
      "learning_rate": 1.092801387684302e-05,
      "loss": 0.6057,
      "step": 630
    },
    {
      "epoch": 0.10947258848022207,
      "grad_norm": 0.684110164642334,
      "learning_rate": 1.0945359930615787e-05,
      "loss": 0.7847,
      "step": 631
    },
    {
      "epoch": 0.10964607911172797,
      "grad_norm": 0.6034166812896729,
      "learning_rate": 1.0962705984388553e-05,
      "loss": 0.7946,
      "step": 632
    },
    {
      "epoch": 0.10981956974323387,
      "grad_norm": 0.9649418592453003,
      "learning_rate": 1.098005203816132e-05,
      "loss": 0.5924,
      "step": 633
    },
    {
      "epoch": 0.10999306037473977,
      "grad_norm": 1.540686011314392,
      "learning_rate": 1.0997398091934087e-05,
      "loss": 0.6996,
      "step": 634
    },
    {
      "epoch": 0.11016655100624566,
      "grad_norm": 0.6039450764656067,
      "learning_rate": 1.1014744145706853e-05,
      "loss": 0.772,
      "step": 635
    },
    {
      "epoch": 0.11034004163775156,
      "grad_norm": 0.7755488753318787,
      "learning_rate": 1.103209019947962e-05,
      "loss": 0.7344,
      "step": 636
    },
    {
      "epoch": 0.11051353226925746,
      "grad_norm": 0.6978097558021545,
      "learning_rate": 1.1049436253252385e-05,
      "loss": 0.8052,
      "step": 637
    },
    {
      "epoch": 0.11068702290076336,
      "grad_norm": 0.9445908665657043,
      "learning_rate": 1.1066782307025152e-05,
      "loss": 0.6602,
      "step": 638
    },
    {
      "epoch": 0.11086051353226926,
      "grad_norm": 3.0431437492370605,
      "learning_rate": 1.1084128360797918e-05,
      "loss": 0.7188,
      "step": 639
    },
    {
      "epoch": 0.11103400416377515,
      "grad_norm": 1.0811216831207275,
      "learning_rate": 1.1101474414570685e-05,
      "loss": 0.7437,
      "step": 640
    },
    {
      "epoch": 0.11120749479528105,
      "grad_norm": 0.763873279094696,
      "learning_rate": 1.1118820468343452e-05,
      "loss": 0.7471,
      "step": 641
    },
    {
      "epoch": 0.11138098542678695,
      "grad_norm": 0.7147685289382935,
      "learning_rate": 1.1136166522116219e-05,
      "loss": 0.6162,
      "step": 642
    },
    {
      "epoch": 0.11155447605829286,
      "grad_norm": 1.0152665376663208,
      "learning_rate": 1.1153512575888985e-05,
      "loss": 0.7397,
      "step": 643
    },
    {
      "epoch": 0.11172796668979876,
      "grad_norm": 0.8622577786445618,
      "learning_rate": 1.1170858629661754e-05,
      "loss": 0.6323,
      "step": 644
    },
    {
      "epoch": 0.11190145732130465,
      "grad_norm": 0.893168032169342,
      "learning_rate": 1.118820468343452e-05,
      "loss": 0.6046,
      "step": 645
    },
    {
      "epoch": 0.11207494795281055,
      "grad_norm": 0.9592801332473755,
      "learning_rate": 1.1205550737207287e-05,
      "loss": 0.621,
      "step": 646
    },
    {
      "epoch": 0.11224843858431645,
      "grad_norm": 0.8946881294250488,
      "learning_rate": 1.1222896790980054e-05,
      "loss": 0.6152,
      "step": 647
    },
    {
      "epoch": 0.11242192921582235,
      "grad_norm": 1.1318047046661377,
      "learning_rate": 1.124024284475282e-05,
      "loss": 0.5914,
      "step": 648
    },
    {
      "epoch": 0.11259541984732824,
      "grad_norm": 0.7299327254295349,
      "learning_rate": 1.1257588898525587e-05,
      "loss": 0.761,
      "step": 649
    },
    {
      "epoch": 0.11276891047883414,
      "grad_norm": 0.785463273525238,
      "learning_rate": 1.1274934952298354e-05,
      "loss": 0.6516,
      "step": 650
    },
    {
      "epoch": 0.11294240111034004,
      "grad_norm": 0.7696375250816345,
      "learning_rate": 1.129228100607112e-05,
      "loss": 0.7043,
      "step": 651
    },
    {
      "epoch": 0.11311589174184594,
      "grad_norm": 1.050863265991211,
      "learning_rate": 1.1309627059843885e-05,
      "loss": 0.7544,
      "step": 652
    },
    {
      "epoch": 0.11328938237335184,
      "grad_norm": 1.3764386177062988,
      "learning_rate": 1.1326973113616652e-05,
      "loss": 0.885,
      "step": 653
    },
    {
      "epoch": 0.11346287300485773,
      "grad_norm": 0.8897507786750793,
      "learning_rate": 1.1344319167389419e-05,
      "loss": 0.6536,
      "step": 654
    },
    {
      "epoch": 0.11363636363636363,
      "grad_norm": 1.1961376667022705,
      "learning_rate": 1.1361665221162186e-05,
      "loss": 0.7806,
      "step": 655
    },
    {
      "epoch": 0.11380985426786953,
      "grad_norm": 1.0161396265029907,
      "learning_rate": 1.1379011274934952e-05,
      "loss": 0.7877,
      "step": 656
    },
    {
      "epoch": 0.11398334489937544,
      "grad_norm": 0.9606118202209473,
      "learning_rate": 1.1396357328707719e-05,
      "loss": 0.6857,
      "step": 657
    },
    {
      "epoch": 0.11415683553088134,
      "grad_norm": 1.3000298738479614,
      "learning_rate": 1.1413703382480487e-05,
      "loss": 0.6887,
      "step": 658
    },
    {
      "epoch": 0.11433032616238722,
      "grad_norm": 0.9779266119003296,
      "learning_rate": 1.1431049436253254e-05,
      "loss": 0.6284,
      "step": 659
    },
    {
      "epoch": 0.11450381679389313,
      "grad_norm": 0.6240894198417664,
      "learning_rate": 1.144839549002602e-05,
      "loss": 0.7858,
      "step": 660
    },
    {
      "epoch": 0.11467730742539903,
      "grad_norm": 1.1069626808166504,
      "learning_rate": 1.1465741543798787e-05,
      "loss": 0.7329,
      "step": 661
    },
    {
      "epoch": 0.11485079805690493,
      "grad_norm": 0.8047638535499573,
      "learning_rate": 1.1483087597571554e-05,
      "loss": 0.7887,
      "step": 662
    },
    {
      "epoch": 0.11502428868841083,
      "grad_norm": 0.7948787808418274,
      "learning_rate": 1.150043365134432e-05,
      "loss": 0.6184,
      "step": 663
    },
    {
      "epoch": 0.11519777931991672,
      "grad_norm": 0.6296604871749878,
      "learning_rate": 1.1517779705117088e-05,
      "loss": 0.7542,
      "step": 664
    },
    {
      "epoch": 0.11537126995142262,
      "grad_norm": 1.9882172346115112,
      "learning_rate": 1.1535125758889854e-05,
      "loss": 0.8248,
      "step": 665
    },
    {
      "epoch": 0.11554476058292852,
      "grad_norm": 0.7945740222930908,
      "learning_rate": 1.1552471812662621e-05,
      "loss": 0.7227,
      "step": 666
    },
    {
      "epoch": 0.11571825121443442,
      "grad_norm": 1.1145966053009033,
      "learning_rate": 1.1569817866435386e-05,
      "loss": 0.7614,
      "step": 667
    },
    {
      "epoch": 0.11589174184594032,
      "grad_norm": 0.8767792582511902,
      "learning_rate": 1.1587163920208153e-05,
      "loss": 0.9128,
      "step": 668
    },
    {
      "epoch": 0.11606523247744621,
      "grad_norm": 0.6942266225814819,
      "learning_rate": 1.160450997398092e-05,
      "loss": 0.8083,
      "step": 669
    },
    {
      "epoch": 0.11623872310895211,
      "grad_norm": 0.8999573588371277,
      "learning_rate": 1.1621856027753686e-05,
      "loss": 0.5856,
      "step": 670
    },
    {
      "epoch": 0.11641221374045801,
      "grad_norm": 0.9616506099700928,
      "learning_rate": 1.1639202081526453e-05,
      "loss": 0.6599,
      "step": 671
    },
    {
      "epoch": 0.11658570437196392,
      "grad_norm": 1.0585647821426392,
      "learning_rate": 1.165654813529922e-05,
      "loss": 0.6489,
      "step": 672
    },
    {
      "epoch": 0.11675919500346982,
      "grad_norm": 0.7967655658721924,
      "learning_rate": 1.1673894189071988e-05,
      "loss": 0.5808,
      "step": 673
    },
    {
      "epoch": 0.1169326856349757,
      "grad_norm": 1.2765015363693237,
      "learning_rate": 1.1691240242844754e-05,
      "loss": 0.6726,
      "step": 674
    },
    {
      "epoch": 0.11710617626648161,
      "grad_norm": 0.666652500629425,
      "learning_rate": 1.1708586296617521e-05,
      "loss": 0.7703,
      "step": 675
    },
    {
      "epoch": 0.11727966689798751,
      "grad_norm": 0.9283226132392883,
      "learning_rate": 1.1725932350390288e-05,
      "loss": 0.6865,
      "step": 676
    },
    {
      "epoch": 0.11745315752949341,
      "grad_norm": 0.8013582229614258,
      "learning_rate": 1.1743278404163055e-05,
      "loss": 0.7915,
      "step": 677
    },
    {
      "epoch": 0.11762664816099931,
      "grad_norm": 1.0678712129592896,
      "learning_rate": 1.1760624457935821e-05,
      "loss": 0.6185,
      "step": 678
    },
    {
      "epoch": 0.1178001387925052,
      "grad_norm": 1.1896597146987915,
      "learning_rate": 1.1777970511708588e-05,
      "loss": 0.6167,
      "step": 679
    },
    {
      "epoch": 0.1179736294240111,
      "grad_norm": 0.7107986211776733,
      "learning_rate": 1.1795316565481355e-05,
      "loss": 0.7207,
      "step": 680
    },
    {
      "epoch": 0.118147120055517,
      "grad_norm": 1.224815011024475,
      "learning_rate": 1.181266261925412e-05,
      "loss": 0.7278,
      "step": 681
    },
    {
      "epoch": 0.1183206106870229,
      "grad_norm": 0.6233601570129395,
      "learning_rate": 1.1830008673026886e-05,
      "loss": 0.7137,
      "step": 682
    },
    {
      "epoch": 0.1184941013185288,
      "grad_norm": 1.0943691730499268,
      "learning_rate": 1.1847354726799653e-05,
      "loss": 0.5635,
      "step": 683
    },
    {
      "epoch": 0.11866759195003469,
      "grad_norm": 1.5748167037963867,
      "learning_rate": 1.186470078057242e-05,
      "loss": 0.8718,
      "step": 684
    },
    {
      "epoch": 0.1188410825815406,
      "grad_norm": 0.5649005174636841,
      "learning_rate": 1.1882046834345186e-05,
      "loss": 0.7893,
      "step": 685
    },
    {
      "epoch": 0.1190145732130465,
      "grad_norm": 2.1243364810943604,
      "learning_rate": 1.1899392888117953e-05,
      "loss": 0.7123,
      "step": 686
    },
    {
      "epoch": 0.1191880638445524,
      "grad_norm": 1.023429274559021,
      "learning_rate": 1.191673894189072e-05,
      "loss": 0.5942,
      "step": 687
    },
    {
      "epoch": 0.1193615544760583,
      "grad_norm": 0.6872596740722656,
      "learning_rate": 1.1934084995663488e-05,
      "loss": 0.7377,
      "step": 688
    },
    {
      "epoch": 0.11953504510756419,
      "grad_norm": 0.7104584574699402,
      "learning_rate": 1.1951431049436255e-05,
      "loss": 0.7969,
      "step": 689
    },
    {
      "epoch": 0.11970853573907009,
      "grad_norm": 0.8265088796615601,
      "learning_rate": 1.1968777103209022e-05,
      "loss": 0.6846,
      "step": 690
    },
    {
      "epoch": 0.11988202637057599,
      "grad_norm": 1.5311951637268066,
      "learning_rate": 1.1986123156981788e-05,
      "loss": 0.5569,
      "step": 691
    },
    {
      "epoch": 0.12005551700208189,
      "grad_norm": 1.1079038381576538,
      "learning_rate": 1.2003469210754555e-05,
      "loss": 0.7522,
      "step": 692
    },
    {
      "epoch": 0.12022900763358779,
      "grad_norm": 1.1485226154327393,
      "learning_rate": 1.2020815264527322e-05,
      "loss": 0.7632,
      "step": 693
    },
    {
      "epoch": 0.12040249826509368,
      "grad_norm": 0.621604323387146,
      "learning_rate": 1.2038161318300088e-05,
      "loss": 0.7302,
      "step": 694
    },
    {
      "epoch": 0.12057598889659958,
      "grad_norm": 0.8844202756881714,
      "learning_rate": 1.2055507372072855e-05,
      "loss": 0.7998,
      "step": 695
    },
    {
      "epoch": 0.12074947952810548,
      "grad_norm": 0.7744740843772888,
      "learning_rate": 1.207285342584562e-05,
      "loss": 0.6853,
      "step": 696
    },
    {
      "epoch": 0.12092297015961138,
      "grad_norm": 0.6762483716011047,
      "learning_rate": 1.2090199479618387e-05,
      "loss": 0.8545,
      "step": 697
    },
    {
      "epoch": 0.12109646079111729,
      "grad_norm": 1.5488284826278687,
      "learning_rate": 1.2107545533391153e-05,
      "loss": 0.6328,
      "step": 698
    },
    {
      "epoch": 0.12126995142262317,
      "grad_norm": 0.716681957244873,
      "learning_rate": 1.212489158716392e-05,
      "loss": 0.7561,
      "step": 699
    },
    {
      "epoch": 0.12144344205412907,
      "grad_norm": 0.8789547681808472,
      "learning_rate": 1.2142237640936687e-05,
      "loss": 0.5847,
      "step": 700
    },
    {
      "epoch": 0.12161693268563498,
      "grad_norm": 1.1625703573226929,
      "learning_rate": 1.2159583694709454e-05,
      "loss": 0.5979,
      "step": 701
    },
    {
      "epoch": 0.12179042331714088,
      "grad_norm": 0.843895435333252,
      "learning_rate": 1.217692974848222e-05,
      "loss": 0.7214,
      "step": 702
    },
    {
      "epoch": 0.12196391394864678,
      "grad_norm": 0.9099196195602417,
      "learning_rate": 1.2194275802254989e-05,
      "loss": 0.6191,
      "step": 703
    },
    {
      "epoch": 0.12213740458015267,
      "grad_norm": 1.0822001695632935,
      "learning_rate": 1.2211621856027755e-05,
      "loss": 0.6824,
      "step": 704
    },
    {
      "epoch": 0.12231089521165857,
      "grad_norm": 1.053365707397461,
      "learning_rate": 1.2228967909800522e-05,
      "loss": 0.7758,
      "step": 705
    },
    {
      "epoch": 0.12248438584316447,
      "grad_norm": 1.6115442514419556,
      "learning_rate": 1.2246313963573289e-05,
      "loss": 0.6538,
      "step": 706
    },
    {
      "epoch": 0.12265787647467037,
      "grad_norm": 0.63421231508255,
      "learning_rate": 1.2263660017346055e-05,
      "loss": 0.7159,
      "step": 707
    },
    {
      "epoch": 0.12283136710617627,
      "grad_norm": 0.7350783944129944,
      "learning_rate": 1.2281006071118822e-05,
      "loss": 0.8186,
      "step": 708
    },
    {
      "epoch": 0.12300485773768216,
      "grad_norm": 0.6793596148490906,
      "learning_rate": 1.2298352124891589e-05,
      "loss": 0.7205,
      "step": 709
    },
    {
      "epoch": 0.12317834836918806,
      "grad_norm": 0.8642478585243225,
      "learning_rate": 1.2315698178664356e-05,
      "loss": 0.7021,
      "step": 710
    },
    {
      "epoch": 0.12335183900069396,
      "grad_norm": 0.7911598086357117,
      "learning_rate": 1.233304423243712e-05,
      "loss": 0.7319,
      "step": 711
    },
    {
      "epoch": 0.12352532963219987,
      "grad_norm": 0.5477712154388428,
      "learning_rate": 1.2350390286209887e-05,
      "loss": 0.6874,
      "step": 712
    },
    {
      "epoch": 0.12369882026370577,
      "grad_norm": 0.5802223682403564,
      "learning_rate": 1.2367736339982654e-05,
      "loss": 0.7007,
      "step": 713
    },
    {
      "epoch": 0.12387231089521165,
      "grad_norm": 0.8528922200202942,
      "learning_rate": 1.238508239375542e-05,
      "loss": 0.6584,
      "step": 714
    },
    {
      "epoch": 0.12404580152671756,
      "grad_norm": 2.9769327640533447,
      "learning_rate": 1.2402428447528187e-05,
      "loss": 0.7139,
      "step": 715
    },
    {
      "epoch": 0.12421929215822346,
      "grad_norm": 1.6489852666854858,
      "learning_rate": 1.2419774501300954e-05,
      "loss": 0.6698,
      "step": 716
    },
    {
      "epoch": 0.12439278278972936,
      "grad_norm": 0.5939946174621582,
      "learning_rate": 1.2437120555073722e-05,
      "loss": 0.9434,
      "step": 717
    },
    {
      "epoch": 0.12456627342123526,
      "grad_norm": 1.066463589668274,
      "learning_rate": 1.2454466608846489e-05,
      "loss": 0.7593,
      "step": 718
    },
    {
      "epoch": 0.12473976405274115,
      "grad_norm": 0.7959454655647278,
      "learning_rate": 1.2471812662619256e-05,
      "loss": 0.7546,
      "step": 719
    },
    {
      "epoch": 0.12491325468424705,
      "grad_norm": 0.763336718082428,
      "learning_rate": 1.2489158716392022e-05,
      "loss": 0.7537,
      "step": 720
    },
    {
      "epoch": 0.12508674531575295,
      "grad_norm": 1.3524776697158813,
      "learning_rate": 1.250650477016479e-05,
      "loss": 0.6158,
      "step": 721
    },
    {
      "epoch": 0.12526023594725885,
      "grad_norm": 0.8605296611785889,
      "learning_rate": 1.2523850823937556e-05,
      "loss": 0.7356,
      "step": 722
    },
    {
      "epoch": 0.12543372657876475,
      "grad_norm": 0.9525180459022522,
      "learning_rate": 1.2541196877710323e-05,
      "loss": 0.6804,
      "step": 723
    },
    {
      "epoch": 0.12560721721027066,
      "grad_norm": 0.6528279185295105,
      "learning_rate": 1.255854293148309e-05,
      "loss": 0.8223,
      "step": 724
    },
    {
      "epoch": 0.12578070784177656,
      "grad_norm": 0.7762887477874756,
      "learning_rate": 1.2575888985255856e-05,
      "loss": 0.6045,
      "step": 725
    },
    {
      "epoch": 0.12595419847328243,
      "grad_norm": 0.918840229511261,
      "learning_rate": 1.2593235039028621e-05,
      "loss": 0.5986,
      "step": 726
    },
    {
      "epoch": 0.12612768910478833,
      "grad_norm": 0.98841392993927,
      "learning_rate": 1.2610581092801388e-05,
      "loss": 0.66,
      "step": 727
    },
    {
      "epoch": 0.12630117973629423,
      "grad_norm": 1.0014210939407349,
      "learning_rate": 1.2627927146574154e-05,
      "loss": 0.748,
      "step": 728
    },
    {
      "epoch": 0.12647467036780013,
      "grad_norm": 0.89778071641922,
      "learning_rate": 1.2645273200346921e-05,
      "loss": 0.611,
      "step": 729
    },
    {
      "epoch": 0.12664816099930604,
      "grad_norm": 1.6311657428741455,
      "learning_rate": 1.2662619254119688e-05,
      "loss": 0.6222,
      "step": 730
    },
    {
      "epoch": 0.12682165163081194,
      "grad_norm": 0.8890562057495117,
      "learning_rate": 1.2679965307892454e-05,
      "loss": 0.7314,
      "step": 731
    },
    {
      "epoch": 0.12699514226231784,
      "grad_norm": 0.7859386801719666,
      "learning_rate": 1.2697311361665223e-05,
      "loss": 0.6956,
      "step": 732
    },
    {
      "epoch": 0.12716863289382374,
      "grad_norm": 1.78225839138031,
      "learning_rate": 1.271465741543799e-05,
      "loss": 0.7323,
      "step": 733
    },
    {
      "epoch": 0.12734212352532964,
      "grad_norm": 1.3801956176757812,
      "learning_rate": 1.2732003469210756e-05,
      "loss": 0.8142,
      "step": 734
    },
    {
      "epoch": 0.12751561415683554,
      "grad_norm": 0.9562146663665771,
      "learning_rate": 1.2749349522983523e-05,
      "loss": 0.6477,
      "step": 735
    },
    {
      "epoch": 0.12768910478834142,
      "grad_norm": 2.3081870079040527,
      "learning_rate": 1.276669557675629e-05,
      "loss": 0.8076,
      "step": 736
    },
    {
      "epoch": 0.12786259541984732,
      "grad_norm": 0.7006694674491882,
      "learning_rate": 1.2784041630529056e-05,
      "loss": 0.7605,
      "step": 737
    },
    {
      "epoch": 0.12803608605135322,
      "grad_norm": 0.8642357587814331,
      "learning_rate": 1.2801387684301823e-05,
      "loss": 0.6172,
      "step": 738
    },
    {
      "epoch": 0.12820957668285912,
      "grad_norm": 1.2619918584823608,
      "learning_rate": 1.281873373807459e-05,
      "loss": 0.5737,
      "step": 739
    },
    {
      "epoch": 0.12838306731436502,
      "grad_norm": 0.7796722650527954,
      "learning_rate": 1.2836079791847356e-05,
      "loss": 0.636,
      "step": 740
    },
    {
      "epoch": 0.12855655794587093,
      "grad_norm": 0.8895028829574585,
      "learning_rate": 1.2853425845620121e-05,
      "loss": 0.6311,
      "step": 741
    },
    {
      "epoch": 0.12873004857737683,
      "grad_norm": 1.0069208145141602,
      "learning_rate": 1.2870771899392888e-05,
      "loss": 0.5934,
      "step": 742
    },
    {
      "epoch": 0.12890353920888273,
      "grad_norm": 0.7322602868080139,
      "learning_rate": 1.2888117953165655e-05,
      "loss": 0.5977,
      "step": 743
    },
    {
      "epoch": 0.12907702984038863,
      "grad_norm": 1.4024487733840942,
      "learning_rate": 1.2905464006938421e-05,
      "loss": 0.5786,
      "step": 744
    },
    {
      "epoch": 0.12925052047189453,
      "grad_norm": 0.7118971943855286,
      "learning_rate": 1.2922810060711188e-05,
      "loss": 0.8091,
      "step": 745
    },
    {
      "epoch": 0.1294240111034004,
      "grad_norm": 0.9231390357017517,
      "learning_rate": 1.2940156114483955e-05,
      "loss": 0.7593,
      "step": 746
    },
    {
      "epoch": 0.1295975017349063,
      "grad_norm": 1.3302514553070068,
      "learning_rate": 1.2957502168256723e-05,
      "loss": 0.7173,
      "step": 747
    },
    {
      "epoch": 0.1297709923664122,
      "grad_norm": 1.1704890727996826,
      "learning_rate": 1.297484822202949e-05,
      "loss": 0.5905,
      "step": 748
    },
    {
      "epoch": 0.1299444829979181,
      "grad_norm": 0.7081783413887024,
      "learning_rate": 1.2992194275802257e-05,
      "loss": 0.7327,
      "step": 749
    },
    {
      "epoch": 0.130117973629424,
      "grad_norm": 0.6808786392211914,
      "learning_rate": 1.3009540329575023e-05,
      "loss": 0.8208,
      "step": 750
    },
    {
      "epoch": 0.1302914642609299,
      "grad_norm": 0.7967669367790222,
      "learning_rate": 1.302688638334779e-05,
      "loss": 0.6287,
      "step": 751
    },
    {
      "epoch": 0.1304649548924358,
      "grad_norm": 0.907132625579834,
      "learning_rate": 1.3044232437120557e-05,
      "loss": 0.7136,
      "step": 752
    },
    {
      "epoch": 0.13063844552394172,
      "grad_norm": 0.8171525597572327,
      "learning_rate": 1.3061578490893323e-05,
      "loss": 0.6865,
      "step": 753
    },
    {
      "epoch": 0.13081193615544762,
      "grad_norm": 0.6056315302848816,
      "learning_rate": 1.307892454466609e-05,
      "loss": 0.7515,
      "step": 754
    },
    {
      "epoch": 0.13098542678695352,
      "grad_norm": 1.0905085802078247,
      "learning_rate": 1.3096270598438857e-05,
      "loss": 0.7351,
      "step": 755
    },
    {
      "epoch": 0.1311589174184594,
      "grad_norm": 1.5320167541503906,
      "learning_rate": 1.3113616652211622e-05,
      "loss": 0.7522,
      "step": 756
    },
    {
      "epoch": 0.1313324080499653,
      "grad_norm": 1.0283844470977783,
      "learning_rate": 1.3130962705984389e-05,
      "loss": 0.6381,
      "step": 757
    },
    {
      "epoch": 0.1315058986814712,
      "grad_norm": 0.7436849474906921,
      "learning_rate": 1.3148308759757155e-05,
      "loss": 0.7091,
      "step": 758
    },
    {
      "epoch": 0.1316793893129771,
      "grad_norm": 0.8111729025840759,
      "learning_rate": 1.3165654813529922e-05,
      "loss": 0.6208,
      "step": 759
    },
    {
      "epoch": 0.131852879944483,
      "grad_norm": 0.7507044076919556,
      "learning_rate": 1.3183000867302689e-05,
      "loss": 0.7162,
      "step": 760
    },
    {
      "epoch": 0.1320263705759889,
      "grad_norm": 1.0076284408569336,
      "learning_rate": 1.3200346921075455e-05,
      "loss": 0.7106,
      "step": 761
    },
    {
      "epoch": 0.1321998612074948,
      "grad_norm": 0.8929764628410339,
      "learning_rate": 1.3217692974848224e-05,
      "loss": 0.7991,
      "step": 762
    },
    {
      "epoch": 0.1323733518390007,
      "grad_norm": 1.046856164932251,
      "learning_rate": 1.323503902862099e-05,
      "loss": 0.6948,
      "step": 763
    },
    {
      "epoch": 0.1325468424705066,
      "grad_norm": 0.8030807375907898,
      "learning_rate": 1.3252385082393757e-05,
      "loss": 0.7334,
      "step": 764
    },
    {
      "epoch": 0.13272033310201248,
      "grad_norm": 0.7715807557106018,
      "learning_rate": 1.3269731136166524e-05,
      "loss": 0.6484,
      "step": 765
    },
    {
      "epoch": 0.13289382373351838,
      "grad_norm": 0.8913213014602661,
      "learning_rate": 1.328707718993929e-05,
      "loss": 0.645,
      "step": 766
    },
    {
      "epoch": 0.13306731436502428,
      "grad_norm": 0.7100571393966675,
      "learning_rate": 1.3304423243712057e-05,
      "loss": 0.5028,
      "step": 767
    },
    {
      "epoch": 0.13324080499653018,
      "grad_norm": 0.858887791633606,
      "learning_rate": 1.3321769297484824e-05,
      "loss": 0.7266,
      "step": 768
    },
    {
      "epoch": 0.13341429562803608,
      "grad_norm": 1.0888160467147827,
      "learning_rate": 1.333911535125759e-05,
      "loss": 0.7002,
      "step": 769
    },
    {
      "epoch": 0.13358778625954199,
      "grad_norm": 0.833756685256958,
      "learning_rate": 1.3356461405030357e-05,
      "loss": 0.6841,
      "step": 770
    },
    {
      "epoch": 0.1337612768910479,
      "grad_norm": 1.0680832862854004,
      "learning_rate": 1.3373807458803122e-05,
      "loss": 0.687,
      "step": 771
    },
    {
      "epoch": 0.1339347675225538,
      "grad_norm": 1.829303503036499,
      "learning_rate": 1.3391153512575889e-05,
      "loss": 0.7256,
      "step": 772
    },
    {
      "epoch": 0.1341082581540597,
      "grad_norm": 0.8159741163253784,
      "learning_rate": 1.3408499566348656e-05,
      "loss": 0.597,
      "step": 773
    },
    {
      "epoch": 0.1342817487855656,
      "grad_norm": 0.8202490210533142,
      "learning_rate": 1.3425845620121422e-05,
      "loss": 0.5811,
      "step": 774
    },
    {
      "epoch": 0.13445523941707146,
      "grad_norm": 1.8445295095443726,
      "learning_rate": 1.3443191673894189e-05,
      "loss": 0.6796,
      "step": 775
    },
    {
      "epoch": 0.13462873004857737,
      "grad_norm": 0.7824861407279968,
      "learning_rate": 1.3460537727666956e-05,
      "loss": 0.6609,
      "step": 776
    },
    {
      "epoch": 0.13480222068008327,
      "grad_norm": 0.9484027624130249,
      "learning_rate": 1.3477883781439724e-05,
      "loss": 0.5861,
      "step": 777
    },
    {
      "epoch": 0.13497571131158917,
      "grad_norm": 0.9183481335639954,
      "learning_rate": 1.349522983521249e-05,
      "loss": 0.5896,
      "step": 778
    },
    {
      "epoch": 0.13514920194309507,
      "grad_norm": 0.8156819343566895,
      "learning_rate": 1.3512575888985258e-05,
      "loss": 0.7158,
      "step": 779
    },
    {
      "epoch": 0.13532269257460097,
      "grad_norm": 0.8175688982009888,
      "learning_rate": 1.3529921942758024e-05,
      "loss": 0.7183,
      "step": 780
    },
    {
      "epoch": 0.13549618320610687,
      "grad_norm": 1.0969560146331787,
      "learning_rate": 1.3547267996530791e-05,
      "loss": 0.7327,
      "step": 781
    },
    {
      "epoch": 0.13566967383761278,
      "grad_norm": 1.166895866394043,
      "learning_rate": 1.3564614050303558e-05,
      "loss": 0.771,
      "step": 782
    },
    {
      "epoch": 0.13584316446911868,
      "grad_norm": 3.4434897899627686,
      "learning_rate": 1.3581960104076324e-05,
      "loss": 0.8831,
      "step": 783
    },
    {
      "epoch": 0.13601665510062458,
      "grad_norm": 1.7552794218063354,
      "learning_rate": 1.3599306157849091e-05,
      "loss": 0.7871,
      "step": 784
    },
    {
      "epoch": 0.13619014573213045,
      "grad_norm": 0.8987719416618347,
      "learning_rate": 1.3616652211621858e-05,
      "loss": 0.7195,
      "step": 785
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 0.5590589642524719,
      "learning_rate": 1.3633998265394623e-05,
      "loss": 0.7832,
      "step": 786
    },
    {
      "epoch": 0.13653712699514226,
      "grad_norm": 0.7459977865219116,
      "learning_rate": 1.365134431916739e-05,
      "loss": 0.6086,
      "step": 787
    },
    {
      "epoch": 0.13671061762664816,
      "grad_norm": 0.7150686979293823,
      "learning_rate": 1.3668690372940156e-05,
      "loss": 0.7173,
      "step": 788
    },
    {
      "epoch": 0.13688410825815406,
      "grad_norm": 1.0562443733215332,
      "learning_rate": 1.3686036426712923e-05,
      "loss": 0.7086,
      "step": 789
    },
    {
      "epoch": 0.13705759888965996,
      "grad_norm": 1.3128288984298706,
      "learning_rate": 1.370338248048569e-05,
      "loss": 0.5895,
      "step": 790
    },
    {
      "epoch": 0.13723108952116586,
      "grad_norm": 1.139470100402832,
      "learning_rate": 1.3720728534258458e-05,
      "loss": 0.635,
      "step": 791
    },
    {
      "epoch": 0.13740458015267176,
      "grad_norm": 0.9284714460372925,
      "learning_rate": 1.3738074588031225e-05,
      "loss": 0.6987,
      "step": 792
    },
    {
      "epoch": 0.13757807078417766,
      "grad_norm": 0.7392175793647766,
      "learning_rate": 1.3755420641803991e-05,
      "loss": 0.5834,
      "step": 793
    },
    {
      "epoch": 0.13775156141568357,
      "grad_norm": 0.5572483539581299,
      "learning_rate": 1.3772766695576758e-05,
      "loss": 0.8179,
      "step": 794
    },
    {
      "epoch": 0.13792505204718944,
      "grad_norm": 1.1708852052688599,
      "learning_rate": 1.3790112749349525e-05,
      "loss": 0.7037,
      "step": 795
    },
    {
      "epoch": 0.13809854267869534,
      "grad_norm": 0.7428001165390015,
      "learning_rate": 1.3807458803122291e-05,
      "loss": 0.7423,
      "step": 796
    },
    {
      "epoch": 0.13827203331020124,
      "grad_norm": 0.8794804215431213,
      "learning_rate": 1.3824804856895058e-05,
      "loss": 0.731,
      "step": 797
    },
    {
      "epoch": 0.13844552394170714,
      "grad_norm": 0.779248058795929,
      "learning_rate": 1.3842150910667825e-05,
      "loss": 0.7002,
      "step": 798
    },
    {
      "epoch": 0.13861901457321305,
      "grad_norm": 0.8916496634483337,
      "learning_rate": 1.3859496964440591e-05,
      "loss": 0.7222,
      "step": 799
    },
    {
      "epoch": 0.13879250520471895,
      "grad_norm": 0.9523424506187439,
      "learning_rate": 1.3876843018213358e-05,
      "loss": 0.6304,
      "step": 800
    },
    {
      "epoch": 0.13896599583622485,
      "grad_norm": 1.1725637912750244,
      "learning_rate": 1.3894189071986123e-05,
      "loss": 0.6189,
      "step": 801
    },
    {
      "epoch": 0.13913948646773075,
      "grad_norm": 0.5944787859916687,
      "learning_rate": 1.391153512575889e-05,
      "loss": 0.8826,
      "step": 802
    },
    {
      "epoch": 0.13931297709923665,
      "grad_norm": 0.7201195955276489,
      "learning_rate": 1.3928881179531657e-05,
      "loss": 0.7395,
      "step": 803
    },
    {
      "epoch": 0.13948646773074255,
      "grad_norm": 0.8105780482292175,
      "learning_rate": 1.3946227233304423e-05,
      "loss": 0.6179,
      "step": 804
    },
    {
      "epoch": 0.13965995836224843,
      "grad_norm": 1.070915699005127,
      "learning_rate": 1.396357328707719e-05,
      "loss": 0.6484,
      "step": 805
    },
    {
      "epoch": 0.13983344899375433,
      "grad_norm": 1.1900025606155396,
      "learning_rate": 1.3980919340849958e-05,
      "loss": 0.5752,
      "step": 806
    },
    {
      "epoch": 0.14000693962526023,
      "grad_norm": 1.0084441900253296,
      "learning_rate": 1.3998265394622725e-05,
      "loss": 0.7759,
      "step": 807
    },
    {
      "epoch": 0.14018043025676613,
      "grad_norm": 1.2829084396362305,
      "learning_rate": 1.4015611448395492e-05,
      "loss": 0.6047,
      "step": 808
    },
    {
      "epoch": 0.14035392088827203,
      "grad_norm": 0.7799201011657715,
      "learning_rate": 1.4032957502168258e-05,
      "loss": 0.7181,
      "step": 809
    },
    {
      "epoch": 0.14052741151977793,
      "grad_norm": 0.8388474583625793,
      "learning_rate": 1.4050303555941025e-05,
      "loss": 0.7703,
      "step": 810
    },
    {
      "epoch": 0.14070090215128384,
      "grad_norm": 0.7399805188179016,
      "learning_rate": 1.4067649609713792e-05,
      "loss": 0.8882,
      "step": 811
    },
    {
      "epoch": 0.14087439278278974,
      "grad_norm": 1.1823039054870605,
      "learning_rate": 1.4084995663486558e-05,
      "loss": 0.8306,
      "step": 812
    },
    {
      "epoch": 0.14104788341429564,
      "grad_norm": 0.8547143340110779,
      "learning_rate": 1.4102341717259325e-05,
      "loss": 0.8206,
      "step": 813
    },
    {
      "epoch": 0.14122137404580154,
      "grad_norm": 0.9972624778747559,
      "learning_rate": 1.4119687771032092e-05,
      "loss": 0.6167,
      "step": 814
    },
    {
      "epoch": 0.1413948646773074,
      "grad_norm": 1.1319990158081055,
      "learning_rate": 1.4137033824804859e-05,
      "loss": 0.5568,
      "step": 815
    },
    {
      "epoch": 0.14156835530881332,
      "grad_norm": 1.930017352104187,
      "learning_rate": 1.4154379878577624e-05,
      "loss": 0.6726,
      "step": 816
    },
    {
      "epoch": 0.14174184594031922,
      "grad_norm": 0.94271320104599,
      "learning_rate": 1.417172593235039e-05,
      "loss": 0.7246,
      "step": 817
    },
    {
      "epoch": 0.14191533657182512,
      "grad_norm": 1.4749341011047363,
      "learning_rate": 1.4189071986123157e-05,
      "loss": 0.7723,
      "step": 818
    },
    {
      "epoch": 0.14208882720333102,
      "grad_norm": 0.8136767745018005,
      "learning_rate": 1.4206418039895924e-05,
      "loss": 0.759,
      "step": 819
    },
    {
      "epoch": 0.14226231783483692,
      "grad_norm": 0.8389351963996887,
      "learning_rate": 1.422376409366869e-05,
      "loss": 0.5598,
      "step": 820
    },
    {
      "epoch": 0.14243580846634282,
      "grad_norm": 0.7450869679450989,
      "learning_rate": 1.4241110147441459e-05,
      "loss": 0.8494,
      "step": 821
    },
    {
      "epoch": 0.14260929909784872,
      "grad_norm": 1.1043566465377808,
      "learning_rate": 1.4258456201214225e-05,
      "loss": 0.6223,
      "step": 822
    },
    {
      "epoch": 0.14278278972935463,
      "grad_norm": 0.7570394277572632,
      "learning_rate": 1.4275802254986992e-05,
      "loss": 0.6093,
      "step": 823
    },
    {
      "epoch": 0.14295628036086053,
      "grad_norm": 0.7042250037193298,
      "learning_rate": 1.4293148308759759e-05,
      "loss": 0.5449,
      "step": 824
    },
    {
      "epoch": 0.1431297709923664,
      "grad_norm": 0.6336850523948669,
      "learning_rate": 1.4310494362532526e-05,
      "loss": 0.7617,
      "step": 825
    },
    {
      "epoch": 0.1433032616238723,
      "grad_norm": 1.6511664390563965,
      "learning_rate": 1.4327840416305292e-05,
      "loss": 0.7332,
      "step": 826
    },
    {
      "epoch": 0.1434767522553782,
      "grad_norm": 1.9736170768737793,
      "learning_rate": 1.4345186470078059e-05,
      "loss": 0.6853,
      "step": 827
    },
    {
      "epoch": 0.1436502428868841,
      "grad_norm": 0.9473209977149963,
      "learning_rate": 1.4362532523850826e-05,
      "loss": 0.6881,
      "step": 828
    },
    {
      "epoch": 0.14382373351839,
      "grad_norm": 0.806874692440033,
      "learning_rate": 1.4379878577623592e-05,
      "loss": 0.5989,
      "step": 829
    },
    {
      "epoch": 0.1439972241498959,
      "grad_norm": 0.7855055928230286,
      "learning_rate": 1.4397224631396359e-05,
      "loss": 0.709,
      "step": 830
    },
    {
      "epoch": 0.1441707147814018,
      "grad_norm": 0.9963886141777039,
      "learning_rate": 1.4414570685169124e-05,
      "loss": 0.6379,
      "step": 831
    },
    {
      "epoch": 0.1443442054129077,
      "grad_norm": 0.7364680767059326,
      "learning_rate": 1.443191673894189e-05,
      "loss": 0.8228,
      "step": 832
    },
    {
      "epoch": 0.1445176960444136,
      "grad_norm": 0.8339455127716064,
      "learning_rate": 1.4449262792714657e-05,
      "loss": 0.6504,
      "step": 833
    },
    {
      "epoch": 0.14469118667591951,
      "grad_norm": 0.9416974782943726,
      "learning_rate": 1.4466608846487424e-05,
      "loss": 0.636,
      "step": 834
    },
    {
      "epoch": 0.1448646773074254,
      "grad_norm": 0.9278693199157715,
      "learning_rate": 1.448395490026019e-05,
      "loss": 0.6721,
      "step": 835
    },
    {
      "epoch": 0.1450381679389313,
      "grad_norm": 2.3351480960845947,
      "learning_rate": 1.450130095403296e-05,
      "loss": 0.6017,
      "step": 836
    },
    {
      "epoch": 0.1452116585704372,
      "grad_norm": 0.6672064661979675,
      "learning_rate": 1.4518647007805726e-05,
      "loss": 0.7229,
      "step": 837
    },
    {
      "epoch": 0.1453851492019431,
      "grad_norm": 0.743253767490387,
      "learning_rate": 1.4535993061578493e-05,
      "loss": 0.6313,
      "step": 838
    },
    {
      "epoch": 0.145558639833449,
      "grad_norm": 1.056763768196106,
      "learning_rate": 1.455333911535126e-05,
      "loss": 0.6968,
      "step": 839
    },
    {
      "epoch": 0.1457321304649549,
      "grad_norm": 0.9053840637207031,
      "learning_rate": 1.4570685169124026e-05,
      "loss": 0.6313,
      "step": 840
    },
    {
      "epoch": 0.1459056210964608,
      "grad_norm": 1.119525671005249,
      "learning_rate": 1.4588031222896793e-05,
      "loss": 0.5525,
      "step": 841
    },
    {
      "epoch": 0.1460791117279667,
      "grad_norm": 1.0211570262908936,
      "learning_rate": 1.460537727666956e-05,
      "loss": 0.6615,
      "step": 842
    },
    {
      "epoch": 0.1462526023594726,
      "grad_norm": 0.7650402188301086,
      "learning_rate": 1.4622723330442326e-05,
      "loss": 0.7095,
      "step": 843
    },
    {
      "epoch": 0.14642609299097847,
      "grad_norm": 1.1841182708740234,
      "learning_rate": 1.4640069384215093e-05,
      "loss": 0.6704,
      "step": 844
    },
    {
      "epoch": 0.14659958362248438,
      "grad_norm": 1.1515867710113525,
      "learning_rate": 1.465741543798786e-05,
      "loss": 0.5818,
      "step": 845
    },
    {
      "epoch": 0.14677307425399028,
      "grad_norm": 0.9013440608978271,
      "learning_rate": 1.4674761491760624e-05,
      "loss": 0.6653,
      "step": 846
    },
    {
      "epoch": 0.14694656488549618,
      "grad_norm": 1.0194121599197388,
      "learning_rate": 1.4692107545533391e-05,
      "loss": 0.7689,
      "step": 847
    },
    {
      "epoch": 0.14712005551700208,
      "grad_norm": 0.7369625568389893,
      "learning_rate": 1.4709453599306158e-05,
      "loss": 0.6931,
      "step": 848
    },
    {
      "epoch": 0.14729354614850798,
      "grad_norm": 0.776012122631073,
      "learning_rate": 1.4726799653078925e-05,
      "loss": 0.7512,
      "step": 849
    },
    {
      "epoch": 0.14746703678001388,
      "grad_norm": 0.8151385188102722,
      "learning_rate": 1.4744145706851693e-05,
      "loss": 0.6995,
      "step": 850
    },
    {
      "epoch": 0.14764052741151978,
      "grad_norm": 0.7555911540985107,
      "learning_rate": 1.476149176062446e-05,
      "loss": 0.611,
      "step": 851
    },
    {
      "epoch": 0.14781401804302569,
      "grad_norm": 1.3994958400726318,
      "learning_rate": 1.4778837814397226e-05,
      "loss": 0.6509,
      "step": 852
    },
    {
      "epoch": 0.1479875086745316,
      "grad_norm": 1.2116901874542236,
      "learning_rate": 1.4796183868169993e-05,
      "loss": 0.626,
      "step": 853
    },
    {
      "epoch": 0.14816099930603746,
      "grad_norm": 1.1880247592926025,
      "learning_rate": 1.481352992194276e-05,
      "loss": 0.6135,
      "step": 854
    },
    {
      "epoch": 0.14833448993754336,
      "grad_norm": 0.6173132061958313,
      "learning_rate": 1.4830875975715526e-05,
      "loss": 0.7172,
      "step": 855
    },
    {
      "epoch": 0.14850798056904926,
      "grad_norm": 0.8529770970344543,
      "learning_rate": 1.4848222029488293e-05,
      "loss": 0.6005,
      "step": 856
    },
    {
      "epoch": 0.14868147120055517,
      "grad_norm": 0.9621935486793518,
      "learning_rate": 1.486556808326106e-05,
      "loss": 0.6149,
      "step": 857
    },
    {
      "epoch": 0.14885496183206107,
      "grad_norm": 1.5569918155670166,
      "learning_rate": 1.4882914137033826e-05,
      "loss": 0.8989,
      "step": 858
    },
    {
      "epoch": 0.14902845246356697,
      "grad_norm": 0.931231677532196,
      "learning_rate": 1.4900260190806593e-05,
      "loss": 0.5619,
      "step": 859
    },
    {
      "epoch": 0.14920194309507287,
      "grad_norm": 1.5929428339004517,
      "learning_rate": 1.491760624457936e-05,
      "loss": 0.5854,
      "step": 860
    },
    {
      "epoch": 0.14937543372657877,
      "grad_norm": 0.9305289387702942,
      "learning_rate": 1.4934952298352125e-05,
      "loss": 0.6697,
      "step": 861
    },
    {
      "epoch": 0.14954892435808467,
      "grad_norm": 0.8200428485870361,
      "learning_rate": 1.4952298352124892e-05,
      "loss": 0.7217,
      "step": 862
    },
    {
      "epoch": 0.14972241498959057,
      "grad_norm": 4.3267107009887695,
      "learning_rate": 1.4969644405897658e-05,
      "loss": 0.5573,
      "step": 863
    },
    {
      "epoch": 0.14989590562109645,
      "grad_norm": 1.2004973888397217,
      "learning_rate": 1.4986990459670425e-05,
      "loss": 0.7554,
      "step": 864
    },
    {
      "epoch": 0.15006939625260235,
      "grad_norm": 0.9642512798309326,
      "learning_rate": 1.5004336513443193e-05,
      "loss": 0.6198,
      "step": 865
    },
    {
      "epoch": 0.15024288688410825,
      "grad_norm": 0.8531646728515625,
      "learning_rate": 1.502168256721596e-05,
      "loss": 0.8345,
      "step": 866
    },
    {
      "epoch": 0.15041637751561415,
      "grad_norm": 1.055066704750061,
      "learning_rate": 1.5039028620988727e-05,
      "loss": 0.749,
      "step": 867
    },
    {
      "epoch": 0.15058986814712005,
      "grad_norm": 1.1706310510635376,
      "learning_rate": 1.5056374674761493e-05,
      "loss": 0.6992,
      "step": 868
    },
    {
      "epoch": 0.15076335877862596,
      "grad_norm": 1.0913740396499634,
      "learning_rate": 1.507372072853426e-05,
      "loss": 0.6554,
      "step": 869
    },
    {
      "epoch": 0.15093684941013186,
      "grad_norm": 0.8652170300483704,
      "learning_rate": 1.5091066782307027e-05,
      "loss": 0.6669,
      "step": 870
    },
    {
      "epoch": 0.15111034004163776,
      "grad_norm": 0.9506787061691284,
      "learning_rate": 1.5108412836079793e-05,
      "loss": 0.7295,
      "step": 871
    },
    {
      "epoch": 0.15128383067314366,
      "grad_norm": 1.150251865386963,
      "learning_rate": 1.512575888985256e-05,
      "loss": 0.733,
      "step": 872
    },
    {
      "epoch": 0.15145732130464956,
      "grad_norm": 0.8466367721557617,
      "learning_rate": 1.5143104943625327e-05,
      "loss": 0.6395,
      "step": 873
    },
    {
      "epoch": 0.15163081193615544,
      "grad_norm": 2.574615240097046,
      "learning_rate": 1.5160450997398094e-05,
      "loss": 0.7278,
      "step": 874
    },
    {
      "epoch": 0.15180430256766134,
      "grad_norm": 1.0064374208450317,
      "learning_rate": 1.517779705117086e-05,
      "loss": 0.7107,
      "step": 875
    },
    {
      "epoch": 0.15197779319916724,
      "grad_norm": 0.8361401557922363,
      "learning_rate": 1.5195143104943625e-05,
      "loss": 0.6068,
      "step": 876
    },
    {
      "epoch": 0.15215128383067314,
      "grad_norm": 0.8473800420761108,
      "learning_rate": 1.5212489158716392e-05,
      "loss": 0.7333,
      "step": 877
    },
    {
      "epoch": 0.15232477446217904,
      "grad_norm": 1.097095012664795,
      "learning_rate": 1.5229835212489159e-05,
      "loss": 0.7101,
      "step": 878
    },
    {
      "epoch": 0.15249826509368494,
      "grad_norm": 0.9678106904029846,
      "learning_rate": 1.5247181266261925e-05,
      "loss": 0.6997,
      "step": 879
    },
    {
      "epoch": 0.15267175572519084,
      "grad_norm": 1.245478630065918,
      "learning_rate": 1.5264527320034695e-05,
      "loss": 0.7833,
      "step": 880
    },
    {
      "epoch": 0.15284524635669675,
      "grad_norm": 1.0961253643035889,
      "learning_rate": 1.528187337380746e-05,
      "loss": 0.7347,
      "step": 881
    },
    {
      "epoch": 0.15301873698820265,
      "grad_norm": 2.126734733581543,
      "learning_rate": 1.529921942758023e-05,
      "loss": 0.7664,
      "step": 882
    },
    {
      "epoch": 0.15319222761970855,
      "grad_norm": 0.8503320813179016,
      "learning_rate": 1.5316565481352994e-05,
      "loss": 0.6718,
      "step": 883
    },
    {
      "epoch": 0.15336571825121442,
      "grad_norm": 0.6812099814414978,
      "learning_rate": 1.5333911535125762e-05,
      "loss": 0.7156,
      "step": 884
    },
    {
      "epoch": 0.15353920888272032,
      "grad_norm": 0.7333630919456482,
      "learning_rate": 1.5351257588898527e-05,
      "loss": 0.6864,
      "step": 885
    },
    {
      "epoch": 0.15371269951422623,
      "grad_norm": 0.8481491208076477,
      "learning_rate": 1.5368603642671292e-05,
      "loss": 0.7781,
      "step": 886
    },
    {
      "epoch": 0.15388619014573213,
      "grad_norm": 1.4921408891677856,
      "learning_rate": 1.538594969644406e-05,
      "loss": 0.6251,
      "step": 887
    },
    {
      "epoch": 0.15405968077723803,
      "grad_norm": 0.9560161232948303,
      "learning_rate": 1.5403295750216826e-05,
      "loss": 0.6892,
      "step": 888
    },
    {
      "epoch": 0.15423317140874393,
      "grad_norm": 1.2779227495193481,
      "learning_rate": 1.5420641803989594e-05,
      "loss": 0.5886,
      "step": 889
    },
    {
      "epoch": 0.15440666204024983,
      "grad_norm": 0.7587497234344482,
      "learning_rate": 1.543798785776236e-05,
      "loss": 0.7178,
      "step": 890
    },
    {
      "epoch": 0.15458015267175573,
      "grad_norm": 0.8463174104690552,
      "learning_rate": 1.5455333911535127e-05,
      "loss": 0.7814,
      "step": 891
    },
    {
      "epoch": 0.15475364330326163,
      "grad_norm": 0.8361932039260864,
      "learning_rate": 1.5472679965307892e-05,
      "loss": 0.5793,
      "step": 892
    },
    {
      "epoch": 0.15492713393476754,
      "grad_norm": 0.8943549990653992,
      "learning_rate": 1.549002601908066e-05,
      "loss": 0.7914,
      "step": 893
    },
    {
      "epoch": 0.1551006245662734,
      "grad_norm": 0.8401117324829102,
      "learning_rate": 1.5507372072853426e-05,
      "loss": 0.6434,
      "step": 894
    },
    {
      "epoch": 0.1552741151977793,
      "grad_norm": 0.961362898349762,
      "learning_rate": 1.5524718126626194e-05,
      "loss": 0.7277,
      "step": 895
    },
    {
      "epoch": 0.1554476058292852,
      "grad_norm": 0.9641268849372864,
      "learning_rate": 1.5542064180398963e-05,
      "loss": 0.7878,
      "step": 896
    },
    {
      "epoch": 0.15562109646079111,
      "grad_norm": 0.8773832321166992,
      "learning_rate": 1.5559410234171728e-05,
      "loss": 0.6924,
      "step": 897
    },
    {
      "epoch": 0.15579458709229702,
      "grad_norm": 0.7039158940315247,
      "learning_rate": 1.5576756287944496e-05,
      "loss": 0.7548,
      "step": 898
    },
    {
      "epoch": 0.15596807772380292,
      "grad_norm": 0.7688771486282349,
      "learning_rate": 1.559410234171726e-05,
      "loss": 0.6881,
      "step": 899
    },
    {
      "epoch": 0.15614156835530882,
      "grad_norm": 0.8563050627708435,
      "learning_rate": 1.5611448395490026e-05,
      "loss": 0.7627,
      "step": 900
    },
    {
      "epoch": 0.15631505898681472,
      "grad_norm": 0.8748807907104492,
      "learning_rate": 1.5628794449262794e-05,
      "loss": 0.6493,
      "step": 901
    },
    {
      "epoch": 0.15648854961832062,
      "grad_norm": 0.7920365929603577,
      "learning_rate": 1.564614050303556e-05,
      "loss": 0.6841,
      "step": 902
    },
    {
      "epoch": 0.15666204024982652,
      "grad_norm": 0.6366187930107117,
      "learning_rate": 1.5663486556808328e-05,
      "loss": 0.6548,
      "step": 903
    },
    {
      "epoch": 0.1568355308813324,
      "grad_norm": 1.085413932800293,
      "learning_rate": 1.5680832610581093e-05,
      "loss": 0.6558,
      "step": 904
    },
    {
      "epoch": 0.1570090215128383,
      "grad_norm": 0.7230006456375122,
      "learning_rate": 1.569817866435386e-05,
      "loss": 0.7896,
      "step": 905
    },
    {
      "epoch": 0.1571825121443442,
      "grad_norm": 0.7782901525497437,
      "learning_rate": 1.5715524718126626e-05,
      "loss": 0.6873,
      "step": 906
    },
    {
      "epoch": 0.1573560027758501,
      "grad_norm": 0.542499303817749,
      "learning_rate": 1.5732870771899395e-05,
      "loss": 0.8535,
      "step": 907
    },
    {
      "epoch": 0.157529493407356,
      "grad_norm": 0.6911293268203735,
      "learning_rate": 1.575021682567216e-05,
      "loss": 0.6926,
      "step": 908
    },
    {
      "epoch": 0.1577029840388619,
      "grad_norm": 0.7596789002418518,
      "learning_rate": 1.5767562879444928e-05,
      "loss": 0.6484,
      "step": 909
    },
    {
      "epoch": 0.1578764746703678,
      "grad_norm": 0.6950908899307251,
      "learning_rate": 1.5784908933217696e-05,
      "loss": 0.6914,
      "step": 910
    },
    {
      "epoch": 0.1580499653018737,
      "grad_norm": 0.8597424030303955,
      "learning_rate": 1.580225498699046e-05,
      "loss": 0.6887,
      "step": 911
    },
    {
      "epoch": 0.1582234559333796,
      "grad_norm": 1.0358662605285645,
      "learning_rate": 1.581960104076323e-05,
      "loss": 0.7153,
      "step": 912
    },
    {
      "epoch": 0.15839694656488548,
      "grad_norm": 1.1343300342559814,
      "learning_rate": 1.5836947094535995e-05,
      "loss": 0.6909,
      "step": 913
    },
    {
      "epoch": 0.15857043719639138,
      "grad_norm": 0.7074477076530457,
      "learning_rate": 1.5854293148308763e-05,
      "loss": 0.6642,
      "step": 914
    },
    {
      "epoch": 0.15874392782789729,
      "grad_norm": 0.7301273345947266,
      "learning_rate": 1.5871639202081528e-05,
      "loss": 0.6434,
      "step": 915
    },
    {
      "epoch": 0.1589174184594032,
      "grad_norm": 1.0247890949249268,
      "learning_rate": 1.5888985255854293e-05,
      "loss": 0.6675,
      "step": 916
    },
    {
      "epoch": 0.1590909090909091,
      "grad_norm": 1.233616828918457,
      "learning_rate": 1.590633130962706e-05,
      "loss": 0.7847,
      "step": 917
    },
    {
      "epoch": 0.159264399722415,
      "grad_norm": 0.9381399154663086,
      "learning_rate": 1.5923677363399826e-05,
      "loss": 0.6826,
      "step": 918
    },
    {
      "epoch": 0.1594378903539209,
      "grad_norm": 0.8014280200004578,
      "learning_rate": 1.5941023417172595e-05,
      "loss": 0.6625,
      "step": 919
    },
    {
      "epoch": 0.1596113809854268,
      "grad_norm": 1.027308702468872,
      "learning_rate": 1.595836947094536e-05,
      "loss": 0.6035,
      "step": 920
    },
    {
      "epoch": 0.1597848716169327,
      "grad_norm": 0.686129629611969,
      "learning_rate": 1.5975715524718128e-05,
      "loss": 0.8022,
      "step": 921
    },
    {
      "epoch": 0.1599583622484386,
      "grad_norm": 0.7895461916923523,
      "learning_rate": 1.5993061578490893e-05,
      "loss": 0.6404,
      "step": 922
    },
    {
      "epoch": 0.16013185287994447,
      "grad_norm": 0.9063572883605957,
      "learning_rate": 1.601040763226366e-05,
      "loss": 0.6422,
      "step": 923
    },
    {
      "epoch": 0.16030534351145037,
      "grad_norm": 0.8540890216827393,
      "learning_rate": 1.602775368603643e-05,
      "loss": 0.6224,
      "step": 924
    },
    {
      "epoch": 0.16047883414295627,
      "grad_norm": 0.9980672001838684,
      "learning_rate": 1.6045099739809195e-05,
      "loss": 0.5835,
      "step": 925
    },
    {
      "epoch": 0.16065232477446217,
      "grad_norm": 0.8262863159179688,
      "learning_rate": 1.6062445793581963e-05,
      "loss": 0.7963,
      "step": 926
    },
    {
      "epoch": 0.16082581540596808,
      "grad_norm": 0.7476772665977478,
      "learning_rate": 1.607979184735473e-05,
      "loss": 0.7102,
      "step": 927
    },
    {
      "epoch": 0.16099930603747398,
      "grad_norm": 0.8359287977218628,
      "learning_rate": 1.6097137901127497e-05,
      "loss": 0.6128,
      "step": 928
    },
    {
      "epoch": 0.16117279666897988,
      "grad_norm": 0.7169504165649414,
      "learning_rate": 1.6114483954900262e-05,
      "loss": 0.7908,
      "step": 929
    },
    {
      "epoch": 0.16134628730048578,
      "grad_norm": 1.0917155742645264,
      "learning_rate": 1.6131830008673027e-05,
      "loss": 0.7098,
      "step": 930
    },
    {
      "epoch": 0.16151977793199168,
      "grad_norm": 0.9570122957229614,
      "learning_rate": 1.6149176062445795e-05,
      "loss": 0.6041,
      "step": 931
    },
    {
      "epoch": 0.16169326856349758,
      "grad_norm": 0.683272123336792,
      "learning_rate": 1.616652211621856e-05,
      "loss": 0.6094,
      "step": 932
    },
    {
      "epoch": 0.16186675919500346,
      "grad_norm": 0.9358447194099426,
      "learning_rate": 1.618386816999133e-05,
      "loss": 0.7856,
      "step": 933
    },
    {
      "epoch": 0.16204024982650936,
      "grad_norm": 0.6288849711418152,
      "learning_rate": 1.6201214223764094e-05,
      "loss": 0.6729,
      "step": 934
    },
    {
      "epoch": 0.16221374045801526,
      "grad_norm": 0.6991991996765137,
      "learning_rate": 1.6218560277536862e-05,
      "loss": 0.723,
      "step": 935
    },
    {
      "epoch": 0.16238723108952116,
      "grad_norm": 0.9126247763633728,
      "learning_rate": 1.6235906331309627e-05,
      "loss": 0.673,
      "step": 936
    },
    {
      "epoch": 0.16256072172102706,
      "grad_norm": 1.2215293645858765,
      "learning_rate": 1.6253252385082395e-05,
      "loss": 0.7214,
      "step": 937
    },
    {
      "epoch": 0.16273421235253296,
      "grad_norm": 0.8412079811096191,
      "learning_rate": 1.627059843885516e-05,
      "loss": 0.7268,
      "step": 938
    },
    {
      "epoch": 0.16290770298403887,
      "grad_norm": 0.8135339021682739,
      "learning_rate": 1.628794449262793e-05,
      "loss": 0.6954,
      "step": 939
    },
    {
      "epoch": 0.16308119361554477,
      "grad_norm": 1.5377135276794434,
      "learning_rate": 1.6305290546400697e-05,
      "loss": 0.6376,
      "step": 940
    },
    {
      "epoch": 0.16325468424705067,
      "grad_norm": 0.8019591569900513,
      "learning_rate": 1.6322636600173462e-05,
      "loss": 0.7319,
      "step": 941
    },
    {
      "epoch": 0.16342817487855657,
      "grad_norm": 0.8551116585731506,
      "learning_rate": 1.633998265394623e-05,
      "loss": 0.6748,
      "step": 942
    },
    {
      "epoch": 0.16360166551006244,
      "grad_norm": 0.7336849570274353,
      "learning_rate": 1.6357328707718996e-05,
      "loss": 0.684,
      "step": 943
    },
    {
      "epoch": 0.16377515614156835,
      "grad_norm": 0.756514310836792,
      "learning_rate": 1.6374674761491764e-05,
      "loss": 0.7261,
      "step": 944
    },
    {
      "epoch": 0.16394864677307425,
      "grad_norm": 0.8664772510528564,
      "learning_rate": 1.639202081526453e-05,
      "loss": 0.6232,
      "step": 945
    },
    {
      "epoch": 0.16412213740458015,
      "grad_norm": 0.9706249833106995,
      "learning_rate": 1.6409366869037294e-05,
      "loss": 0.7896,
      "step": 946
    },
    {
      "epoch": 0.16429562803608605,
      "grad_norm": 1.3999273777008057,
      "learning_rate": 1.6426712922810062e-05,
      "loss": 0.6323,
      "step": 947
    },
    {
      "epoch": 0.16446911866759195,
      "grad_norm": 0.9607291221618652,
      "learning_rate": 1.6444058976582827e-05,
      "loss": 0.6218,
      "step": 948
    },
    {
      "epoch": 0.16464260929909785,
      "grad_norm": 0.9746975302696228,
      "learning_rate": 1.6461405030355596e-05,
      "loss": 0.6968,
      "step": 949
    },
    {
      "epoch": 0.16481609993060375,
      "grad_norm": 0.8557673096656799,
      "learning_rate": 1.647875108412836e-05,
      "loss": 0.6807,
      "step": 950
    },
    {
      "epoch": 0.16498959056210966,
      "grad_norm": 1.3538912534713745,
      "learning_rate": 1.649609713790113e-05,
      "loss": 0.6714,
      "step": 951
    },
    {
      "epoch": 0.16516308119361556,
      "grad_norm": 0.7352486848831177,
      "learning_rate": 1.6513443191673894e-05,
      "loss": 0.679,
      "step": 952
    },
    {
      "epoch": 0.16533657182512143,
      "grad_norm": 1.0318317413330078,
      "learning_rate": 1.6530789245446663e-05,
      "loss": 0.8328,
      "step": 953
    },
    {
      "epoch": 0.16551006245662733,
      "grad_norm": 0.9433748126029968,
      "learning_rate": 1.654813529921943e-05,
      "loss": 0.6721,
      "step": 954
    },
    {
      "epoch": 0.16568355308813323,
      "grad_norm": 1.44105863571167,
      "learning_rate": 1.6565481352992196e-05,
      "loss": 0.7727,
      "step": 955
    },
    {
      "epoch": 0.16585704371963914,
      "grad_norm": 1.178808569908142,
      "learning_rate": 1.6582827406764964e-05,
      "loss": 0.6615,
      "step": 956
    },
    {
      "epoch": 0.16603053435114504,
      "grad_norm": 0.7474299073219299,
      "learning_rate": 1.660017346053773e-05,
      "loss": 0.6886,
      "step": 957
    },
    {
      "epoch": 0.16620402498265094,
      "grad_norm": 1.0822902917861938,
      "learning_rate": 1.6617519514310498e-05,
      "loss": 0.636,
      "step": 958
    },
    {
      "epoch": 0.16637751561415684,
      "grad_norm": 0.7172223329544067,
      "learning_rate": 1.6634865568083263e-05,
      "loss": 0.7212,
      "step": 959
    },
    {
      "epoch": 0.16655100624566274,
      "grad_norm": 0.789437472820282,
      "learning_rate": 1.6652211621856028e-05,
      "loss": 0.666,
      "step": 960
    },
    {
      "epoch": 0.16672449687716864,
      "grad_norm": 1.3900312185287476,
      "learning_rate": 1.6669557675628796e-05,
      "loss": 0.6277,
      "step": 961
    },
    {
      "epoch": 0.16689798750867454,
      "grad_norm": 0.7792858481407166,
      "learning_rate": 1.668690372940156e-05,
      "loss": 0.7878,
      "step": 962
    },
    {
      "epoch": 0.16707147814018042,
      "grad_norm": 0.6243866086006165,
      "learning_rate": 1.670424978317433e-05,
      "loss": 0.6951,
      "step": 963
    },
    {
      "epoch": 0.16724496877168632,
      "grad_norm": 0.7946894764900208,
      "learning_rate": 1.6721595836947094e-05,
      "loss": 0.621,
      "step": 964
    },
    {
      "epoch": 0.16741845940319222,
      "grad_norm": 1.2947770357131958,
      "learning_rate": 1.6738941890719863e-05,
      "loss": 0.676,
      "step": 965
    },
    {
      "epoch": 0.16759195003469812,
      "grad_norm": 1.1111828088760376,
      "learning_rate": 1.6756287944492628e-05,
      "loss": 0.7052,
      "step": 966
    },
    {
      "epoch": 0.16776544066620402,
      "grad_norm": 0.7376157641410828,
      "learning_rate": 1.6773633998265396e-05,
      "loss": 0.6995,
      "step": 967
    },
    {
      "epoch": 0.16793893129770993,
      "grad_norm": 1.1629353761672974,
      "learning_rate": 1.679098005203816e-05,
      "loss": 0.7086,
      "step": 968
    },
    {
      "epoch": 0.16811242192921583,
      "grad_norm": 0.9028804302215576,
      "learning_rate": 1.680832610581093e-05,
      "loss": 0.5719,
      "step": 969
    },
    {
      "epoch": 0.16828591256072173,
      "grad_norm": 1.0768271684646606,
      "learning_rate": 1.6825672159583698e-05,
      "loss": 0.6647,
      "step": 970
    },
    {
      "epoch": 0.16845940319222763,
      "grad_norm": 1.0081478357315063,
      "learning_rate": 1.6843018213356463e-05,
      "loss": 0.6356,
      "step": 971
    },
    {
      "epoch": 0.16863289382373353,
      "grad_norm": 1.476508378982544,
      "learning_rate": 1.686036426712923e-05,
      "loss": 0.8376,
      "step": 972
    },
    {
      "epoch": 0.1688063844552394,
      "grad_norm": 0.6738322377204895,
      "learning_rate": 1.6877710320901996e-05,
      "loss": 0.8074,
      "step": 973
    },
    {
      "epoch": 0.1689798750867453,
      "grad_norm": 0.8401079773902893,
      "learning_rate": 1.689505637467476e-05,
      "loss": 0.77,
      "step": 974
    },
    {
      "epoch": 0.1691533657182512,
      "grad_norm": 2.7095425128936768,
      "learning_rate": 1.691240242844753e-05,
      "loss": 0.6422,
      "step": 975
    },
    {
      "epoch": 0.1693268563497571,
      "grad_norm": 1.0661002397537231,
      "learning_rate": 1.6929748482220295e-05,
      "loss": 0.7129,
      "step": 976
    },
    {
      "epoch": 0.169500346981263,
      "grad_norm": 0.741974949836731,
      "learning_rate": 1.6947094535993063e-05,
      "loss": 0.7639,
      "step": 977
    },
    {
      "epoch": 0.1696738376127689,
      "grad_norm": 0.8547816276550293,
      "learning_rate": 1.6964440589765828e-05,
      "loss": 0.7159,
      "step": 978
    },
    {
      "epoch": 0.16984732824427481,
      "grad_norm": 0.731971025466919,
      "learning_rate": 1.6981786643538597e-05,
      "loss": 0.7456,
      "step": 979
    },
    {
      "epoch": 0.17002081887578072,
      "grad_norm": 1.1164857149124146,
      "learning_rate": 1.699913269731136e-05,
      "loss": 0.7031,
      "step": 980
    },
    {
      "epoch": 0.17019430950728662,
      "grad_norm": 0.8163184523582458,
      "learning_rate": 1.701647875108413e-05,
      "loss": 0.6238,
      "step": 981
    },
    {
      "epoch": 0.17036780013879252,
      "grad_norm": 0.7814517021179199,
      "learning_rate": 1.7033824804856895e-05,
      "loss": 0.6144,
      "step": 982
    },
    {
      "epoch": 0.1705412907702984,
      "grad_norm": 1.004395842552185,
      "learning_rate": 1.7051170858629663e-05,
      "loss": 0.5923,
      "step": 983
    },
    {
      "epoch": 0.1707147814018043,
      "grad_norm": 1.3236013650894165,
      "learning_rate": 1.7068516912402432e-05,
      "loss": 0.7759,
      "step": 984
    },
    {
      "epoch": 0.1708882720333102,
      "grad_norm": 0.7804445028305054,
      "learning_rate": 1.7085862966175197e-05,
      "loss": 0.7361,
      "step": 985
    },
    {
      "epoch": 0.1710617626648161,
      "grad_norm": 1.4555665254592896,
      "learning_rate": 1.7103209019947965e-05,
      "loss": 0.6619,
      "step": 986
    },
    {
      "epoch": 0.171235253296322,
      "grad_norm": 0.7560045123100281,
      "learning_rate": 1.712055507372073e-05,
      "loss": 0.728,
      "step": 987
    },
    {
      "epoch": 0.1714087439278279,
      "grad_norm": 1.0049445629119873,
      "learning_rate": 1.71379011274935e-05,
      "loss": 0.5458,
      "step": 988
    },
    {
      "epoch": 0.1715822345593338,
      "grad_norm": 1.0618914365768433,
      "learning_rate": 1.7155247181266264e-05,
      "loss": 0.6783,
      "step": 989
    },
    {
      "epoch": 0.1717557251908397,
      "grad_norm": 1.0983893871307373,
      "learning_rate": 1.717259323503903e-05,
      "loss": 0.6158,
      "step": 990
    },
    {
      "epoch": 0.1719292158223456,
      "grad_norm": 0.742469847202301,
      "learning_rate": 1.7189939288811797e-05,
      "loss": 0.6021,
      "step": 991
    },
    {
      "epoch": 0.17210270645385148,
      "grad_norm": 0.7174831628799438,
      "learning_rate": 1.7207285342584562e-05,
      "loss": 0.782,
      "step": 992
    },
    {
      "epoch": 0.17227619708535738,
      "grad_norm": 1.3333544731140137,
      "learning_rate": 1.722463139635733e-05,
      "loss": 0.5986,
      "step": 993
    },
    {
      "epoch": 0.17244968771686328,
      "grad_norm": 2.129117727279663,
      "learning_rate": 1.7241977450130095e-05,
      "loss": 0.7273,
      "step": 994
    },
    {
      "epoch": 0.17262317834836918,
      "grad_norm": 0.9279270172119141,
      "learning_rate": 1.7259323503902864e-05,
      "loss": 0.7134,
      "step": 995
    },
    {
      "epoch": 0.17279666897987508,
      "grad_norm": 0.8577216267585754,
      "learning_rate": 1.727666955767563e-05,
      "loss": 0.6736,
      "step": 996
    },
    {
      "epoch": 0.17297015961138099,
      "grad_norm": 0.8197943568229675,
      "learning_rate": 1.7294015611448397e-05,
      "loss": 0.7085,
      "step": 997
    },
    {
      "epoch": 0.1731436502428869,
      "grad_norm": 1.0312492847442627,
      "learning_rate": 1.7311361665221166e-05,
      "loss": 0.6168,
      "step": 998
    },
    {
      "epoch": 0.1733171408743928,
      "grad_norm": 1.2503159046173096,
      "learning_rate": 1.732870771899393e-05,
      "loss": 0.6034,
      "step": 999
    },
    {
      "epoch": 0.1734906315058987,
      "grad_norm": 1.1078702211380005,
      "learning_rate": 1.73460537727667e-05,
      "loss": 0.6769,
      "step": 1000
    },
    {
      "epoch": 0.1736641221374046,
      "grad_norm": 0.7119059562683105,
      "learning_rate": 1.7363399826539464e-05,
      "loss": 0.6415,
      "step": 1001
    },
    {
      "epoch": 0.17383761276891047,
      "grad_norm": 0.8156967163085938,
      "learning_rate": 1.7380745880312232e-05,
      "loss": 0.589,
      "step": 1002
    },
    {
      "epoch": 0.17401110340041637,
      "grad_norm": 1.3735185861587524,
      "learning_rate": 1.7398091934084997e-05,
      "loss": 0.7031,
      "step": 1003
    },
    {
      "epoch": 0.17418459403192227,
      "grad_norm": 0.8018829822540283,
      "learning_rate": 1.7415437987857762e-05,
      "loss": 0.6443,
      "step": 1004
    },
    {
      "epoch": 0.17435808466342817,
      "grad_norm": 0.7476117014884949,
      "learning_rate": 1.743278404163053e-05,
      "loss": 0.7019,
      "step": 1005
    },
    {
      "epoch": 0.17453157529493407,
      "grad_norm": 3.358261823654175,
      "learning_rate": 1.7450130095403296e-05,
      "loss": 0.7725,
      "step": 1006
    },
    {
      "epoch": 0.17470506592643997,
      "grad_norm": 0.6919204592704773,
      "learning_rate": 1.7467476149176064e-05,
      "loss": 0.7163,
      "step": 1007
    },
    {
      "epoch": 0.17487855655794587,
      "grad_norm": 1.4837769269943237,
      "learning_rate": 1.748482220294883e-05,
      "loss": 0.5562,
      "step": 1008
    },
    {
      "epoch": 0.17505204718945178,
      "grad_norm": 0.7044925689697266,
      "learning_rate": 1.7502168256721597e-05,
      "loss": 0.7113,
      "step": 1009
    },
    {
      "epoch": 0.17522553782095768,
      "grad_norm": 1.0583440065383911,
      "learning_rate": 1.7519514310494362e-05,
      "loss": 0.6238,
      "step": 1010
    },
    {
      "epoch": 0.17539902845246358,
      "grad_norm": 0.7085881233215332,
      "learning_rate": 1.753686036426713e-05,
      "loss": 0.7446,
      "step": 1011
    },
    {
      "epoch": 0.17557251908396945,
      "grad_norm": 1.2622864246368408,
      "learning_rate": 1.7554206418039896e-05,
      "loss": 0.6267,
      "step": 1012
    },
    {
      "epoch": 0.17574600971547535,
      "grad_norm": 0.6951441168785095,
      "learning_rate": 1.7571552471812664e-05,
      "loss": 0.7543,
      "step": 1013
    },
    {
      "epoch": 0.17591950034698126,
      "grad_norm": 0.9578819274902344,
      "learning_rate": 1.7588898525585433e-05,
      "loss": 0.7092,
      "step": 1014
    },
    {
      "epoch": 0.17609299097848716,
      "grad_norm": 0.8530898690223694,
      "learning_rate": 1.7606244579358198e-05,
      "loss": 0.7383,
      "step": 1015
    },
    {
      "epoch": 0.17626648160999306,
      "grad_norm": 0.7052402496337891,
      "learning_rate": 1.7623590633130966e-05,
      "loss": 0.7693,
      "step": 1016
    },
    {
      "epoch": 0.17643997224149896,
      "grad_norm": 1.1654605865478516,
      "learning_rate": 1.764093668690373e-05,
      "loss": 0.6918,
      "step": 1017
    },
    {
      "epoch": 0.17661346287300486,
      "grad_norm": 0.8184143900871277,
      "learning_rate": 1.76582827406765e-05,
      "loss": 0.6852,
      "step": 1018
    },
    {
      "epoch": 0.17678695350451076,
      "grad_norm": 1.003745675086975,
      "learning_rate": 1.7675628794449264e-05,
      "loss": 0.6572,
      "step": 1019
    },
    {
      "epoch": 0.17696044413601666,
      "grad_norm": 1.0442581176757812,
      "learning_rate": 1.769297484822203e-05,
      "loss": 0.5624,
      "step": 1020
    },
    {
      "epoch": 0.17713393476752257,
      "grad_norm": 2.1000277996063232,
      "learning_rate": 1.7710320901994798e-05,
      "loss": 0.724,
      "step": 1021
    },
    {
      "epoch": 0.17730742539902844,
      "grad_norm": 1.0788401365280151,
      "learning_rate": 1.7727666955767563e-05,
      "loss": 0.7444,
      "step": 1022
    },
    {
      "epoch": 0.17748091603053434,
      "grad_norm": 1.9666985273361206,
      "learning_rate": 1.774501300954033e-05,
      "loss": 0.707,
      "step": 1023
    },
    {
      "epoch": 0.17765440666204024,
      "grad_norm": 1.935426950454712,
      "learning_rate": 1.7762359063313096e-05,
      "loss": 0.7725,
      "step": 1024
    },
    {
      "epoch": 0.17782789729354614,
      "grad_norm": 0.8484199047088623,
      "learning_rate": 1.7779705117085865e-05,
      "loss": 0.6472,
      "step": 1025
    },
    {
      "epoch": 0.17800138792505205,
      "grad_norm": 0.953031599521637,
      "learning_rate": 1.779705117085863e-05,
      "loss": 0.6996,
      "step": 1026
    },
    {
      "epoch": 0.17817487855655795,
      "grad_norm": 0.6190481185913086,
      "learning_rate": 1.7814397224631398e-05,
      "loss": 0.7246,
      "step": 1027
    },
    {
      "epoch": 0.17834836918806385,
      "grad_norm": 0.7665278315544128,
      "learning_rate": 1.7831743278404166e-05,
      "loss": 0.7291,
      "step": 1028
    },
    {
      "epoch": 0.17852185981956975,
      "grad_norm": 1.5020960569381714,
      "learning_rate": 1.784908933217693e-05,
      "loss": 0.6891,
      "step": 1029
    },
    {
      "epoch": 0.17869535045107565,
      "grad_norm": 0.690614640712738,
      "learning_rate": 1.78664353859497e-05,
      "loss": 0.6453,
      "step": 1030
    },
    {
      "epoch": 0.17886884108258155,
      "grad_norm": 1.2899292707443237,
      "learning_rate": 1.7883781439722465e-05,
      "loss": 0.6635,
      "step": 1031
    },
    {
      "epoch": 0.17904233171408743,
      "grad_norm": 1.8824074268341064,
      "learning_rate": 1.7901127493495233e-05,
      "loss": 0.6235,
      "step": 1032
    },
    {
      "epoch": 0.17921582234559333,
      "grad_norm": 1.0645564794540405,
      "learning_rate": 1.7918473547267998e-05,
      "loss": 0.7957,
      "step": 1033
    },
    {
      "epoch": 0.17938931297709923,
      "grad_norm": 0.8005107641220093,
      "learning_rate": 1.7935819601040763e-05,
      "loss": 0.6934,
      "step": 1034
    },
    {
      "epoch": 0.17956280360860513,
      "grad_norm": 0.8759954571723938,
      "learning_rate": 1.795316565481353e-05,
      "loss": 0.6537,
      "step": 1035
    },
    {
      "epoch": 0.17973629424011103,
      "grad_norm": 1.047008752822876,
      "learning_rate": 1.7970511708586297e-05,
      "loss": 0.6056,
      "step": 1036
    },
    {
      "epoch": 0.17990978487161693,
      "grad_norm": 0.831057608127594,
      "learning_rate": 1.7987857762359065e-05,
      "loss": 0.6173,
      "step": 1037
    },
    {
      "epoch": 0.18008327550312284,
      "grad_norm": 1.0177525281906128,
      "learning_rate": 1.800520381613183e-05,
      "loss": 0.6202,
      "step": 1038
    },
    {
      "epoch": 0.18025676613462874,
      "grad_norm": 0.9185387492179871,
      "learning_rate": 1.80225498699046e-05,
      "loss": 0.7454,
      "step": 1039
    },
    {
      "epoch": 0.18043025676613464,
      "grad_norm": 0.8477778434753418,
      "learning_rate": 1.8039895923677363e-05,
      "loss": 0.6642,
      "step": 1040
    },
    {
      "epoch": 0.18060374739764054,
      "grad_norm": 1.2702401876449585,
      "learning_rate": 1.8057241977450132e-05,
      "loss": 0.7075,
      "step": 1041
    },
    {
      "epoch": 0.18077723802914641,
      "grad_norm": 0.902235209941864,
      "learning_rate": 1.8074588031222897e-05,
      "loss": 0.6028,
      "step": 1042
    },
    {
      "epoch": 0.18095072866065232,
      "grad_norm": 1.9682577848434448,
      "learning_rate": 1.8091934084995665e-05,
      "loss": 0.5509,
      "step": 1043
    },
    {
      "epoch": 0.18112421929215822,
      "grad_norm": 0.7393232583999634,
      "learning_rate": 1.8109280138768434e-05,
      "loss": 0.6578,
      "step": 1044
    },
    {
      "epoch": 0.18129770992366412,
      "grad_norm": 0.9041082262992859,
      "learning_rate": 1.81266261925412e-05,
      "loss": 0.5804,
      "step": 1045
    },
    {
      "epoch": 0.18147120055517002,
      "grad_norm": 0.6570352911949158,
      "learning_rate": 1.8143972246313967e-05,
      "loss": 0.7332,
      "step": 1046
    },
    {
      "epoch": 0.18164469118667592,
      "grad_norm": 1.3145637512207031,
      "learning_rate": 1.8161318300086732e-05,
      "loss": 0.5594,
      "step": 1047
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.9379259943962097,
      "learning_rate": 1.81786643538595e-05,
      "loss": 0.6917,
      "step": 1048
    },
    {
      "epoch": 0.18199167244968772,
      "grad_norm": 1.066662311553955,
      "learning_rate": 1.8196010407632265e-05,
      "loss": 0.676,
      "step": 1049
    },
    {
      "epoch": 0.18216516308119363,
      "grad_norm": 1.6131765842437744,
      "learning_rate": 1.821335646140503e-05,
      "loss": 0.7084,
      "step": 1050
    },
    {
      "epoch": 0.18233865371269953,
      "grad_norm": 0.7636017203330994,
      "learning_rate": 1.82307025151778e-05,
      "loss": 0.7501,
      "step": 1051
    },
    {
      "epoch": 0.1825121443442054,
      "grad_norm": 1.3714878559112549,
      "learning_rate": 1.8248048568950564e-05,
      "loss": 0.6624,
      "step": 1052
    },
    {
      "epoch": 0.1826856349757113,
      "grad_norm": 0.9227768778800964,
      "learning_rate": 1.8265394622723332e-05,
      "loss": 0.6272,
      "step": 1053
    },
    {
      "epoch": 0.1828591256072172,
      "grad_norm": 2.468353271484375,
      "learning_rate": 1.8282740676496097e-05,
      "loss": 0.5273,
      "step": 1054
    },
    {
      "epoch": 0.1830326162387231,
      "grad_norm": 0.7945556640625,
      "learning_rate": 1.8300086730268865e-05,
      "loss": 0.6641,
      "step": 1055
    },
    {
      "epoch": 0.183206106870229,
      "grad_norm": 1.8361812829971313,
      "learning_rate": 1.831743278404163e-05,
      "loss": 0.5623,
      "step": 1056
    },
    {
      "epoch": 0.1833795975017349,
      "grad_norm": 0.9092496037483215,
      "learning_rate": 1.83347788378144e-05,
      "loss": 0.6288,
      "step": 1057
    },
    {
      "epoch": 0.1835530881332408,
      "grad_norm": 1.0778285264968872,
      "learning_rate": 1.8352124891587167e-05,
      "loss": 0.6891,
      "step": 1058
    },
    {
      "epoch": 0.1837265787647467,
      "grad_norm": 0.9915594458580017,
      "learning_rate": 1.8369470945359932e-05,
      "loss": 0.7334,
      "step": 1059
    },
    {
      "epoch": 0.1839000693962526,
      "grad_norm": 0.8176564574241638,
      "learning_rate": 1.83868169991327e-05,
      "loss": 0.6724,
      "step": 1060
    },
    {
      "epoch": 0.18407356002775851,
      "grad_norm": 0.9000877737998962,
      "learning_rate": 1.8404163052905466e-05,
      "loss": 0.74,
      "step": 1061
    },
    {
      "epoch": 0.1842470506592644,
      "grad_norm": 0.617867112159729,
      "learning_rate": 1.8421509106678234e-05,
      "loss": 0.7555,
      "step": 1062
    },
    {
      "epoch": 0.1844205412907703,
      "grad_norm": 0.8626068234443665,
      "learning_rate": 1.8438855160451e-05,
      "loss": 0.6428,
      "step": 1063
    },
    {
      "epoch": 0.1845940319222762,
      "grad_norm": 0.8504796028137207,
      "learning_rate": 1.8456201214223764e-05,
      "loss": 0.7659,
      "step": 1064
    },
    {
      "epoch": 0.1847675225537821,
      "grad_norm": 0.8839169144630432,
      "learning_rate": 1.8473547267996532e-05,
      "loss": 0.7334,
      "step": 1065
    },
    {
      "epoch": 0.184941013185288,
      "grad_norm": 0.7100916504859924,
      "learning_rate": 1.8490893321769297e-05,
      "loss": 0.7568,
      "step": 1066
    },
    {
      "epoch": 0.1851145038167939,
      "grad_norm": 1.005721926689148,
      "learning_rate": 1.8508239375542066e-05,
      "loss": 0.5739,
      "step": 1067
    },
    {
      "epoch": 0.1852879944482998,
      "grad_norm": 1.0025429725646973,
      "learning_rate": 1.852558542931483e-05,
      "loss": 0.5717,
      "step": 1068
    },
    {
      "epoch": 0.1854614850798057,
      "grad_norm": 0.7993957996368408,
      "learning_rate": 1.85429314830876e-05,
      "loss": 0.6868,
      "step": 1069
    },
    {
      "epoch": 0.1856349757113116,
      "grad_norm": 1.636637806892395,
      "learning_rate": 1.8560277536860364e-05,
      "loss": 0.7036,
      "step": 1070
    },
    {
      "epoch": 0.18580846634281747,
      "grad_norm": 0.7456915378570557,
      "learning_rate": 1.8577623590633133e-05,
      "loss": 0.6378,
      "step": 1071
    },
    {
      "epoch": 0.18598195697432338,
      "grad_norm": 2.336636543273926,
      "learning_rate": 1.85949696444059e-05,
      "loss": 0.7042,
      "step": 1072
    },
    {
      "epoch": 0.18615544760582928,
      "grad_norm": 0.7203032970428467,
      "learning_rate": 1.8612315698178666e-05,
      "loss": 0.6406,
      "step": 1073
    },
    {
      "epoch": 0.18632893823733518,
      "grad_norm": 1.2361599206924438,
      "learning_rate": 1.8629661751951434e-05,
      "loss": 0.7024,
      "step": 1074
    },
    {
      "epoch": 0.18650242886884108,
      "grad_norm": 0.71657794713974,
      "learning_rate": 1.86470078057242e-05,
      "loss": 0.6204,
      "step": 1075
    },
    {
      "epoch": 0.18667591950034698,
      "grad_norm": 1.2553610801696777,
      "learning_rate": 1.8664353859496968e-05,
      "loss": 0.6677,
      "step": 1076
    },
    {
      "epoch": 0.18684941013185288,
      "grad_norm": 0.9711937308311462,
      "learning_rate": 1.8681699913269733e-05,
      "loss": 0.614,
      "step": 1077
    },
    {
      "epoch": 0.18702290076335878,
      "grad_norm": 0.6472023129463196,
      "learning_rate": 1.86990459670425e-05,
      "loss": 0.7666,
      "step": 1078
    },
    {
      "epoch": 0.1871963913948647,
      "grad_norm": 1.3309922218322754,
      "learning_rate": 1.8716392020815266e-05,
      "loss": 0.6407,
      "step": 1079
    },
    {
      "epoch": 0.1873698820263706,
      "grad_norm": 0.9143368601799011,
      "learning_rate": 1.873373807458803e-05,
      "loss": 0.6594,
      "step": 1080
    },
    {
      "epoch": 0.18754337265787646,
      "grad_norm": 1.003936529159546,
      "learning_rate": 1.87510841283608e-05,
      "loss": 0.6857,
      "step": 1081
    },
    {
      "epoch": 0.18771686328938236,
      "grad_norm": 0.7457190752029419,
      "learning_rate": 1.8768430182133565e-05,
      "loss": 0.7072,
      "step": 1082
    },
    {
      "epoch": 0.18789035392088826,
      "grad_norm": 0.8663756251335144,
      "learning_rate": 1.8785776235906333e-05,
      "loss": 0.6554,
      "step": 1083
    },
    {
      "epoch": 0.18806384455239417,
      "grad_norm": 0.7410752177238464,
      "learning_rate": 1.8803122289679098e-05,
      "loss": 0.7334,
      "step": 1084
    },
    {
      "epoch": 0.18823733518390007,
      "grad_norm": 0.8176028728485107,
      "learning_rate": 1.8820468343451866e-05,
      "loss": 0.7578,
      "step": 1085
    },
    {
      "epoch": 0.18841082581540597,
      "grad_norm": 1.0696377754211426,
      "learning_rate": 1.883781439722463e-05,
      "loss": 0.7449,
      "step": 1086
    },
    {
      "epoch": 0.18858431644691187,
      "grad_norm": 0.7168681621551514,
      "learning_rate": 1.88551604509974e-05,
      "loss": 0.7106,
      "step": 1087
    },
    {
      "epoch": 0.18875780707841777,
      "grad_norm": 0.7650038599967957,
      "learning_rate": 1.8872506504770168e-05,
      "loss": 0.6597,
      "step": 1088
    },
    {
      "epoch": 0.18893129770992367,
      "grad_norm": 1.2690316438674927,
      "learning_rate": 1.8889852558542933e-05,
      "loss": 0.6167,
      "step": 1089
    },
    {
      "epoch": 0.18910478834142957,
      "grad_norm": 1.2952897548675537,
      "learning_rate": 1.89071986123157e-05,
      "loss": 0.611,
      "step": 1090
    },
    {
      "epoch": 0.18927827897293545,
      "grad_norm": 2.8345625400543213,
      "learning_rate": 1.8924544666088467e-05,
      "loss": 0.7052,
      "step": 1091
    },
    {
      "epoch": 0.18945176960444135,
      "grad_norm": 0.8807550668716431,
      "learning_rate": 1.8941890719861235e-05,
      "loss": 0.598,
      "step": 1092
    },
    {
      "epoch": 0.18962526023594725,
      "grad_norm": 0.739676296710968,
      "learning_rate": 1.8959236773634e-05,
      "loss": 0.6504,
      "step": 1093
    },
    {
      "epoch": 0.18979875086745315,
      "grad_norm": 0.7994574904441833,
      "learning_rate": 1.8976582827406765e-05,
      "loss": 0.7261,
      "step": 1094
    },
    {
      "epoch": 0.18997224149895905,
      "grad_norm": 0.6968166828155518,
      "learning_rate": 1.8993928881179533e-05,
      "loss": 0.698,
      "step": 1095
    },
    {
      "epoch": 0.19014573213046496,
      "grad_norm": 0.8572284579277039,
      "learning_rate": 1.9011274934952298e-05,
      "loss": 0.6481,
      "step": 1096
    },
    {
      "epoch": 0.19031922276197086,
      "grad_norm": 0.714260995388031,
      "learning_rate": 1.9028620988725067e-05,
      "loss": 0.7607,
      "step": 1097
    },
    {
      "epoch": 0.19049271339347676,
      "grad_norm": 0.815679132938385,
      "learning_rate": 1.904596704249783e-05,
      "loss": 0.5797,
      "step": 1098
    },
    {
      "epoch": 0.19066620402498266,
      "grad_norm": 0.9675291180610657,
      "learning_rate": 1.90633130962706e-05,
      "loss": 0.6279,
      "step": 1099
    },
    {
      "epoch": 0.19083969465648856,
      "grad_norm": 2.4207892417907715,
      "learning_rate": 1.9080659150043365e-05,
      "loss": 0.5724,
      "step": 1100
    },
    {
      "epoch": 0.19101318528799444,
      "grad_norm": 1.2507901191711426,
      "learning_rate": 1.9098005203816133e-05,
      "loss": 0.8313,
      "step": 1101
    },
    {
      "epoch": 0.19118667591950034,
      "grad_norm": 0.9661257266998291,
      "learning_rate": 1.9115351257588902e-05,
      "loss": 0.6674,
      "step": 1102
    },
    {
      "epoch": 0.19136016655100624,
      "grad_norm": 1.0192809104919434,
      "learning_rate": 1.9132697311361667e-05,
      "loss": 0.5961,
      "step": 1103
    },
    {
      "epoch": 0.19153365718251214,
      "grad_norm": 0.9285260438919067,
      "learning_rate": 1.9150043365134435e-05,
      "loss": 0.5983,
      "step": 1104
    },
    {
      "epoch": 0.19170714781401804,
      "grad_norm": 0.9810802340507507,
      "learning_rate": 1.91673894189072e-05,
      "loss": 0.6393,
      "step": 1105
    },
    {
      "epoch": 0.19188063844552394,
      "grad_norm": 1.281665563583374,
      "learning_rate": 1.918473547267997e-05,
      "loss": 0.7402,
      "step": 1106
    },
    {
      "epoch": 0.19205412907702984,
      "grad_norm": 1.100449562072754,
      "learning_rate": 1.9202081526452734e-05,
      "loss": 0.6868,
      "step": 1107
    },
    {
      "epoch": 0.19222761970853575,
      "grad_norm": 1.5075404644012451,
      "learning_rate": 1.9219427580225502e-05,
      "loss": 0.615,
      "step": 1108
    },
    {
      "epoch": 0.19240111034004165,
      "grad_norm": 1.5390740633010864,
      "learning_rate": 1.9236773633998267e-05,
      "loss": 0.577,
      "step": 1109
    },
    {
      "epoch": 0.19257460097154755,
      "grad_norm": 0.8646528124809265,
      "learning_rate": 1.9254119687771032e-05,
      "loss": 0.6506,
      "step": 1110
    },
    {
      "epoch": 0.19274809160305342,
      "grad_norm": 0.8771539926528931,
      "learning_rate": 1.92714657415438e-05,
      "loss": 0.6888,
      "step": 1111
    },
    {
      "epoch": 0.19292158223455932,
      "grad_norm": 0.824918270111084,
      "learning_rate": 1.9288811795316565e-05,
      "loss": 0.6956,
      "step": 1112
    },
    {
      "epoch": 0.19309507286606523,
      "grad_norm": 0.8597473502159119,
      "learning_rate": 1.9306157849089334e-05,
      "loss": 0.6765,
      "step": 1113
    },
    {
      "epoch": 0.19326856349757113,
      "grad_norm": 1.0198392868041992,
      "learning_rate": 1.93235039028621e-05,
      "loss": 0.7966,
      "step": 1114
    },
    {
      "epoch": 0.19344205412907703,
      "grad_norm": 1.0613431930541992,
      "learning_rate": 1.9340849956634867e-05,
      "loss": 0.6799,
      "step": 1115
    },
    {
      "epoch": 0.19361554476058293,
      "grad_norm": 1.0472047328948975,
      "learning_rate": 1.9358196010407632e-05,
      "loss": 0.5466,
      "step": 1116
    },
    {
      "epoch": 0.19378903539208883,
      "grad_norm": 1.0182578563690186,
      "learning_rate": 1.93755420641804e-05,
      "loss": 0.6404,
      "step": 1117
    },
    {
      "epoch": 0.19396252602359473,
      "grad_norm": 1.1602972745895386,
      "learning_rate": 1.939288811795317e-05,
      "loss": 0.7067,
      "step": 1118
    },
    {
      "epoch": 0.19413601665510063,
      "grad_norm": 0.7067492008209229,
      "learning_rate": 1.9410234171725934e-05,
      "loss": 0.6321,
      "step": 1119
    },
    {
      "epoch": 0.19430950728660654,
      "grad_norm": 0.6538707613945007,
      "learning_rate": 1.9427580225498702e-05,
      "loss": 0.6466,
      "step": 1120
    },
    {
      "epoch": 0.1944829979181124,
      "grad_norm": 0.7539204955101013,
      "learning_rate": 1.9444926279271467e-05,
      "loss": 0.6584,
      "step": 1121
    },
    {
      "epoch": 0.1946564885496183,
      "grad_norm": 0.9606083631515503,
      "learning_rate": 1.9462272333044236e-05,
      "loss": 0.7148,
      "step": 1122
    },
    {
      "epoch": 0.1948299791811242,
      "grad_norm": 1.1239814758300781,
      "learning_rate": 1.9479618386817e-05,
      "loss": 0.7046,
      "step": 1123
    },
    {
      "epoch": 0.19500346981263011,
      "grad_norm": 0.8692232370376587,
      "learning_rate": 1.9496964440589766e-05,
      "loss": 0.5984,
      "step": 1124
    },
    {
      "epoch": 0.19517696044413602,
      "grad_norm": 0.9197940826416016,
      "learning_rate": 1.9514310494362534e-05,
      "loss": 0.7756,
      "step": 1125
    },
    {
      "epoch": 0.19535045107564192,
      "grad_norm": 0.8177411556243896,
      "learning_rate": 1.95316565481353e-05,
      "loss": 0.7275,
      "step": 1126
    },
    {
      "epoch": 0.19552394170714782,
      "grad_norm": 1.1024653911590576,
      "learning_rate": 1.9549002601908068e-05,
      "loss": 0.7601,
      "step": 1127
    },
    {
      "epoch": 0.19569743233865372,
      "grad_norm": 2.704585552215576,
      "learning_rate": 1.9566348655680833e-05,
      "loss": 0.7224,
      "step": 1128
    },
    {
      "epoch": 0.19587092297015962,
      "grad_norm": 0.9784897565841675,
      "learning_rate": 1.95836947094536e-05,
      "loss": 0.6036,
      "step": 1129
    },
    {
      "epoch": 0.19604441360166552,
      "grad_norm": 1.8710535764694214,
      "learning_rate": 1.9601040763226366e-05,
      "loss": 0.7267,
      "step": 1130
    },
    {
      "epoch": 0.1962179042331714,
      "grad_norm": 1.8200170993804932,
      "learning_rate": 1.9618386816999134e-05,
      "loss": 0.6038,
      "step": 1131
    },
    {
      "epoch": 0.1963913948646773,
      "grad_norm": 1.568758249282837,
      "learning_rate": 1.9635732870771903e-05,
      "loss": 0.5441,
      "step": 1132
    },
    {
      "epoch": 0.1965648854961832,
      "grad_norm": 2.1508874893188477,
      "learning_rate": 1.9653078924544668e-05,
      "loss": 0.6168,
      "step": 1133
    },
    {
      "epoch": 0.1967383761276891,
      "grad_norm": 1.4580485820770264,
      "learning_rate": 1.9670424978317436e-05,
      "loss": 0.6018,
      "step": 1134
    },
    {
      "epoch": 0.196911866759195,
      "grad_norm": 0.9399105310440063,
      "learning_rate": 1.96877710320902e-05,
      "loss": 0.6045,
      "step": 1135
    },
    {
      "epoch": 0.1970853573907009,
      "grad_norm": 1.007959008216858,
      "learning_rate": 1.970511708586297e-05,
      "loss": 0.6597,
      "step": 1136
    },
    {
      "epoch": 0.1972588480222068,
      "grad_norm": 1.3925546407699585,
      "learning_rate": 1.9722463139635735e-05,
      "loss": 0.658,
      "step": 1137
    },
    {
      "epoch": 0.1974323386537127,
      "grad_norm": 1.083231806755066,
      "learning_rate": 1.9739809193408503e-05,
      "loss": 0.5747,
      "step": 1138
    },
    {
      "epoch": 0.1976058292852186,
      "grad_norm": 1.03508460521698,
      "learning_rate": 1.9757155247181268e-05,
      "loss": 0.5938,
      "step": 1139
    },
    {
      "epoch": 0.19777931991672448,
      "grad_norm": 0.9703415632247925,
      "learning_rate": 1.9774501300954033e-05,
      "loss": 0.6207,
      "step": 1140
    },
    {
      "epoch": 0.19795281054823038,
      "grad_norm": 0.9999934434890747,
      "learning_rate": 1.97918473547268e-05,
      "loss": 0.6732,
      "step": 1141
    },
    {
      "epoch": 0.19812630117973629,
      "grad_norm": 1.4494614601135254,
      "learning_rate": 1.9809193408499566e-05,
      "loss": 0.5573,
      "step": 1142
    },
    {
      "epoch": 0.1982997918112422,
      "grad_norm": 2.6835134029388428,
      "learning_rate": 1.9826539462272335e-05,
      "loss": 0.6957,
      "step": 1143
    },
    {
      "epoch": 0.1984732824427481,
      "grad_norm": 1.1297166347503662,
      "learning_rate": 1.98438855160451e-05,
      "loss": 0.6608,
      "step": 1144
    },
    {
      "epoch": 0.198646773074254,
      "grad_norm": 0.9641552567481995,
      "learning_rate": 1.9861231569817868e-05,
      "loss": 0.7238,
      "step": 1145
    },
    {
      "epoch": 0.1988202637057599,
      "grad_norm": 0.6486564874649048,
      "learning_rate": 1.9878577623590636e-05,
      "loss": 0.7588,
      "step": 1146
    },
    {
      "epoch": 0.1989937543372658,
      "grad_norm": 1.3994876146316528,
      "learning_rate": 1.98959236773634e-05,
      "loss": 0.5844,
      "step": 1147
    },
    {
      "epoch": 0.1991672449687717,
      "grad_norm": 1.1556226015090942,
      "learning_rate": 1.991326973113617e-05,
      "loss": 0.6755,
      "step": 1148
    },
    {
      "epoch": 0.1993407356002776,
      "grad_norm": 0.6605205535888672,
      "learning_rate": 1.9930615784908935e-05,
      "loss": 0.7324,
      "step": 1149
    },
    {
      "epoch": 0.19951422623178347,
      "grad_norm": 0.8294914960861206,
      "learning_rate": 1.9947961838681703e-05,
      "loss": 0.6222,
      "step": 1150
    },
    {
      "epoch": 0.19968771686328937,
      "grad_norm": 0.6706499457359314,
      "learning_rate": 1.9965307892454468e-05,
      "loss": 0.7544,
      "step": 1151
    },
    {
      "epoch": 0.19986120749479527,
      "grad_norm": 1.0484025478363037,
      "learning_rate": 1.9982653946227237e-05,
      "loss": 0.6194,
      "step": 1152
    },
    {
      "epoch": 0.20003469812630117,
      "grad_norm": 0.8263928294181824,
      "learning_rate": 2e-05,
      "loss": 0.6472,
      "step": 1153
    },
    {
      "epoch": 0.20020818875780708,
      "grad_norm": 1.6001183986663818,
      "learning_rate": 2.0017346053772767e-05,
      "loss": 0.646,
      "step": 1154
    },
    {
      "epoch": 0.20038167938931298,
      "grad_norm": 1.1579716205596924,
      "learning_rate": 2.0034692107545535e-05,
      "loss": 0.6827,
      "step": 1155
    },
    {
      "epoch": 0.20055517002081888,
      "grad_norm": 1.0516129732131958,
      "learning_rate": 2.00520381613183e-05,
      "loss": 0.6022,
      "step": 1156
    },
    {
      "epoch": 0.20072866065232478,
      "grad_norm": 0.8496077656745911,
      "learning_rate": 2.006938421509107e-05,
      "loss": 0.5554,
      "step": 1157
    },
    {
      "epoch": 0.20090215128383068,
      "grad_norm": 1.4365261793136597,
      "learning_rate": 2.0086730268863833e-05,
      "loss": 0.7332,
      "step": 1158
    },
    {
      "epoch": 0.20107564191533658,
      "grad_norm": 0.7282311320304871,
      "learning_rate": 2.0104076322636602e-05,
      "loss": 0.7644,
      "step": 1159
    },
    {
      "epoch": 0.20124913254684246,
      "grad_norm": 0.7782039642333984,
      "learning_rate": 2.0121422376409367e-05,
      "loss": 0.6063,
      "step": 1160
    },
    {
      "epoch": 0.20142262317834836,
      "grad_norm": 0.5990754961967468,
      "learning_rate": 2.0138768430182135e-05,
      "loss": 0.7354,
      "step": 1161
    },
    {
      "epoch": 0.20159611380985426,
      "grad_norm": 0.8651938438415527,
      "learning_rate": 2.01561144839549e-05,
      "loss": 0.5518,
      "step": 1162
    },
    {
      "epoch": 0.20176960444136016,
      "grad_norm": 1.5074180364608765,
      "learning_rate": 2.017346053772767e-05,
      "loss": 0.7057,
      "step": 1163
    },
    {
      "epoch": 0.20194309507286606,
      "grad_norm": 0.8941850662231445,
      "learning_rate": 2.0190806591500434e-05,
      "loss": 0.7015,
      "step": 1164
    },
    {
      "epoch": 0.20211658570437196,
      "grad_norm": 0.8068917989730835,
      "learning_rate": 2.02081526452732e-05,
      "loss": 0.649,
      "step": 1165
    },
    {
      "epoch": 0.20229007633587787,
      "grad_norm": 1.2736928462982178,
      "learning_rate": 2.0225498699045967e-05,
      "loss": 0.6069,
      "step": 1166
    },
    {
      "epoch": 0.20246356696738377,
      "grad_norm": 0.8033802509307861,
      "learning_rate": 2.0242844752818732e-05,
      "loss": 0.6732,
      "step": 1167
    },
    {
      "epoch": 0.20263705759888967,
      "grad_norm": 0.9226799607276917,
      "learning_rate": 2.0260190806591504e-05,
      "loss": 0.5758,
      "step": 1168
    },
    {
      "epoch": 0.20281054823039557,
      "grad_norm": 0.889927089214325,
      "learning_rate": 2.0277536860364272e-05,
      "loss": 0.635,
      "step": 1169
    },
    {
      "epoch": 0.20298403886190144,
      "grad_norm": 1.000593662261963,
      "learning_rate": 2.0294882914137037e-05,
      "loss": 0.6479,
      "step": 1170
    },
    {
      "epoch": 0.20315752949340735,
      "grad_norm": 0.9482126235961914,
      "learning_rate": 2.0312228967909806e-05,
      "loss": 0.7021,
      "step": 1171
    },
    {
      "epoch": 0.20333102012491325,
      "grad_norm": 0.7132568955421448,
      "learning_rate": 2.032957502168257e-05,
      "loss": 0.6385,
      "step": 1172
    },
    {
      "epoch": 0.20350451075641915,
      "grad_norm": 0.9106391072273254,
      "learning_rate": 2.0346921075455336e-05,
      "loss": 0.6313,
      "step": 1173
    },
    {
      "epoch": 0.20367800138792505,
      "grad_norm": 0.8777891993522644,
      "learning_rate": 2.0364267129228104e-05,
      "loss": 0.6902,
      "step": 1174
    },
    {
      "epoch": 0.20385149201943095,
      "grad_norm": 0.7946134209632874,
      "learning_rate": 2.038161318300087e-05,
      "loss": 0.5692,
      "step": 1175
    },
    {
      "epoch": 0.20402498265093685,
      "grad_norm": 0.7996445298194885,
      "learning_rate": 2.0398959236773637e-05,
      "loss": 0.6609,
      "step": 1176
    },
    {
      "epoch": 0.20419847328244276,
      "grad_norm": 0.9031609892845154,
      "learning_rate": 2.0416305290546402e-05,
      "loss": 0.6771,
      "step": 1177
    },
    {
      "epoch": 0.20437196391394866,
      "grad_norm": 1.042405605316162,
      "learning_rate": 2.043365134431917e-05,
      "loss": 0.6077,
      "step": 1178
    },
    {
      "epoch": 0.20454545454545456,
      "grad_norm": 0.8185474276542664,
      "learning_rate": 2.0450997398091936e-05,
      "loss": 0.6809,
      "step": 1179
    },
    {
      "epoch": 0.20471894517696043,
      "grad_norm": 0.7025969624519348,
      "learning_rate": 2.0468343451864704e-05,
      "loss": 0.6758,
      "step": 1180
    },
    {
      "epoch": 0.20489243580846633,
      "grad_norm": 0.7403766512870789,
      "learning_rate": 2.048568950563747e-05,
      "loss": 0.635,
      "step": 1181
    },
    {
      "epoch": 0.20506592643997223,
      "grad_norm": 1.3111224174499512,
      "learning_rate": 2.0503035559410238e-05,
      "loss": 0.6948,
      "step": 1182
    },
    {
      "epoch": 0.20523941707147814,
      "grad_norm": 0.9487437009811401,
      "learning_rate": 2.0520381613183003e-05,
      "loss": 0.6877,
      "step": 1183
    },
    {
      "epoch": 0.20541290770298404,
      "grad_norm": 0.8493991494178772,
      "learning_rate": 2.0537727666955768e-05,
      "loss": 0.6831,
      "step": 1184
    },
    {
      "epoch": 0.20558639833448994,
      "grad_norm": 0.9466814398765564,
      "learning_rate": 2.0555073720728536e-05,
      "loss": 0.824,
      "step": 1185
    },
    {
      "epoch": 0.20575988896599584,
      "grad_norm": 1.1064136028289795,
      "learning_rate": 2.05724197745013e-05,
      "loss": 0.6874,
      "step": 1186
    },
    {
      "epoch": 0.20593337959750174,
      "grad_norm": 1.2074017524719238,
      "learning_rate": 2.058976582827407e-05,
      "loss": 0.8259,
      "step": 1187
    },
    {
      "epoch": 0.20610687022900764,
      "grad_norm": 0.7420201301574707,
      "learning_rate": 2.0607111882046834e-05,
      "loss": 0.7096,
      "step": 1188
    },
    {
      "epoch": 0.20628036086051355,
      "grad_norm": 1.0953930616378784,
      "learning_rate": 2.0624457935819603e-05,
      "loss": 0.6317,
      "step": 1189
    },
    {
      "epoch": 0.20645385149201942,
      "grad_norm": 0.8365890979766846,
      "learning_rate": 2.0641803989592368e-05,
      "loss": 0.7136,
      "step": 1190
    },
    {
      "epoch": 0.20662734212352532,
      "grad_norm": 0.860747218132019,
      "learning_rate": 2.0659150043365136e-05,
      "loss": 0.6864,
      "step": 1191
    },
    {
      "epoch": 0.20680083275503122,
      "grad_norm": 0.7875193357467651,
      "learning_rate": 2.06764960971379e-05,
      "loss": 0.7091,
      "step": 1192
    },
    {
      "epoch": 0.20697432338653712,
      "grad_norm": 0.8169761896133423,
      "learning_rate": 2.069384215091067e-05,
      "loss": 0.5524,
      "step": 1193
    },
    {
      "epoch": 0.20714781401804302,
      "grad_norm": 0.8388202786445618,
      "learning_rate": 2.0711188204683434e-05,
      "loss": 0.6893,
      "step": 1194
    },
    {
      "epoch": 0.20732130464954893,
      "grad_norm": 1.2917083501815796,
      "learning_rate": 2.07285342584562e-05,
      "loss": 0.5945,
      "step": 1195
    },
    {
      "epoch": 0.20749479528105483,
      "grad_norm": 0.9487801194190979,
      "learning_rate": 2.0745880312228968e-05,
      "loss": 0.6343,
      "step": 1196
    },
    {
      "epoch": 0.20766828591256073,
      "grad_norm": 0.9186866283416748,
      "learning_rate": 2.0763226366001733e-05,
      "loss": 0.6746,
      "step": 1197
    },
    {
      "epoch": 0.20784177654406663,
      "grad_norm": 1.1883459091186523,
      "learning_rate": 2.0780572419774505e-05,
      "loss": 0.6559,
      "step": 1198
    },
    {
      "epoch": 0.20801526717557253,
      "grad_norm": 0.8712085485458374,
      "learning_rate": 2.0797918473547273e-05,
      "loss": 0.6472,
      "step": 1199
    },
    {
      "epoch": 0.2081887578070784,
      "grad_norm": 0.7856881618499756,
      "learning_rate": 2.0815264527320038e-05,
      "loss": 0.8369,
      "step": 1200
    },
    {
      "epoch": 0.2083622484385843,
      "grad_norm": 1.4384877681732178,
      "learning_rate": 2.0832610581092806e-05,
      "loss": 0.6841,
      "step": 1201
    },
    {
      "epoch": 0.2085357390700902,
      "grad_norm": 1.0486500263214111,
      "learning_rate": 2.084995663486557e-05,
      "loss": 0.6429,
      "step": 1202
    },
    {
      "epoch": 0.2087092297015961,
      "grad_norm": 0.9659835696220398,
      "learning_rate": 2.0867302688638336e-05,
      "loss": 0.634,
      "step": 1203
    },
    {
      "epoch": 0.208882720333102,
      "grad_norm": 0.7448798418045044,
      "learning_rate": 2.0884648742411105e-05,
      "loss": 0.6191,
      "step": 1204
    },
    {
      "epoch": 0.2090562109646079,
      "grad_norm": 0.9722678065299988,
      "learning_rate": 2.090199479618387e-05,
      "loss": 0.7395,
      "step": 1205
    },
    {
      "epoch": 0.20922970159611382,
      "grad_norm": 0.8635239005088806,
      "learning_rate": 2.0919340849956638e-05,
      "loss": 0.7175,
      "step": 1206
    },
    {
      "epoch": 0.20940319222761972,
      "grad_norm": 0.7446389198303223,
      "learning_rate": 2.0936686903729403e-05,
      "loss": 0.8132,
      "step": 1207
    },
    {
      "epoch": 0.20957668285912562,
      "grad_norm": 0.9571833610534668,
      "learning_rate": 2.095403295750217e-05,
      "loss": 0.783,
      "step": 1208
    },
    {
      "epoch": 0.20975017349063152,
      "grad_norm": 1.470550775527954,
      "learning_rate": 2.0971379011274937e-05,
      "loss": 0.5852,
      "step": 1209
    },
    {
      "epoch": 0.2099236641221374,
      "grad_norm": 0.820504903793335,
      "learning_rate": 2.0988725065047705e-05,
      "loss": 0.6759,
      "step": 1210
    },
    {
      "epoch": 0.2100971547536433,
      "grad_norm": 1.0833055973052979,
      "learning_rate": 2.100607111882047e-05,
      "loss": 0.6273,
      "step": 1211
    },
    {
      "epoch": 0.2102706453851492,
      "grad_norm": 0.8269476890563965,
      "learning_rate": 2.102341717259324e-05,
      "loss": 0.7256,
      "step": 1212
    },
    {
      "epoch": 0.2104441360166551,
      "grad_norm": 0.9470596313476562,
      "learning_rate": 2.1040763226366003e-05,
      "loss": 0.6292,
      "step": 1213
    },
    {
      "epoch": 0.210617626648161,
      "grad_norm": 2.144624710083008,
      "learning_rate": 2.105810928013877e-05,
      "loss": 0.5927,
      "step": 1214
    },
    {
      "epoch": 0.2107911172796669,
      "grad_norm": 0.7160753607749939,
      "learning_rate": 2.1075455333911537e-05,
      "loss": 0.6057,
      "step": 1215
    },
    {
      "epoch": 0.2109646079111728,
      "grad_norm": 0.8870556354522705,
      "learning_rate": 2.1092801387684302e-05,
      "loss": 0.6117,
      "step": 1216
    },
    {
      "epoch": 0.2111380985426787,
      "grad_norm": 1.2170469760894775,
      "learning_rate": 2.111014744145707e-05,
      "loss": 0.6293,
      "step": 1217
    },
    {
      "epoch": 0.2113115891741846,
      "grad_norm": 1.373672604560852,
      "learning_rate": 2.1127493495229835e-05,
      "loss": 0.5717,
      "step": 1218
    },
    {
      "epoch": 0.21148507980569048,
      "grad_norm": 1.0951582193374634,
      "learning_rate": 2.1144839549002604e-05,
      "loss": 0.6418,
      "step": 1219
    },
    {
      "epoch": 0.21165857043719638,
      "grad_norm": 0.8446418642997742,
      "learning_rate": 2.116218560277537e-05,
      "loss": 0.5923,
      "step": 1220
    },
    {
      "epoch": 0.21183206106870228,
      "grad_norm": 0.7682600617408752,
      "learning_rate": 2.1179531656548137e-05,
      "loss": 0.7159,
      "step": 1221
    },
    {
      "epoch": 0.21200555170020818,
      "grad_norm": 0.8873063921928406,
      "learning_rate": 2.1196877710320902e-05,
      "loss": 0.7673,
      "step": 1222
    },
    {
      "epoch": 0.21217904233171409,
      "grad_norm": 0.7964252829551697,
      "learning_rate": 2.121422376409367e-05,
      "loss": 0.692,
      "step": 1223
    },
    {
      "epoch": 0.21235253296322,
      "grad_norm": 0.831866443157196,
      "learning_rate": 2.1231569817866435e-05,
      "loss": 0.73,
      "step": 1224
    },
    {
      "epoch": 0.2125260235947259,
      "grad_norm": 1.1679537296295166,
      "learning_rate": 2.12489158716392e-05,
      "loss": 0.7026,
      "step": 1225
    },
    {
      "epoch": 0.2126995142262318,
      "grad_norm": 1.1698815822601318,
      "learning_rate": 2.126626192541197e-05,
      "loss": 0.5995,
      "step": 1226
    },
    {
      "epoch": 0.2128730048577377,
      "grad_norm": 1.1372910737991333,
      "learning_rate": 2.128360797918474e-05,
      "loss": 0.6755,
      "step": 1227
    },
    {
      "epoch": 0.2130464954892436,
      "grad_norm": 0.8028494119644165,
      "learning_rate": 2.1300954032957505e-05,
      "loss": 0.5986,
      "step": 1228
    },
    {
      "epoch": 0.21321998612074947,
      "grad_norm": 0.8080164790153503,
      "learning_rate": 2.1318300086730274e-05,
      "loss": 0.5869,
      "step": 1229
    },
    {
      "epoch": 0.21339347675225537,
      "grad_norm": 1.5431324243545532,
      "learning_rate": 2.133564614050304e-05,
      "loss": 0.6377,
      "step": 1230
    },
    {
      "epoch": 0.21356696738376127,
      "grad_norm": 0.7453404068946838,
      "learning_rate": 2.1352992194275807e-05,
      "loss": 0.7156,
      "step": 1231
    },
    {
      "epoch": 0.21374045801526717,
      "grad_norm": 1.216294765472412,
      "learning_rate": 2.1370338248048572e-05,
      "loss": 0.5951,
      "step": 1232
    },
    {
      "epoch": 0.21391394864677307,
      "grad_norm": 1.3197087049484253,
      "learning_rate": 2.1387684301821337e-05,
      "loss": 0.533,
      "step": 1233
    },
    {
      "epoch": 0.21408743927827897,
      "grad_norm": 0.753345787525177,
      "learning_rate": 2.1405030355594106e-05,
      "loss": 0.5773,
      "step": 1234
    },
    {
      "epoch": 0.21426092990978488,
      "grad_norm": 1.1952768564224243,
      "learning_rate": 2.142237640936687e-05,
      "loss": 0.6125,
      "step": 1235
    },
    {
      "epoch": 0.21443442054129078,
      "grad_norm": 3.2357378005981445,
      "learning_rate": 2.143972246313964e-05,
      "loss": 0.6302,
      "step": 1236
    },
    {
      "epoch": 0.21460791117279668,
      "grad_norm": 1.2979215383529663,
      "learning_rate": 2.1457068516912404e-05,
      "loss": 0.6006,
      "step": 1237
    },
    {
      "epoch": 0.21478140180430258,
      "grad_norm": 0.8040156364440918,
      "learning_rate": 2.1474414570685172e-05,
      "loss": 0.771,
      "step": 1238
    },
    {
      "epoch": 0.21495489243580845,
      "grad_norm": 0.8429499268531799,
      "learning_rate": 2.1491760624457937e-05,
      "loss": 0.7969,
      "step": 1239
    },
    {
      "epoch": 0.21512838306731435,
      "grad_norm": 1.141281247138977,
      "learning_rate": 2.1509106678230706e-05,
      "loss": 0.6262,
      "step": 1240
    },
    {
      "epoch": 0.21530187369882026,
      "grad_norm": 0.8168627023696899,
      "learning_rate": 2.152645273200347e-05,
      "loss": 0.7534,
      "step": 1241
    },
    {
      "epoch": 0.21547536433032616,
      "grad_norm": 0.8681612610816956,
      "learning_rate": 2.154379878577624e-05,
      "loss": 0.7527,
      "step": 1242
    },
    {
      "epoch": 0.21564885496183206,
      "grad_norm": 1.500241756439209,
      "learning_rate": 2.1561144839549004e-05,
      "loss": 0.6274,
      "step": 1243
    },
    {
      "epoch": 0.21582234559333796,
      "grad_norm": 2.348919630050659,
      "learning_rate": 2.157849089332177e-05,
      "loss": 0.65,
      "step": 1244
    },
    {
      "epoch": 0.21599583622484386,
      "grad_norm": 0.8949500918388367,
      "learning_rate": 2.1595836947094538e-05,
      "loss": 0.5729,
      "step": 1245
    },
    {
      "epoch": 0.21616932685634976,
      "grad_norm": 0.706988513469696,
      "learning_rate": 2.1613183000867303e-05,
      "loss": 0.7266,
      "step": 1246
    },
    {
      "epoch": 0.21634281748785567,
      "grad_norm": 0.9260134696960449,
      "learning_rate": 2.163052905464007e-05,
      "loss": 0.6777,
      "step": 1247
    },
    {
      "epoch": 0.21651630811936157,
      "grad_norm": 1.1210914850234985,
      "learning_rate": 2.1647875108412836e-05,
      "loss": 0.5737,
      "step": 1248
    },
    {
      "epoch": 0.21668979875086744,
      "grad_norm": 0.857086718082428,
      "learning_rate": 2.1665221162185604e-05,
      "loss": 0.7446,
      "step": 1249
    },
    {
      "epoch": 0.21686328938237334,
      "grad_norm": 0.6255559325218201,
      "learning_rate": 2.168256721595837e-05,
      "loss": 0.7583,
      "step": 1250
    },
    {
      "epoch": 0.21703678001387924,
      "grad_norm": 0.9600679874420166,
      "learning_rate": 2.1699913269731138e-05,
      "loss": 0.7446,
      "step": 1251
    },
    {
      "epoch": 0.21721027064538515,
      "grad_norm": 0.932635486125946,
      "learning_rate": 2.1717259323503903e-05,
      "loss": 0.6096,
      "step": 1252
    },
    {
      "epoch": 0.21738376127689105,
      "grad_norm": 1.0339999198913574,
      "learning_rate": 2.173460537727667e-05,
      "loss": 0.6318,
      "step": 1253
    },
    {
      "epoch": 0.21755725190839695,
      "grad_norm": 1.0141346454620361,
      "learning_rate": 2.1751951431049436e-05,
      "loss": 0.5837,
      "step": 1254
    },
    {
      "epoch": 0.21773074253990285,
      "grad_norm": 1.782895565032959,
      "learning_rate": 2.17692974848222e-05,
      "loss": 0.5902,
      "step": 1255
    },
    {
      "epoch": 0.21790423317140875,
      "grad_norm": 0.7666664719581604,
      "learning_rate": 2.178664353859497e-05,
      "loss": 0.6473,
      "step": 1256
    },
    {
      "epoch": 0.21807772380291465,
      "grad_norm": 0.7407053112983704,
      "learning_rate": 2.180398959236774e-05,
      "loss": 0.5646,
      "step": 1257
    },
    {
      "epoch": 0.21825121443442055,
      "grad_norm": 0.742052435874939,
      "learning_rate": 2.1821335646140506e-05,
      "loss": 0.6991,
      "step": 1258
    },
    {
      "epoch": 0.21842470506592643,
      "grad_norm": 0.9027974009513855,
      "learning_rate": 2.1838681699913275e-05,
      "loss": 0.7266,
      "step": 1259
    },
    {
      "epoch": 0.21859819569743233,
      "grad_norm": 0.6909552216529846,
      "learning_rate": 2.185602775368604e-05,
      "loss": 0.7407,
      "step": 1260
    },
    {
      "epoch": 0.21877168632893823,
      "grad_norm": 0.9069677591323853,
      "learning_rate": 2.1873373807458808e-05,
      "loss": 0.6567,
      "step": 1261
    },
    {
      "epoch": 0.21894517696044413,
      "grad_norm": 0.8825785517692566,
      "learning_rate": 2.1890719861231573e-05,
      "loss": 0.6696,
      "step": 1262
    },
    {
      "epoch": 0.21911866759195003,
      "grad_norm": 0.6216393709182739,
      "learning_rate": 2.1908065915004338e-05,
      "loss": 0.6754,
      "step": 1263
    },
    {
      "epoch": 0.21929215822345594,
      "grad_norm": 0.7830954790115356,
      "learning_rate": 2.1925411968777107e-05,
      "loss": 0.7622,
      "step": 1264
    },
    {
      "epoch": 0.21946564885496184,
      "grad_norm": 0.897305428981781,
      "learning_rate": 2.194275802254987e-05,
      "loss": 0.7532,
      "step": 1265
    },
    {
      "epoch": 0.21963913948646774,
      "grad_norm": 0.8858644366264343,
      "learning_rate": 2.196010407632264e-05,
      "loss": 0.6171,
      "step": 1266
    },
    {
      "epoch": 0.21981263011797364,
      "grad_norm": 1.420778512954712,
      "learning_rate": 2.1977450130095405e-05,
      "loss": 0.5931,
      "step": 1267
    },
    {
      "epoch": 0.21998612074947954,
      "grad_norm": 0.7846259474754333,
      "learning_rate": 2.1994796183868173e-05,
      "loss": 0.7,
      "step": 1268
    },
    {
      "epoch": 0.22015961138098541,
      "grad_norm": 0.9707682132720947,
      "learning_rate": 2.201214223764094e-05,
      "loss": 0.7571,
      "step": 1269
    },
    {
      "epoch": 0.22033310201249132,
      "grad_norm": 0.917709469795227,
      "learning_rate": 2.2029488291413707e-05,
      "loss": 0.7003,
      "step": 1270
    },
    {
      "epoch": 0.22050659264399722,
      "grad_norm": 0.8586825728416443,
      "learning_rate": 2.2046834345186472e-05,
      "loss": 0.697,
      "step": 1271
    },
    {
      "epoch": 0.22068008327550312,
      "grad_norm": 0.9106934666633606,
      "learning_rate": 2.206418039895924e-05,
      "loss": 0.6313,
      "step": 1272
    },
    {
      "epoch": 0.22085357390700902,
      "grad_norm": 0.8944715857505798,
      "learning_rate": 2.2081526452732005e-05,
      "loss": 0.6823,
      "step": 1273
    },
    {
      "epoch": 0.22102706453851492,
      "grad_norm": 0.9109546542167664,
      "learning_rate": 2.209887250650477e-05,
      "loss": 0.5646,
      "step": 1274
    },
    {
      "epoch": 0.22120055517002082,
      "grad_norm": 0.8423030376434326,
      "learning_rate": 2.211621856027754e-05,
      "loss": 0.6578,
      "step": 1275
    },
    {
      "epoch": 0.22137404580152673,
      "grad_norm": 0.8873112201690674,
      "learning_rate": 2.2133564614050303e-05,
      "loss": 0.6309,
      "step": 1276
    },
    {
      "epoch": 0.22154753643303263,
      "grad_norm": 1.1744393110275269,
      "learning_rate": 2.2150910667823072e-05,
      "loss": 0.7184,
      "step": 1277
    },
    {
      "epoch": 0.22172102706453853,
      "grad_norm": 0.6773825883865356,
      "learning_rate": 2.2168256721595837e-05,
      "loss": 0.7881,
      "step": 1278
    },
    {
      "epoch": 0.2218945176960444,
      "grad_norm": 0.8595862984657288,
      "learning_rate": 2.2185602775368605e-05,
      "loss": 0.6566,
      "step": 1279
    },
    {
      "epoch": 0.2220680083275503,
      "grad_norm": 0.8502959609031677,
      "learning_rate": 2.220294882914137e-05,
      "loss": 0.5854,
      "step": 1280
    },
    {
      "epoch": 0.2222414989590562,
      "grad_norm": 0.8511287569999695,
      "learning_rate": 2.222029488291414e-05,
      "loss": 0.6636,
      "step": 1281
    },
    {
      "epoch": 0.2224149895905621,
      "grad_norm": 0.9587035775184631,
      "learning_rate": 2.2237640936686904e-05,
      "loss": 0.6709,
      "step": 1282
    },
    {
      "epoch": 0.222588480222068,
      "grad_norm": 0.8576658964157104,
      "learning_rate": 2.2254986990459672e-05,
      "loss": 0.5322,
      "step": 1283
    },
    {
      "epoch": 0.2227619708535739,
      "grad_norm": 1.0752559900283813,
      "learning_rate": 2.2272333044232437e-05,
      "loss": 0.705,
      "step": 1284
    },
    {
      "epoch": 0.2229354614850798,
      "grad_norm": 1.1957930326461792,
      "learning_rate": 2.2289679098005202e-05,
      "loss": 0.5844,
      "step": 1285
    },
    {
      "epoch": 0.2231089521165857,
      "grad_norm": 1.0012633800506592,
      "learning_rate": 2.230702515177797e-05,
      "loss": 0.6102,
      "step": 1286
    },
    {
      "epoch": 0.22328244274809161,
      "grad_norm": 1.117314100265503,
      "learning_rate": 2.2324371205550742e-05,
      "loss": 0.6372,
      "step": 1287
    },
    {
      "epoch": 0.22345593337959752,
      "grad_norm": 0.8368109464645386,
      "learning_rate": 2.2341717259323507e-05,
      "loss": 0.5454,
      "step": 1288
    },
    {
      "epoch": 0.2236294240111034,
      "grad_norm": 1.178607702255249,
      "learning_rate": 2.2359063313096276e-05,
      "loss": 0.6875,
      "step": 1289
    },
    {
      "epoch": 0.2238029146426093,
      "grad_norm": 0.7383229732513428,
      "learning_rate": 2.237640936686904e-05,
      "loss": 0.6957,
      "step": 1290
    },
    {
      "epoch": 0.2239764052741152,
      "grad_norm": 1.0060182809829712,
      "learning_rate": 2.2393755420641806e-05,
      "loss": 0.6543,
      "step": 1291
    },
    {
      "epoch": 0.2241498959056211,
      "grad_norm": 0.8270978927612305,
      "learning_rate": 2.2411101474414574e-05,
      "loss": 0.6302,
      "step": 1292
    },
    {
      "epoch": 0.224323386537127,
      "grad_norm": 1.0428016185760498,
      "learning_rate": 2.242844752818734e-05,
      "loss": 0.741,
      "step": 1293
    },
    {
      "epoch": 0.2244968771686329,
      "grad_norm": 0.7089052796363831,
      "learning_rate": 2.2445793581960107e-05,
      "loss": 0.7869,
      "step": 1294
    },
    {
      "epoch": 0.2246703678001388,
      "grad_norm": 1.201980710029602,
      "learning_rate": 2.2463139635732872e-05,
      "loss": 0.6082,
      "step": 1295
    },
    {
      "epoch": 0.2248438584316447,
      "grad_norm": 1.013190507888794,
      "learning_rate": 2.248048568950564e-05,
      "loss": 0.6641,
      "step": 1296
    },
    {
      "epoch": 0.2250173490631506,
      "grad_norm": 0.9957582950592041,
      "learning_rate": 2.2497831743278406e-05,
      "loss": 0.6294,
      "step": 1297
    },
    {
      "epoch": 0.22519083969465647,
      "grad_norm": 0.8996625542640686,
      "learning_rate": 2.2515177797051174e-05,
      "loss": 0.6589,
      "step": 1298
    },
    {
      "epoch": 0.22536433032616238,
      "grad_norm": 1.442975640296936,
      "learning_rate": 2.253252385082394e-05,
      "loss": 0.6422,
      "step": 1299
    },
    {
      "epoch": 0.22553782095766828,
      "grad_norm": 0.7491903305053711,
      "learning_rate": 2.2549869904596708e-05,
      "loss": 0.6428,
      "step": 1300
    },
    {
      "epoch": 0.22571131158917418,
      "grad_norm": 0.7685233950614929,
      "learning_rate": 2.2567215958369473e-05,
      "loss": 0.6646,
      "step": 1301
    },
    {
      "epoch": 0.22588480222068008,
      "grad_norm": 1.3932480812072754,
      "learning_rate": 2.258456201214224e-05,
      "loss": 0.6388,
      "step": 1302
    },
    {
      "epoch": 0.22605829285218598,
      "grad_norm": 0.8122157454490662,
      "learning_rate": 2.2601908065915006e-05,
      "loss": 0.6692,
      "step": 1303
    },
    {
      "epoch": 0.22623178348369188,
      "grad_norm": 0.857111930847168,
      "learning_rate": 2.261925411968777e-05,
      "loss": 0.8053,
      "step": 1304
    },
    {
      "epoch": 0.22640527411519779,
      "grad_norm": 1.4582233428955078,
      "learning_rate": 2.263660017346054e-05,
      "loss": 0.6735,
      "step": 1305
    },
    {
      "epoch": 0.2265787647467037,
      "grad_norm": 1.2034339904785156,
      "learning_rate": 2.2653946227233304e-05,
      "loss": 0.7019,
      "step": 1306
    },
    {
      "epoch": 0.2267522553782096,
      "grad_norm": 0.9479273557662964,
      "learning_rate": 2.2671292281006073e-05,
      "loss": 0.6255,
      "step": 1307
    },
    {
      "epoch": 0.22692574600971546,
      "grad_norm": 0.9462427496910095,
      "learning_rate": 2.2688638334778838e-05,
      "loss": 0.6827,
      "step": 1308
    },
    {
      "epoch": 0.22709923664122136,
      "grad_norm": 1.025986909866333,
      "learning_rate": 2.2705984388551606e-05,
      "loss": 0.5671,
      "step": 1309
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 1.1107748746871948,
      "learning_rate": 2.272333044232437e-05,
      "loss": 0.7192,
      "step": 1310
    },
    {
      "epoch": 0.22744621790423317,
      "grad_norm": 1.1183154582977295,
      "learning_rate": 2.274067649609714e-05,
      "loss": 0.741,
      "step": 1311
    },
    {
      "epoch": 0.22761970853573907,
      "grad_norm": 0.8567702174186707,
      "learning_rate": 2.2758022549869905e-05,
      "loss": 0.769,
      "step": 1312
    },
    {
      "epoch": 0.22779319916724497,
      "grad_norm": 0.747605562210083,
      "learning_rate": 2.2775368603642673e-05,
      "loss": 0.6871,
      "step": 1313
    },
    {
      "epoch": 0.22796668979875087,
      "grad_norm": 1.1219127178192139,
      "learning_rate": 2.2792714657415438e-05,
      "loss": 0.6958,
      "step": 1314
    },
    {
      "epoch": 0.22814018043025677,
      "grad_norm": 0.9096964597702026,
      "learning_rate": 2.2810060711188203e-05,
      "loss": 0.5948,
      "step": 1315
    },
    {
      "epoch": 0.22831367106176267,
      "grad_norm": 0.7826958298683167,
      "learning_rate": 2.2827406764960975e-05,
      "loss": 0.6185,
      "step": 1316
    },
    {
      "epoch": 0.22848716169326858,
      "grad_norm": 1.2576274871826172,
      "learning_rate": 2.2844752818733743e-05,
      "loss": 0.6299,
      "step": 1317
    },
    {
      "epoch": 0.22866065232477445,
      "grad_norm": 1.082729458808899,
      "learning_rate": 2.2862098872506508e-05,
      "loss": 0.6459,
      "step": 1318
    },
    {
      "epoch": 0.22883414295628035,
      "grad_norm": 0.7407817840576172,
      "learning_rate": 2.2879444926279276e-05,
      "loss": 0.6615,
      "step": 1319
    },
    {
      "epoch": 0.22900763358778625,
      "grad_norm": 0.8147691488265991,
      "learning_rate": 2.289679098005204e-05,
      "loss": 0.671,
      "step": 1320
    },
    {
      "epoch": 0.22918112421929215,
      "grad_norm": 0.7441374063491821,
      "learning_rate": 2.2914137033824806e-05,
      "loss": 0.6202,
      "step": 1321
    },
    {
      "epoch": 0.22935461485079806,
      "grad_norm": 2.1673049926757812,
      "learning_rate": 2.2931483087597575e-05,
      "loss": 0.6427,
      "step": 1322
    },
    {
      "epoch": 0.22952810548230396,
      "grad_norm": 0.8096586465835571,
      "learning_rate": 2.294882914137034e-05,
      "loss": 0.5872,
      "step": 1323
    },
    {
      "epoch": 0.22970159611380986,
      "grad_norm": 0.8028786778450012,
      "learning_rate": 2.2966175195143108e-05,
      "loss": 0.6263,
      "step": 1324
    },
    {
      "epoch": 0.22987508674531576,
      "grad_norm": 0.6502306461334229,
      "learning_rate": 2.2983521248915873e-05,
      "loss": 0.6902,
      "step": 1325
    },
    {
      "epoch": 0.23004857737682166,
      "grad_norm": 0.8782081007957458,
      "learning_rate": 2.300086730268864e-05,
      "loss": 0.6886,
      "step": 1326
    },
    {
      "epoch": 0.23022206800832756,
      "grad_norm": 1.0046528577804565,
      "learning_rate": 2.3018213356461407e-05,
      "loss": 0.6178,
      "step": 1327
    },
    {
      "epoch": 0.23039555863983344,
      "grad_norm": 1.015367031097412,
      "learning_rate": 2.3035559410234175e-05,
      "loss": 0.559,
      "step": 1328
    },
    {
      "epoch": 0.23056904927133934,
      "grad_norm": 0.9212409853935242,
      "learning_rate": 2.305290546400694e-05,
      "loss": 0.6744,
      "step": 1329
    },
    {
      "epoch": 0.23074253990284524,
      "grad_norm": 0.669114351272583,
      "learning_rate": 2.307025151777971e-05,
      "loss": 0.7456,
      "step": 1330
    },
    {
      "epoch": 0.23091603053435114,
      "grad_norm": 0.755664587020874,
      "learning_rate": 2.3087597571552473e-05,
      "loss": 0.7069,
      "step": 1331
    },
    {
      "epoch": 0.23108952116585704,
      "grad_norm": 0.6906589269638062,
      "learning_rate": 2.3104943625325242e-05,
      "loss": 0.7749,
      "step": 1332
    },
    {
      "epoch": 0.23126301179736294,
      "grad_norm": 0.6546650528907776,
      "learning_rate": 2.3122289679098007e-05,
      "loss": 0.7385,
      "step": 1333
    },
    {
      "epoch": 0.23143650242886885,
      "grad_norm": 1.0618844032287598,
      "learning_rate": 2.3139635732870772e-05,
      "loss": 0.7479,
      "step": 1334
    },
    {
      "epoch": 0.23160999306037475,
      "grad_norm": 0.8666930198669434,
      "learning_rate": 2.315698178664354e-05,
      "loss": 0.6582,
      "step": 1335
    },
    {
      "epoch": 0.23178348369188065,
      "grad_norm": 0.6178314685821533,
      "learning_rate": 2.3174327840416305e-05,
      "loss": 0.6498,
      "step": 1336
    },
    {
      "epoch": 0.23195697432338655,
      "grad_norm": 0.9197214245796204,
      "learning_rate": 2.3191673894189074e-05,
      "loss": 0.5626,
      "step": 1337
    },
    {
      "epoch": 0.23213046495489242,
      "grad_norm": 1.4937176704406738,
      "learning_rate": 2.320901994796184e-05,
      "loss": 0.7056,
      "step": 1338
    },
    {
      "epoch": 0.23230395558639833,
      "grad_norm": 0.8406773209571838,
      "learning_rate": 2.3226366001734607e-05,
      "loss": 0.7026,
      "step": 1339
    },
    {
      "epoch": 0.23247744621790423,
      "grad_norm": 0.6967365145683289,
      "learning_rate": 2.3243712055507372e-05,
      "loss": 0.6952,
      "step": 1340
    },
    {
      "epoch": 0.23265093684941013,
      "grad_norm": 1.1950923204421997,
      "learning_rate": 2.326105810928014e-05,
      "loss": 0.7312,
      "step": 1341
    },
    {
      "epoch": 0.23282442748091603,
      "grad_norm": 0.819696843624115,
      "learning_rate": 2.3278404163052905e-05,
      "loss": 0.5697,
      "step": 1342
    },
    {
      "epoch": 0.23299791811242193,
      "grad_norm": 1.7255115509033203,
      "learning_rate": 2.3295750216825674e-05,
      "loss": 0.5286,
      "step": 1343
    },
    {
      "epoch": 0.23317140874392783,
      "grad_norm": 1.2118375301361084,
      "learning_rate": 2.331309627059844e-05,
      "loss": 0.7142,
      "step": 1344
    },
    {
      "epoch": 0.23334489937543373,
      "grad_norm": 1.0448431968688965,
      "learning_rate": 2.3330442324371204e-05,
      "loss": 0.6359,
      "step": 1345
    },
    {
      "epoch": 0.23351839000693964,
      "grad_norm": 1.05678129196167,
      "learning_rate": 2.3347788378143976e-05,
      "loss": 0.5945,
      "step": 1346
    },
    {
      "epoch": 0.23369188063844554,
      "grad_norm": 0.7752077579498291,
      "learning_rate": 2.3365134431916744e-05,
      "loss": 0.7537,
      "step": 1347
    },
    {
      "epoch": 0.2338653712699514,
      "grad_norm": 0.6281952261924744,
      "learning_rate": 2.338248048568951e-05,
      "loss": 0.627,
      "step": 1348
    },
    {
      "epoch": 0.2340388619014573,
      "grad_norm": 0.9324553608894348,
      "learning_rate": 2.3399826539462277e-05,
      "loss": 0.6323,
      "step": 1349
    },
    {
      "epoch": 0.23421235253296321,
      "grad_norm": 0.7048712372779846,
      "learning_rate": 2.3417172593235042e-05,
      "loss": 0.7174,
      "step": 1350
    },
    {
      "epoch": 0.23438584316446912,
      "grad_norm": 1.0396593809127808,
      "learning_rate": 2.3434518647007807e-05,
      "loss": 0.5864,
      "step": 1351
    },
    {
      "epoch": 0.23455933379597502,
      "grad_norm": 0.9440403580665588,
      "learning_rate": 2.3451864700780576e-05,
      "loss": 0.6034,
      "step": 1352
    },
    {
      "epoch": 0.23473282442748092,
      "grad_norm": 0.8695996403694153,
      "learning_rate": 2.346921075455334e-05,
      "loss": 0.587,
      "step": 1353
    },
    {
      "epoch": 0.23490631505898682,
      "grad_norm": 0.8083711862564087,
      "learning_rate": 2.348655680832611e-05,
      "loss": 0.7283,
      "step": 1354
    },
    {
      "epoch": 0.23507980569049272,
      "grad_norm": 0.8822265863418579,
      "learning_rate": 2.3503902862098874e-05,
      "loss": 0.6411,
      "step": 1355
    },
    {
      "epoch": 0.23525329632199862,
      "grad_norm": 0.7099165320396423,
      "learning_rate": 2.3521248915871643e-05,
      "loss": 0.6519,
      "step": 1356
    },
    {
      "epoch": 0.23542678695350452,
      "grad_norm": 0.6208922863006592,
      "learning_rate": 2.3538594969644408e-05,
      "loss": 0.6401,
      "step": 1357
    },
    {
      "epoch": 0.2356002775850104,
      "grad_norm": 0.8681457042694092,
      "learning_rate": 2.3555941023417176e-05,
      "loss": 0.674,
      "step": 1358
    },
    {
      "epoch": 0.2357737682165163,
      "grad_norm": 0.9446276426315308,
      "learning_rate": 2.357328707718994e-05,
      "loss": 0.6056,
      "step": 1359
    },
    {
      "epoch": 0.2359472588480222,
      "grad_norm": 3.0077075958251953,
      "learning_rate": 2.359063313096271e-05,
      "loss": 0.5176,
      "step": 1360
    },
    {
      "epoch": 0.2361207494795281,
      "grad_norm": 1.011677861213684,
      "learning_rate": 2.3607979184735474e-05,
      "loss": 0.5721,
      "step": 1361
    },
    {
      "epoch": 0.236294240111034,
      "grad_norm": 1.2111024856567383,
      "learning_rate": 2.362532523850824e-05,
      "loss": 0.6101,
      "step": 1362
    },
    {
      "epoch": 0.2364677307425399,
      "grad_norm": 0.6885407567024231,
      "learning_rate": 2.3642671292281008e-05,
      "loss": 0.7671,
      "step": 1363
    },
    {
      "epoch": 0.2366412213740458,
      "grad_norm": 0.8285648226737976,
      "learning_rate": 2.3660017346053773e-05,
      "loss": 0.7073,
      "step": 1364
    },
    {
      "epoch": 0.2368147120055517,
      "grad_norm": 0.75030916929245,
      "learning_rate": 2.367736339982654e-05,
      "loss": 0.6289,
      "step": 1365
    },
    {
      "epoch": 0.2369882026370576,
      "grad_norm": 0.5976499915122986,
      "learning_rate": 2.3694709453599306e-05,
      "loss": 0.8052,
      "step": 1366
    },
    {
      "epoch": 0.23716169326856348,
      "grad_norm": 0.5793814063072205,
      "learning_rate": 2.3712055507372074e-05,
      "loss": 0.7625,
      "step": 1367
    },
    {
      "epoch": 0.23733518390006939,
      "grad_norm": 0.7182748317718506,
      "learning_rate": 2.372940156114484e-05,
      "loss": 0.6562,
      "step": 1368
    },
    {
      "epoch": 0.2375086745315753,
      "grad_norm": 0.8480749726295471,
      "learning_rate": 2.3746747614917608e-05,
      "loss": 0.6787,
      "step": 1369
    },
    {
      "epoch": 0.2376821651630812,
      "grad_norm": 1.211319088935852,
      "learning_rate": 2.3764093668690373e-05,
      "loss": 0.6803,
      "step": 1370
    },
    {
      "epoch": 0.2378556557945871,
      "grad_norm": 0.9534586668014526,
      "learning_rate": 2.378143972246314e-05,
      "loss": 0.7072,
      "step": 1371
    },
    {
      "epoch": 0.238029146426093,
      "grad_norm": 1.0700056552886963,
      "learning_rate": 2.3798785776235906e-05,
      "loss": 0.7094,
      "step": 1372
    },
    {
      "epoch": 0.2382026370575989,
      "grad_norm": 0.8894718885421753,
      "learning_rate": 2.3816131830008675e-05,
      "loss": 0.5952,
      "step": 1373
    },
    {
      "epoch": 0.2383761276891048,
      "grad_norm": 0.8361707925796509,
      "learning_rate": 2.383347788378144e-05,
      "loss": 0.6184,
      "step": 1374
    },
    {
      "epoch": 0.2385496183206107,
      "grad_norm": 0.9295462369918823,
      "learning_rate": 2.385082393755421e-05,
      "loss": 0.6335,
      "step": 1375
    },
    {
      "epoch": 0.2387231089521166,
      "grad_norm": 0.7952740788459778,
      "learning_rate": 2.3868169991326976e-05,
      "loss": 0.6921,
      "step": 1376
    },
    {
      "epoch": 0.23889659958362247,
      "grad_norm": 0.7641652226448059,
      "learning_rate": 2.3885516045099745e-05,
      "loss": 0.6609,
      "step": 1377
    },
    {
      "epoch": 0.23907009021512837,
      "grad_norm": 1.317805290222168,
      "learning_rate": 2.390286209887251e-05,
      "loss": 0.6741,
      "step": 1378
    },
    {
      "epoch": 0.23924358084663427,
      "grad_norm": 0.9185047149658203,
      "learning_rate": 2.3920208152645278e-05,
      "loss": 0.5428,
      "step": 1379
    },
    {
      "epoch": 0.23941707147814018,
      "grad_norm": 0.9052122235298157,
      "learning_rate": 2.3937554206418043e-05,
      "loss": 0.6556,
      "step": 1380
    },
    {
      "epoch": 0.23959056210964608,
      "grad_norm": 0.8003539443016052,
      "learning_rate": 2.3954900260190808e-05,
      "loss": 0.7117,
      "step": 1381
    },
    {
      "epoch": 0.23976405274115198,
      "grad_norm": 1.0138602256774902,
      "learning_rate": 2.3972246313963577e-05,
      "loss": 0.6655,
      "step": 1382
    },
    {
      "epoch": 0.23993754337265788,
      "grad_norm": 0.8630676865577698,
      "learning_rate": 2.398959236773634e-05,
      "loss": 0.7196,
      "step": 1383
    },
    {
      "epoch": 0.24011103400416378,
      "grad_norm": 0.9065120220184326,
      "learning_rate": 2.400693842150911e-05,
      "loss": 0.8337,
      "step": 1384
    },
    {
      "epoch": 0.24028452463566968,
      "grad_norm": 0.9683220982551575,
      "learning_rate": 2.4024284475281875e-05,
      "loss": 0.6088,
      "step": 1385
    },
    {
      "epoch": 0.24045801526717558,
      "grad_norm": 1.0982496738433838,
      "learning_rate": 2.4041630529054643e-05,
      "loss": 0.5911,
      "step": 1386
    },
    {
      "epoch": 0.24063150589868146,
      "grad_norm": 1.0864150524139404,
      "learning_rate": 2.405897658282741e-05,
      "loss": 0.6309,
      "step": 1387
    },
    {
      "epoch": 0.24080499653018736,
      "grad_norm": 1.4859693050384521,
      "learning_rate": 2.4076322636600177e-05,
      "loss": 0.6058,
      "step": 1388
    },
    {
      "epoch": 0.24097848716169326,
      "grad_norm": 0.7649808526039124,
      "learning_rate": 2.4093668690372942e-05,
      "loss": 0.6556,
      "step": 1389
    },
    {
      "epoch": 0.24115197779319916,
      "grad_norm": 0.8709794282913208,
      "learning_rate": 2.411101474414571e-05,
      "loss": 0.6486,
      "step": 1390
    },
    {
      "epoch": 0.24132546842470506,
      "grad_norm": 0.8978895545005798,
      "learning_rate": 2.4128360797918475e-05,
      "loss": 0.5717,
      "step": 1391
    },
    {
      "epoch": 0.24149895905621097,
      "grad_norm": 0.870811939239502,
      "learning_rate": 2.414570685169124e-05,
      "loss": 0.7159,
      "step": 1392
    },
    {
      "epoch": 0.24167244968771687,
      "grad_norm": 0.8962591290473938,
      "learning_rate": 2.416305290546401e-05,
      "loss": 0.6539,
      "step": 1393
    },
    {
      "epoch": 0.24184594031922277,
      "grad_norm": 0.8146924376487732,
      "learning_rate": 2.4180398959236774e-05,
      "loss": 0.7087,
      "step": 1394
    },
    {
      "epoch": 0.24201943095072867,
      "grad_norm": 1.1942592859268188,
      "learning_rate": 2.4197745013009542e-05,
      "loss": 0.5968,
      "step": 1395
    },
    {
      "epoch": 0.24219292158223457,
      "grad_norm": 0.867985188961029,
      "learning_rate": 2.4215091066782307e-05,
      "loss": 0.6763,
      "step": 1396
    },
    {
      "epoch": 0.24236641221374045,
      "grad_norm": 0.7588414549827576,
      "learning_rate": 2.4232437120555075e-05,
      "loss": 0.5974,
      "step": 1397
    },
    {
      "epoch": 0.24253990284524635,
      "grad_norm": 1.1507147550582886,
      "learning_rate": 2.424978317432784e-05,
      "loss": 0.7734,
      "step": 1398
    },
    {
      "epoch": 0.24271339347675225,
      "grad_norm": 0.8985394835472107,
      "learning_rate": 2.426712922810061e-05,
      "loss": 0.6003,
      "step": 1399
    },
    {
      "epoch": 0.24288688410825815,
      "grad_norm": 0.8671207427978516,
      "learning_rate": 2.4284475281873374e-05,
      "loss": 0.6472,
      "step": 1400
    },
    {
      "epoch": 0.24306037473976405,
      "grad_norm": 0.7705470323562622,
      "learning_rate": 2.4301821335646142e-05,
      "loss": 0.7676,
      "step": 1401
    },
    {
      "epoch": 0.24323386537126995,
      "grad_norm": 1.5794962644577026,
      "learning_rate": 2.4319167389418907e-05,
      "loss": 0.7045,
      "step": 1402
    },
    {
      "epoch": 0.24340735600277585,
      "grad_norm": 0.8116719126701355,
      "learning_rate": 2.4336513443191676e-05,
      "loss": 0.6333,
      "step": 1403
    },
    {
      "epoch": 0.24358084663428176,
      "grad_norm": 0.8300854563713074,
      "learning_rate": 2.435385949696444e-05,
      "loss": 0.6934,
      "step": 1404
    },
    {
      "epoch": 0.24375433726578766,
      "grad_norm": 0.7705616354942322,
      "learning_rate": 2.4371205550737212e-05,
      "loss": 0.6648,
      "step": 1405
    },
    {
      "epoch": 0.24392782789729356,
      "grad_norm": 0.9937134385108948,
      "learning_rate": 2.4388551604509977e-05,
      "loss": 0.6571,
      "step": 1406
    },
    {
      "epoch": 0.24410131852879943,
      "grad_norm": 1.039591908454895,
      "learning_rate": 2.4405897658282746e-05,
      "loss": 0.6,
      "step": 1407
    },
    {
      "epoch": 0.24427480916030533,
      "grad_norm": 1.7142187356948853,
      "learning_rate": 2.442324371205551e-05,
      "loss": 0.7238,
      "step": 1408
    },
    {
      "epoch": 0.24444829979181124,
      "grad_norm": 1.2775105237960815,
      "learning_rate": 2.444058976582828e-05,
      "loss": 0.6331,
      "step": 1409
    },
    {
      "epoch": 0.24462179042331714,
      "grad_norm": 2.677325487136841,
      "learning_rate": 2.4457935819601044e-05,
      "loss": 0.6667,
      "step": 1410
    },
    {
      "epoch": 0.24479528105482304,
      "grad_norm": 0.825202226638794,
      "learning_rate": 2.447528187337381e-05,
      "loss": 0.7372,
      "step": 1411
    },
    {
      "epoch": 0.24496877168632894,
      "grad_norm": 1.1789332628250122,
      "learning_rate": 2.4492627927146577e-05,
      "loss": 0.5078,
      "step": 1412
    },
    {
      "epoch": 0.24514226231783484,
      "grad_norm": 0.6857295036315918,
      "learning_rate": 2.4509973980919342e-05,
      "loss": 0.5812,
      "step": 1413
    },
    {
      "epoch": 0.24531575294934074,
      "grad_norm": 0.7828952074050903,
      "learning_rate": 2.452732003469211e-05,
      "loss": 0.7065,
      "step": 1414
    },
    {
      "epoch": 0.24548924358084664,
      "grad_norm": 0.6869457364082336,
      "learning_rate": 2.4544666088464876e-05,
      "loss": 0.8044,
      "step": 1415
    },
    {
      "epoch": 0.24566273421235255,
      "grad_norm": 0.7021481990814209,
      "learning_rate": 2.4562012142237644e-05,
      "loss": 0.6599,
      "step": 1416
    },
    {
      "epoch": 0.24583622484385842,
      "grad_norm": 0.9084320068359375,
      "learning_rate": 2.457935819601041e-05,
      "loss": 0.6476,
      "step": 1417
    },
    {
      "epoch": 0.24600971547536432,
      "grad_norm": 1.0445537567138672,
      "learning_rate": 2.4596704249783178e-05,
      "loss": 0.7562,
      "step": 1418
    },
    {
      "epoch": 0.24618320610687022,
      "grad_norm": 0.7493770122528076,
      "learning_rate": 2.4614050303555943e-05,
      "loss": 0.631,
      "step": 1419
    },
    {
      "epoch": 0.24635669673837612,
      "grad_norm": 0.9136198163032532,
      "learning_rate": 2.463139635732871e-05,
      "loss": 0.5819,
      "step": 1420
    },
    {
      "epoch": 0.24653018736988203,
      "grad_norm": 1.6267791986465454,
      "learning_rate": 2.4648742411101476e-05,
      "loss": 0.5837,
      "step": 1421
    },
    {
      "epoch": 0.24670367800138793,
      "grad_norm": 1.2293862104415894,
      "learning_rate": 2.466608846487424e-05,
      "loss": 0.7241,
      "step": 1422
    },
    {
      "epoch": 0.24687716863289383,
      "grad_norm": 0.7609236836433411,
      "learning_rate": 2.468343451864701e-05,
      "loss": 0.6497,
      "step": 1423
    },
    {
      "epoch": 0.24705065926439973,
      "grad_norm": 0.693767786026001,
      "learning_rate": 2.4700780572419774e-05,
      "loss": 0.6656,
      "step": 1424
    },
    {
      "epoch": 0.24722414989590563,
      "grad_norm": 0.6940293908119202,
      "learning_rate": 2.4718126626192543e-05,
      "loss": 0.7141,
      "step": 1425
    },
    {
      "epoch": 0.24739764052741153,
      "grad_norm": 0.6686576008796692,
      "learning_rate": 2.4735472679965308e-05,
      "loss": 0.6167,
      "step": 1426
    },
    {
      "epoch": 0.2475711311589174,
      "grad_norm": 0.7655903697013855,
      "learning_rate": 2.4752818733738076e-05,
      "loss": 0.6821,
      "step": 1427
    },
    {
      "epoch": 0.2477446217904233,
      "grad_norm": 1.346714735031128,
      "learning_rate": 2.477016478751084e-05,
      "loss": 0.5706,
      "step": 1428
    },
    {
      "epoch": 0.2479181124219292,
      "grad_norm": 1.171101450920105,
      "learning_rate": 2.478751084128361e-05,
      "loss": 0.4973,
      "step": 1429
    },
    {
      "epoch": 0.2480916030534351,
      "grad_norm": 0.6083464622497559,
      "learning_rate": 2.4804856895056375e-05,
      "loss": 0.7275,
      "step": 1430
    },
    {
      "epoch": 0.248265093684941,
      "grad_norm": 1.08219575881958,
      "learning_rate": 2.4822202948829143e-05,
      "loss": 0.7886,
      "step": 1431
    },
    {
      "epoch": 0.24843858431644691,
      "grad_norm": 0.7164523601531982,
      "learning_rate": 2.4839549002601908e-05,
      "loss": 0.7415,
      "step": 1432
    },
    {
      "epoch": 0.24861207494795282,
      "grad_norm": 0.8163508772850037,
      "learning_rate": 2.4856895056374676e-05,
      "loss": 0.7021,
      "step": 1433
    },
    {
      "epoch": 0.24878556557945872,
      "grad_norm": 0.6706337928771973,
      "learning_rate": 2.4874241110147445e-05,
      "loss": 0.6196,
      "step": 1434
    },
    {
      "epoch": 0.24895905621096462,
      "grad_norm": 1.0622358322143555,
      "learning_rate": 2.4891587163920213e-05,
      "loss": 0.5898,
      "step": 1435
    },
    {
      "epoch": 0.24913254684247052,
      "grad_norm": 1.08125901222229,
      "learning_rate": 2.4908933217692978e-05,
      "loss": 0.6,
      "step": 1436
    },
    {
      "epoch": 0.2493060374739764,
      "grad_norm": 0.7177915573120117,
      "learning_rate": 2.4926279271465747e-05,
      "loss": 0.6243,
      "step": 1437
    },
    {
      "epoch": 0.2494795281054823,
      "grad_norm": 0.9025813937187195,
      "learning_rate": 2.494362532523851e-05,
      "loss": 0.5082,
      "step": 1438
    },
    {
      "epoch": 0.2496530187369882,
      "grad_norm": 2.0147836208343506,
      "learning_rate": 2.496097137901128e-05,
      "loss": 0.6139,
      "step": 1439
    },
    {
      "epoch": 0.2498265093684941,
      "grad_norm": 0.5875595211982727,
      "learning_rate": 2.4978317432784045e-05,
      "loss": 0.7036,
      "step": 1440
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0961546897888184,
      "learning_rate": 2.499566348655681e-05,
      "loss": 0.5535,
      "step": 1441
    },
    {
      "epoch": 0.2501734906315059,
      "grad_norm": 0.606073796749115,
      "learning_rate": 2.501300954032958e-05,
      "loss": 0.6997,
      "step": 1442
    },
    {
      "epoch": 0.2503469812630118,
      "grad_norm": 0.9950074553489685,
      "learning_rate": 2.5030355594102343e-05,
      "loss": 0.606,
      "step": 1443
    },
    {
      "epoch": 0.2505204718945177,
      "grad_norm": 1.1742510795593262,
      "learning_rate": 2.5047701647875112e-05,
      "loss": 0.616,
      "step": 1444
    },
    {
      "epoch": 0.2506939625260236,
      "grad_norm": 0.967836856842041,
      "learning_rate": 2.5065047701647877e-05,
      "loss": 0.7161,
      "step": 1445
    },
    {
      "epoch": 0.2508674531575295,
      "grad_norm": 1.021001935005188,
      "learning_rate": 2.5082393755420645e-05,
      "loss": 0.6312,
      "step": 1446
    },
    {
      "epoch": 0.2510409437890354,
      "grad_norm": 1.0591120719909668,
      "learning_rate": 2.509973980919341e-05,
      "loss": 0.594,
      "step": 1447
    },
    {
      "epoch": 0.2512144344205413,
      "grad_norm": 0.6812381744384766,
      "learning_rate": 2.511708586296618e-05,
      "loss": 0.6257,
      "step": 1448
    },
    {
      "epoch": 0.2513879250520472,
      "grad_norm": 1.3215664625167847,
      "learning_rate": 2.5134431916738944e-05,
      "loss": 0.6381,
      "step": 1449
    },
    {
      "epoch": 0.2515614156835531,
      "grad_norm": 0.818316638469696,
      "learning_rate": 2.5151777970511712e-05,
      "loss": 0.6444,
      "step": 1450
    },
    {
      "epoch": 0.25173490631505896,
      "grad_norm": 0.6277799010276794,
      "learning_rate": 2.5169124024284477e-05,
      "loss": 0.6661,
      "step": 1451
    },
    {
      "epoch": 0.25190839694656486,
      "grad_norm": 1.578797459602356,
      "learning_rate": 2.5186470078057242e-05,
      "loss": 0.8,
      "step": 1452
    },
    {
      "epoch": 0.25208188757807076,
      "grad_norm": 1.7333303689956665,
      "learning_rate": 2.520381613183001e-05,
      "loss": 0.8259,
      "step": 1453
    },
    {
      "epoch": 0.25225537820957666,
      "grad_norm": 1.0634512901306152,
      "learning_rate": 2.5221162185602775e-05,
      "loss": 0.7368,
      "step": 1454
    },
    {
      "epoch": 0.25242886884108257,
      "grad_norm": 0.7911814451217651,
      "learning_rate": 2.5238508239375544e-05,
      "loss": 0.5845,
      "step": 1455
    },
    {
      "epoch": 0.25260235947258847,
      "grad_norm": 1.1727575063705444,
      "learning_rate": 2.525585429314831e-05,
      "loss": 0.5659,
      "step": 1456
    },
    {
      "epoch": 0.25277585010409437,
      "grad_norm": 1.3862674236297607,
      "learning_rate": 2.5273200346921077e-05,
      "loss": 0.8508,
      "step": 1457
    },
    {
      "epoch": 0.25294934073560027,
      "grad_norm": 2.0431294441223145,
      "learning_rate": 2.5290546400693842e-05,
      "loss": 0.7278,
      "step": 1458
    },
    {
      "epoch": 0.25312283136710617,
      "grad_norm": 0.8106757998466492,
      "learning_rate": 2.530789245446661e-05,
      "loss": 0.5602,
      "step": 1459
    },
    {
      "epoch": 0.2532963219986121,
      "grad_norm": 0.6260269284248352,
      "learning_rate": 2.5325238508239375e-05,
      "loss": 0.7405,
      "step": 1460
    },
    {
      "epoch": 0.253469812630118,
      "grad_norm": 0.846102237701416,
      "learning_rate": 2.5342584562012144e-05,
      "loss": 0.6794,
      "step": 1461
    },
    {
      "epoch": 0.2536433032616239,
      "grad_norm": 0.7476555705070496,
      "learning_rate": 2.535993061578491e-05,
      "loss": 0.6829,
      "step": 1462
    },
    {
      "epoch": 0.2538167938931298,
      "grad_norm": 1.4022223949432373,
      "learning_rate": 2.5377276669557674e-05,
      "loss": 0.5559,
      "step": 1463
    },
    {
      "epoch": 0.2539902845246357,
      "grad_norm": 0.8950586915016174,
      "learning_rate": 2.5394622723330446e-05,
      "loss": 0.6235,
      "step": 1464
    },
    {
      "epoch": 0.2541637751561416,
      "grad_norm": 0.9371039271354675,
      "learning_rate": 2.5411968777103214e-05,
      "loss": 0.5811,
      "step": 1465
    },
    {
      "epoch": 0.2543372657876475,
      "grad_norm": 0.819125771522522,
      "learning_rate": 2.542931483087598e-05,
      "loss": 0.6378,
      "step": 1466
    },
    {
      "epoch": 0.2545107564191534,
      "grad_norm": 0.9547324180603027,
      "learning_rate": 2.5446660884648747e-05,
      "loss": 0.6903,
      "step": 1467
    },
    {
      "epoch": 0.2546842470506593,
      "grad_norm": 0.8518248796463013,
      "learning_rate": 2.5464006938421512e-05,
      "loss": 0.584,
      "step": 1468
    },
    {
      "epoch": 0.2548577376821652,
      "grad_norm": 0.8062921762466431,
      "learning_rate": 2.548135299219428e-05,
      "loss": 0.6942,
      "step": 1469
    },
    {
      "epoch": 0.2550312283136711,
      "grad_norm": 1.0598679780960083,
      "learning_rate": 2.5498699045967046e-05,
      "loss": 0.5164,
      "step": 1470
    },
    {
      "epoch": 0.25520471894517693,
      "grad_norm": 1.1887106895446777,
      "learning_rate": 2.551604509973981e-05,
      "loss": 0.6323,
      "step": 1471
    },
    {
      "epoch": 0.25537820957668284,
      "grad_norm": 1.7033594846725464,
      "learning_rate": 2.553339115351258e-05,
      "loss": 0.6974,
      "step": 1472
    },
    {
      "epoch": 0.25555170020818874,
      "grad_norm": 0.6564008593559265,
      "learning_rate": 2.5550737207285344e-05,
      "loss": 0.6829,
      "step": 1473
    },
    {
      "epoch": 0.25572519083969464,
      "grad_norm": 1.0994482040405273,
      "learning_rate": 2.5568083261058113e-05,
      "loss": 0.5802,
      "step": 1474
    },
    {
      "epoch": 0.25589868147120054,
      "grad_norm": 0.7509127259254456,
      "learning_rate": 2.5585429314830878e-05,
      "loss": 0.5813,
      "step": 1475
    },
    {
      "epoch": 0.25607217210270644,
      "grad_norm": 1.0179251432418823,
      "learning_rate": 2.5602775368603646e-05,
      "loss": 0.6151,
      "step": 1476
    },
    {
      "epoch": 0.25624566273421234,
      "grad_norm": 0.833226203918457,
      "learning_rate": 2.562012142237641e-05,
      "loss": 0.6606,
      "step": 1477
    },
    {
      "epoch": 0.25641915336571824,
      "grad_norm": 0.8429149985313416,
      "learning_rate": 2.563746747614918e-05,
      "loss": 0.5847,
      "step": 1478
    },
    {
      "epoch": 0.25659264399722415,
      "grad_norm": 0.82933509349823,
      "learning_rate": 2.5654813529921944e-05,
      "loss": 0.5924,
      "step": 1479
    },
    {
      "epoch": 0.25676613462873005,
      "grad_norm": 0.7633499503135681,
      "learning_rate": 2.5672159583694713e-05,
      "loss": 0.606,
      "step": 1480
    },
    {
      "epoch": 0.25693962526023595,
      "grad_norm": 0.8920152187347412,
      "learning_rate": 2.5689505637467478e-05,
      "loss": 0.6642,
      "step": 1481
    },
    {
      "epoch": 0.25711311589174185,
      "grad_norm": 1.372825026512146,
      "learning_rate": 2.5706851691240243e-05,
      "loss": 0.7399,
      "step": 1482
    },
    {
      "epoch": 0.25728660652324775,
      "grad_norm": 0.8788719177246094,
      "learning_rate": 2.572419774501301e-05,
      "loss": 0.6349,
      "step": 1483
    },
    {
      "epoch": 0.25746009715475365,
      "grad_norm": 0.9693309664726257,
      "learning_rate": 2.5741543798785776e-05,
      "loss": 0.6376,
      "step": 1484
    },
    {
      "epoch": 0.25763358778625955,
      "grad_norm": 0.7878218293190002,
      "learning_rate": 2.5758889852558545e-05,
      "loss": 0.5402,
      "step": 1485
    },
    {
      "epoch": 0.25780707841776546,
      "grad_norm": 0.8361150026321411,
      "learning_rate": 2.577623590633131e-05,
      "loss": 0.598,
      "step": 1486
    },
    {
      "epoch": 0.25798056904927136,
      "grad_norm": 0.7228373289108276,
      "learning_rate": 2.5793581960104078e-05,
      "loss": 0.7197,
      "step": 1487
    },
    {
      "epoch": 0.25815405968077726,
      "grad_norm": 4.787299156188965,
      "learning_rate": 2.5810928013876843e-05,
      "loss": 0.6613,
      "step": 1488
    },
    {
      "epoch": 0.25832755031228316,
      "grad_norm": 0.8377718329429626,
      "learning_rate": 2.582827406764961e-05,
      "loss": 0.712,
      "step": 1489
    },
    {
      "epoch": 0.25850104094378906,
      "grad_norm": 1.3003144264221191,
      "learning_rate": 2.5845620121422376e-05,
      "loss": 0.6018,
      "step": 1490
    },
    {
      "epoch": 0.2586745315752949,
      "grad_norm": 0.7731533646583557,
      "learning_rate": 2.5862966175195145e-05,
      "loss": 0.6294,
      "step": 1491
    },
    {
      "epoch": 0.2588480222068008,
      "grad_norm": 0.8601460456848145,
      "learning_rate": 2.588031222896791e-05,
      "loss": 0.6941,
      "step": 1492
    },
    {
      "epoch": 0.2590215128383067,
      "grad_norm": 0.8807269334793091,
      "learning_rate": 2.589765828274068e-05,
      "loss": 0.5585,
      "step": 1493
    },
    {
      "epoch": 0.2591950034698126,
      "grad_norm": 0.9132180213928223,
      "learning_rate": 2.5915004336513447e-05,
      "loss": 0.5447,
      "step": 1494
    },
    {
      "epoch": 0.2593684941013185,
      "grad_norm": 1.032853364944458,
      "learning_rate": 2.5932350390286215e-05,
      "loss": 0.6312,
      "step": 1495
    },
    {
      "epoch": 0.2595419847328244,
      "grad_norm": 0.8208354711532593,
      "learning_rate": 2.594969644405898e-05,
      "loss": 0.5564,
      "step": 1496
    },
    {
      "epoch": 0.2597154753643303,
      "grad_norm": 0.7401357889175415,
      "learning_rate": 2.5967042497831748e-05,
      "loss": 0.6831,
      "step": 1497
    },
    {
      "epoch": 0.2598889659958362,
      "grad_norm": 0.6182374358177185,
      "learning_rate": 2.5984388551604513e-05,
      "loss": 0.7275,
      "step": 1498
    },
    {
      "epoch": 0.2600624566273421,
      "grad_norm": 0.8194847702980042,
      "learning_rate": 2.600173460537728e-05,
      "loss": 0.5455,
      "step": 1499
    },
    {
      "epoch": 0.260235947258848,
      "grad_norm": 0.8735288381576538,
      "learning_rate": 2.6019080659150047e-05,
      "loss": 0.5653,
      "step": 1500
    },
    {
      "epoch": 0.2604094378903539,
      "grad_norm": 0.9586334824562073,
      "learning_rate": 2.603642671292281e-05,
      "loss": 0.6289,
      "step": 1501
    },
    {
      "epoch": 0.2605829285218598,
      "grad_norm": 2.6775732040405273,
      "learning_rate": 2.605377276669558e-05,
      "loss": 0.5901,
      "step": 1502
    },
    {
      "epoch": 0.2607564191533657,
      "grad_norm": 0.7645121216773987,
      "learning_rate": 2.6071118820468345e-05,
      "loss": 0.7407,
      "step": 1503
    },
    {
      "epoch": 0.2609299097848716,
      "grad_norm": 0.8603814840316772,
      "learning_rate": 2.6088464874241113e-05,
      "loss": 0.6304,
      "step": 1504
    },
    {
      "epoch": 0.26110340041637753,
      "grad_norm": 0.8805418014526367,
      "learning_rate": 2.610581092801388e-05,
      "loss": 0.6538,
      "step": 1505
    },
    {
      "epoch": 0.26127689104788343,
      "grad_norm": 0.7971379160881042,
      "learning_rate": 2.6123156981786647e-05,
      "loss": 0.6194,
      "step": 1506
    },
    {
      "epoch": 0.26145038167938933,
      "grad_norm": 1.210202932357788,
      "learning_rate": 2.6140503035559412e-05,
      "loss": 0.6835,
      "step": 1507
    },
    {
      "epoch": 0.26162387231089523,
      "grad_norm": 0.8352237939834595,
      "learning_rate": 2.615784908933218e-05,
      "loss": 0.6198,
      "step": 1508
    },
    {
      "epoch": 0.26179736294240114,
      "grad_norm": 1.833086371421814,
      "learning_rate": 2.6175195143104945e-05,
      "loss": 0.7163,
      "step": 1509
    },
    {
      "epoch": 0.26197085357390704,
      "grad_norm": 0.8619792461395264,
      "learning_rate": 2.6192541196877714e-05,
      "loss": 0.6056,
      "step": 1510
    },
    {
      "epoch": 0.2621443442054129,
      "grad_norm": 1.010326862335205,
      "learning_rate": 2.620988725065048e-05,
      "loss": 0.74,
      "step": 1511
    },
    {
      "epoch": 0.2623178348369188,
      "grad_norm": 0.6972298622131348,
      "learning_rate": 2.6227233304423244e-05,
      "loss": 0.7306,
      "step": 1512
    },
    {
      "epoch": 0.2624913254684247,
      "grad_norm": 0.8568160533905029,
      "learning_rate": 2.6244579358196012e-05,
      "loss": 0.6146,
      "step": 1513
    },
    {
      "epoch": 0.2626648160999306,
      "grad_norm": 0.7707713842391968,
      "learning_rate": 2.6261925411968777e-05,
      "loss": 0.6111,
      "step": 1514
    },
    {
      "epoch": 0.2628383067314365,
      "grad_norm": 0.9823936820030212,
      "learning_rate": 2.6279271465741545e-05,
      "loss": 0.6719,
      "step": 1515
    },
    {
      "epoch": 0.2630117973629424,
      "grad_norm": 0.9398236274719238,
      "learning_rate": 2.629661751951431e-05,
      "loss": 0.5513,
      "step": 1516
    },
    {
      "epoch": 0.2631852879944483,
      "grad_norm": 0.8186742067337036,
      "learning_rate": 2.631396357328708e-05,
      "loss": 0.5397,
      "step": 1517
    },
    {
      "epoch": 0.2633587786259542,
      "grad_norm": 0.7485017776489258,
      "learning_rate": 2.6331309627059844e-05,
      "loss": 0.637,
      "step": 1518
    },
    {
      "epoch": 0.2635322692574601,
      "grad_norm": 1.1243284940719604,
      "learning_rate": 2.6348655680832612e-05,
      "loss": 0.5631,
      "step": 1519
    },
    {
      "epoch": 0.263705759888966,
      "grad_norm": 0.7137185335159302,
      "learning_rate": 2.6366001734605377e-05,
      "loss": 0.5881,
      "step": 1520
    },
    {
      "epoch": 0.2638792505204719,
      "grad_norm": 4.920660972595215,
      "learning_rate": 2.6383347788378146e-05,
      "loss": 0.5968,
      "step": 1521
    },
    {
      "epoch": 0.2640527411519778,
      "grad_norm": 0.6850613355636597,
      "learning_rate": 2.640069384215091e-05,
      "loss": 0.6979,
      "step": 1522
    },
    {
      "epoch": 0.2642262317834837,
      "grad_norm": 1.2354705333709717,
      "learning_rate": 2.6418039895923682e-05,
      "loss": 0.5844,
      "step": 1523
    },
    {
      "epoch": 0.2643997224149896,
      "grad_norm": 3.75435471534729,
      "learning_rate": 2.6435385949696447e-05,
      "loss": 0.5521,
      "step": 1524
    },
    {
      "epoch": 0.2645732130464955,
      "grad_norm": 0.9748126268386841,
      "learning_rate": 2.6452732003469216e-05,
      "loss": 0.6317,
      "step": 1525
    },
    {
      "epoch": 0.2647467036780014,
      "grad_norm": 0.8047106862068176,
      "learning_rate": 2.647007805724198e-05,
      "loss": 0.6383,
      "step": 1526
    },
    {
      "epoch": 0.2649201943095073,
      "grad_norm": 0.9678272008895874,
      "learning_rate": 2.648742411101475e-05,
      "loss": 0.5825,
      "step": 1527
    },
    {
      "epoch": 0.2650936849410132,
      "grad_norm": 0.713713526725769,
      "learning_rate": 2.6504770164787514e-05,
      "loss": 0.6475,
      "step": 1528
    },
    {
      "epoch": 0.2652671755725191,
      "grad_norm": 0.851690411567688,
      "learning_rate": 2.6522116218560283e-05,
      "loss": 0.7598,
      "step": 1529
    },
    {
      "epoch": 0.26544066620402496,
      "grad_norm": 1.8548156023025513,
      "learning_rate": 2.6539462272333048e-05,
      "loss": 0.618,
      "step": 1530
    },
    {
      "epoch": 0.26561415683553086,
      "grad_norm": 0.7332980036735535,
      "learning_rate": 2.6556808326105813e-05,
      "loss": 0.791,
      "step": 1531
    },
    {
      "epoch": 0.26578764746703676,
      "grad_norm": 3.4511547088623047,
      "learning_rate": 2.657415437987858e-05,
      "loss": 0.6028,
      "step": 1532
    },
    {
      "epoch": 0.26596113809854266,
      "grad_norm": 0.759696364402771,
      "learning_rate": 2.6591500433651346e-05,
      "loss": 0.614,
      "step": 1533
    },
    {
      "epoch": 0.26613462873004856,
      "grad_norm": 0.8297407031059265,
      "learning_rate": 2.6608846487424114e-05,
      "loss": 0.626,
      "step": 1534
    },
    {
      "epoch": 0.26630811936155446,
      "grad_norm": 0.936953067779541,
      "learning_rate": 2.662619254119688e-05,
      "loss": 0.6895,
      "step": 1535
    },
    {
      "epoch": 0.26648160999306036,
      "grad_norm": 1.5277049541473389,
      "learning_rate": 2.6643538594969648e-05,
      "loss": 0.623,
      "step": 1536
    },
    {
      "epoch": 0.26665510062456627,
      "grad_norm": 1.0148879289627075,
      "learning_rate": 2.6660884648742413e-05,
      "loss": 0.6768,
      "step": 1537
    },
    {
      "epoch": 0.26682859125607217,
      "grad_norm": 0.8687107563018799,
      "learning_rate": 2.667823070251518e-05,
      "loss": 0.5505,
      "step": 1538
    },
    {
      "epoch": 0.26700208188757807,
      "grad_norm": 2.721482038497925,
      "learning_rate": 2.6695576756287946e-05,
      "loss": 0.6248,
      "step": 1539
    },
    {
      "epoch": 0.26717557251908397,
      "grad_norm": 1.6190732717514038,
      "learning_rate": 2.6712922810060714e-05,
      "loss": 0.5505,
      "step": 1540
    },
    {
      "epoch": 0.26734906315058987,
      "grad_norm": 0.9620177745819092,
      "learning_rate": 2.673026886383348e-05,
      "loss": 0.5226,
      "step": 1541
    },
    {
      "epoch": 0.2675225537820958,
      "grad_norm": 0.9856328368186951,
      "learning_rate": 2.6747614917606245e-05,
      "loss": 0.5323,
      "step": 1542
    },
    {
      "epoch": 0.2676960444136017,
      "grad_norm": 1.3991506099700928,
      "learning_rate": 2.6764960971379013e-05,
      "loss": 0.5851,
      "step": 1543
    },
    {
      "epoch": 0.2678695350451076,
      "grad_norm": 0.6915868520736694,
      "learning_rate": 2.6782307025151778e-05,
      "loss": 0.7461,
      "step": 1544
    },
    {
      "epoch": 0.2680430256766135,
      "grad_norm": 0.7301471829414368,
      "learning_rate": 2.6799653078924546e-05,
      "loss": 0.6425,
      "step": 1545
    },
    {
      "epoch": 0.2682165163081194,
      "grad_norm": 0.7267754077911377,
      "learning_rate": 2.681699913269731e-05,
      "loss": 0.6333,
      "step": 1546
    },
    {
      "epoch": 0.2683900069396253,
      "grad_norm": 1.0410354137420654,
      "learning_rate": 2.683434518647008e-05,
      "loss": 0.6542,
      "step": 1547
    },
    {
      "epoch": 0.2685634975711312,
      "grad_norm": 1.1848565340042114,
      "learning_rate": 2.6851691240242845e-05,
      "loss": 0.6288,
      "step": 1548
    },
    {
      "epoch": 0.2687369882026371,
      "grad_norm": 0.6985346078872681,
      "learning_rate": 2.6869037294015613e-05,
      "loss": 0.7563,
      "step": 1549
    },
    {
      "epoch": 0.26891047883414293,
      "grad_norm": 0.7319064140319824,
      "learning_rate": 2.6886383347788378e-05,
      "loss": 0.7356,
      "step": 1550
    },
    {
      "epoch": 0.26908396946564883,
      "grad_norm": 0.8546984195709229,
      "learning_rate": 2.6903729401561146e-05,
      "loss": 0.6451,
      "step": 1551
    },
    {
      "epoch": 0.26925746009715473,
      "grad_norm": 0.7902388572692871,
      "learning_rate": 2.692107545533391e-05,
      "loss": 0.7434,
      "step": 1552
    },
    {
      "epoch": 0.26943095072866063,
      "grad_norm": 0.9439723491668701,
      "learning_rate": 2.6938421509106683e-05,
      "loss": 0.6918,
      "step": 1553
    },
    {
      "epoch": 0.26960444136016654,
      "grad_norm": 1.1431201696395874,
      "learning_rate": 2.6955767562879448e-05,
      "loss": 0.5726,
      "step": 1554
    },
    {
      "epoch": 0.26977793199167244,
      "grad_norm": 0.9502336978912354,
      "learning_rate": 2.6973113616652217e-05,
      "loss": 0.6831,
      "step": 1555
    },
    {
      "epoch": 0.26995142262317834,
      "grad_norm": 0.6996731162071228,
      "learning_rate": 2.699045967042498e-05,
      "loss": 0.5747,
      "step": 1556
    },
    {
      "epoch": 0.27012491325468424,
      "grad_norm": 0.8544343113899231,
      "learning_rate": 2.700780572419775e-05,
      "loss": 0.6241,
      "step": 1557
    },
    {
      "epoch": 0.27029840388619014,
      "grad_norm": 1.1212594509124756,
      "learning_rate": 2.7025151777970515e-05,
      "loss": 0.6632,
      "step": 1558
    },
    {
      "epoch": 0.27047189451769604,
      "grad_norm": 0.8834282159805298,
      "learning_rate": 2.7042497831743283e-05,
      "loss": 0.6011,
      "step": 1559
    },
    {
      "epoch": 0.27064538514920194,
      "grad_norm": 0.7834562063217163,
      "learning_rate": 2.705984388551605e-05,
      "loss": 0.6708,
      "step": 1560
    },
    {
      "epoch": 0.27081887578070785,
      "grad_norm": 0.6939826607704163,
      "learning_rate": 2.7077189939288813e-05,
      "loss": 0.6743,
      "step": 1561
    },
    {
      "epoch": 0.27099236641221375,
      "grad_norm": 0.9736182689666748,
      "learning_rate": 2.7094535993061582e-05,
      "loss": 0.5938,
      "step": 1562
    },
    {
      "epoch": 0.27116585704371965,
      "grad_norm": 1.8810274600982666,
      "learning_rate": 2.7111882046834347e-05,
      "loss": 0.7124,
      "step": 1563
    },
    {
      "epoch": 0.27133934767522555,
      "grad_norm": 1.272519588470459,
      "learning_rate": 2.7129228100607115e-05,
      "loss": 0.6328,
      "step": 1564
    },
    {
      "epoch": 0.27151283830673145,
      "grad_norm": 0.8820292949676514,
      "learning_rate": 2.714657415437988e-05,
      "loss": 0.5922,
      "step": 1565
    },
    {
      "epoch": 0.27168632893823735,
      "grad_norm": 0.808553159236908,
      "learning_rate": 2.716392020815265e-05,
      "loss": 0.5687,
      "step": 1566
    },
    {
      "epoch": 0.27185981956974326,
      "grad_norm": 0.79135662317276,
      "learning_rate": 2.7181266261925414e-05,
      "loss": 0.5488,
      "step": 1567
    },
    {
      "epoch": 0.27203331020124916,
      "grad_norm": 0.8797126412391663,
      "learning_rate": 2.7198612315698182e-05,
      "loss": 0.5465,
      "step": 1568
    },
    {
      "epoch": 0.27220680083275506,
      "grad_norm": 1.229600191116333,
      "learning_rate": 2.7215958369470947e-05,
      "loss": 0.6514,
      "step": 1569
    },
    {
      "epoch": 0.2723802914642609,
      "grad_norm": 0.7109715342521667,
      "learning_rate": 2.7233304423243715e-05,
      "loss": 0.6831,
      "step": 1570
    },
    {
      "epoch": 0.2725537820957668,
      "grad_norm": 1.7438849210739136,
      "learning_rate": 2.725065047701648e-05,
      "loss": 0.548,
      "step": 1571
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 1.8204153776168823,
      "learning_rate": 2.7267996530789245e-05,
      "loss": 0.6622,
      "step": 1572
    },
    {
      "epoch": 0.2729007633587786,
      "grad_norm": 0.8696473240852356,
      "learning_rate": 2.7285342584562014e-05,
      "loss": 0.6555,
      "step": 1573
    },
    {
      "epoch": 0.2730742539902845,
      "grad_norm": 0.8713580965995789,
      "learning_rate": 2.730268863833478e-05,
      "loss": 0.6252,
      "step": 1574
    },
    {
      "epoch": 0.2732477446217904,
      "grad_norm": 0.9230552911758423,
      "learning_rate": 2.7320034692107547e-05,
      "loss": 0.7051,
      "step": 1575
    },
    {
      "epoch": 0.2734212352532963,
      "grad_norm": 0.9991665482521057,
      "learning_rate": 2.7337380745880312e-05,
      "loss": 0.6189,
      "step": 1576
    },
    {
      "epoch": 0.2735947258848022,
      "grad_norm": 1.474080204963684,
      "learning_rate": 2.735472679965308e-05,
      "loss": 0.6189,
      "step": 1577
    },
    {
      "epoch": 0.2737682165163081,
      "grad_norm": 0.8902952671051025,
      "learning_rate": 2.7372072853425846e-05,
      "loss": 0.6364,
      "step": 1578
    },
    {
      "epoch": 0.273941707147814,
      "grad_norm": 1.0150178670883179,
      "learning_rate": 2.7389418907198614e-05,
      "loss": 0.7532,
      "step": 1579
    },
    {
      "epoch": 0.2741151977793199,
      "grad_norm": 3.1479642391204834,
      "learning_rate": 2.740676496097138e-05,
      "loss": 0.6188,
      "step": 1580
    },
    {
      "epoch": 0.2742886884108258,
      "grad_norm": 0.7297648191452026,
      "learning_rate": 2.7424111014744147e-05,
      "loss": 0.7668,
      "step": 1581
    },
    {
      "epoch": 0.2744621790423317,
      "grad_norm": 0.6790329217910767,
      "learning_rate": 2.7441457068516916e-05,
      "loss": 0.6566,
      "step": 1582
    },
    {
      "epoch": 0.2746356696738376,
      "grad_norm": 0.7647008299827576,
      "learning_rate": 2.7458803122289684e-05,
      "loss": 0.6169,
      "step": 1583
    },
    {
      "epoch": 0.2748091603053435,
      "grad_norm": 0.9886143207550049,
      "learning_rate": 2.747614917606245e-05,
      "loss": 0.6399,
      "step": 1584
    },
    {
      "epoch": 0.2749826509368494,
      "grad_norm": 0.8156219720840454,
      "learning_rate": 2.7493495229835217e-05,
      "loss": 0.6313,
      "step": 1585
    },
    {
      "epoch": 0.27515614156835533,
      "grad_norm": 0.7117592096328735,
      "learning_rate": 2.7510841283607982e-05,
      "loss": 0.6704,
      "step": 1586
    },
    {
      "epoch": 0.27532963219986123,
      "grad_norm": 0.8192128539085388,
      "learning_rate": 2.752818733738075e-05,
      "loss": 0.5906,
      "step": 1587
    },
    {
      "epoch": 0.27550312283136713,
      "grad_norm": 0.9606592655181885,
      "learning_rate": 2.7545533391153516e-05,
      "loss": 0.5538,
      "step": 1588
    },
    {
      "epoch": 0.27567661346287303,
      "grad_norm": 1.0025795698165894,
      "learning_rate": 2.7562879444926284e-05,
      "loss": 0.567,
      "step": 1589
    },
    {
      "epoch": 0.2758501040943789,
      "grad_norm": 0.6096360087394714,
      "learning_rate": 2.758022549869905e-05,
      "loss": 0.708,
      "step": 1590
    },
    {
      "epoch": 0.2760235947258848,
      "grad_norm": 0.6717957854270935,
      "learning_rate": 2.7597571552471814e-05,
      "loss": 0.6978,
      "step": 1591
    },
    {
      "epoch": 0.2761970853573907,
      "grad_norm": 0.9949507713317871,
      "learning_rate": 2.7614917606244583e-05,
      "loss": 0.5021,
      "step": 1592
    },
    {
      "epoch": 0.2763705759888966,
      "grad_norm": 0.8604462742805481,
      "learning_rate": 2.7632263660017348e-05,
      "loss": 0.5776,
      "step": 1593
    },
    {
      "epoch": 0.2765440666204025,
      "grad_norm": 1.1553502082824707,
      "learning_rate": 2.7649609713790116e-05,
      "loss": 0.5986,
      "step": 1594
    },
    {
      "epoch": 0.2767175572519084,
      "grad_norm": 0.768334686756134,
      "learning_rate": 2.766695576756288e-05,
      "loss": 0.6689,
      "step": 1595
    },
    {
      "epoch": 0.2768910478834143,
      "grad_norm": 0.8372496366500854,
      "learning_rate": 2.768430182133565e-05,
      "loss": 0.6992,
      "step": 1596
    },
    {
      "epoch": 0.2770645385149202,
      "grad_norm": 1.0399575233459473,
      "learning_rate": 2.7701647875108414e-05,
      "loss": 0.6202,
      "step": 1597
    },
    {
      "epoch": 0.2772380291464261,
      "grad_norm": 1.0247970819473267,
      "learning_rate": 2.7718993928881183e-05,
      "loss": 0.76,
      "step": 1598
    },
    {
      "epoch": 0.277411519777932,
      "grad_norm": 0.810613214969635,
      "learning_rate": 2.7736339982653948e-05,
      "loss": 0.6338,
      "step": 1599
    },
    {
      "epoch": 0.2775850104094379,
      "grad_norm": 1.4586679935455322,
      "learning_rate": 2.7753686036426716e-05,
      "loss": 0.6869,
      "step": 1600
    },
    {
      "epoch": 0.2777585010409438,
      "grad_norm": 0.9183869957923889,
      "learning_rate": 2.777103209019948e-05,
      "loss": 0.6377,
      "step": 1601
    },
    {
      "epoch": 0.2779319916724497,
      "grad_norm": 0.9354574084281921,
      "learning_rate": 2.7788378143972246e-05,
      "loss": 0.6223,
      "step": 1602
    },
    {
      "epoch": 0.2781054823039556,
      "grad_norm": 0.6997314095497131,
      "learning_rate": 2.7805724197745015e-05,
      "loss": 0.6864,
      "step": 1603
    },
    {
      "epoch": 0.2782789729354615,
      "grad_norm": 0.981356680393219,
      "learning_rate": 2.782307025151778e-05,
      "loss": 0.5828,
      "step": 1604
    },
    {
      "epoch": 0.2784524635669674,
      "grad_norm": 0.7175416350364685,
      "learning_rate": 2.7840416305290548e-05,
      "loss": 0.5935,
      "step": 1605
    },
    {
      "epoch": 0.2786259541984733,
      "grad_norm": 0.7980823516845703,
      "learning_rate": 2.7857762359063313e-05,
      "loss": 0.5989,
      "step": 1606
    },
    {
      "epoch": 0.2787994448299792,
      "grad_norm": 0.7140958905220032,
      "learning_rate": 2.787510841283608e-05,
      "loss": 0.6205,
      "step": 1607
    },
    {
      "epoch": 0.2789729354614851,
      "grad_norm": 0.7325379252433777,
      "learning_rate": 2.7892454466608846e-05,
      "loss": 0.6599,
      "step": 1608
    },
    {
      "epoch": 0.27914642609299095,
      "grad_norm": 1.1121330261230469,
      "learning_rate": 2.7909800520381615e-05,
      "loss": 0.7002,
      "step": 1609
    },
    {
      "epoch": 0.27931991672449685,
      "grad_norm": 0.8870050311088562,
      "learning_rate": 2.792714657415438e-05,
      "loss": 0.5525,
      "step": 1610
    },
    {
      "epoch": 0.27949340735600275,
      "grad_norm": 1.183931827545166,
      "learning_rate": 2.7944492627927148e-05,
      "loss": 0.6285,
      "step": 1611
    },
    {
      "epoch": 0.27966689798750866,
      "grad_norm": 1.1623479127883911,
      "learning_rate": 2.7961838681699917e-05,
      "loss": 0.5328,
      "step": 1612
    },
    {
      "epoch": 0.27984038861901456,
      "grad_norm": 1.0510090589523315,
      "learning_rate": 2.7979184735472685e-05,
      "loss": 0.5796,
      "step": 1613
    },
    {
      "epoch": 0.28001387925052046,
      "grad_norm": 0.9020524024963379,
      "learning_rate": 2.799653078924545e-05,
      "loss": 0.5897,
      "step": 1614
    },
    {
      "epoch": 0.28018736988202636,
      "grad_norm": 2.9960711002349854,
      "learning_rate": 2.801387684301822e-05,
      "loss": 0.6235,
      "step": 1615
    },
    {
      "epoch": 0.28036086051353226,
      "grad_norm": 0.8520289063453674,
      "learning_rate": 2.8031222896790983e-05,
      "loss": 0.6537,
      "step": 1616
    },
    {
      "epoch": 0.28053435114503816,
      "grad_norm": 1.2885127067565918,
      "learning_rate": 2.8048568950563752e-05,
      "loss": 0.5812,
      "step": 1617
    },
    {
      "epoch": 0.28070784177654406,
      "grad_norm": 0.9299231171607971,
      "learning_rate": 2.8065915004336517e-05,
      "loss": 0.5502,
      "step": 1618
    },
    {
      "epoch": 0.28088133240804997,
      "grad_norm": 0.761440098285675,
      "learning_rate": 2.8083261058109285e-05,
      "loss": 0.6263,
      "step": 1619
    },
    {
      "epoch": 0.28105482303955587,
      "grad_norm": 0.8906003832817078,
      "learning_rate": 2.810060711188205e-05,
      "loss": 0.6716,
      "step": 1620
    },
    {
      "epoch": 0.28122831367106177,
      "grad_norm": 0.9329630732536316,
      "learning_rate": 2.8117953165654815e-05,
      "loss": 0.712,
      "step": 1621
    },
    {
      "epoch": 0.28140180430256767,
      "grad_norm": 0.9584671854972839,
      "learning_rate": 2.8135299219427584e-05,
      "loss": 0.5972,
      "step": 1622
    },
    {
      "epoch": 0.28157529493407357,
      "grad_norm": 2.436326026916504,
      "learning_rate": 2.815264527320035e-05,
      "loss": 0.5856,
      "step": 1623
    },
    {
      "epoch": 0.2817487855655795,
      "grad_norm": 0.8243262767791748,
      "learning_rate": 2.8169991326973117e-05,
      "loss": 0.5275,
      "step": 1624
    },
    {
      "epoch": 0.2819222761970854,
      "grad_norm": 0.7558211088180542,
      "learning_rate": 2.8187337380745882e-05,
      "loss": 0.7168,
      "step": 1625
    },
    {
      "epoch": 0.2820957668285913,
      "grad_norm": 0.6931042075157166,
      "learning_rate": 2.820468343451865e-05,
      "loss": 0.6819,
      "step": 1626
    },
    {
      "epoch": 0.2822692574600972,
      "grad_norm": 1.038571834564209,
      "learning_rate": 2.8222029488291415e-05,
      "loss": 0.5641,
      "step": 1627
    },
    {
      "epoch": 0.2824427480916031,
      "grad_norm": 1.068108320236206,
      "learning_rate": 2.8239375542064184e-05,
      "loss": 0.537,
      "step": 1628
    },
    {
      "epoch": 0.2826162387231089,
      "grad_norm": 1.0364161729812622,
      "learning_rate": 2.825672159583695e-05,
      "loss": 0.676,
      "step": 1629
    },
    {
      "epoch": 0.2827897293546148,
      "grad_norm": 1.1429134607315063,
      "learning_rate": 2.8274067649609717e-05,
      "loss": 0.5065,
      "step": 1630
    },
    {
      "epoch": 0.28296321998612073,
      "grad_norm": 0.7850894331932068,
      "learning_rate": 2.8291413703382482e-05,
      "loss": 0.6429,
      "step": 1631
    },
    {
      "epoch": 0.28313671061762663,
      "grad_norm": 0.8497446775436401,
      "learning_rate": 2.8308759757155247e-05,
      "loss": 0.5441,
      "step": 1632
    },
    {
      "epoch": 0.28331020124913253,
      "grad_norm": 1.1594936847686768,
      "learning_rate": 2.8326105810928015e-05,
      "loss": 0.5991,
      "step": 1633
    },
    {
      "epoch": 0.28348369188063843,
      "grad_norm": 0.780320405960083,
      "learning_rate": 2.834345186470078e-05,
      "loss": 0.6357,
      "step": 1634
    },
    {
      "epoch": 0.28365718251214433,
      "grad_norm": 0.816405177116394,
      "learning_rate": 2.836079791847355e-05,
      "loss": 0.5743,
      "step": 1635
    },
    {
      "epoch": 0.28383067314365024,
      "grad_norm": 1.3144505023956299,
      "learning_rate": 2.8378143972246314e-05,
      "loss": 0.5634,
      "step": 1636
    },
    {
      "epoch": 0.28400416377515614,
      "grad_norm": 0.9545460343360901,
      "learning_rate": 2.8395490026019082e-05,
      "loss": 0.5889,
      "step": 1637
    },
    {
      "epoch": 0.28417765440666204,
      "grad_norm": 1.016923189163208,
      "learning_rate": 2.8412836079791847e-05,
      "loss": 0.6056,
      "step": 1638
    },
    {
      "epoch": 0.28435114503816794,
      "grad_norm": 1.6665234565734863,
      "learning_rate": 2.8430182133564616e-05,
      "loss": 0.6687,
      "step": 1639
    },
    {
      "epoch": 0.28452463566967384,
      "grad_norm": 1.1342241764068604,
      "learning_rate": 2.844752818733738e-05,
      "loss": 0.5743,
      "step": 1640
    },
    {
      "epoch": 0.28469812630117974,
      "grad_norm": 0.7941092252731323,
      "learning_rate": 2.8464874241110152e-05,
      "loss": 0.6633,
      "step": 1641
    },
    {
      "epoch": 0.28487161693268565,
      "grad_norm": 0.8998805284500122,
      "learning_rate": 2.8482220294882917e-05,
      "loss": 0.6926,
      "step": 1642
    },
    {
      "epoch": 0.28504510756419155,
      "grad_norm": 1.1339398622512817,
      "learning_rate": 2.8499566348655686e-05,
      "loss": 0.7589,
      "step": 1643
    },
    {
      "epoch": 0.28521859819569745,
      "grad_norm": 0.6638227701187134,
      "learning_rate": 2.851691240242845e-05,
      "loss": 0.7004,
      "step": 1644
    },
    {
      "epoch": 0.28539208882720335,
      "grad_norm": 1.128553867340088,
      "learning_rate": 2.853425845620122e-05,
      "loss": 0.5492,
      "step": 1645
    },
    {
      "epoch": 0.28556557945870925,
      "grad_norm": 0.9050806164741516,
      "learning_rate": 2.8551604509973984e-05,
      "loss": 0.627,
      "step": 1646
    },
    {
      "epoch": 0.28573907009021515,
      "grad_norm": 0.7931371331214905,
      "learning_rate": 2.8568950563746753e-05,
      "loss": 0.6644,
      "step": 1647
    },
    {
      "epoch": 0.28591256072172105,
      "grad_norm": 0.6553512811660767,
      "learning_rate": 2.8586296617519518e-05,
      "loss": 0.6649,
      "step": 1648
    },
    {
      "epoch": 0.2860860513532269,
      "grad_norm": 0.7781357765197754,
      "learning_rate": 2.8603642671292286e-05,
      "loss": 0.7988,
      "step": 1649
    },
    {
      "epoch": 0.2862595419847328,
      "grad_norm": 0.7760520577430725,
      "learning_rate": 2.862098872506505e-05,
      "loss": 0.6641,
      "step": 1650
    },
    {
      "epoch": 0.2864330326162387,
      "grad_norm": 1.9323515892028809,
      "learning_rate": 2.8638334778837816e-05,
      "loss": 0.6757,
      "step": 1651
    },
    {
      "epoch": 0.2866065232477446,
      "grad_norm": 0.767174243927002,
      "learning_rate": 2.8655680832610584e-05,
      "loss": 0.6904,
      "step": 1652
    },
    {
      "epoch": 0.2867800138792505,
      "grad_norm": 0.9943397641181946,
      "learning_rate": 2.867302688638335e-05,
      "loss": 0.7026,
      "step": 1653
    },
    {
      "epoch": 0.2869535045107564,
      "grad_norm": 0.8809281587600708,
      "learning_rate": 2.8690372940156118e-05,
      "loss": 0.4868,
      "step": 1654
    },
    {
      "epoch": 0.2871269951422623,
      "grad_norm": 0.758124053478241,
      "learning_rate": 2.8707718993928883e-05,
      "loss": 0.5481,
      "step": 1655
    },
    {
      "epoch": 0.2873004857737682,
      "grad_norm": 0.978379487991333,
      "learning_rate": 2.872506504770165e-05,
      "loss": 0.7194,
      "step": 1656
    },
    {
      "epoch": 0.2874739764052741,
      "grad_norm": 0.686726450920105,
      "learning_rate": 2.8742411101474416e-05,
      "loss": 0.7852,
      "step": 1657
    },
    {
      "epoch": 0.28764746703678,
      "grad_norm": 0.8651068210601807,
      "learning_rate": 2.8759757155247185e-05,
      "loss": 0.582,
      "step": 1658
    },
    {
      "epoch": 0.2878209576682859,
      "grad_norm": 1.214903712272644,
      "learning_rate": 2.877710320901995e-05,
      "loss": 0.5227,
      "step": 1659
    },
    {
      "epoch": 0.2879944482997918,
      "grad_norm": 0.8349252939224243,
      "learning_rate": 2.8794449262792718e-05,
      "loss": 0.5859,
      "step": 1660
    },
    {
      "epoch": 0.2881679389312977,
      "grad_norm": 1.026668906211853,
      "learning_rate": 2.8811795316565483e-05,
      "loss": 0.52,
      "step": 1661
    },
    {
      "epoch": 0.2883414295628036,
      "grad_norm": 2.8020482063293457,
      "learning_rate": 2.8829141370338248e-05,
      "loss": 0.7031,
      "step": 1662
    },
    {
      "epoch": 0.2885149201943095,
      "grad_norm": 1.0019539594650269,
      "learning_rate": 2.8846487424111016e-05,
      "loss": 0.641,
      "step": 1663
    },
    {
      "epoch": 0.2886884108258154,
      "grad_norm": 0.8635918498039246,
      "learning_rate": 2.886383347788378e-05,
      "loss": 0.572,
      "step": 1664
    },
    {
      "epoch": 0.2888619014573213,
      "grad_norm": 1.5796480178833008,
      "learning_rate": 2.888117953165655e-05,
      "loss": 0.5725,
      "step": 1665
    },
    {
      "epoch": 0.2890353920888272,
      "grad_norm": 0.8226332068443298,
      "learning_rate": 2.8898525585429315e-05,
      "loss": 0.6166,
      "step": 1666
    },
    {
      "epoch": 0.2892088827203331,
      "grad_norm": 1.6667702198028564,
      "learning_rate": 2.8915871639202083e-05,
      "loss": 0.5413,
      "step": 1667
    },
    {
      "epoch": 0.28938237335183903,
      "grad_norm": 0.9546117186546326,
      "learning_rate": 2.8933217692974848e-05,
      "loss": 0.5764,
      "step": 1668
    },
    {
      "epoch": 0.2895558639833449,
      "grad_norm": 0.8718425631523132,
      "learning_rate": 2.8950563746747617e-05,
      "loss": 0.554,
      "step": 1669
    },
    {
      "epoch": 0.2897293546148508,
      "grad_norm": 1.003496527671814,
      "learning_rate": 2.896790980052038e-05,
      "loss": 0.547,
      "step": 1670
    },
    {
      "epoch": 0.2899028452463567,
      "grad_norm": 0.9424927234649658,
      "learning_rate": 2.8985255854293153e-05,
      "loss": 0.6836,
      "step": 1671
    },
    {
      "epoch": 0.2900763358778626,
      "grad_norm": 0.6520414352416992,
      "learning_rate": 2.900260190806592e-05,
      "loss": 0.6129,
      "step": 1672
    },
    {
      "epoch": 0.2902498265093685,
      "grad_norm": 0.6672245860099792,
      "learning_rate": 2.9019947961838687e-05,
      "loss": 0.6251,
      "step": 1673
    },
    {
      "epoch": 0.2904233171408744,
      "grad_norm": 0.8402858376502991,
      "learning_rate": 2.9037294015611452e-05,
      "loss": 0.7013,
      "step": 1674
    },
    {
      "epoch": 0.2905968077723803,
      "grad_norm": 0.6859344244003296,
      "learning_rate": 2.905464006938422e-05,
      "loss": 0.6455,
      "step": 1675
    },
    {
      "epoch": 0.2907702984038862,
      "grad_norm": 0.8473714590072632,
      "learning_rate": 2.9071986123156985e-05,
      "loss": 0.6072,
      "step": 1676
    },
    {
      "epoch": 0.2909437890353921,
      "grad_norm": 0.8517741560935974,
      "learning_rate": 2.9089332176929753e-05,
      "loss": 0.6248,
      "step": 1677
    },
    {
      "epoch": 0.291117279666898,
      "grad_norm": 1.1224453449249268,
      "learning_rate": 2.910667823070252e-05,
      "loss": 0.6846,
      "step": 1678
    },
    {
      "epoch": 0.2912907702984039,
      "grad_norm": 0.8093835115432739,
      "learning_rate": 2.9124024284475287e-05,
      "loss": 0.7305,
      "step": 1679
    },
    {
      "epoch": 0.2914642609299098,
      "grad_norm": 1.2922654151916504,
      "learning_rate": 2.9141370338248052e-05,
      "loss": 0.7435,
      "step": 1680
    },
    {
      "epoch": 0.2916377515614157,
      "grad_norm": 0.858190655708313,
      "learning_rate": 2.9158716392020817e-05,
      "loss": 0.6708,
      "step": 1681
    },
    {
      "epoch": 0.2918112421929216,
      "grad_norm": 0.7695908546447754,
      "learning_rate": 2.9176062445793585e-05,
      "loss": 0.7864,
      "step": 1682
    },
    {
      "epoch": 0.2919847328244275,
      "grad_norm": 0.981799840927124,
      "learning_rate": 2.919340849956635e-05,
      "loss": 0.673,
      "step": 1683
    },
    {
      "epoch": 0.2921582234559334,
      "grad_norm": 0.9595302939414978,
      "learning_rate": 2.921075455333912e-05,
      "loss": 0.7332,
      "step": 1684
    },
    {
      "epoch": 0.2923317140874393,
      "grad_norm": 0.9652135372161865,
      "learning_rate": 2.9228100607111884e-05,
      "loss": 0.5652,
      "step": 1685
    },
    {
      "epoch": 0.2925052047189452,
      "grad_norm": 1.3736512660980225,
      "learning_rate": 2.9245446660884652e-05,
      "loss": 0.6237,
      "step": 1686
    },
    {
      "epoch": 0.2926786953504511,
      "grad_norm": 0.8677798509597778,
      "learning_rate": 2.9262792714657417e-05,
      "loss": 0.632,
      "step": 1687
    },
    {
      "epoch": 0.29285218598195695,
      "grad_norm": 1.0175774097442627,
      "learning_rate": 2.9280138768430185e-05,
      "loss": 0.5555,
      "step": 1688
    },
    {
      "epoch": 0.29302567661346285,
      "grad_norm": 0.856152355670929,
      "learning_rate": 2.929748482220295e-05,
      "loss": 0.7053,
      "step": 1689
    },
    {
      "epoch": 0.29319916724496875,
      "grad_norm": 0.8478633165359497,
      "learning_rate": 2.931483087597572e-05,
      "loss": 0.6868,
      "step": 1690
    },
    {
      "epoch": 0.29337265787647465,
      "grad_norm": 0.8868247866630554,
      "learning_rate": 2.9332176929748484e-05,
      "loss": 0.6127,
      "step": 1691
    },
    {
      "epoch": 0.29354614850798055,
      "grad_norm": 0.8794786334037781,
      "learning_rate": 2.934952298352125e-05,
      "loss": 0.6006,
      "step": 1692
    },
    {
      "epoch": 0.29371963913948645,
      "grad_norm": 0.72084641456604,
      "learning_rate": 2.9366869037294017e-05,
      "loss": 0.7935,
      "step": 1693
    },
    {
      "epoch": 0.29389312977099236,
      "grad_norm": 0.9005531668663025,
      "learning_rate": 2.9384215091066782e-05,
      "loss": 0.6073,
      "step": 1694
    },
    {
      "epoch": 0.29406662040249826,
      "grad_norm": 0.8129706978797913,
      "learning_rate": 2.940156114483955e-05,
      "loss": 0.6124,
      "step": 1695
    },
    {
      "epoch": 0.29424011103400416,
      "grad_norm": 1.0862834453582764,
      "learning_rate": 2.9418907198612316e-05,
      "loss": 0.6075,
      "step": 1696
    },
    {
      "epoch": 0.29441360166551006,
      "grad_norm": 1.0599271059036255,
      "learning_rate": 2.9436253252385084e-05,
      "loss": 0.5315,
      "step": 1697
    },
    {
      "epoch": 0.29458709229701596,
      "grad_norm": 1.2996435165405273,
      "learning_rate": 2.945359930615785e-05,
      "loss": 0.5306,
      "step": 1698
    },
    {
      "epoch": 0.29476058292852186,
      "grad_norm": 0.9041792750358582,
      "learning_rate": 2.9470945359930617e-05,
      "loss": 0.536,
      "step": 1699
    },
    {
      "epoch": 0.29493407356002777,
      "grad_norm": 0.914151132106781,
      "learning_rate": 2.9488291413703386e-05,
      "loss": 0.5747,
      "step": 1700
    },
    {
      "epoch": 0.29510756419153367,
      "grad_norm": 0.828359067440033,
      "learning_rate": 2.9505637467476154e-05,
      "loss": 0.6954,
      "step": 1701
    },
    {
      "epoch": 0.29528105482303957,
      "grad_norm": 1.7115951776504517,
      "learning_rate": 2.952298352124892e-05,
      "loss": 0.5409,
      "step": 1702
    },
    {
      "epoch": 0.29545454545454547,
      "grad_norm": 0.9483741521835327,
      "learning_rate": 2.9540329575021688e-05,
      "loss": 0.6135,
      "step": 1703
    },
    {
      "epoch": 0.29562803608605137,
      "grad_norm": 0.8337392210960388,
      "learning_rate": 2.9557675628794453e-05,
      "loss": 0.6631,
      "step": 1704
    },
    {
      "epoch": 0.2958015267175573,
      "grad_norm": 0.7385498881340027,
      "learning_rate": 2.957502168256722e-05,
      "loss": 0.7437,
      "step": 1705
    },
    {
      "epoch": 0.2959750173490632,
      "grad_norm": 0.9643473625183105,
      "learning_rate": 2.9592367736339986e-05,
      "loss": 0.5907,
      "step": 1706
    },
    {
      "epoch": 0.2961485079805691,
      "grad_norm": 0.8231221437454224,
      "learning_rate": 2.9609713790112754e-05,
      "loss": 0.6471,
      "step": 1707
    },
    {
      "epoch": 0.2963219986120749,
      "grad_norm": 0.6505579352378845,
      "learning_rate": 2.962705984388552e-05,
      "loss": 0.6716,
      "step": 1708
    },
    {
      "epoch": 0.2964954892435808,
      "grad_norm": 0.8700175881385803,
      "learning_rate": 2.9644405897658284e-05,
      "loss": 0.6149,
      "step": 1709
    },
    {
      "epoch": 0.2966689798750867,
      "grad_norm": 0.7063069939613342,
      "learning_rate": 2.9661751951431053e-05,
      "loss": 0.6666,
      "step": 1710
    },
    {
      "epoch": 0.2968424705065926,
      "grad_norm": 0.8879273533821106,
      "learning_rate": 2.9679098005203818e-05,
      "loss": 0.6432,
      "step": 1711
    },
    {
      "epoch": 0.2970159611380985,
      "grad_norm": 0.7333043813705444,
      "learning_rate": 2.9696444058976586e-05,
      "loss": 0.5787,
      "step": 1712
    },
    {
      "epoch": 0.29718945176960443,
      "grad_norm": 0.6548595428466797,
      "learning_rate": 2.971379011274935e-05,
      "loss": 0.7612,
      "step": 1713
    },
    {
      "epoch": 0.29736294240111033,
      "grad_norm": 1.1210254430770874,
      "learning_rate": 2.973113616652212e-05,
      "loss": 0.6273,
      "step": 1714
    },
    {
      "epoch": 0.29753643303261623,
      "grad_norm": 0.7752652168273926,
      "learning_rate": 2.9748482220294885e-05,
      "loss": 0.781,
      "step": 1715
    },
    {
      "epoch": 0.29770992366412213,
      "grad_norm": 0.9227986931800842,
      "learning_rate": 2.9765828274067653e-05,
      "loss": 0.5767,
      "step": 1716
    },
    {
      "epoch": 0.29788341429562804,
      "grad_norm": 1.5093612670898438,
      "learning_rate": 2.9783174327840418e-05,
      "loss": 0.5686,
      "step": 1717
    },
    {
      "epoch": 0.29805690492713394,
      "grad_norm": 0.694484293460846,
      "learning_rate": 2.9800520381613186e-05,
      "loss": 0.7233,
      "step": 1718
    },
    {
      "epoch": 0.29823039555863984,
      "grad_norm": 0.8226703405380249,
      "learning_rate": 2.981786643538595e-05,
      "loss": 0.6245,
      "step": 1719
    },
    {
      "epoch": 0.29840388619014574,
      "grad_norm": 0.7781656384468079,
      "learning_rate": 2.983521248915872e-05,
      "loss": 0.6234,
      "step": 1720
    },
    {
      "epoch": 0.29857737682165164,
      "grad_norm": 0.7376680374145508,
      "learning_rate": 2.9852558542931485e-05,
      "loss": 0.632,
      "step": 1721
    },
    {
      "epoch": 0.29875086745315754,
      "grad_norm": 1.330215573310852,
      "learning_rate": 2.986990459670425e-05,
      "loss": 0.5302,
      "step": 1722
    },
    {
      "epoch": 0.29892435808466344,
      "grad_norm": 0.9246023297309875,
      "learning_rate": 2.9887250650477018e-05,
      "loss": 0.5,
      "step": 1723
    },
    {
      "epoch": 0.29909784871616935,
      "grad_norm": 0.6971983909606934,
      "learning_rate": 2.9904596704249783e-05,
      "loss": 0.5924,
      "step": 1724
    },
    {
      "epoch": 0.29927133934767525,
      "grad_norm": 1.3495919704437256,
      "learning_rate": 2.992194275802255e-05,
      "loss": 0.6742,
      "step": 1725
    },
    {
      "epoch": 0.29944482997918115,
      "grad_norm": 0.8896782994270325,
      "learning_rate": 2.9939288811795316e-05,
      "loss": 0.5295,
      "step": 1726
    },
    {
      "epoch": 0.29961832061068705,
      "grad_norm": 1.149600625038147,
      "learning_rate": 2.9956634865568085e-05,
      "loss": 0.6323,
      "step": 1727
    },
    {
      "epoch": 0.2997918112421929,
      "grad_norm": 0.6024688482284546,
      "learning_rate": 2.997398091934085e-05,
      "loss": 0.745,
      "step": 1728
    },
    {
      "epoch": 0.2999653018736988,
      "grad_norm": 0.7137205600738525,
      "learning_rate": 2.9991326973113618e-05,
      "loss": 0.5896,
      "step": 1729
    },
    {
      "epoch": 0.3001387925052047,
      "grad_norm": 0.5844210982322693,
      "learning_rate": 3.0008673026886387e-05,
      "loss": 0.72,
      "step": 1730
    },
    {
      "epoch": 0.3003122831367106,
      "grad_norm": 0.9469254612922668,
      "learning_rate": 3.0026019080659155e-05,
      "loss": 0.7002,
      "step": 1731
    },
    {
      "epoch": 0.3004857737682165,
      "grad_norm": 1.4320896863937378,
      "learning_rate": 3.004336513443192e-05,
      "loss": 0.5857,
      "step": 1732
    },
    {
      "epoch": 0.3006592643997224,
      "grad_norm": 0.819429337978363,
      "learning_rate": 3.006071118820469e-05,
      "loss": 0.6317,
      "step": 1733
    },
    {
      "epoch": 0.3008327550312283,
      "grad_norm": 0.8208953738212585,
      "learning_rate": 3.0078057241977453e-05,
      "loss": 0.5531,
      "step": 1734
    },
    {
      "epoch": 0.3010062456627342,
      "grad_norm": 1.2594075202941895,
      "learning_rate": 3.0095403295750222e-05,
      "loss": 0.6279,
      "step": 1735
    },
    {
      "epoch": 0.3011797362942401,
      "grad_norm": 0.8851017355918884,
      "learning_rate": 3.0112749349522987e-05,
      "loss": 0.5364,
      "step": 1736
    },
    {
      "epoch": 0.301353226925746,
      "grad_norm": 0.8470969796180725,
      "learning_rate": 3.0130095403295755e-05,
      "loss": 0.6034,
      "step": 1737
    },
    {
      "epoch": 0.3015267175572519,
      "grad_norm": 0.8872317671775818,
      "learning_rate": 3.014744145706852e-05,
      "loss": 0.7142,
      "step": 1738
    },
    {
      "epoch": 0.3017002081887578,
      "grad_norm": 0.6019777059555054,
      "learning_rate": 3.0164787510841285e-05,
      "loss": 0.7383,
      "step": 1739
    },
    {
      "epoch": 0.3018736988202637,
      "grad_norm": 0.7203370332717896,
      "learning_rate": 3.0182133564614054e-05,
      "loss": 0.6733,
      "step": 1740
    },
    {
      "epoch": 0.3020471894517696,
      "grad_norm": 0.9027160406112671,
      "learning_rate": 3.019947961838682e-05,
      "loss": 0.5337,
      "step": 1741
    },
    {
      "epoch": 0.3022206800832755,
      "grad_norm": 1.074172019958496,
      "learning_rate": 3.0216825672159587e-05,
      "loss": 0.6599,
      "step": 1742
    },
    {
      "epoch": 0.3023941707147814,
      "grad_norm": 0.9253829121589661,
      "learning_rate": 3.0234171725932352e-05,
      "loss": 0.6835,
      "step": 1743
    },
    {
      "epoch": 0.3025676613462873,
      "grad_norm": 0.9681320786476135,
      "learning_rate": 3.025151777970512e-05,
      "loss": 0.5883,
      "step": 1744
    },
    {
      "epoch": 0.3027411519777932,
      "grad_norm": 0.7276582717895508,
      "learning_rate": 3.0268863833477885e-05,
      "loss": 0.6689,
      "step": 1745
    },
    {
      "epoch": 0.3029146426092991,
      "grad_norm": 0.7069199681282043,
      "learning_rate": 3.0286209887250654e-05,
      "loss": 0.6892,
      "step": 1746
    },
    {
      "epoch": 0.30308813324080497,
      "grad_norm": 0.8422447443008423,
      "learning_rate": 3.030355594102342e-05,
      "loss": 0.6062,
      "step": 1747
    },
    {
      "epoch": 0.30326162387231087,
      "grad_norm": 2.7167630195617676,
      "learning_rate": 3.0320901994796187e-05,
      "loss": 0.7922,
      "step": 1748
    },
    {
      "epoch": 0.30343511450381677,
      "grad_norm": 0.8047178983688354,
      "learning_rate": 3.0338248048568952e-05,
      "loss": 0.7277,
      "step": 1749
    },
    {
      "epoch": 0.3036086051353227,
      "grad_norm": 0.9170768857002258,
      "learning_rate": 3.035559410234172e-05,
      "loss": 0.6207,
      "step": 1750
    },
    {
      "epoch": 0.3037820957668286,
      "grad_norm": 0.6820874810218811,
      "learning_rate": 3.0372940156114486e-05,
      "loss": 0.6847,
      "step": 1751
    },
    {
      "epoch": 0.3039555863983345,
      "grad_norm": 1.0593931674957275,
      "learning_rate": 3.039028620988725e-05,
      "loss": 0.6033,
      "step": 1752
    },
    {
      "epoch": 0.3041290770298404,
      "grad_norm": 1.0743542909622192,
      "learning_rate": 3.040763226366002e-05,
      "loss": 0.6537,
      "step": 1753
    },
    {
      "epoch": 0.3043025676613463,
      "grad_norm": 0.7283003330230713,
      "learning_rate": 3.0424978317432784e-05,
      "loss": 0.5768,
      "step": 1754
    },
    {
      "epoch": 0.3044760582928522,
      "grad_norm": 0.8777532577514648,
      "learning_rate": 3.0442324371205552e-05,
      "loss": 0.559,
      "step": 1755
    },
    {
      "epoch": 0.3046495489243581,
      "grad_norm": 0.8460490107536316,
      "learning_rate": 3.0459670424978317e-05,
      "loss": 0.6449,
      "step": 1756
    },
    {
      "epoch": 0.304823039555864,
      "grad_norm": 0.7011381983757019,
      "learning_rate": 3.0477016478751086e-05,
      "loss": 0.6785,
      "step": 1757
    },
    {
      "epoch": 0.3049965301873699,
      "grad_norm": 0.7739598751068115,
      "learning_rate": 3.049436253252385e-05,
      "loss": 0.6609,
      "step": 1758
    },
    {
      "epoch": 0.3051700208188758,
      "grad_norm": 0.6341903209686279,
      "learning_rate": 3.051170858629662e-05,
      "loss": 0.7423,
      "step": 1759
    },
    {
      "epoch": 0.3053435114503817,
      "grad_norm": 0.8954048752784729,
      "learning_rate": 3.052905464006939e-05,
      "loss": 0.5604,
      "step": 1760
    },
    {
      "epoch": 0.3055170020818876,
      "grad_norm": 0.7283293008804321,
      "learning_rate": 3.0546400693842156e-05,
      "loss": 0.6965,
      "step": 1761
    },
    {
      "epoch": 0.3056904927133935,
      "grad_norm": 0.7408492565155029,
      "learning_rate": 3.056374674761492e-05,
      "loss": 0.6975,
      "step": 1762
    },
    {
      "epoch": 0.3058639833448994,
      "grad_norm": 0.70777428150177,
      "learning_rate": 3.0581092801387686e-05,
      "loss": 0.6914,
      "step": 1763
    },
    {
      "epoch": 0.3060374739764053,
      "grad_norm": 0.6914678812026978,
      "learning_rate": 3.059843885516046e-05,
      "loss": 0.6989,
      "step": 1764
    },
    {
      "epoch": 0.3062109646079112,
      "grad_norm": 1.1912055015563965,
      "learning_rate": 3.061578490893322e-05,
      "loss": 0.5037,
      "step": 1765
    },
    {
      "epoch": 0.3063844552394171,
      "grad_norm": 0.9246727824211121,
      "learning_rate": 3.063313096270599e-05,
      "loss": 0.6804,
      "step": 1766
    },
    {
      "epoch": 0.30655794587092294,
      "grad_norm": 0.6906810402870178,
      "learning_rate": 3.065047701647875e-05,
      "loss": 0.7336,
      "step": 1767
    },
    {
      "epoch": 0.30673143650242884,
      "grad_norm": 0.787849485874176,
      "learning_rate": 3.0667823070251524e-05,
      "loss": 0.6996,
      "step": 1768
    },
    {
      "epoch": 0.30690492713393475,
      "grad_norm": 1.0751699209213257,
      "learning_rate": 3.068516912402429e-05,
      "loss": 0.7373,
      "step": 1769
    },
    {
      "epoch": 0.30707841776544065,
      "grad_norm": 1.01177978515625,
      "learning_rate": 3.0702515177797054e-05,
      "loss": 0.5951,
      "step": 1770
    },
    {
      "epoch": 0.30725190839694655,
      "grad_norm": 0.9446077346801758,
      "learning_rate": 3.071986123156982e-05,
      "loss": 0.5613,
      "step": 1771
    },
    {
      "epoch": 0.30742539902845245,
      "grad_norm": 0.9041101336479187,
      "learning_rate": 3.0737207285342584e-05,
      "loss": 0.6337,
      "step": 1772
    },
    {
      "epoch": 0.30759888965995835,
      "grad_norm": 2.196101188659668,
      "learning_rate": 3.0754553339115356e-05,
      "loss": 0.5615,
      "step": 1773
    },
    {
      "epoch": 0.30777238029146425,
      "grad_norm": 1.4874036312103271,
      "learning_rate": 3.077189939288812e-05,
      "loss": 0.5897,
      "step": 1774
    },
    {
      "epoch": 0.30794587092297016,
      "grad_norm": 0.8694006204605103,
      "learning_rate": 3.0789245446660886e-05,
      "loss": 0.6401,
      "step": 1775
    },
    {
      "epoch": 0.30811936155447606,
      "grad_norm": 0.5794796943664551,
      "learning_rate": 3.080659150043365e-05,
      "loss": 0.7235,
      "step": 1776
    },
    {
      "epoch": 0.30829285218598196,
      "grad_norm": 0.9236095547676086,
      "learning_rate": 3.082393755420642e-05,
      "loss": 0.7251,
      "step": 1777
    },
    {
      "epoch": 0.30846634281748786,
      "grad_norm": 0.7490460872650146,
      "learning_rate": 3.084128360797919e-05,
      "loss": 0.718,
      "step": 1778
    },
    {
      "epoch": 0.30863983344899376,
      "grad_norm": 0.9465798139572144,
      "learning_rate": 3.085862966175195e-05,
      "loss": 0.691,
      "step": 1779
    },
    {
      "epoch": 0.30881332408049966,
      "grad_norm": 0.7532344460487366,
      "learning_rate": 3.087597571552472e-05,
      "loss": 0.6631,
      "step": 1780
    },
    {
      "epoch": 0.30898681471200556,
      "grad_norm": 1.433074712753296,
      "learning_rate": 3.089332176929748e-05,
      "loss": 0.6438,
      "step": 1781
    },
    {
      "epoch": 0.30916030534351147,
      "grad_norm": 1.390010952949524,
      "learning_rate": 3.0910667823070255e-05,
      "loss": 0.7295,
      "step": 1782
    },
    {
      "epoch": 0.30933379597501737,
      "grad_norm": 0.8118870258331299,
      "learning_rate": 3.092801387684302e-05,
      "loss": 0.572,
      "step": 1783
    },
    {
      "epoch": 0.30950728660652327,
      "grad_norm": 1.1158040761947632,
      "learning_rate": 3.0945359930615785e-05,
      "loss": 0.627,
      "step": 1784
    },
    {
      "epoch": 0.30968077723802917,
      "grad_norm": 0.7021401524543762,
      "learning_rate": 3.096270598438855e-05,
      "loss": 0.6309,
      "step": 1785
    },
    {
      "epoch": 0.30985426786953507,
      "grad_norm": 0.9394262433052063,
      "learning_rate": 3.098005203816132e-05,
      "loss": 0.5513,
      "step": 1786
    },
    {
      "epoch": 0.3100277585010409,
      "grad_norm": 0.9389492869377136,
      "learning_rate": 3.0997398091934087e-05,
      "loss": 0.6484,
      "step": 1787
    },
    {
      "epoch": 0.3102012491325468,
      "grad_norm": 0.6000140309333801,
      "learning_rate": 3.101474414570685e-05,
      "loss": 0.6553,
      "step": 1788
    },
    {
      "epoch": 0.3103747397640527,
      "grad_norm": 0.8362985849380493,
      "learning_rate": 3.103209019947962e-05,
      "loss": 0.6587,
      "step": 1789
    },
    {
      "epoch": 0.3105482303955586,
      "grad_norm": 0.7522281408309937,
      "learning_rate": 3.104943625325239e-05,
      "loss": 0.708,
      "step": 1790
    },
    {
      "epoch": 0.3107217210270645,
      "grad_norm": 0.6182814836502075,
      "learning_rate": 3.106678230702515e-05,
      "loss": 0.765,
      "step": 1791
    },
    {
      "epoch": 0.3108952116585704,
      "grad_norm": 0.9737197160720825,
      "learning_rate": 3.1084128360797925e-05,
      "loss": 0.7123,
      "step": 1792
    },
    {
      "epoch": 0.3110687022900763,
      "grad_norm": 2.1213455200195312,
      "learning_rate": 3.110147441457069e-05,
      "loss": 0.5731,
      "step": 1793
    },
    {
      "epoch": 0.31124219292158223,
      "grad_norm": 1.0937367677688599,
      "learning_rate": 3.1118820468343455e-05,
      "loss": 0.5756,
      "step": 1794
    },
    {
      "epoch": 0.31141568355308813,
      "grad_norm": 0.7878339290618896,
      "learning_rate": 3.113616652211622e-05,
      "loss": 0.6467,
      "step": 1795
    },
    {
      "epoch": 0.31158917418459403,
      "grad_norm": 1.0251662731170654,
      "learning_rate": 3.115351257588899e-05,
      "loss": 0.6653,
      "step": 1796
    },
    {
      "epoch": 0.31176266481609993,
      "grad_norm": 0.618768572807312,
      "learning_rate": 3.117085862966176e-05,
      "loss": 0.7207,
      "step": 1797
    },
    {
      "epoch": 0.31193615544760583,
      "grad_norm": 0.9752748012542725,
      "learning_rate": 3.118820468343452e-05,
      "loss": 0.5526,
      "step": 1798
    },
    {
      "epoch": 0.31210964607911174,
      "grad_norm": 1.442103385925293,
      "learning_rate": 3.120555073720729e-05,
      "loss": 0.5773,
      "step": 1799
    },
    {
      "epoch": 0.31228313671061764,
      "grad_norm": 1.0630028247833252,
      "learning_rate": 3.122289679098005e-05,
      "loss": 0.6183,
      "step": 1800
    },
    {
      "epoch": 0.31245662734212354,
      "grad_norm": 0.7093740105628967,
      "learning_rate": 3.1240242844752824e-05,
      "loss": 0.6361,
      "step": 1801
    },
    {
      "epoch": 0.31263011797362944,
      "grad_norm": 0.9507588744163513,
      "learning_rate": 3.125758889852559e-05,
      "loss": 0.5803,
      "step": 1802
    },
    {
      "epoch": 0.31280360860513534,
      "grad_norm": 1.4544535875320435,
      "learning_rate": 3.1274934952298354e-05,
      "loss": 0.5353,
      "step": 1803
    },
    {
      "epoch": 0.31297709923664124,
      "grad_norm": 0.9902476668357849,
      "learning_rate": 3.129228100607112e-05,
      "loss": 0.6809,
      "step": 1804
    },
    {
      "epoch": 0.31315058986814714,
      "grad_norm": 0.9561774134635925,
      "learning_rate": 3.130962705984389e-05,
      "loss": 0.5793,
      "step": 1805
    },
    {
      "epoch": 0.31332408049965305,
      "grad_norm": 0.9205996990203857,
      "learning_rate": 3.1326973113616656e-05,
      "loss": 0.6047,
      "step": 1806
    },
    {
      "epoch": 0.3134975711311589,
      "grad_norm": 0.9781830310821533,
      "learning_rate": 3.134431916738942e-05,
      "loss": 0.5886,
      "step": 1807
    },
    {
      "epoch": 0.3136710617626648,
      "grad_norm": 0.8477849364280701,
      "learning_rate": 3.1361665221162186e-05,
      "loss": 0.6198,
      "step": 1808
    },
    {
      "epoch": 0.3138445523941707,
      "grad_norm": 0.8550189137458801,
      "learning_rate": 3.137901127493496e-05,
      "loss": 0.5726,
      "step": 1809
    },
    {
      "epoch": 0.3140180430256766,
      "grad_norm": 0.8213677406311035,
      "learning_rate": 3.139635732870772e-05,
      "loss": 0.6034,
      "step": 1810
    },
    {
      "epoch": 0.3141915336571825,
      "grad_norm": 0.7088372707366943,
      "learning_rate": 3.141370338248049e-05,
      "loss": 0.6888,
      "step": 1811
    },
    {
      "epoch": 0.3143650242886884,
      "grad_norm": 0.7192822694778442,
      "learning_rate": 3.143104943625325e-05,
      "loss": 0.7292,
      "step": 1812
    },
    {
      "epoch": 0.3145385149201943,
      "grad_norm": 0.9309755563735962,
      "learning_rate": 3.144839549002602e-05,
      "loss": 0.5566,
      "step": 1813
    },
    {
      "epoch": 0.3147120055517002,
      "grad_norm": 0.8870457410812378,
      "learning_rate": 3.146574154379879e-05,
      "loss": 0.556,
      "step": 1814
    },
    {
      "epoch": 0.3148854961832061,
      "grad_norm": 0.7572647333145142,
      "learning_rate": 3.1483087597571554e-05,
      "loss": 0.5516,
      "step": 1815
    },
    {
      "epoch": 0.315058986814712,
      "grad_norm": 0.8539776802062988,
      "learning_rate": 3.150043365134432e-05,
      "loss": 0.6786,
      "step": 1816
    },
    {
      "epoch": 0.3152324774462179,
      "grad_norm": 2.1104016304016113,
      "learning_rate": 3.1517779705117084e-05,
      "loss": 0.6483,
      "step": 1817
    },
    {
      "epoch": 0.3154059680777238,
      "grad_norm": 0.6436271071434021,
      "learning_rate": 3.1535125758889856e-05,
      "loss": 0.7158,
      "step": 1818
    },
    {
      "epoch": 0.3155794587092297,
      "grad_norm": 0.956113874912262,
      "learning_rate": 3.155247181266262e-05,
      "loss": 0.5658,
      "step": 1819
    },
    {
      "epoch": 0.3157529493407356,
      "grad_norm": 0.8388386368751526,
      "learning_rate": 3.156981786643539e-05,
      "loss": 0.67,
      "step": 1820
    },
    {
      "epoch": 0.3159264399722415,
      "grad_norm": 0.764722466468811,
      "learning_rate": 3.158716392020816e-05,
      "loss": 0.5817,
      "step": 1821
    },
    {
      "epoch": 0.3160999306037474,
      "grad_norm": 0.8697858452796936,
      "learning_rate": 3.160450997398092e-05,
      "loss": 0.6027,
      "step": 1822
    },
    {
      "epoch": 0.3162734212352533,
      "grad_norm": 0.9201172590255737,
      "learning_rate": 3.162185602775369e-05,
      "loss": 0.6139,
      "step": 1823
    },
    {
      "epoch": 0.3164469118667592,
      "grad_norm": 1.1344634294509888,
      "learning_rate": 3.163920208152646e-05,
      "loss": 0.7426,
      "step": 1824
    },
    {
      "epoch": 0.3166204024982651,
      "grad_norm": 1.032973051071167,
      "learning_rate": 3.1656548135299224e-05,
      "loss": 0.6969,
      "step": 1825
    },
    {
      "epoch": 0.31679389312977096,
      "grad_norm": 1.1614660024642944,
      "learning_rate": 3.167389418907199e-05,
      "loss": 0.6847,
      "step": 1826
    },
    {
      "epoch": 0.31696738376127687,
      "grad_norm": 0.8547880053520203,
      "learning_rate": 3.1691240242844754e-05,
      "loss": 0.7056,
      "step": 1827
    },
    {
      "epoch": 0.31714087439278277,
      "grad_norm": 0.7039558291435242,
      "learning_rate": 3.1708586296617526e-05,
      "loss": 0.8013,
      "step": 1828
    },
    {
      "epoch": 0.31731436502428867,
      "grad_norm": 0.7111900448799133,
      "learning_rate": 3.172593235039029e-05,
      "loss": 0.7852,
      "step": 1829
    },
    {
      "epoch": 0.31748785565579457,
      "grad_norm": 1.033314824104309,
      "learning_rate": 3.1743278404163056e-05,
      "loss": 0.7156,
      "step": 1830
    },
    {
      "epoch": 0.3176613462873005,
      "grad_norm": 0.6807761788368225,
      "learning_rate": 3.176062445793582e-05,
      "loss": 0.6707,
      "step": 1831
    },
    {
      "epoch": 0.3178348369188064,
      "grad_norm": 0.9980964660644531,
      "learning_rate": 3.1777970511708586e-05,
      "loss": 0.5842,
      "step": 1832
    },
    {
      "epoch": 0.3180083275503123,
      "grad_norm": 0.7964962720870972,
      "learning_rate": 3.179531656548136e-05,
      "loss": 0.6732,
      "step": 1833
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 1.0412522554397583,
      "learning_rate": 3.181266261925412e-05,
      "loss": 0.606,
      "step": 1834
    },
    {
      "epoch": 0.3183553088133241,
      "grad_norm": 0.8972187638282776,
      "learning_rate": 3.183000867302689e-05,
      "loss": 0.6263,
      "step": 1835
    },
    {
      "epoch": 0.31852879944483,
      "grad_norm": 0.7858732342720032,
      "learning_rate": 3.184735472679965e-05,
      "loss": 0.5792,
      "step": 1836
    },
    {
      "epoch": 0.3187022900763359,
      "grad_norm": 0.7865777015686035,
      "learning_rate": 3.1864700780572425e-05,
      "loss": 0.5364,
      "step": 1837
    },
    {
      "epoch": 0.3188757807078418,
      "grad_norm": 0.9523298144340515,
      "learning_rate": 3.188204683434519e-05,
      "loss": 0.683,
      "step": 1838
    },
    {
      "epoch": 0.3190492713393477,
      "grad_norm": 1.0407299995422363,
      "learning_rate": 3.1899392888117955e-05,
      "loss": 0.5481,
      "step": 1839
    },
    {
      "epoch": 0.3192227619708536,
      "grad_norm": 1.987041711807251,
      "learning_rate": 3.191673894189072e-05,
      "loss": 0.5403,
      "step": 1840
    },
    {
      "epoch": 0.3193962526023595,
      "grad_norm": 0.820220410823822,
      "learning_rate": 3.1934084995663485e-05,
      "loss": 0.6174,
      "step": 1841
    },
    {
      "epoch": 0.3195697432338654,
      "grad_norm": 0.73539137840271,
      "learning_rate": 3.1951431049436257e-05,
      "loss": 0.6836,
      "step": 1842
    },
    {
      "epoch": 0.3197432338653713,
      "grad_norm": 1.3756539821624756,
      "learning_rate": 3.196877710320902e-05,
      "loss": 0.7025,
      "step": 1843
    },
    {
      "epoch": 0.3199167244968772,
      "grad_norm": 0.6816329956054688,
      "learning_rate": 3.1986123156981787e-05,
      "loss": 0.6943,
      "step": 1844
    },
    {
      "epoch": 0.3200902151283831,
      "grad_norm": 0.7202216982841492,
      "learning_rate": 3.200346921075455e-05,
      "loss": 0.7222,
      "step": 1845
    },
    {
      "epoch": 0.32026370575988894,
      "grad_norm": 0.676460325717926,
      "learning_rate": 3.202081526452732e-05,
      "loss": 0.6368,
      "step": 1846
    },
    {
      "epoch": 0.32043719639139484,
      "grad_norm": 0.8167012333869934,
      "learning_rate": 3.203816131830009e-05,
      "loss": 0.7476,
      "step": 1847
    },
    {
      "epoch": 0.32061068702290074,
      "grad_norm": 0.8015779256820679,
      "learning_rate": 3.205550737207286e-05,
      "loss": 0.6105,
      "step": 1848
    },
    {
      "epoch": 0.32078417765440664,
      "grad_norm": 0.6521061658859253,
      "learning_rate": 3.2072853425845625e-05,
      "loss": 0.7544,
      "step": 1849
    },
    {
      "epoch": 0.32095766828591255,
      "grad_norm": 0.9332058429718018,
      "learning_rate": 3.209019947961839e-05,
      "loss": 0.5671,
      "step": 1850
    },
    {
      "epoch": 0.32113115891741845,
      "grad_norm": 0.8357752561569214,
      "learning_rate": 3.2107545533391155e-05,
      "loss": 0.6456,
      "step": 1851
    },
    {
      "epoch": 0.32130464954892435,
      "grad_norm": 0.7136110067367554,
      "learning_rate": 3.212489158716393e-05,
      "loss": 0.626,
      "step": 1852
    },
    {
      "epoch": 0.32147814018043025,
      "grad_norm": 0.8159222602844238,
      "learning_rate": 3.214223764093669e-05,
      "loss": 0.6195,
      "step": 1853
    },
    {
      "epoch": 0.32165163081193615,
      "grad_norm": 0.8828167915344238,
      "learning_rate": 3.215958369470946e-05,
      "loss": 0.5812,
      "step": 1854
    },
    {
      "epoch": 0.32182512144344205,
      "grad_norm": 0.844054102897644,
      "learning_rate": 3.217692974848222e-05,
      "loss": 0.7297,
      "step": 1855
    },
    {
      "epoch": 0.32199861207494795,
      "grad_norm": 0.9942353963851929,
      "learning_rate": 3.2194275802254994e-05,
      "loss": 0.6737,
      "step": 1856
    },
    {
      "epoch": 0.32217210270645386,
      "grad_norm": 0.8586241006851196,
      "learning_rate": 3.221162185602776e-05,
      "loss": 0.5699,
      "step": 1857
    },
    {
      "epoch": 0.32234559333795976,
      "grad_norm": 1.068901777267456,
      "learning_rate": 3.2228967909800524e-05,
      "loss": 0.6454,
      "step": 1858
    },
    {
      "epoch": 0.32251908396946566,
      "grad_norm": 0.94094318151474,
      "learning_rate": 3.224631396357329e-05,
      "loss": 0.6882,
      "step": 1859
    },
    {
      "epoch": 0.32269257460097156,
      "grad_norm": 1.031432032585144,
      "learning_rate": 3.2263660017346054e-05,
      "loss": 0.5964,
      "step": 1860
    },
    {
      "epoch": 0.32286606523247746,
      "grad_norm": 1.1422418355941772,
      "learning_rate": 3.2281006071118825e-05,
      "loss": 0.7263,
      "step": 1861
    },
    {
      "epoch": 0.32303955586398336,
      "grad_norm": 1.1631206274032593,
      "learning_rate": 3.229835212489159e-05,
      "loss": 0.665,
      "step": 1862
    },
    {
      "epoch": 0.32321304649548926,
      "grad_norm": 0.7793134450912476,
      "learning_rate": 3.2315698178664355e-05,
      "loss": 0.6418,
      "step": 1863
    },
    {
      "epoch": 0.32338653712699517,
      "grad_norm": 0.861473560333252,
      "learning_rate": 3.233304423243712e-05,
      "loss": 0.6372,
      "step": 1864
    },
    {
      "epoch": 0.32356002775850107,
      "grad_norm": 0.8410269618034363,
      "learning_rate": 3.235039028620989e-05,
      "loss": 0.5958,
      "step": 1865
    },
    {
      "epoch": 0.3237335183900069,
      "grad_norm": 0.71745365858078,
      "learning_rate": 3.236773633998266e-05,
      "loss": 0.5424,
      "step": 1866
    },
    {
      "epoch": 0.3239070090215128,
      "grad_norm": 0.6829073429107666,
      "learning_rate": 3.238508239375542e-05,
      "loss": 0.7207,
      "step": 1867
    },
    {
      "epoch": 0.3240804996530187,
      "grad_norm": 0.9286072254180908,
      "learning_rate": 3.240242844752819e-05,
      "loss": 0.6857,
      "step": 1868
    },
    {
      "epoch": 0.3242539902845246,
      "grad_norm": 0.9504925608634949,
      "learning_rate": 3.241977450130096e-05,
      "loss": 0.5566,
      "step": 1869
    },
    {
      "epoch": 0.3244274809160305,
      "grad_norm": 0.8263635635375977,
      "learning_rate": 3.2437120555073724e-05,
      "loss": 0.7421,
      "step": 1870
    },
    {
      "epoch": 0.3246009715475364,
      "grad_norm": 1.0866979360580444,
      "learning_rate": 3.245446660884649e-05,
      "loss": 0.5236,
      "step": 1871
    },
    {
      "epoch": 0.3247744621790423,
      "grad_norm": 1.151369571685791,
      "learning_rate": 3.2471812662619254e-05,
      "loss": 0.6591,
      "step": 1872
    },
    {
      "epoch": 0.3249479528105482,
      "grad_norm": 1.141158938407898,
      "learning_rate": 3.248915871639202e-05,
      "loss": 0.7085,
      "step": 1873
    },
    {
      "epoch": 0.3251214434420541,
      "grad_norm": 0.7256845235824585,
      "learning_rate": 3.250650477016479e-05,
      "loss": 0.5247,
      "step": 1874
    },
    {
      "epoch": 0.32529493407356,
      "grad_norm": 0.6779228448867798,
      "learning_rate": 3.2523850823937556e-05,
      "loss": 0.686,
      "step": 1875
    },
    {
      "epoch": 0.32546842470506593,
      "grad_norm": 0.6428731679916382,
      "learning_rate": 3.254119687771032e-05,
      "loss": 0.6531,
      "step": 1876
    },
    {
      "epoch": 0.32564191533657183,
      "grad_norm": 0.7823044061660767,
      "learning_rate": 3.2558542931483086e-05,
      "loss": 0.621,
      "step": 1877
    },
    {
      "epoch": 0.32581540596807773,
      "grad_norm": 1.2496860027313232,
      "learning_rate": 3.257588898525586e-05,
      "loss": 0.6849,
      "step": 1878
    },
    {
      "epoch": 0.32598889659958363,
      "grad_norm": 0.825401246547699,
      "learning_rate": 3.259323503902862e-05,
      "loss": 0.5262,
      "step": 1879
    },
    {
      "epoch": 0.32616238723108953,
      "grad_norm": 0.8637552261352539,
      "learning_rate": 3.2610581092801394e-05,
      "loss": 0.6127,
      "step": 1880
    },
    {
      "epoch": 0.32633587786259544,
      "grad_norm": 0.5828275084495544,
      "learning_rate": 3.262792714657416e-05,
      "loss": 0.6902,
      "step": 1881
    },
    {
      "epoch": 0.32650936849410134,
      "grad_norm": 1.9262351989746094,
      "learning_rate": 3.2645273200346924e-05,
      "loss": 0.6202,
      "step": 1882
    },
    {
      "epoch": 0.32668285912560724,
      "grad_norm": 0.9045750498771667,
      "learning_rate": 3.266261925411969e-05,
      "loss": 0.6622,
      "step": 1883
    },
    {
      "epoch": 0.32685634975711314,
      "grad_norm": 0.6373817920684814,
      "learning_rate": 3.267996530789246e-05,
      "loss": 0.7363,
      "step": 1884
    },
    {
      "epoch": 0.32702984038861904,
      "grad_norm": 0.6378402709960938,
      "learning_rate": 3.2697311361665226e-05,
      "loss": 0.6084,
      "step": 1885
    },
    {
      "epoch": 0.3272033310201249,
      "grad_norm": 0.9899991750717163,
      "learning_rate": 3.271465741543799e-05,
      "loss": 0.6056,
      "step": 1886
    },
    {
      "epoch": 0.3273768216516308,
      "grad_norm": 0.7649429440498352,
      "learning_rate": 3.2732003469210756e-05,
      "loss": 0.5872,
      "step": 1887
    },
    {
      "epoch": 0.3275503122831367,
      "grad_norm": 0.9158859252929688,
      "learning_rate": 3.274934952298353e-05,
      "loss": 0.5853,
      "step": 1888
    },
    {
      "epoch": 0.3277238029146426,
      "grad_norm": 0.8275113701820374,
      "learning_rate": 3.276669557675629e-05,
      "loss": 0.6613,
      "step": 1889
    },
    {
      "epoch": 0.3278972935461485,
      "grad_norm": 0.880486011505127,
      "learning_rate": 3.278404163052906e-05,
      "loss": 0.561,
      "step": 1890
    },
    {
      "epoch": 0.3280707841776544,
      "grad_norm": 0.7612427473068237,
      "learning_rate": 3.280138768430182e-05,
      "loss": 0.6125,
      "step": 1891
    },
    {
      "epoch": 0.3282442748091603,
      "grad_norm": 0.7106688022613525,
      "learning_rate": 3.281873373807459e-05,
      "loss": 0.743,
      "step": 1892
    },
    {
      "epoch": 0.3284177654406662,
      "grad_norm": 0.831951379776001,
      "learning_rate": 3.283607979184736e-05,
      "loss": 0.5771,
      "step": 1893
    },
    {
      "epoch": 0.3285912560721721,
      "grad_norm": 1.737910270690918,
      "learning_rate": 3.2853425845620125e-05,
      "loss": 0.6971,
      "step": 1894
    },
    {
      "epoch": 0.328764746703678,
      "grad_norm": 0.9117143750190735,
      "learning_rate": 3.287077189939289e-05,
      "loss": 0.5634,
      "step": 1895
    },
    {
      "epoch": 0.3289382373351839,
      "grad_norm": 0.7156542539596558,
      "learning_rate": 3.2888117953165655e-05,
      "loss": 0.6713,
      "step": 1896
    },
    {
      "epoch": 0.3291117279666898,
      "grad_norm": 0.9128335118293762,
      "learning_rate": 3.2905464006938426e-05,
      "loss": 0.5865,
      "step": 1897
    },
    {
      "epoch": 0.3292852185981957,
      "grad_norm": 0.9140109419822693,
      "learning_rate": 3.292281006071119e-05,
      "loss": 0.5771,
      "step": 1898
    },
    {
      "epoch": 0.3294587092297016,
      "grad_norm": 0.8033643364906311,
      "learning_rate": 3.2940156114483956e-05,
      "loss": 0.528,
      "step": 1899
    },
    {
      "epoch": 0.3296321998612075,
      "grad_norm": 1.2055332660675049,
      "learning_rate": 3.295750216825672e-05,
      "loss": 0.52,
      "step": 1900
    },
    {
      "epoch": 0.3298056904927134,
      "grad_norm": 0.8967594504356384,
      "learning_rate": 3.2974848222029487e-05,
      "loss": 0.637,
      "step": 1901
    },
    {
      "epoch": 0.3299791811242193,
      "grad_norm": 0.7392427325248718,
      "learning_rate": 3.299219427580226e-05,
      "loss": 0.5469,
      "step": 1902
    },
    {
      "epoch": 0.3301526717557252,
      "grad_norm": 0.6878560185432434,
      "learning_rate": 3.300954032957502e-05,
      "loss": 0.676,
      "step": 1903
    },
    {
      "epoch": 0.3303261623872311,
      "grad_norm": 0.8250359296798706,
      "learning_rate": 3.302688638334779e-05,
      "loss": 0.6309,
      "step": 1904
    },
    {
      "epoch": 0.33049965301873696,
      "grad_norm": 0.9844671487808228,
      "learning_rate": 3.304423243712055e-05,
      "loss": 0.541,
      "step": 1905
    },
    {
      "epoch": 0.33067314365024286,
      "grad_norm": 1.5574495792388916,
      "learning_rate": 3.3061578490893325e-05,
      "loss": 0.7218,
      "step": 1906
    },
    {
      "epoch": 0.33084663428174876,
      "grad_norm": 0.6480480432510376,
      "learning_rate": 3.307892454466609e-05,
      "loss": 0.8379,
      "step": 1907
    },
    {
      "epoch": 0.33102012491325467,
      "grad_norm": 0.8829315304756165,
      "learning_rate": 3.309627059843886e-05,
      "loss": 0.6604,
      "step": 1908
    },
    {
      "epoch": 0.33119361554476057,
      "grad_norm": 0.7221475839614868,
      "learning_rate": 3.311361665221163e-05,
      "loss": 0.6022,
      "step": 1909
    },
    {
      "epoch": 0.33136710617626647,
      "grad_norm": 0.7198967933654785,
      "learning_rate": 3.313096270598439e-05,
      "loss": 0.5106,
      "step": 1910
    },
    {
      "epoch": 0.33154059680777237,
      "grad_norm": 1.0678820610046387,
      "learning_rate": 3.314830875975716e-05,
      "loss": 0.5536,
      "step": 1911
    },
    {
      "epoch": 0.33171408743927827,
      "grad_norm": 0.940387487411499,
      "learning_rate": 3.316565481352993e-05,
      "loss": 0.6127,
      "step": 1912
    },
    {
      "epoch": 0.3318875780707842,
      "grad_norm": 1.3829164505004883,
      "learning_rate": 3.3183000867302694e-05,
      "loss": 0.6375,
      "step": 1913
    },
    {
      "epoch": 0.3320610687022901,
      "grad_norm": 0.8988991975784302,
      "learning_rate": 3.320034692107546e-05,
      "loss": 0.6748,
      "step": 1914
    },
    {
      "epoch": 0.332234559333796,
      "grad_norm": 0.6385945677757263,
      "learning_rate": 3.3217692974848224e-05,
      "loss": 0.6494,
      "step": 1915
    },
    {
      "epoch": 0.3324080499653019,
      "grad_norm": 0.7207217812538147,
      "learning_rate": 3.3235039028620995e-05,
      "loss": 0.6686,
      "step": 1916
    },
    {
      "epoch": 0.3325815405968078,
      "grad_norm": 0.9134792685508728,
      "learning_rate": 3.325238508239376e-05,
      "loss": 0.6448,
      "step": 1917
    },
    {
      "epoch": 0.3327550312283137,
      "grad_norm": 0.9056723713874817,
      "learning_rate": 3.3269731136166525e-05,
      "loss": 0.7886,
      "step": 1918
    },
    {
      "epoch": 0.3329285218598196,
      "grad_norm": 0.916094183921814,
      "learning_rate": 3.328707718993929e-05,
      "loss": 0.7588,
      "step": 1919
    },
    {
      "epoch": 0.3331020124913255,
      "grad_norm": 0.7988206744194031,
      "learning_rate": 3.3304423243712055e-05,
      "loss": 0.6619,
      "step": 1920
    },
    {
      "epoch": 0.3332755031228314,
      "grad_norm": 0.7587366104125977,
      "learning_rate": 3.332176929748483e-05,
      "loss": 0.6594,
      "step": 1921
    },
    {
      "epoch": 0.3334489937543373,
      "grad_norm": 0.9271337985992432,
      "learning_rate": 3.333911535125759e-05,
      "loss": 0.5983,
      "step": 1922
    },
    {
      "epoch": 0.3336224843858432,
      "grad_norm": 0.8258846998214722,
      "learning_rate": 3.335646140503036e-05,
      "loss": 0.6704,
      "step": 1923
    },
    {
      "epoch": 0.3337959750173491,
      "grad_norm": 0.9814427495002747,
      "learning_rate": 3.337380745880312e-05,
      "loss": 0.5504,
      "step": 1924
    },
    {
      "epoch": 0.33396946564885494,
      "grad_norm": 0.7927559614181519,
      "learning_rate": 3.3391153512575894e-05,
      "loss": 0.5997,
      "step": 1925
    },
    {
      "epoch": 0.33414295628036084,
      "grad_norm": 1.139303207397461,
      "learning_rate": 3.340849956634866e-05,
      "loss": 0.7278,
      "step": 1926
    },
    {
      "epoch": 0.33431644691186674,
      "grad_norm": 0.993246853351593,
      "learning_rate": 3.3425845620121424e-05,
      "loss": 0.6042,
      "step": 1927
    },
    {
      "epoch": 0.33448993754337264,
      "grad_norm": 1.020550012588501,
      "learning_rate": 3.344319167389419e-05,
      "loss": 0.5493,
      "step": 1928
    },
    {
      "epoch": 0.33466342817487854,
      "grad_norm": 0.8157418370246887,
      "learning_rate": 3.346053772766696e-05,
      "loss": 0.6833,
      "step": 1929
    },
    {
      "epoch": 0.33483691880638444,
      "grad_norm": 0.8034759759902954,
      "learning_rate": 3.3477883781439726e-05,
      "loss": 0.707,
      "step": 1930
    },
    {
      "epoch": 0.33501040943789034,
      "grad_norm": 0.6415484547615051,
      "learning_rate": 3.349522983521249e-05,
      "loss": 0.7014,
      "step": 1931
    },
    {
      "epoch": 0.33518390006939625,
      "grad_norm": 0.9110195636749268,
      "learning_rate": 3.3512575888985256e-05,
      "loss": 0.663,
      "step": 1932
    },
    {
      "epoch": 0.33535739070090215,
      "grad_norm": 3.435911178588867,
      "learning_rate": 3.352992194275802e-05,
      "loss": 0.5925,
      "step": 1933
    },
    {
      "epoch": 0.33553088133240805,
      "grad_norm": 0.6798315644264221,
      "learning_rate": 3.354726799653079e-05,
      "loss": 0.7164,
      "step": 1934
    },
    {
      "epoch": 0.33570437196391395,
      "grad_norm": 1.6037698984146118,
      "learning_rate": 3.356461405030356e-05,
      "loss": 0.6107,
      "step": 1935
    },
    {
      "epoch": 0.33587786259541985,
      "grad_norm": 0.7971057295799255,
      "learning_rate": 3.358196010407632e-05,
      "loss": 0.7469,
      "step": 1936
    },
    {
      "epoch": 0.33605135322692575,
      "grad_norm": 1.0822875499725342,
      "learning_rate": 3.3599306157849094e-05,
      "loss": 0.5383,
      "step": 1937
    },
    {
      "epoch": 0.33622484385843165,
      "grad_norm": 0.7857166528701782,
      "learning_rate": 3.361665221162186e-05,
      "loss": 0.7449,
      "step": 1938
    },
    {
      "epoch": 0.33639833448993756,
      "grad_norm": 0.6329569816589355,
      "learning_rate": 3.3633998265394624e-05,
      "loss": 0.7579,
      "step": 1939
    },
    {
      "epoch": 0.33657182512144346,
      "grad_norm": 0.7199830412864685,
      "learning_rate": 3.3651344319167396e-05,
      "loss": 0.7336,
      "step": 1940
    },
    {
      "epoch": 0.33674531575294936,
      "grad_norm": 0.8670171499252319,
      "learning_rate": 3.366869037294016e-05,
      "loss": 0.6079,
      "step": 1941
    },
    {
      "epoch": 0.33691880638445526,
      "grad_norm": 0.7961511611938477,
      "learning_rate": 3.3686036426712926e-05,
      "loss": 0.5454,
      "step": 1942
    },
    {
      "epoch": 0.33709229701596116,
      "grad_norm": 0.8177798986434937,
      "learning_rate": 3.370338248048569e-05,
      "loss": 0.6361,
      "step": 1943
    },
    {
      "epoch": 0.33726578764746706,
      "grad_norm": 0.8449214696884155,
      "learning_rate": 3.372072853425846e-05,
      "loss": 0.7,
      "step": 1944
    },
    {
      "epoch": 0.3374392782789729,
      "grad_norm": 1.0306763648986816,
      "learning_rate": 3.373807458803123e-05,
      "loss": 0.6237,
      "step": 1945
    },
    {
      "epoch": 0.3376127689104788,
      "grad_norm": 0.7724658250808716,
      "learning_rate": 3.375542064180399e-05,
      "loss": 0.6016,
      "step": 1946
    },
    {
      "epoch": 0.3377862595419847,
      "grad_norm": 0.8948357701301575,
      "learning_rate": 3.377276669557676e-05,
      "loss": 0.7991,
      "step": 1947
    },
    {
      "epoch": 0.3379597501734906,
      "grad_norm": 0.7655358910560608,
      "learning_rate": 3.379011274934952e-05,
      "loss": 0.6937,
      "step": 1948
    },
    {
      "epoch": 0.3381332408049965,
      "grad_norm": 0.8109301328659058,
      "learning_rate": 3.3807458803122295e-05,
      "loss": 0.6616,
      "step": 1949
    },
    {
      "epoch": 0.3383067314365024,
      "grad_norm": 0.8285800218582153,
      "learning_rate": 3.382480485689506e-05,
      "loss": 0.7268,
      "step": 1950
    },
    {
      "epoch": 0.3384802220680083,
      "grad_norm": 0.7333600521087646,
      "learning_rate": 3.3842150910667825e-05,
      "loss": 0.6385,
      "step": 1951
    },
    {
      "epoch": 0.3386537126995142,
      "grad_norm": 0.5980129241943359,
      "learning_rate": 3.385949696444059e-05,
      "loss": 0.7239,
      "step": 1952
    },
    {
      "epoch": 0.3388272033310201,
      "grad_norm": 2.1083834171295166,
      "learning_rate": 3.387684301821336e-05,
      "loss": 0.6746,
      "step": 1953
    },
    {
      "epoch": 0.339000693962526,
      "grad_norm": 0.774154543876648,
      "learning_rate": 3.3894189071986126e-05,
      "loss": 0.671,
      "step": 1954
    },
    {
      "epoch": 0.3391741845940319,
      "grad_norm": 0.8086047172546387,
      "learning_rate": 3.391153512575889e-05,
      "loss": 0.8333,
      "step": 1955
    },
    {
      "epoch": 0.3393476752255378,
      "grad_norm": 0.9128818511962891,
      "learning_rate": 3.3928881179531656e-05,
      "loss": 0.6677,
      "step": 1956
    },
    {
      "epoch": 0.3395211658570437,
      "grad_norm": 0.8383493423461914,
      "learning_rate": 3.394622723330443e-05,
      "loss": 0.6716,
      "step": 1957
    },
    {
      "epoch": 0.33969465648854963,
      "grad_norm": 2.182744264602661,
      "learning_rate": 3.396357328707719e-05,
      "loss": 0.6908,
      "step": 1958
    },
    {
      "epoch": 0.33986814712005553,
      "grad_norm": 0.8761880397796631,
      "learning_rate": 3.398091934084996e-05,
      "loss": 0.6628,
      "step": 1959
    },
    {
      "epoch": 0.34004163775156143,
      "grad_norm": 1.249531865119934,
      "learning_rate": 3.399826539462272e-05,
      "loss": 0.6277,
      "step": 1960
    },
    {
      "epoch": 0.34021512838306733,
      "grad_norm": 1.0625807046890259,
      "learning_rate": 3.401561144839549e-05,
      "loss": 0.7263,
      "step": 1961
    },
    {
      "epoch": 0.34038861901457323,
      "grad_norm": 0.9543647766113281,
      "learning_rate": 3.403295750216826e-05,
      "loss": 0.7351,
      "step": 1962
    },
    {
      "epoch": 0.34056210964607914,
      "grad_norm": 0.7973102927207947,
      "learning_rate": 3.4050303555941025e-05,
      "loss": 0.5649,
      "step": 1963
    },
    {
      "epoch": 0.34073560027758504,
      "grad_norm": 1.3753342628479004,
      "learning_rate": 3.406764960971379e-05,
      "loss": 0.5652,
      "step": 1964
    },
    {
      "epoch": 0.3409090909090909,
      "grad_norm": 1.9109654426574707,
      "learning_rate": 3.4084995663486555e-05,
      "loss": 0.5186,
      "step": 1965
    },
    {
      "epoch": 0.3410825815405968,
      "grad_norm": 1.8508704900741577,
      "learning_rate": 3.410234171725933e-05,
      "loss": 0.5887,
      "step": 1966
    },
    {
      "epoch": 0.3412560721721027,
      "grad_norm": 0.8658906817436218,
      "learning_rate": 3.411968777103209e-05,
      "loss": 0.7981,
      "step": 1967
    },
    {
      "epoch": 0.3414295628036086,
      "grad_norm": 1.0544072389602661,
      "learning_rate": 3.4137033824804864e-05,
      "loss": 0.5813,
      "step": 1968
    },
    {
      "epoch": 0.3416030534351145,
      "grad_norm": 1.6280932426452637,
      "learning_rate": 3.415437987857763e-05,
      "loss": 0.7124,
      "step": 1969
    },
    {
      "epoch": 0.3417765440666204,
      "grad_norm": 1.027541995048523,
      "learning_rate": 3.4171725932350394e-05,
      "loss": 0.6947,
      "step": 1970
    },
    {
      "epoch": 0.3419500346981263,
      "grad_norm": 1.1156625747680664,
      "learning_rate": 3.418907198612316e-05,
      "loss": 0.6039,
      "step": 1971
    },
    {
      "epoch": 0.3421235253296322,
      "grad_norm": 1.0394290685653687,
      "learning_rate": 3.420641803989593e-05,
      "loss": 0.7434,
      "step": 1972
    },
    {
      "epoch": 0.3422970159611381,
      "grad_norm": 0.9105010628700256,
      "learning_rate": 3.4223764093668695e-05,
      "loss": 0.6143,
      "step": 1973
    },
    {
      "epoch": 0.342470506592644,
      "grad_norm": 0.7156486511230469,
      "learning_rate": 3.424111014744146e-05,
      "loss": 0.6757,
      "step": 1974
    },
    {
      "epoch": 0.3426439972241499,
      "grad_norm": 0.7882575392723083,
      "learning_rate": 3.4258456201214225e-05,
      "loss": 0.5669,
      "step": 1975
    },
    {
      "epoch": 0.3428174878556558,
      "grad_norm": 0.7309373617172241,
      "learning_rate": 3.4275802254987e-05,
      "loss": 0.6183,
      "step": 1976
    },
    {
      "epoch": 0.3429909784871617,
      "grad_norm": 1.8437594175338745,
      "learning_rate": 3.429314830875976e-05,
      "loss": 0.6377,
      "step": 1977
    },
    {
      "epoch": 0.3431644691186676,
      "grad_norm": 1.0819367170333862,
      "learning_rate": 3.431049436253253e-05,
      "loss": 0.6041,
      "step": 1978
    },
    {
      "epoch": 0.3433379597501735,
      "grad_norm": 0.8812955617904663,
      "learning_rate": 3.432784041630529e-05,
      "loss": 0.4973,
      "step": 1979
    },
    {
      "epoch": 0.3435114503816794,
      "grad_norm": 0.6564153432846069,
      "learning_rate": 3.434518647007806e-05,
      "loss": 0.7019,
      "step": 1980
    },
    {
      "epoch": 0.3436849410131853,
      "grad_norm": 1.0282037258148193,
      "learning_rate": 3.436253252385083e-05,
      "loss": 0.658,
      "step": 1981
    },
    {
      "epoch": 0.3438584316446912,
      "grad_norm": 0.7225372195243835,
      "learning_rate": 3.4379878577623594e-05,
      "loss": 0.6141,
      "step": 1982
    },
    {
      "epoch": 0.3440319222761971,
      "grad_norm": 0.6336444616317749,
      "learning_rate": 3.439722463139636e-05,
      "loss": 0.7566,
      "step": 1983
    },
    {
      "epoch": 0.34420541290770296,
      "grad_norm": 0.9767608046531677,
      "learning_rate": 3.4414570685169124e-05,
      "loss": 0.6304,
      "step": 1984
    },
    {
      "epoch": 0.34437890353920886,
      "grad_norm": 1.0240918397903442,
      "learning_rate": 3.4431916738941896e-05,
      "loss": 0.6415,
      "step": 1985
    },
    {
      "epoch": 0.34455239417071476,
      "grad_norm": 0.7765363454818726,
      "learning_rate": 3.444926279271466e-05,
      "loss": 0.6307,
      "step": 1986
    },
    {
      "epoch": 0.34472588480222066,
      "grad_norm": 0.6897956728935242,
      "learning_rate": 3.4466608846487426e-05,
      "loss": 0.6045,
      "step": 1987
    },
    {
      "epoch": 0.34489937543372656,
      "grad_norm": 0.7995160222053528,
      "learning_rate": 3.448395490026019e-05,
      "loss": 0.6552,
      "step": 1988
    },
    {
      "epoch": 0.34507286606523246,
      "grad_norm": 0.7897160649299622,
      "learning_rate": 3.4501300954032956e-05,
      "loss": 0.6372,
      "step": 1989
    },
    {
      "epoch": 0.34524635669673837,
      "grad_norm": 0.9942298531532288,
      "learning_rate": 3.451864700780573e-05,
      "loss": 0.6865,
      "step": 1990
    },
    {
      "epoch": 0.34541984732824427,
      "grad_norm": 0.7617659568786621,
      "learning_rate": 3.453599306157849e-05,
      "loss": 0.7029,
      "step": 1991
    },
    {
      "epoch": 0.34559333795975017,
      "grad_norm": 0.9708228707313538,
      "learning_rate": 3.455333911535126e-05,
      "loss": 0.6744,
      "step": 1992
    },
    {
      "epoch": 0.34576682859125607,
      "grad_norm": 0.9564141035079956,
      "learning_rate": 3.457068516912402e-05,
      "loss": 0.6765,
      "step": 1993
    },
    {
      "epoch": 0.34594031922276197,
      "grad_norm": 1.037091851234436,
      "learning_rate": 3.4588031222896794e-05,
      "loss": 0.702,
      "step": 1994
    },
    {
      "epoch": 0.3461138098542679,
      "grad_norm": 0.8068506717681885,
      "learning_rate": 3.460537727666956e-05,
      "loss": 0.6803,
      "step": 1995
    },
    {
      "epoch": 0.3462873004857738,
      "grad_norm": 0.7901633381843567,
      "learning_rate": 3.462272333044233e-05,
      "loss": 0.6143,
      "step": 1996
    },
    {
      "epoch": 0.3464607911172797,
      "grad_norm": 2.00464129447937,
      "learning_rate": 3.4640069384215096e-05,
      "loss": 0.5975,
      "step": 1997
    },
    {
      "epoch": 0.3466342817487856,
      "grad_norm": 0.9880556464195251,
      "learning_rate": 3.465741543798786e-05,
      "loss": 0.5653,
      "step": 1998
    },
    {
      "epoch": 0.3468077723802915,
      "grad_norm": 0.989421010017395,
      "learning_rate": 3.4674761491760626e-05,
      "loss": 0.6676,
      "step": 1999
    },
    {
      "epoch": 0.3469812630117974,
      "grad_norm": 0.8350646495819092,
      "learning_rate": 3.46921075455334e-05,
      "loss": 0.5137,
      "step": 2000
    },
    {
      "epoch": 0.3471547536433033,
      "grad_norm": 1.1387163400650024,
      "learning_rate": 3.470945359930616e-05,
      "loss": 0.5483,
      "step": 2001
    },
    {
      "epoch": 0.3473282442748092,
      "grad_norm": 2.10577392578125,
      "learning_rate": 3.472679965307893e-05,
      "loss": 0.6202,
      "step": 2002
    },
    {
      "epoch": 0.3475017349063151,
      "grad_norm": 0.8258737921714783,
      "learning_rate": 3.474414570685169e-05,
      "loss": 0.6392,
      "step": 2003
    },
    {
      "epoch": 0.34767522553782093,
      "grad_norm": 1.2150181531906128,
      "learning_rate": 3.4761491760624465e-05,
      "loss": 0.7468,
      "step": 2004
    },
    {
      "epoch": 0.34784871616932683,
      "grad_norm": 0.7214828729629517,
      "learning_rate": 3.477883781439723e-05,
      "loss": 0.6019,
      "step": 2005
    },
    {
      "epoch": 0.34802220680083273,
      "grad_norm": 0.6550180315971375,
      "learning_rate": 3.4796183868169995e-05,
      "loss": 0.7052,
      "step": 2006
    },
    {
      "epoch": 0.34819569743233864,
      "grad_norm": 2.2414355278015137,
      "learning_rate": 3.481352992194276e-05,
      "loss": 0.6088,
      "step": 2007
    },
    {
      "epoch": 0.34836918806384454,
      "grad_norm": 0.8592984676361084,
      "learning_rate": 3.4830875975715525e-05,
      "loss": 0.6058,
      "step": 2008
    },
    {
      "epoch": 0.34854267869535044,
      "grad_norm": 0.7922565937042236,
      "learning_rate": 3.4848222029488296e-05,
      "loss": 0.7596,
      "step": 2009
    },
    {
      "epoch": 0.34871616932685634,
      "grad_norm": 0.7546785473823547,
      "learning_rate": 3.486556808326106e-05,
      "loss": 0.6936,
      "step": 2010
    },
    {
      "epoch": 0.34888965995836224,
      "grad_norm": 0.831555187702179,
      "learning_rate": 3.4882914137033826e-05,
      "loss": 0.5795,
      "step": 2011
    },
    {
      "epoch": 0.34906315058986814,
      "grad_norm": 0.9311597943305969,
      "learning_rate": 3.490026019080659e-05,
      "loss": 0.6919,
      "step": 2012
    },
    {
      "epoch": 0.34923664122137404,
      "grad_norm": 0.9510216116905212,
      "learning_rate": 3.491760624457936e-05,
      "loss": 0.5664,
      "step": 2013
    },
    {
      "epoch": 0.34941013185287995,
      "grad_norm": 0.7857622504234314,
      "learning_rate": 3.493495229835213e-05,
      "loss": 0.6205,
      "step": 2014
    },
    {
      "epoch": 0.34958362248438585,
      "grad_norm": 1.2958338260650635,
      "learning_rate": 3.495229835212489e-05,
      "loss": 0.5803,
      "step": 2015
    },
    {
      "epoch": 0.34975711311589175,
      "grad_norm": 0.7746779918670654,
      "learning_rate": 3.496964440589766e-05,
      "loss": 0.6233,
      "step": 2016
    },
    {
      "epoch": 0.34993060374739765,
      "grad_norm": 0.6474652886390686,
      "learning_rate": 3.498699045967043e-05,
      "loss": 0.739,
      "step": 2017
    },
    {
      "epoch": 0.35010409437890355,
      "grad_norm": 0.7284134030342102,
      "learning_rate": 3.5004336513443195e-05,
      "loss": 0.8052,
      "step": 2018
    },
    {
      "epoch": 0.35027758501040945,
      "grad_norm": 0.6557030081748962,
      "learning_rate": 3.502168256721596e-05,
      "loss": 0.6912,
      "step": 2019
    },
    {
      "epoch": 0.35045107564191535,
      "grad_norm": 0.8244677782058716,
      "learning_rate": 3.5039028620988725e-05,
      "loss": 0.6118,
      "step": 2020
    },
    {
      "epoch": 0.35062456627342126,
      "grad_norm": 1.0263489484786987,
      "learning_rate": 3.505637467476149e-05,
      "loss": 0.5574,
      "step": 2021
    },
    {
      "epoch": 0.35079805690492716,
      "grad_norm": 1.0217803716659546,
      "learning_rate": 3.507372072853426e-05,
      "loss": 0.5685,
      "step": 2022
    },
    {
      "epoch": 0.35097154753643306,
      "grad_norm": 3.0454044342041016,
      "learning_rate": 3.509106678230703e-05,
      "loss": 0.5991,
      "step": 2023
    },
    {
      "epoch": 0.3511450381679389,
      "grad_norm": 0.8422743082046509,
      "learning_rate": 3.510841283607979e-05,
      "loss": 0.6758,
      "step": 2024
    },
    {
      "epoch": 0.3513185287994448,
      "grad_norm": 0.6514055728912354,
      "learning_rate": 3.512575888985256e-05,
      "loss": 0.7083,
      "step": 2025
    },
    {
      "epoch": 0.3514920194309507,
      "grad_norm": 0.6344184875488281,
      "learning_rate": 3.514310494362533e-05,
      "loss": 0.7272,
      "step": 2026
    },
    {
      "epoch": 0.3516655100624566,
      "grad_norm": 0.5242217183113098,
      "learning_rate": 3.5160450997398094e-05,
      "loss": 0.7678,
      "step": 2027
    },
    {
      "epoch": 0.3518390006939625,
      "grad_norm": 1.2077232599258423,
      "learning_rate": 3.5177797051170865e-05,
      "loss": 0.595,
      "step": 2028
    },
    {
      "epoch": 0.3520124913254684,
      "grad_norm": 0.8761990666389465,
      "learning_rate": 3.519514310494363e-05,
      "loss": 0.5796,
      "step": 2029
    },
    {
      "epoch": 0.3521859819569743,
      "grad_norm": 1.412538766860962,
      "learning_rate": 3.5212489158716395e-05,
      "loss": 0.6929,
      "step": 2030
    },
    {
      "epoch": 0.3523594725884802,
      "grad_norm": 0.9181896448135376,
      "learning_rate": 3.522983521248916e-05,
      "loss": 0.5657,
      "step": 2031
    },
    {
      "epoch": 0.3525329632199861,
      "grad_norm": 0.7443928718566895,
      "learning_rate": 3.524718126626193e-05,
      "loss": 0.6014,
      "step": 2032
    },
    {
      "epoch": 0.352706453851492,
      "grad_norm": 0.6138537526130676,
      "learning_rate": 3.52645273200347e-05,
      "loss": 0.7174,
      "step": 2033
    },
    {
      "epoch": 0.3528799444829979,
      "grad_norm": 1.006217360496521,
      "learning_rate": 3.528187337380746e-05,
      "loss": 0.549,
      "step": 2034
    },
    {
      "epoch": 0.3530534351145038,
      "grad_norm": 0.8398961424827576,
      "learning_rate": 3.529921942758023e-05,
      "loss": 0.6268,
      "step": 2035
    },
    {
      "epoch": 0.3532269257460097,
      "grad_norm": 0.6724855303764343,
      "learning_rate": 3.5316565481353e-05,
      "loss": 0.6564,
      "step": 2036
    },
    {
      "epoch": 0.3534004163775156,
      "grad_norm": 0.8726453185081482,
      "learning_rate": 3.5333911535125764e-05,
      "loss": 0.6663,
      "step": 2037
    },
    {
      "epoch": 0.3535739070090215,
      "grad_norm": 1.2279284000396729,
      "learning_rate": 3.535125758889853e-05,
      "loss": 0.6401,
      "step": 2038
    },
    {
      "epoch": 0.35374739764052743,
      "grad_norm": 0.6089798808097839,
      "learning_rate": 3.5368603642671294e-05,
      "loss": 0.6617,
      "step": 2039
    },
    {
      "epoch": 0.35392088827203333,
      "grad_norm": 0.7994186878204346,
      "learning_rate": 3.538594969644406e-05,
      "loss": 0.632,
      "step": 2040
    },
    {
      "epoch": 0.35409437890353923,
      "grad_norm": 0.7627372145652771,
      "learning_rate": 3.540329575021683e-05,
      "loss": 0.5959,
      "step": 2041
    },
    {
      "epoch": 0.35426786953504513,
      "grad_norm": 0.6806637644767761,
      "learning_rate": 3.5420641803989596e-05,
      "loss": 0.6321,
      "step": 2042
    },
    {
      "epoch": 0.35444136016655103,
      "grad_norm": 0.8860881924629211,
      "learning_rate": 3.543798785776236e-05,
      "loss": 0.6539,
      "step": 2043
    },
    {
      "epoch": 0.3546148507980569,
      "grad_norm": 1.116827368736267,
      "learning_rate": 3.5455333911535126e-05,
      "loss": 0.694,
      "step": 2044
    },
    {
      "epoch": 0.3547883414295628,
      "grad_norm": 1.0090057849884033,
      "learning_rate": 3.54726799653079e-05,
      "loss": 0.6637,
      "step": 2045
    },
    {
      "epoch": 0.3549618320610687,
      "grad_norm": 0.9741466641426086,
      "learning_rate": 3.549002601908066e-05,
      "loss": 0.6384,
      "step": 2046
    },
    {
      "epoch": 0.3551353226925746,
      "grad_norm": 0.7068755030632019,
      "learning_rate": 3.550737207285343e-05,
      "loss": 0.6006,
      "step": 2047
    },
    {
      "epoch": 0.3553088133240805,
      "grad_norm": 1.0879136323928833,
      "learning_rate": 3.552471812662619e-05,
      "loss": 0.6957,
      "step": 2048
    },
    {
      "epoch": 0.3554823039555864,
      "grad_norm": 0.9993870854377747,
      "learning_rate": 3.554206418039896e-05,
      "loss": 0.6057,
      "step": 2049
    },
    {
      "epoch": 0.3556557945870923,
      "grad_norm": 0.7793830037117004,
      "learning_rate": 3.555941023417173e-05,
      "loss": 0.6934,
      "step": 2050
    },
    {
      "epoch": 0.3558292852185982,
      "grad_norm": 1.8161706924438477,
      "learning_rate": 3.5576756287944494e-05,
      "loss": 0.7069,
      "step": 2051
    },
    {
      "epoch": 0.3560027758501041,
      "grad_norm": 0.8136652112007141,
      "learning_rate": 3.559410234171726e-05,
      "loss": 0.5745,
      "step": 2052
    },
    {
      "epoch": 0.35617626648161,
      "grad_norm": 0.7486060857772827,
      "learning_rate": 3.5611448395490024e-05,
      "loss": 0.5691,
      "step": 2053
    },
    {
      "epoch": 0.3563497571131159,
      "grad_norm": 1.0805221796035767,
      "learning_rate": 3.5628794449262796e-05,
      "loss": 0.595,
      "step": 2054
    },
    {
      "epoch": 0.3565232477446218,
      "grad_norm": 0.8663496971130371,
      "learning_rate": 3.564614050303557e-05,
      "loss": 0.5597,
      "step": 2055
    },
    {
      "epoch": 0.3566967383761277,
      "grad_norm": 1.3997595310211182,
      "learning_rate": 3.566348655680833e-05,
      "loss": 0.5537,
      "step": 2056
    },
    {
      "epoch": 0.3568702290076336,
      "grad_norm": 1.1377601623535156,
      "learning_rate": 3.56808326105811e-05,
      "loss": 0.5688,
      "step": 2057
    },
    {
      "epoch": 0.3570437196391395,
      "grad_norm": 0.7313696146011353,
      "learning_rate": 3.569817866435386e-05,
      "loss": 0.7212,
      "step": 2058
    },
    {
      "epoch": 0.3572172102706454,
      "grad_norm": 0.9031012654304504,
      "learning_rate": 3.571552471812663e-05,
      "loss": 0.5355,
      "step": 2059
    },
    {
      "epoch": 0.3573907009021513,
      "grad_norm": 1.2563116550445557,
      "learning_rate": 3.57328707718994e-05,
      "loss": 0.5861,
      "step": 2060
    },
    {
      "epoch": 0.3575641915336572,
      "grad_norm": 0.6538774371147156,
      "learning_rate": 3.5750216825672165e-05,
      "loss": 0.7454,
      "step": 2061
    },
    {
      "epoch": 0.3577376821651631,
      "grad_norm": 0.9398738145828247,
      "learning_rate": 3.576756287944493e-05,
      "loss": 0.5988,
      "step": 2062
    },
    {
      "epoch": 0.35791117279666895,
      "grad_norm": 0.8730028867721558,
      "learning_rate": 3.5784908933217695e-05,
      "loss": 0.511,
      "step": 2063
    },
    {
      "epoch": 0.35808466342817485,
      "grad_norm": 1.0857669115066528,
      "learning_rate": 3.5802254986990466e-05,
      "loss": 0.5911,
      "step": 2064
    },
    {
      "epoch": 0.35825815405968076,
      "grad_norm": 0.6124600172042847,
      "learning_rate": 3.581960104076323e-05,
      "loss": 0.6644,
      "step": 2065
    },
    {
      "epoch": 0.35843164469118666,
      "grad_norm": 0.7109978795051575,
      "learning_rate": 3.5836947094535996e-05,
      "loss": 0.6721,
      "step": 2066
    },
    {
      "epoch": 0.35860513532269256,
      "grad_norm": 0.7086520195007324,
      "learning_rate": 3.585429314830876e-05,
      "loss": 0.647,
      "step": 2067
    },
    {
      "epoch": 0.35877862595419846,
      "grad_norm": 0.7438195943832397,
      "learning_rate": 3.5871639202081526e-05,
      "loss": 0.5869,
      "step": 2068
    },
    {
      "epoch": 0.35895211658570436,
      "grad_norm": 0.965003490447998,
      "learning_rate": 3.58889852558543e-05,
      "loss": 0.5374,
      "step": 2069
    },
    {
      "epoch": 0.35912560721721026,
      "grad_norm": 0.8170627355575562,
      "learning_rate": 3.590633130962706e-05,
      "loss": 0.5886,
      "step": 2070
    },
    {
      "epoch": 0.35929909784871616,
      "grad_norm": 2.7678844928741455,
      "learning_rate": 3.592367736339983e-05,
      "loss": 0.6692,
      "step": 2071
    },
    {
      "epoch": 0.35947258848022207,
      "grad_norm": 0.6812294125556946,
      "learning_rate": 3.594102341717259e-05,
      "loss": 0.6429,
      "step": 2072
    },
    {
      "epoch": 0.35964607911172797,
      "grad_norm": 0.8648760318756104,
      "learning_rate": 3.5958369470945365e-05,
      "loss": 0.551,
      "step": 2073
    },
    {
      "epoch": 0.35981956974323387,
      "grad_norm": 0.8667991161346436,
      "learning_rate": 3.597571552471813e-05,
      "loss": 0.5897,
      "step": 2074
    },
    {
      "epoch": 0.35999306037473977,
      "grad_norm": 2.108598232269287,
      "learning_rate": 3.5993061578490895e-05,
      "loss": 0.5443,
      "step": 2075
    },
    {
      "epoch": 0.36016655100624567,
      "grad_norm": 0.8901035189628601,
      "learning_rate": 3.601040763226366e-05,
      "loss": 0.6406,
      "step": 2076
    },
    {
      "epoch": 0.3603400416377516,
      "grad_norm": 0.6625940203666687,
      "learning_rate": 3.602775368603643e-05,
      "loss": 0.6876,
      "step": 2077
    },
    {
      "epoch": 0.3605135322692575,
      "grad_norm": 1.0242830514907837,
      "learning_rate": 3.60450997398092e-05,
      "loss": 0.6406,
      "step": 2078
    },
    {
      "epoch": 0.3606870229007634,
      "grad_norm": 0.9576663374900818,
      "learning_rate": 3.606244579358196e-05,
      "loss": 0.613,
      "step": 2079
    },
    {
      "epoch": 0.3608605135322693,
      "grad_norm": 1.330987811088562,
      "learning_rate": 3.607979184735473e-05,
      "loss": 0.5693,
      "step": 2080
    },
    {
      "epoch": 0.3610340041637752,
      "grad_norm": 0.9956892132759094,
      "learning_rate": 3.609713790112749e-05,
      "loss": 0.6139,
      "step": 2081
    },
    {
      "epoch": 0.3612074947952811,
      "grad_norm": 0.786950945854187,
      "learning_rate": 3.6114483954900263e-05,
      "loss": 0.6703,
      "step": 2082
    },
    {
      "epoch": 0.3613809854267869,
      "grad_norm": 1.808635950088501,
      "learning_rate": 3.613183000867303e-05,
      "loss": 0.5145,
      "step": 2083
    },
    {
      "epoch": 0.36155447605829283,
      "grad_norm": 0.9782841801643372,
      "learning_rate": 3.6149176062445793e-05,
      "loss": 0.5297,
      "step": 2084
    },
    {
      "epoch": 0.36172796668979873,
      "grad_norm": 0.9573488235473633,
      "learning_rate": 3.6166522116218565e-05,
      "loss": 0.6165,
      "step": 2085
    },
    {
      "epoch": 0.36190145732130463,
      "grad_norm": 0.8779812455177307,
      "learning_rate": 3.618386816999133e-05,
      "loss": 0.5905,
      "step": 2086
    },
    {
      "epoch": 0.36207494795281053,
      "grad_norm": 0.8574431538581848,
      "learning_rate": 3.6201214223764095e-05,
      "loss": 0.5747,
      "step": 2087
    },
    {
      "epoch": 0.36224843858431643,
      "grad_norm": 0.9366618990898132,
      "learning_rate": 3.621856027753687e-05,
      "loss": 0.6127,
      "step": 2088
    },
    {
      "epoch": 0.36242192921582234,
      "grad_norm": 1.0383944511413574,
      "learning_rate": 3.623590633130963e-05,
      "loss": 0.5952,
      "step": 2089
    },
    {
      "epoch": 0.36259541984732824,
      "grad_norm": 0.8221400380134583,
      "learning_rate": 3.62532523850824e-05,
      "loss": 0.5695,
      "step": 2090
    },
    {
      "epoch": 0.36276891047883414,
      "grad_norm": 0.8370509743690491,
      "learning_rate": 3.627059843885516e-05,
      "loss": 0.5514,
      "step": 2091
    },
    {
      "epoch": 0.36294240111034004,
      "grad_norm": 0.7895044684410095,
      "learning_rate": 3.6287944492627934e-05,
      "loss": 0.645,
      "step": 2092
    },
    {
      "epoch": 0.36311589174184594,
      "grad_norm": 0.8655766248703003,
      "learning_rate": 3.63052905464007e-05,
      "loss": 0.7417,
      "step": 2093
    },
    {
      "epoch": 0.36328938237335184,
      "grad_norm": 0.60938960313797,
      "learning_rate": 3.6322636600173464e-05,
      "loss": 0.7515,
      "step": 2094
    },
    {
      "epoch": 0.36346287300485774,
      "grad_norm": 0.906223714351654,
      "learning_rate": 3.633998265394623e-05,
      "loss": 0.7554,
      "step": 2095
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 1.096692681312561,
      "learning_rate": 3.6357328707719e-05,
      "loss": 0.6246,
      "step": 2096
    },
    {
      "epoch": 0.36380985426786955,
      "grad_norm": 0.6672376394271851,
      "learning_rate": 3.6374674761491766e-05,
      "loss": 0.7445,
      "step": 2097
    },
    {
      "epoch": 0.36398334489937545,
      "grad_norm": 0.7774509191513062,
      "learning_rate": 3.639202081526453e-05,
      "loss": 0.5376,
      "step": 2098
    },
    {
      "epoch": 0.36415683553088135,
      "grad_norm": 1.0610190629959106,
      "learning_rate": 3.6409366869037296e-05,
      "loss": 0.62,
      "step": 2099
    },
    {
      "epoch": 0.36433032616238725,
      "grad_norm": 0.8590747714042664,
      "learning_rate": 3.642671292281006e-05,
      "loss": 0.5973,
      "step": 2100
    },
    {
      "epoch": 0.36450381679389315,
      "grad_norm": 0.8628801703453064,
      "learning_rate": 3.644405897658283e-05,
      "loss": 0.6003,
      "step": 2101
    },
    {
      "epoch": 0.36467730742539906,
      "grad_norm": 0.56028813123703,
      "learning_rate": 3.64614050303556e-05,
      "loss": 0.7511,
      "step": 2102
    },
    {
      "epoch": 0.3648507980569049,
      "grad_norm": 0.7692236304283142,
      "learning_rate": 3.647875108412836e-05,
      "loss": 0.7776,
      "step": 2103
    },
    {
      "epoch": 0.3650242886884108,
      "grad_norm": 0.5963106155395508,
      "learning_rate": 3.649609713790113e-05,
      "loss": 0.7537,
      "step": 2104
    },
    {
      "epoch": 0.3651977793199167,
      "grad_norm": 0.9407646656036377,
      "learning_rate": 3.65134431916739e-05,
      "loss": 0.577,
      "step": 2105
    },
    {
      "epoch": 0.3653712699514226,
      "grad_norm": 0.6260008215904236,
      "learning_rate": 3.6530789245446664e-05,
      "loss": 0.6802,
      "step": 2106
    },
    {
      "epoch": 0.3655447605829285,
      "grad_norm": 0.7514500617980957,
      "learning_rate": 3.654813529921943e-05,
      "loss": 0.6919,
      "step": 2107
    },
    {
      "epoch": 0.3657182512144344,
      "grad_norm": 0.6525545120239258,
      "learning_rate": 3.6565481352992194e-05,
      "loss": 0.7079,
      "step": 2108
    },
    {
      "epoch": 0.3658917418459403,
      "grad_norm": 1.0511095523834229,
      "learning_rate": 3.658282740676496e-05,
      "loss": 0.663,
      "step": 2109
    },
    {
      "epoch": 0.3660652324774462,
      "grad_norm": 0.7916147708892822,
      "learning_rate": 3.660017346053773e-05,
      "loss": 0.5444,
      "step": 2110
    },
    {
      "epoch": 0.3662387231089521,
      "grad_norm": 0.689697265625,
      "learning_rate": 3.6617519514310496e-05,
      "loss": 0.6318,
      "step": 2111
    },
    {
      "epoch": 0.366412213740458,
      "grad_norm": 0.7486228346824646,
      "learning_rate": 3.663486556808326e-05,
      "loss": 0.5714,
      "step": 2112
    },
    {
      "epoch": 0.3665857043719639,
      "grad_norm": 0.8919068574905396,
      "learning_rate": 3.6652211621856026e-05,
      "loss": 0.6603,
      "step": 2113
    },
    {
      "epoch": 0.3667591950034698,
      "grad_norm": 2.1213366985321045,
      "learning_rate": 3.66695576756288e-05,
      "loss": 0.5385,
      "step": 2114
    },
    {
      "epoch": 0.3669326856349757,
      "grad_norm": 1.0311821699142456,
      "learning_rate": 3.668690372940157e-05,
      "loss": 0.5289,
      "step": 2115
    },
    {
      "epoch": 0.3671061762664816,
      "grad_norm": 0.9412194490432739,
      "learning_rate": 3.6704249783174335e-05,
      "loss": 0.546,
      "step": 2116
    },
    {
      "epoch": 0.3672796668979875,
      "grad_norm": 0.9072990417480469,
      "learning_rate": 3.67215958369471e-05,
      "loss": 0.545,
      "step": 2117
    },
    {
      "epoch": 0.3674531575294934,
      "grad_norm": 0.994905948638916,
      "learning_rate": 3.6738941890719865e-05,
      "loss": 0.6029,
      "step": 2118
    },
    {
      "epoch": 0.3676266481609993,
      "grad_norm": 1.0085077285766602,
      "learning_rate": 3.675628794449263e-05,
      "loss": 0.6748,
      "step": 2119
    },
    {
      "epoch": 0.3678001387925052,
      "grad_norm": 0.7583150267601013,
      "learning_rate": 3.67736339982654e-05,
      "loss": 0.6609,
      "step": 2120
    },
    {
      "epoch": 0.36797362942401113,
      "grad_norm": 0.9201546907424927,
      "learning_rate": 3.6790980052038166e-05,
      "loss": 0.4989,
      "step": 2121
    },
    {
      "epoch": 0.36814712005551703,
      "grad_norm": 0.6300575137138367,
      "learning_rate": 3.680832610581093e-05,
      "loss": 0.5645,
      "step": 2122
    },
    {
      "epoch": 0.3683206106870229,
      "grad_norm": 1.0931706428527832,
      "learning_rate": 3.6825672159583696e-05,
      "loss": 0.6112,
      "step": 2123
    },
    {
      "epoch": 0.3684941013185288,
      "grad_norm": 0.850147008895874,
      "learning_rate": 3.684301821335647e-05,
      "loss": 0.5879,
      "step": 2124
    },
    {
      "epoch": 0.3686675919500347,
      "grad_norm": 0.8981357216835022,
      "learning_rate": 3.686036426712923e-05,
      "loss": 0.6051,
      "step": 2125
    },
    {
      "epoch": 0.3688410825815406,
      "grad_norm": 0.7931495308876038,
      "learning_rate": 3.6877710320902e-05,
      "loss": 0.7314,
      "step": 2126
    },
    {
      "epoch": 0.3690145732130465,
      "grad_norm": 0.832497775554657,
      "learning_rate": 3.689505637467476e-05,
      "loss": 0.7456,
      "step": 2127
    },
    {
      "epoch": 0.3691880638445524,
      "grad_norm": 0.9802011251449585,
      "learning_rate": 3.691240242844753e-05,
      "loss": 0.6581,
      "step": 2128
    },
    {
      "epoch": 0.3693615544760583,
      "grad_norm": 0.8732492327690125,
      "learning_rate": 3.69297484822203e-05,
      "loss": 0.6853,
      "step": 2129
    },
    {
      "epoch": 0.3695350451075642,
      "grad_norm": 2.1102023124694824,
      "learning_rate": 3.6947094535993065e-05,
      "loss": 0.7095,
      "step": 2130
    },
    {
      "epoch": 0.3697085357390701,
      "grad_norm": 0.8344833254814148,
      "learning_rate": 3.696444058976583e-05,
      "loss": 0.5624,
      "step": 2131
    },
    {
      "epoch": 0.369882026370576,
      "grad_norm": 0.7335735559463501,
      "learning_rate": 3.6981786643538595e-05,
      "loss": 0.6384,
      "step": 2132
    },
    {
      "epoch": 0.3700555170020819,
      "grad_norm": 0.808387815952301,
      "learning_rate": 3.699913269731137e-05,
      "loss": 0.5967,
      "step": 2133
    },
    {
      "epoch": 0.3702290076335878,
      "grad_norm": 0.8561879992485046,
      "learning_rate": 3.701647875108413e-05,
      "loss": 0.6221,
      "step": 2134
    },
    {
      "epoch": 0.3704024982650937,
      "grad_norm": 1.3011711835861206,
      "learning_rate": 3.70338248048569e-05,
      "loss": 0.6879,
      "step": 2135
    },
    {
      "epoch": 0.3705759888965996,
      "grad_norm": 0.8561848998069763,
      "learning_rate": 3.705117085862966e-05,
      "loss": 0.5869,
      "step": 2136
    },
    {
      "epoch": 0.3707494795281055,
      "grad_norm": 3.0262610912323,
      "learning_rate": 3.7068516912402433e-05,
      "loss": 0.6707,
      "step": 2137
    },
    {
      "epoch": 0.3709229701596114,
      "grad_norm": 0.6116523742675781,
      "learning_rate": 3.70858629661752e-05,
      "loss": 0.6661,
      "step": 2138
    },
    {
      "epoch": 0.3710964607911173,
      "grad_norm": 0.6808441281318665,
      "learning_rate": 3.7103209019947963e-05,
      "loss": 0.6449,
      "step": 2139
    },
    {
      "epoch": 0.3712699514226232,
      "grad_norm": 0.9481117129325867,
      "learning_rate": 3.712055507372073e-05,
      "loss": 0.5366,
      "step": 2140
    },
    {
      "epoch": 0.3714434420541291,
      "grad_norm": 3.1375207901000977,
      "learning_rate": 3.7137901127493493e-05,
      "loss": 0.6429,
      "step": 2141
    },
    {
      "epoch": 0.37161693268563495,
      "grad_norm": 0.6729617714881897,
      "learning_rate": 3.7155247181266265e-05,
      "loss": 0.6078,
      "step": 2142
    },
    {
      "epoch": 0.37179042331714085,
      "grad_norm": 0.6703296899795532,
      "learning_rate": 3.717259323503903e-05,
      "loss": 0.6831,
      "step": 2143
    },
    {
      "epoch": 0.37196391394864675,
      "grad_norm": 0.8146363496780396,
      "learning_rate": 3.71899392888118e-05,
      "loss": 0.6257,
      "step": 2144
    },
    {
      "epoch": 0.37213740458015265,
      "grad_norm": 0.7584531903266907,
      "learning_rate": 3.720728534258457e-05,
      "loss": 0.6805,
      "step": 2145
    },
    {
      "epoch": 0.37231089521165855,
      "grad_norm": 0.8820028901100159,
      "learning_rate": 3.722463139635733e-05,
      "loss": 0.5426,
      "step": 2146
    },
    {
      "epoch": 0.37248438584316446,
      "grad_norm": 0.6640870571136475,
      "learning_rate": 3.72419774501301e-05,
      "loss": 0.6062,
      "step": 2147
    },
    {
      "epoch": 0.37265787647467036,
      "grad_norm": 0.8160288333892822,
      "learning_rate": 3.725932350390287e-05,
      "loss": 0.653,
      "step": 2148
    },
    {
      "epoch": 0.37283136710617626,
      "grad_norm": 0.9772170782089233,
      "learning_rate": 3.7276669557675634e-05,
      "loss": 0.5541,
      "step": 2149
    },
    {
      "epoch": 0.37300485773768216,
      "grad_norm": 0.8377646207809448,
      "learning_rate": 3.72940156114484e-05,
      "loss": 0.569,
      "step": 2150
    },
    {
      "epoch": 0.37317834836918806,
      "grad_norm": 0.7429654002189636,
      "learning_rate": 3.7311361665221164e-05,
      "loss": 0.545,
      "step": 2151
    },
    {
      "epoch": 0.37335183900069396,
      "grad_norm": 0.965164065361023,
      "learning_rate": 3.7328707718993936e-05,
      "loss": 0.5671,
      "step": 2152
    },
    {
      "epoch": 0.37352532963219987,
      "grad_norm": 0.7786089181900024,
      "learning_rate": 3.73460537727667e-05,
      "loss": 0.7173,
      "step": 2153
    },
    {
      "epoch": 0.37369882026370577,
      "grad_norm": 1.0826573371887207,
      "learning_rate": 3.7363399826539466e-05,
      "loss": 0.6929,
      "step": 2154
    },
    {
      "epoch": 0.37387231089521167,
      "grad_norm": 1.2865610122680664,
      "learning_rate": 3.738074588031223e-05,
      "loss": 0.5865,
      "step": 2155
    },
    {
      "epoch": 0.37404580152671757,
      "grad_norm": 0.9463958144187927,
      "learning_rate": 3.7398091934085e-05,
      "loss": 0.5681,
      "step": 2156
    },
    {
      "epoch": 0.37421929215822347,
      "grad_norm": 0.966681182384491,
      "learning_rate": 3.741543798785777e-05,
      "loss": 0.6877,
      "step": 2157
    },
    {
      "epoch": 0.3743927827897294,
      "grad_norm": 0.8081557154655457,
      "learning_rate": 3.743278404163053e-05,
      "loss": 0.7155,
      "step": 2158
    },
    {
      "epoch": 0.3745662734212353,
      "grad_norm": 0.5800716280937195,
      "learning_rate": 3.74501300954033e-05,
      "loss": 0.687,
      "step": 2159
    },
    {
      "epoch": 0.3747397640527412,
      "grad_norm": 0.863257646560669,
      "learning_rate": 3.746747614917606e-05,
      "loss": 0.5291,
      "step": 2160
    },
    {
      "epoch": 0.3749132546842471,
      "grad_norm": 0.7452552914619446,
      "learning_rate": 3.7484822202948834e-05,
      "loss": 0.637,
      "step": 2161
    },
    {
      "epoch": 0.3750867453157529,
      "grad_norm": 0.9935975074768066,
      "learning_rate": 3.75021682567216e-05,
      "loss": 0.5657,
      "step": 2162
    },
    {
      "epoch": 0.3752602359472588,
      "grad_norm": 0.7122426629066467,
      "learning_rate": 3.7519514310494364e-05,
      "loss": 0.58,
      "step": 2163
    },
    {
      "epoch": 0.3754337265787647,
      "grad_norm": 0.6864352226257324,
      "learning_rate": 3.753686036426713e-05,
      "loss": 0.666,
      "step": 2164
    },
    {
      "epoch": 0.3756072172102706,
      "grad_norm": 0.7913309335708618,
      "learning_rate": 3.75542064180399e-05,
      "loss": 0.5684,
      "step": 2165
    },
    {
      "epoch": 0.37578070784177653,
      "grad_norm": 0.819497287273407,
      "learning_rate": 3.7571552471812666e-05,
      "loss": 0.8098,
      "step": 2166
    },
    {
      "epoch": 0.37595419847328243,
      "grad_norm": 0.7337230443954468,
      "learning_rate": 3.758889852558543e-05,
      "loss": 0.6859,
      "step": 2167
    },
    {
      "epoch": 0.37612768910478833,
      "grad_norm": 1.0212225914001465,
      "learning_rate": 3.7606244579358196e-05,
      "loss": 0.5035,
      "step": 2168
    },
    {
      "epoch": 0.37630117973629423,
      "grad_norm": 0.9265641570091248,
      "learning_rate": 3.762359063313096e-05,
      "loss": 0.5811,
      "step": 2169
    },
    {
      "epoch": 0.37647467036780013,
      "grad_norm": 0.9080912470817566,
      "learning_rate": 3.764093668690373e-05,
      "loss": 0.7583,
      "step": 2170
    },
    {
      "epoch": 0.37664816099930604,
      "grad_norm": 1.1696559190750122,
      "learning_rate": 3.76582827406765e-05,
      "loss": 0.5392,
      "step": 2171
    },
    {
      "epoch": 0.37682165163081194,
      "grad_norm": 0.7206262946128845,
      "learning_rate": 3.767562879444926e-05,
      "loss": 0.6002,
      "step": 2172
    },
    {
      "epoch": 0.37699514226231784,
      "grad_norm": 0.7880340814590454,
      "learning_rate": 3.7692974848222034e-05,
      "loss": 0.6541,
      "step": 2173
    },
    {
      "epoch": 0.37716863289382374,
      "grad_norm": 0.7043308019638062,
      "learning_rate": 3.77103209019948e-05,
      "loss": 0.6379,
      "step": 2174
    },
    {
      "epoch": 0.37734212352532964,
      "grad_norm": 1.0066626071929932,
      "learning_rate": 3.772766695576757e-05,
      "loss": 0.6377,
      "step": 2175
    },
    {
      "epoch": 0.37751561415683554,
      "grad_norm": 0.7290460467338562,
      "learning_rate": 3.7745013009540336e-05,
      "loss": 0.6566,
      "step": 2176
    },
    {
      "epoch": 0.37768910478834145,
      "grad_norm": 0.8432542681694031,
      "learning_rate": 3.77623590633131e-05,
      "loss": 0.5568,
      "step": 2177
    },
    {
      "epoch": 0.37786259541984735,
      "grad_norm": 1.1949617862701416,
      "learning_rate": 3.7779705117085866e-05,
      "loss": 0.6653,
      "step": 2178
    },
    {
      "epoch": 0.37803608605135325,
      "grad_norm": 0.9642347097396851,
      "learning_rate": 3.779705117085863e-05,
      "loss": 0.6412,
      "step": 2179
    },
    {
      "epoch": 0.37820957668285915,
      "grad_norm": 0.8353285193443298,
      "learning_rate": 3.78143972246314e-05,
      "loss": 0.6458,
      "step": 2180
    },
    {
      "epoch": 0.37838306731436505,
      "grad_norm": 0.7169858813285828,
      "learning_rate": 3.783174327840417e-05,
      "loss": 0.7122,
      "step": 2181
    },
    {
      "epoch": 0.3785565579458709,
      "grad_norm": 0.6582884788513184,
      "learning_rate": 3.784908933217693e-05,
      "loss": 0.6321,
      "step": 2182
    },
    {
      "epoch": 0.3787300485773768,
      "grad_norm": 0.9503910541534424,
      "learning_rate": 3.78664353859497e-05,
      "loss": 0.5455,
      "step": 2183
    },
    {
      "epoch": 0.3789035392088827,
      "grad_norm": 0.8722729682922363,
      "learning_rate": 3.788378143972247e-05,
      "loss": 0.7169,
      "step": 2184
    },
    {
      "epoch": 0.3790770298403886,
      "grad_norm": 0.6877573728561401,
      "learning_rate": 3.7901127493495235e-05,
      "loss": 0.6674,
      "step": 2185
    },
    {
      "epoch": 0.3792505204718945,
      "grad_norm": 0.8797959685325623,
      "learning_rate": 3.7918473547268e-05,
      "loss": 0.6737,
      "step": 2186
    },
    {
      "epoch": 0.3794240111034004,
      "grad_norm": 0.9650540351867676,
      "learning_rate": 3.7935819601040765e-05,
      "loss": 0.7108,
      "step": 2187
    },
    {
      "epoch": 0.3795975017349063,
      "grad_norm": 0.8198643326759338,
      "learning_rate": 3.795316565481353e-05,
      "loss": 0.6161,
      "step": 2188
    },
    {
      "epoch": 0.3797709923664122,
      "grad_norm": 1.009889841079712,
      "learning_rate": 3.79705117085863e-05,
      "loss": 0.601,
      "step": 2189
    },
    {
      "epoch": 0.3799444829979181,
      "grad_norm": 0.7123552560806274,
      "learning_rate": 3.7987857762359067e-05,
      "loss": 0.6511,
      "step": 2190
    },
    {
      "epoch": 0.380117973629424,
      "grad_norm": 0.8433806896209717,
      "learning_rate": 3.800520381613183e-05,
      "loss": 0.5269,
      "step": 2191
    },
    {
      "epoch": 0.3802914642609299,
      "grad_norm": 0.9886137247085571,
      "learning_rate": 3.8022549869904597e-05,
      "loss": 0.5284,
      "step": 2192
    },
    {
      "epoch": 0.3804649548924358,
      "grad_norm": 1.609928011894226,
      "learning_rate": 3.803989592367737e-05,
      "loss": 0.6855,
      "step": 2193
    },
    {
      "epoch": 0.3806384455239417,
      "grad_norm": 0.5896164178848267,
      "learning_rate": 3.805724197745013e-05,
      "loss": 0.7257,
      "step": 2194
    },
    {
      "epoch": 0.3808119361554476,
      "grad_norm": 0.7190706729888916,
      "learning_rate": 3.80745880312229e-05,
      "loss": 0.6545,
      "step": 2195
    },
    {
      "epoch": 0.3809854267869535,
      "grad_norm": 0.7872779369354248,
      "learning_rate": 3.809193408499566e-05,
      "loss": 0.5746,
      "step": 2196
    },
    {
      "epoch": 0.3811589174184594,
      "grad_norm": 0.9195467233657837,
      "learning_rate": 3.8109280138768435e-05,
      "loss": 0.6157,
      "step": 2197
    },
    {
      "epoch": 0.3813324080499653,
      "grad_norm": 1.1707857847213745,
      "learning_rate": 3.81266261925412e-05,
      "loss": 0.6997,
      "step": 2198
    },
    {
      "epoch": 0.3815058986814712,
      "grad_norm": 0.7408986687660217,
      "learning_rate": 3.8143972246313965e-05,
      "loss": 0.618,
      "step": 2199
    },
    {
      "epoch": 0.3816793893129771,
      "grad_norm": 0.7133910059928894,
      "learning_rate": 3.816131830008673e-05,
      "loss": 0.7285,
      "step": 2200
    },
    {
      "epoch": 0.38185287994448297,
      "grad_norm": 0.6581571698188782,
      "learning_rate": 3.8178664353859495e-05,
      "loss": 0.6235,
      "step": 2201
    },
    {
      "epoch": 0.38202637057598887,
      "grad_norm": 0.792733907699585,
      "learning_rate": 3.819601040763227e-05,
      "loss": 0.5786,
      "step": 2202
    },
    {
      "epoch": 0.3821998612074948,
      "grad_norm": 0.9168846607208252,
      "learning_rate": 3.821335646140504e-05,
      "loss": 0.573,
      "step": 2203
    },
    {
      "epoch": 0.3823733518390007,
      "grad_norm": 0.6650002002716064,
      "learning_rate": 3.8230702515177804e-05,
      "loss": 0.6843,
      "step": 2204
    },
    {
      "epoch": 0.3825468424705066,
      "grad_norm": 0.8428970575332642,
      "learning_rate": 3.824804856895057e-05,
      "loss": 0.5229,
      "step": 2205
    },
    {
      "epoch": 0.3827203331020125,
      "grad_norm": 0.8378449082374573,
      "learning_rate": 3.8265394622723334e-05,
      "loss": 0.5596,
      "step": 2206
    },
    {
      "epoch": 0.3828938237335184,
      "grad_norm": 0.6730502843856812,
      "learning_rate": 3.82827406764961e-05,
      "loss": 0.5841,
      "step": 2207
    },
    {
      "epoch": 0.3830673143650243,
      "grad_norm": 0.9363775849342346,
      "learning_rate": 3.830008673026887e-05,
      "loss": 0.6696,
      "step": 2208
    },
    {
      "epoch": 0.3832408049965302,
      "grad_norm": 0.9516682624816895,
      "learning_rate": 3.8317432784041636e-05,
      "loss": 0.6219,
      "step": 2209
    },
    {
      "epoch": 0.3834142956280361,
      "grad_norm": 0.9467013478279114,
      "learning_rate": 3.83347788378144e-05,
      "loss": 0.5372,
      "step": 2210
    },
    {
      "epoch": 0.383587786259542,
      "grad_norm": 1.103040099143982,
      "learning_rate": 3.8352124891587166e-05,
      "loss": 0.5616,
      "step": 2211
    },
    {
      "epoch": 0.3837612768910479,
      "grad_norm": 0.8702182769775391,
      "learning_rate": 3.836947094535994e-05,
      "loss": 0.5455,
      "step": 2212
    },
    {
      "epoch": 0.3839347675225538,
      "grad_norm": 1.0493888854980469,
      "learning_rate": 3.83868169991327e-05,
      "loss": 0.6152,
      "step": 2213
    },
    {
      "epoch": 0.3841082581540597,
      "grad_norm": 0.7445110082626343,
      "learning_rate": 3.840416305290547e-05,
      "loss": 0.5646,
      "step": 2214
    },
    {
      "epoch": 0.3842817487855656,
      "grad_norm": 0.9141197204589844,
      "learning_rate": 3.842150910667823e-05,
      "loss": 0.5713,
      "step": 2215
    },
    {
      "epoch": 0.3844552394170715,
      "grad_norm": 0.7753279209136963,
      "learning_rate": 3.8438855160451004e-05,
      "loss": 0.7032,
      "step": 2216
    },
    {
      "epoch": 0.3846287300485774,
      "grad_norm": 0.8179315328598022,
      "learning_rate": 3.845620121422377e-05,
      "loss": 0.6094,
      "step": 2217
    },
    {
      "epoch": 0.3848022206800833,
      "grad_norm": 1.1109416484832764,
      "learning_rate": 3.8473547267996534e-05,
      "loss": 0.5817,
      "step": 2218
    },
    {
      "epoch": 0.3849757113115892,
      "grad_norm": 0.778189480304718,
      "learning_rate": 3.84908933217693e-05,
      "loss": 0.6007,
      "step": 2219
    },
    {
      "epoch": 0.3851492019430951,
      "grad_norm": 0.5885578393936157,
      "learning_rate": 3.8508239375542064e-05,
      "loss": 0.7057,
      "step": 2220
    },
    {
      "epoch": 0.38532269257460094,
      "grad_norm": 0.9650107026100159,
      "learning_rate": 3.8525585429314836e-05,
      "loss": 0.6461,
      "step": 2221
    },
    {
      "epoch": 0.38549618320610685,
      "grad_norm": 0.7100480198860168,
      "learning_rate": 3.85429314830876e-05,
      "loss": 0.6464,
      "step": 2222
    },
    {
      "epoch": 0.38566967383761275,
      "grad_norm": 0.7356059551239014,
      "learning_rate": 3.8560277536860366e-05,
      "loss": 0.7244,
      "step": 2223
    },
    {
      "epoch": 0.38584316446911865,
      "grad_norm": 0.8487452268600464,
      "learning_rate": 3.857762359063313e-05,
      "loss": 0.6389,
      "step": 2224
    },
    {
      "epoch": 0.38601665510062455,
      "grad_norm": 2.1440274715423584,
      "learning_rate": 3.85949696444059e-05,
      "loss": 0.5741,
      "step": 2225
    },
    {
      "epoch": 0.38619014573213045,
      "grad_norm": 1.0137948989868164,
      "learning_rate": 3.861231569817867e-05,
      "loss": 0.5248,
      "step": 2226
    },
    {
      "epoch": 0.38636363636363635,
      "grad_norm": 2.05366849899292,
      "learning_rate": 3.862966175195143e-05,
      "loss": 0.6107,
      "step": 2227
    },
    {
      "epoch": 0.38653712699514226,
      "grad_norm": 0.663736879825592,
      "learning_rate": 3.86470078057242e-05,
      "loss": 0.6602,
      "step": 2228
    },
    {
      "epoch": 0.38671061762664816,
      "grad_norm": 0.797764778137207,
      "learning_rate": 3.866435385949696e-05,
      "loss": 0.6271,
      "step": 2229
    },
    {
      "epoch": 0.38688410825815406,
      "grad_norm": 1.0539405345916748,
      "learning_rate": 3.8681699913269734e-05,
      "loss": 0.7142,
      "step": 2230
    },
    {
      "epoch": 0.38705759888965996,
      "grad_norm": 0.8324370384216309,
      "learning_rate": 3.86990459670425e-05,
      "loss": 0.6418,
      "step": 2231
    },
    {
      "epoch": 0.38723108952116586,
      "grad_norm": 1.3873273134231567,
      "learning_rate": 3.8716392020815264e-05,
      "loss": 0.5942,
      "step": 2232
    },
    {
      "epoch": 0.38740458015267176,
      "grad_norm": 1.3949419260025024,
      "learning_rate": 3.8733738074588036e-05,
      "loss": 0.6537,
      "step": 2233
    },
    {
      "epoch": 0.38757807078417766,
      "grad_norm": 0.9070783257484436,
      "learning_rate": 3.87510841283608e-05,
      "loss": 0.6234,
      "step": 2234
    },
    {
      "epoch": 0.38775156141568357,
      "grad_norm": 0.8216388821601868,
      "learning_rate": 3.8768430182133566e-05,
      "loss": 0.6975,
      "step": 2235
    },
    {
      "epoch": 0.38792505204718947,
      "grad_norm": 0.7318658828735352,
      "learning_rate": 3.878577623590634e-05,
      "loss": 0.7463,
      "step": 2236
    },
    {
      "epoch": 0.38809854267869537,
      "grad_norm": 1.1542892456054688,
      "learning_rate": 3.88031222896791e-05,
      "loss": 0.5834,
      "step": 2237
    },
    {
      "epoch": 0.38827203331020127,
      "grad_norm": 0.7708222270011902,
      "learning_rate": 3.882046834345187e-05,
      "loss": 0.6576,
      "step": 2238
    },
    {
      "epoch": 0.38844552394170717,
      "grad_norm": 0.8864904642105103,
      "learning_rate": 3.883781439722463e-05,
      "loss": 0.6471,
      "step": 2239
    },
    {
      "epoch": 0.3886190145732131,
      "grad_norm": 0.9263260960578918,
      "learning_rate": 3.8855160450997405e-05,
      "loss": 0.6013,
      "step": 2240
    },
    {
      "epoch": 0.3887925052047189,
      "grad_norm": 0.9893810153007507,
      "learning_rate": 3.887250650477017e-05,
      "loss": 0.6096,
      "step": 2241
    },
    {
      "epoch": 0.3889659958362248,
      "grad_norm": 1.1212811470031738,
      "learning_rate": 3.8889852558542935e-05,
      "loss": 0.5563,
      "step": 2242
    },
    {
      "epoch": 0.3891394864677307,
      "grad_norm": 0.6433964371681213,
      "learning_rate": 3.89071986123157e-05,
      "loss": 0.7335,
      "step": 2243
    },
    {
      "epoch": 0.3893129770992366,
      "grad_norm": 1.5604666471481323,
      "learning_rate": 3.892454466608847e-05,
      "loss": 0.5704,
      "step": 2244
    },
    {
      "epoch": 0.3894864677307425,
      "grad_norm": 2.013279438018799,
      "learning_rate": 3.8941890719861237e-05,
      "loss": 0.6403,
      "step": 2245
    },
    {
      "epoch": 0.3896599583622484,
      "grad_norm": 0.7819218039512634,
      "learning_rate": 3.8959236773634e-05,
      "loss": 0.5585,
      "step": 2246
    },
    {
      "epoch": 0.38983344899375433,
      "grad_norm": 1.1008758544921875,
      "learning_rate": 3.8976582827406767e-05,
      "loss": 0.5536,
      "step": 2247
    },
    {
      "epoch": 0.39000693962526023,
      "grad_norm": 1.0997666120529175,
      "learning_rate": 3.899392888117953e-05,
      "loss": 0.6185,
      "step": 2248
    },
    {
      "epoch": 0.39018043025676613,
      "grad_norm": 1.2280099391937256,
      "learning_rate": 3.90112749349523e-05,
      "loss": 0.6738,
      "step": 2249
    },
    {
      "epoch": 0.39035392088827203,
      "grad_norm": 0.8849902749061584,
      "learning_rate": 3.902862098872507e-05,
      "loss": 0.6309,
      "step": 2250
    },
    {
      "epoch": 0.39052741151977793,
      "grad_norm": 0.9604009389877319,
      "learning_rate": 3.904596704249783e-05,
      "loss": 0.5986,
      "step": 2251
    },
    {
      "epoch": 0.39070090215128384,
      "grad_norm": 0.6952579617500305,
      "learning_rate": 3.90633130962706e-05,
      "loss": 0.6101,
      "step": 2252
    },
    {
      "epoch": 0.39087439278278974,
      "grad_norm": 1.2840449810028076,
      "learning_rate": 3.908065915004337e-05,
      "loss": 0.6625,
      "step": 2253
    },
    {
      "epoch": 0.39104788341429564,
      "grad_norm": 0.8691678047180176,
      "learning_rate": 3.9098005203816135e-05,
      "loss": 0.5641,
      "step": 2254
    },
    {
      "epoch": 0.39122137404580154,
      "grad_norm": 1.1951429843902588,
      "learning_rate": 3.91153512575889e-05,
      "loss": 0.5901,
      "step": 2255
    },
    {
      "epoch": 0.39139486467730744,
      "grad_norm": 1.7035776376724243,
      "learning_rate": 3.9132697311361665e-05,
      "loss": 0.5787,
      "step": 2256
    },
    {
      "epoch": 0.39156835530881334,
      "grad_norm": 0.8311005234718323,
      "learning_rate": 3.915004336513444e-05,
      "loss": 0.6257,
      "step": 2257
    },
    {
      "epoch": 0.39174184594031924,
      "grad_norm": 0.9469131827354431,
      "learning_rate": 3.91673894189072e-05,
      "loss": 0.5909,
      "step": 2258
    },
    {
      "epoch": 0.39191533657182515,
      "grad_norm": 1.100370168685913,
      "learning_rate": 3.918473547267997e-05,
      "loss": 0.6981,
      "step": 2259
    },
    {
      "epoch": 0.39208882720333105,
      "grad_norm": 0.7783090472221375,
      "learning_rate": 3.920208152645273e-05,
      "loss": 0.6653,
      "step": 2260
    },
    {
      "epoch": 0.3922623178348369,
      "grad_norm": 1.446513295173645,
      "learning_rate": 3.92194275802255e-05,
      "loss": 0.6958,
      "step": 2261
    },
    {
      "epoch": 0.3924358084663428,
      "grad_norm": 0.6650233864784241,
      "learning_rate": 3.923677363399827e-05,
      "loss": 0.7476,
      "step": 2262
    },
    {
      "epoch": 0.3926092990978487,
      "grad_norm": 0.9387511014938354,
      "learning_rate": 3.925411968777104e-05,
      "loss": 0.533,
      "step": 2263
    },
    {
      "epoch": 0.3927827897293546,
      "grad_norm": 0.9248531460762024,
      "learning_rate": 3.9271465741543805e-05,
      "loss": 0.6064,
      "step": 2264
    },
    {
      "epoch": 0.3929562803608605,
      "grad_norm": 0.8621205687522888,
      "learning_rate": 3.928881179531657e-05,
      "loss": 0.639,
      "step": 2265
    },
    {
      "epoch": 0.3931297709923664,
      "grad_norm": 0.9346652030944824,
      "learning_rate": 3.9306157849089335e-05,
      "loss": 0.6278,
      "step": 2266
    },
    {
      "epoch": 0.3933032616238723,
      "grad_norm": 0.9083680510520935,
      "learning_rate": 3.93235039028621e-05,
      "loss": 0.5778,
      "step": 2267
    },
    {
      "epoch": 0.3934767522553782,
      "grad_norm": 1.2940402030944824,
      "learning_rate": 3.934084995663487e-05,
      "loss": 0.6317,
      "step": 2268
    },
    {
      "epoch": 0.3936502428868841,
      "grad_norm": 1.2326910495758057,
      "learning_rate": 3.935819601040764e-05,
      "loss": 0.5909,
      "step": 2269
    },
    {
      "epoch": 0.39382373351839,
      "grad_norm": 0.8138448596000671,
      "learning_rate": 3.93755420641804e-05,
      "loss": 0.6462,
      "step": 2270
    },
    {
      "epoch": 0.3939972241498959,
      "grad_norm": 0.7268115878105164,
      "learning_rate": 3.939288811795317e-05,
      "loss": 0.6428,
      "step": 2271
    },
    {
      "epoch": 0.3941707147814018,
      "grad_norm": 0.7852060198783875,
      "learning_rate": 3.941023417172594e-05,
      "loss": 0.6348,
      "step": 2272
    },
    {
      "epoch": 0.3943442054129077,
      "grad_norm": 0.7785980105400085,
      "learning_rate": 3.9427580225498704e-05,
      "loss": 0.5223,
      "step": 2273
    },
    {
      "epoch": 0.3945176960444136,
      "grad_norm": 0.8455443978309631,
      "learning_rate": 3.944492627927147e-05,
      "loss": 0.5814,
      "step": 2274
    },
    {
      "epoch": 0.3946911866759195,
      "grad_norm": 0.972460150718689,
      "learning_rate": 3.9462272333044234e-05,
      "loss": 0.6086,
      "step": 2275
    },
    {
      "epoch": 0.3948646773074254,
      "grad_norm": 1.0304902791976929,
      "learning_rate": 3.9479618386817006e-05,
      "loss": 0.7295,
      "step": 2276
    },
    {
      "epoch": 0.3950381679389313,
      "grad_norm": 0.7572159767150879,
      "learning_rate": 3.949696444058977e-05,
      "loss": 0.7764,
      "step": 2277
    },
    {
      "epoch": 0.3952116585704372,
      "grad_norm": 1.0979856252670288,
      "learning_rate": 3.9514310494362536e-05,
      "loss": 0.6222,
      "step": 2278
    },
    {
      "epoch": 0.3953851492019431,
      "grad_norm": 0.7576919198036194,
      "learning_rate": 3.95316565481353e-05,
      "loss": 0.6149,
      "step": 2279
    },
    {
      "epoch": 0.39555863983344897,
      "grad_norm": 0.682855486869812,
      "learning_rate": 3.9549002601908066e-05,
      "loss": 0.6792,
      "step": 2280
    },
    {
      "epoch": 0.39573213046495487,
      "grad_norm": 0.8346561193466187,
      "learning_rate": 3.956634865568084e-05,
      "loss": 0.6987,
      "step": 2281
    },
    {
      "epoch": 0.39590562109646077,
      "grad_norm": 0.7672377824783325,
      "learning_rate": 3.95836947094536e-05,
      "loss": 0.7587,
      "step": 2282
    },
    {
      "epoch": 0.39607911172796667,
      "grad_norm": 0.830708920955658,
      "learning_rate": 3.960104076322637e-05,
      "loss": 0.5814,
      "step": 2283
    },
    {
      "epoch": 0.39625260235947257,
      "grad_norm": 0.7514001131057739,
      "learning_rate": 3.961838681699913e-05,
      "loss": 0.5084,
      "step": 2284
    },
    {
      "epoch": 0.3964260929909785,
      "grad_norm": 0.5804221630096436,
      "learning_rate": 3.9635732870771904e-05,
      "loss": 0.6553,
      "step": 2285
    },
    {
      "epoch": 0.3965995836224844,
      "grad_norm": 1.231408953666687,
      "learning_rate": 3.965307892454467e-05,
      "loss": 0.5291,
      "step": 2286
    },
    {
      "epoch": 0.3967730742539903,
      "grad_norm": 0.8249265551567078,
      "learning_rate": 3.9670424978317434e-05,
      "loss": 0.5966,
      "step": 2287
    },
    {
      "epoch": 0.3969465648854962,
      "grad_norm": 0.8786765336990356,
      "learning_rate": 3.96877710320902e-05,
      "loss": 0.6144,
      "step": 2288
    },
    {
      "epoch": 0.3971200555170021,
      "grad_norm": 0.9649869203567505,
      "learning_rate": 3.9705117085862964e-05,
      "loss": 0.5176,
      "step": 2289
    },
    {
      "epoch": 0.397293546148508,
      "grad_norm": 0.6982461214065552,
      "learning_rate": 3.9722463139635736e-05,
      "loss": 0.5854,
      "step": 2290
    },
    {
      "epoch": 0.3974670367800139,
      "grad_norm": 0.9511545896530151,
      "learning_rate": 3.97398091934085e-05,
      "loss": 0.6732,
      "step": 2291
    },
    {
      "epoch": 0.3976405274115198,
      "grad_norm": 0.6593926548957825,
      "learning_rate": 3.975715524718127e-05,
      "loss": 0.8062,
      "step": 2292
    },
    {
      "epoch": 0.3978140180430257,
      "grad_norm": 0.7266292572021484,
      "learning_rate": 3.977450130095404e-05,
      "loss": 0.6166,
      "step": 2293
    },
    {
      "epoch": 0.3979875086745316,
      "grad_norm": 1.044750452041626,
      "learning_rate": 3.97918473547268e-05,
      "loss": 0.7644,
      "step": 2294
    },
    {
      "epoch": 0.3981609993060375,
      "grad_norm": 0.8857741355895996,
      "learning_rate": 3.980919340849957e-05,
      "loss": 0.5856,
      "step": 2295
    },
    {
      "epoch": 0.3983344899375434,
      "grad_norm": 0.6424846649169922,
      "learning_rate": 3.982653946227234e-05,
      "loss": 0.6316,
      "step": 2296
    },
    {
      "epoch": 0.3985079805690493,
      "grad_norm": 1.01796555519104,
      "learning_rate": 3.9843885516045105e-05,
      "loss": 0.5682,
      "step": 2297
    },
    {
      "epoch": 0.3986814712005552,
      "grad_norm": 0.8512812256813049,
      "learning_rate": 3.986123156981787e-05,
      "loss": 0.5811,
      "step": 2298
    },
    {
      "epoch": 0.3988549618320611,
      "grad_norm": 1.0336955785751343,
      "learning_rate": 3.9878577623590635e-05,
      "loss": 0.5056,
      "step": 2299
    },
    {
      "epoch": 0.39902845246356694,
      "grad_norm": 1.2619831562042236,
      "learning_rate": 3.9895923677363406e-05,
      "loss": 0.5651,
      "step": 2300
    },
    {
      "epoch": 0.39920194309507284,
      "grad_norm": 0.985720694065094,
      "learning_rate": 3.991326973113617e-05,
      "loss": 0.5935,
      "step": 2301
    },
    {
      "epoch": 0.39937543372657874,
      "grad_norm": 1.572470784187317,
      "learning_rate": 3.9930615784908936e-05,
      "loss": 0.7886,
      "step": 2302
    },
    {
      "epoch": 0.39954892435808465,
      "grad_norm": 1.0250521898269653,
      "learning_rate": 3.99479618386817e-05,
      "loss": 0.6128,
      "step": 2303
    },
    {
      "epoch": 0.39972241498959055,
      "grad_norm": 0.7787879705429077,
      "learning_rate": 3.996530789245447e-05,
      "loss": 0.6128,
      "step": 2304
    },
    {
      "epoch": 0.39989590562109645,
      "grad_norm": 0.8346136212348938,
      "learning_rate": 3.998265394622724e-05,
      "loss": 0.5773,
      "step": 2305
    },
    {
      "epoch": 0.40006939625260235,
      "grad_norm": 0.7871057391166687,
      "learning_rate": 4e-05,
      "loss": 0.6619,
      "step": 2306
    },
    {
      "epoch": 0.40024288688410825,
      "grad_norm": 0.909366250038147,
      "learning_rate": 3.9999998839488355e-05,
      "loss": 0.658,
      "step": 2307
    },
    {
      "epoch": 0.40041637751561415,
      "grad_norm": 0.6434295177459717,
      "learning_rate": 3.999999535795353e-05,
      "loss": 0.6929,
      "step": 2308
    },
    {
      "epoch": 0.40058986814712005,
      "grad_norm": 0.7157797813415527,
      "learning_rate": 3.999998955539594e-05,
      "loss": 0.6593,
      "step": 2309
    },
    {
      "epoch": 0.40076335877862596,
      "grad_norm": 0.8653424382209778,
      "learning_rate": 3.9999981431816256e-05,
      "loss": 0.7112,
      "step": 2310
    },
    {
      "epoch": 0.40093684941013186,
      "grad_norm": 0.9613097310066223,
      "learning_rate": 3.999997098721543e-05,
      "loss": 0.5836,
      "step": 2311
    },
    {
      "epoch": 0.40111034004163776,
      "grad_norm": 1.0514494180679321,
      "learning_rate": 3.999995822159466e-05,
      "loss": 0.762,
      "step": 2312
    },
    {
      "epoch": 0.40128383067314366,
      "grad_norm": 0.7570793032646179,
      "learning_rate": 3.9999943134955436e-05,
      "loss": 0.6372,
      "step": 2313
    },
    {
      "epoch": 0.40145732130464956,
      "grad_norm": 0.8194951415061951,
      "learning_rate": 3.9999925727299505e-05,
      "loss": 0.6851,
      "step": 2314
    },
    {
      "epoch": 0.40163081193615546,
      "grad_norm": 0.7198679447174072,
      "learning_rate": 3.999990599862889e-05,
      "loss": 0.5604,
      "step": 2315
    },
    {
      "epoch": 0.40180430256766136,
      "grad_norm": 0.9790434837341309,
      "learning_rate": 3.999988394894588e-05,
      "loss": 0.6573,
      "step": 2316
    },
    {
      "epoch": 0.40197779319916727,
      "grad_norm": 0.7335717082023621,
      "learning_rate": 3.999985957825303e-05,
      "loss": 0.6274,
      "step": 2317
    },
    {
      "epoch": 0.40215128383067317,
      "grad_norm": 0.8409496545791626,
      "learning_rate": 3.999983288655318e-05,
      "loss": 0.6637,
      "step": 2318
    },
    {
      "epoch": 0.40232477446217907,
      "grad_norm": 1.3137797117233276,
      "learning_rate": 3.999980387384941e-05,
      "loss": 0.6411,
      "step": 2319
    },
    {
      "epoch": 0.4024982650936849,
      "grad_norm": 0.6379951238632202,
      "learning_rate": 3.9999772540145104e-05,
      "loss": 0.6807,
      "step": 2320
    },
    {
      "epoch": 0.4026717557251908,
      "grad_norm": 0.7253673672676086,
      "learning_rate": 3.9999738885443885e-05,
      "loss": 0.6781,
      "step": 2321
    },
    {
      "epoch": 0.4028452463566967,
      "grad_norm": 0.7153136134147644,
      "learning_rate": 3.999970290974967e-05,
      "loss": 0.6554,
      "step": 2322
    },
    {
      "epoch": 0.4030187369882026,
      "grad_norm": 2.572360038757324,
      "learning_rate": 3.9999664613066615e-05,
      "loss": 0.6593,
      "step": 2323
    },
    {
      "epoch": 0.4031922276197085,
      "grad_norm": 0.7657721042633057,
      "learning_rate": 3.999962399539919e-05,
      "loss": 0.6445,
      "step": 2324
    },
    {
      "epoch": 0.4033657182512144,
      "grad_norm": 2.1976494789123535,
      "learning_rate": 3.9999581056752085e-05,
      "loss": 0.5967,
      "step": 2325
    },
    {
      "epoch": 0.4035392088827203,
      "grad_norm": 0.9684531092643738,
      "learning_rate": 3.9999535797130304e-05,
      "loss": 0.6548,
      "step": 2326
    },
    {
      "epoch": 0.4037126995142262,
      "grad_norm": 0.7884881496429443,
      "learning_rate": 3.999948821653908e-05,
      "loss": 0.6665,
      "step": 2327
    },
    {
      "epoch": 0.4038861901457321,
      "grad_norm": 0.6548993587493896,
      "learning_rate": 3.999943831498395e-05,
      "loss": 0.6117,
      "step": 2328
    },
    {
      "epoch": 0.40405968077723803,
      "grad_norm": 0.7876485586166382,
      "learning_rate": 3.99993860924707e-05,
      "loss": 0.6306,
      "step": 2329
    },
    {
      "epoch": 0.40423317140874393,
      "grad_norm": 0.7914551496505737,
      "learning_rate": 3.9999331549005394e-05,
      "loss": 0.6366,
      "step": 2330
    },
    {
      "epoch": 0.40440666204024983,
      "grad_norm": 0.9158231019973755,
      "learning_rate": 3.999927468459435e-05,
      "loss": 0.5707,
      "step": 2331
    },
    {
      "epoch": 0.40458015267175573,
      "grad_norm": 0.8651548624038696,
      "learning_rate": 3.999921549924418e-05,
      "loss": 0.5966,
      "step": 2332
    },
    {
      "epoch": 0.40475364330326163,
      "grad_norm": 0.7848436236381531,
      "learning_rate": 3.999915399296175e-05,
      "loss": 0.6642,
      "step": 2333
    },
    {
      "epoch": 0.40492713393476754,
      "grad_norm": 0.721215009689331,
      "learning_rate": 3.999909016575419e-05,
      "loss": 0.6163,
      "step": 2334
    },
    {
      "epoch": 0.40510062456627344,
      "grad_norm": 0.8471186757087708,
      "learning_rate": 3.999902401762892e-05,
      "loss": 0.7141,
      "step": 2335
    },
    {
      "epoch": 0.40527411519777934,
      "grad_norm": 0.7264529466629028,
      "learning_rate": 3.99989555485936e-05,
      "loss": 0.7271,
      "step": 2336
    },
    {
      "epoch": 0.40544760582928524,
      "grad_norm": 0.7853133082389832,
      "learning_rate": 3.999888475865619e-05,
      "loss": 0.6869,
      "step": 2337
    },
    {
      "epoch": 0.40562109646079114,
      "grad_norm": 1.0160361528396606,
      "learning_rate": 3.99988116478249e-05,
      "loss": 0.6592,
      "step": 2338
    },
    {
      "epoch": 0.40579458709229704,
      "grad_norm": 1.0822649002075195,
      "learning_rate": 3.999873621610822e-05,
      "loss": 0.554,
      "step": 2339
    },
    {
      "epoch": 0.4059680777238029,
      "grad_norm": 0.8811540007591248,
      "learning_rate": 3.999865846351489e-05,
      "loss": 0.6926,
      "step": 2340
    },
    {
      "epoch": 0.4061415683553088,
      "grad_norm": 0.7317081689834595,
      "learning_rate": 3.999857839005395e-05,
      "loss": 0.5583,
      "step": 2341
    },
    {
      "epoch": 0.4063150589868147,
      "grad_norm": 0.7234801650047302,
      "learning_rate": 3.9998495995734677e-05,
      "loss": 0.6863,
      "step": 2342
    },
    {
      "epoch": 0.4064885496183206,
      "grad_norm": 0.7849730253219604,
      "learning_rate": 3.999841128056664e-05,
      "loss": 0.6707,
      "step": 2343
    },
    {
      "epoch": 0.4066620402498265,
      "grad_norm": 0.6559509634971619,
      "learning_rate": 3.999832424455968e-05,
      "loss": 0.6493,
      "step": 2344
    },
    {
      "epoch": 0.4068355308813324,
      "grad_norm": 0.7420839071273804,
      "learning_rate": 3.999823488772388e-05,
      "loss": 0.5861,
      "step": 2345
    },
    {
      "epoch": 0.4070090215128383,
      "grad_norm": 0.8226618766784668,
      "learning_rate": 3.999814321006963e-05,
      "loss": 0.5984,
      "step": 2346
    },
    {
      "epoch": 0.4071825121443442,
      "grad_norm": 0.8272169828414917,
      "learning_rate": 3.9998049211607546e-05,
      "loss": 0.7593,
      "step": 2347
    },
    {
      "epoch": 0.4073560027758501,
      "grad_norm": 0.6217381954193115,
      "learning_rate": 3.999795289234856e-05,
      "loss": 0.6067,
      "step": 2348
    },
    {
      "epoch": 0.407529493407356,
      "grad_norm": 1.3326365947723389,
      "learning_rate": 3.9997854252303826e-05,
      "loss": 0.524,
      "step": 2349
    },
    {
      "epoch": 0.4077029840388619,
      "grad_norm": 0.9312711954116821,
      "learning_rate": 3.9997753291484816e-05,
      "loss": 0.6646,
      "step": 2350
    },
    {
      "epoch": 0.4078764746703678,
      "grad_norm": 0.7296499609947205,
      "learning_rate": 3.9997650009903226e-05,
      "loss": 0.722,
      "step": 2351
    },
    {
      "epoch": 0.4080499653018737,
      "grad_norm": 0.6808332800865173,
      "learning_rate": 3.999754440757105e-05,
      "loss": 0.6744,
      "step": 2352
    },
    {
      "epoch": 0.4082234559333796,
      "grad_norm": 1.1659239530563354,
      "learning_rate": 3.999743648450055e-05,
      "loss": 0.5813,
      "step": 2353
    },
    {
      "epoch": 0.4083969465648855,
      "grad_norm": 1.0359535217285156,
      "learning_rate": 3.999732624070424e-05,
      "loss": 0.5223,
      "step": 2354
    },
    {
      "epoch": 0.4085704371963914,
      "grad_norm": 0.7763608694076538,
      "learning_rate": 3.999721367619492e-05,
      "loss": 0.6569,
      "step": 2355
    },
    {
      "epoch": 0.4087439278278973,
      "grad_norm": 0.8380506038665771,
      "learning_rate": 3.999709879098565e-05,
      "loss": 0.7256,
      "step": 2356
    },
    {
      "epoch": 0.4089174184594032,
      "grad_norm": 1.13246488571167,
      "learning_rate": 3.999698158508977e-05,
      "loss": 0.5399,
      "step": 2357
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 1.2012083530426025,
      "learning_rate": 3.999686205852087e-05,
      "loss": 0.5126,
      "step": 2358
    },
    {
      "epoch": 0.40926439972241496,
      "grad_norm": 1.7218319177627563,
      "learning_rate": 3.999674021129283e-05,
      "loss": 0.5927,
      "step": 2359
    },
    {
      "epoch": 0.40943789035392086,
      "grad_norm": 1.1242047548294067,
      "learning_rate": 3.999661604341978e-05,
      "loss": 0.4895,
      "step": 2360
    },
    {
      "epoch": 0.40961138098542677,
      "grad_norm": 0.841090977191925,
      "learning_rate": 3.9996489554916145e-05,
      "loss": 0.6304,
      "step": 2361
    },
    {
      "epoch": 0.40978487161693267,
      "grad_norm": 1.1248053312301636,
      "learning_rate": 3.99963607457966e-05,
      "loss": 0.6895,
      "step": 2362
    },
    {
      "epoch": 0.40995836224843857,
      "grad_norm": 0.9558753967285156,
      "learning_rate": 3.9996229616076086e-05,
      "loss": 0.7213,
      "step": 2363
    },
    {
      "epoch": 0.41013185287994447,
      "grad_norm": 1.1117990016937256,
      "learning_rate": 3.999609616576982e-05,
      "loss": 0.699,
      "step": 2364
    },
    {
      "epoch": 0.41030534351145037,
      "grad_norm": 0.8498603701591492,
      "learning_rate": 3.99959603948933e-05,
      "loss": 0.6716,
      "step": 2365
    },
    {
      "epoch": 0.4104788341429563,
      "grad_norm": 0.6780929565429688,
      "learning_rate": 3.9995822303462273e-05,
      "loss": 0.6489,
      "step": 2366
    },
    {
      "epoch": 0.4106523247744622,
      "grad_norm": 1.1830792427062988,
      "learning_rate": 3.9995681891492774e-05,
      "loss": 0.7004,
      "step": 2367
    },
    {
      "epoch": 0.4108258154059681,
      "grad_norm": 0.7626097202301025,
      "learning_rate": 3.9995539159001074e-05,
      "loss": 0.6608,
      "step": 2368
    },
    {
      "epoch": 0.410999306037474,
      "grad_norm": 0.8115872740745544,
      "learning_rate": 3.999539410600378e-05,
      "loss": 0.6238,
      "step": 2369
    },
    {
      "epoch": 0.4111727966689799,
      "grad_norm": 0.9422028660774231,
      "learning_rate": 3.999524673251768e-05,
      "loss": 0.6399,
      "step": 2370
    },
    {
      "epoch": 0.4113462873004858,
      "grad_norm": 0.8287187218666077,
      "learning_rate": 3.999509703855991e-05,
      "loss": 0.7007,
      "step": 2371
    },
    {
      "epoch": 0.4115197779319917,
      "grad_norm": 0.9184843301773071,
      "learning_rate": 3.999494502414783e-05,
      "loss": 0.6715,
      "step": 2372
    },
    {
      "epoch": 0.4116932685634976,
      "grad_norm": 0.680824339389801,
      "learning_rate": 3.999479068929907e-05,
      "loss": 0.6639,
      "step": 2373
    },
    {
      "epoch": 0.4118667591950035,
      "grad_norm": 1.0199631452560425,
      "learning_rate": 3.999463403403156e-05,
      "loss": 0.6597,
      "step": 2374
    },
    {
      "epoch": 0.4120402498265094,
      "grad_norm": 0.8106906414031982,
      "learning_rate": 3.999447505836347e-05,
      "loss": 0.6565,
      "step": 2375
    },
    {
      "epoch": 0.4122137404580153,
      "grad_norm": 0.8393969535827637,
      "learning_rate": 3.999431376231326e-05,
      "loss": 0.679,
      "step": 2376
    },
    {
      "epoch": 0.4123872310895212,
      "grad_norm": 1.0296335220336914,
      "learning_rate": 3.999415014589963e-05,
      "loss": 0.5259,
      "step": 2377
    },
    {
      "epoch": 0.4125607217210271,
      "grad_norm": 1.2408167123794556,
      "learning_rate": 3.9993984209141576e-05,
      "loss": 0.4758,
      "step": 2378
    },
    {
      "epoch": 0.41273421235253294,
      "grad_norm": 0.6542762517929077,
      "learning_rate": 3.999381595205836e-05,
      "loss": 0.6488,
      "step": 2379
    },
    {
      "epoch": 0.41290770298403884,
      "grad_norm": 0.9614208340644836,
      "learning_rate": 3.999364537466951e-05,
      "loss": 0.5717,
      "step": 2380
    },
    {
      "epoch": 0.41308119361554474,
      "grad_norm": 0.9062381386756897,
      "learning_rate": 3.999347247699481e-05,
      "loss": 0.6156,
      "step": 2381
    },
    {
      "epoch": 0.41325468424705064,
      "grad_norm": 0.7734870314598083,
      "learning_rate": 3.999329725905434e-05,
      "loss": 0.5577,
      "step": 2382
    },
    {
      "epoch": 0.41342817487855654,
      "grad_norm": 0.7082153558731079,
      "learning_rate": 3.999311972086842e-05,
      "loss": 0.5798,
      "step": 2383
    },
    {
      "epoch": 0.41360166551006244,
      "grad_norm": 0.83006751537323,
      "learning_rate": 3.999293986245766e-05,
      "loss": 0.6296,
      "step": 2384
    },
    {
      "epoch": 0.41377515614156835,
      "grad_norm": 1.2090071439743042,
      "learning_rate": 3.999275768384294e-05,
      "loss": 0.5714,
      "step": 2385
    },
    {
      "epoch": 0.41394864677307425,
      "grad_norm": 0.6968817710876465,
      "learning_rate": 3.9992573185045386e-05,
      "loss": 0.5883,
      "step": 2386
    },
    {
      "epoch": 0.41412213740458015,
      "grad_norm": 1.0211082696914673,
      "learning_rate": 3.9992386366086415e-05,
      "loss": 0.693,
      "step": 2387
    },
    {
      "epoch": 0.41429562803608605,
      "grad_norm": 1.071164846420288,
      "learning_rate": 3.9992197226987725e-05,
      "loss": 0.5638,
      "step": 2388
    },
    {
      "epoch": 0.41446911866759195,
      "grad_norm": 1.1762923002243042,
      "learning_rate": 3.9992005767771236e-05,
      "loss": 0.616,
      "step": 2389
    },
    {
      "epoch": 0.41464260929909785,
      "grad_norm": 1.2997300624847412,
      "learning_rate": 3.999181198845919e-05,
      "loss": 0.571,
      "step": 2390
    },
    {
      "epoch": 0.41481609993060375,
      "grad_norm": 0.7499904632568359,
      "learning_rate": 3.9991615889074065e-05,
      "loss": 0.7039,
      "step": 2391
    },
    {
      "epoch": 0.41498959056210966,
      "grad_norm": 1.0127307176589966,
      "learning_rate": 3.999141746963862e-05,
      "loss": 0.6185,
      "step": 2392
    },
    {
      "epoch": 0.41516308119361556,
      "grad_norm": 1.1571502685546875,
      "learning_rate": 3.999121673017589e-05,
      "loss": 0.5881,
      "step": 2393
    },
    {
      "epoch": 0.41533657182512146,
      "grad_norm": 0.8128833770751953,
      "learning_rate": 3.999101367070916e-05,
      "loss": 0.5934,
      "step": 2394
    },
    {
      "epoch": 0.41551006245662736,
      "grad_norm": 1.1349921226501465,
      "learning_rate": 3.9990808291262e-05,
      "loss": 0.5839,
      "step": 2395
    },
    {
      "epoch": 0.41568355308813326,
      "grad_norm": 1.5129714012145996,
      "learning_rate": 3.9990600591858244e-05,
      "loss": 0.676,
      "step": 2396
    },
    {
      "epoch": 0.41585704371963916,
      "grad_norm": 0.7934991717338562,
      "learning_rate": 3.9990390572522e-05,
      "loss": 0.6084,
      "step": 2397
    },
    {
      "epoch": 0.41603053435114506,
      "grad_norm": 0.8803777694702148,
      "learning_rate": 3.999017823327762e-05,
      "loss": 0.6567,
      "step": 2398
    },
    {
      "epoch": 0.4162040249826509,
      "grad_norm": 0.8170890808105469,
      "learning_rate": 3.998996357414978e-05,
      "loss": 0.7646,
      "step": 2399
    },
    {
      "epoch": 0.4163775156141568,
      "grad_norm": 0.7423005700111389,
      "learning_rate": 3.9989746595163364e-05,
      "loss": 0.587,
      "step": 2400
    },
    {
      "epoch": 0.4165510062456627,
      "grad_norm": 1.7788679599761963,
      "learning_rate": 3.998952729634357e-05,
      "loss": 0.6564,
      "step": 2401
    },
    {
      "epoch": 0.4167244968771686,
      "grad_norm": 1.6209176778793335,
      "learning_rate": 3.998930567771583e-05,
      "loss": 0.666,
      "step": 2402
    },
    {
      "epoch": 0.4168979875086745,
      "grad_norm": 0.9754631519317627,
      "learning_rate": 3.998908173930589e-05,
      "loss": 0.5767,
      "step": 2403
    },
    {
      "epoch": 0.4170714781401804,
      "grad_norm": 1.4091540575027466,
      "learning_rate": 3.998885548113971e-05,
      "loss": 0.5999,
      "step": 2404
    },
    {
      "epoch": 0.4172449687716863,
      "grad_norm": 2.70163631439209,
      "learning_rate": 3.998862690324357e-05,
      "loss": 0.7211,
      "step": 2405
    },
    {
      "epoch": 0.4174184594031922,
      "grad_norm": 1.098078727722168,
      "learning_rate": 3.998839600564398e-05,
      "loss": 0.5623,
      "step": 2406
    },
    {
      "epoch": 0.4175919500346981,
      "grad_norm": 1.1452560424804688,
      "learning_rate": 3.9988162788367744e-05,
      "loss": 0.6671,
      "step": 2407
    },
    {
      "epoch": 0.417765440666204,
      "grad_norm": 1.117812156677246,
      "learning_rate": 3.998792725144192e-05,
      "loss": 0.5447,
      "step": 2408
    },
    {
      "epoch": 0.4179389312977099,
      "grad_norm": 1.441041350364685,
      "learning_rate": 3.9987689394893855e-05,
      "loss": 0.6554,
      "step": 2409
    },
    {
      "epoch": 0.4181124219292158,
      "grad_norm": 1.0478700399398804,
      "learning_rate": 3.9987449218751134e-05,
      "loss": 0.66,
      "step": 2410
    },
    {
      "epoch": 0.41828591256072173,
      "grad_norm": 1.5800617933273315,
      "learning_rate": 3.9987206723041654e-05,
      "loss": 0.6804,
      "step": 2411
    },
    {
      "epoch": 0.41845940319222763,
      "grad_norm": 1.2367937564849854,
      "learning_rate": 3.998696190779354e-05,
      "loss": 0.5527,
      "step": 2412
    },
    {
      "epoch": 0.41863289382373353,
      "grad_norm": 1.0617979764938354,
      "learning_rate": 3.9986714773035207e-05,
      "loss": 0.6426,
      "step": 2413
    },
    {
      "epoch": 0.41880638445523943,
      "grad_norm": 1.3675651550292969,
      "learning_rate": 3.9986465318795336e-05,
      "loss": 0.6241,
      "step": 2414
    },
    {
      "epoch": 0.41897987508674533,
      "grad_norm": 1.568328619003296,
      "learning_rate": 3.998621354510288e-05,
      "loss": 0.5544,
      "step": 2415
    },
    {
      "epoch": 0.41915336571825124,
      "grad_norm": 0.8526886105537415,
      "learning_rate": 3.998595945198705e-05,
      "loss": 0.7128,
      "step": 2416
    },
    {
      "epoch": 0.41932685634975714,
      "grad_norm": 0.8320168852806091,
      "learning_rate": 3.998570303947733e-05,
      "loss": 0.6757,
      "step": 2417
    },
    {
      "epoch": 0.41950034698126304,
      "grad_norm": 0.9623420834541321,
      "learning_rate": 3.9985444307603497e-05,
      "loss": 0.5509,
      "step": 2418
    },
    {
      "epoch": 0.4196738376127689,
      "grad_norm": 0.8692437410354614,
      "learning_rate": 3.998518325639556e-05,
      "loss": 0.6351,
      "step": 2419
    },
    {
      "epoch": 0.4198473282442748,
      "grad_norm": 1.0957146883010864,
      "learning_rate": 3.998491988588381e-05,
      "loss": 0.6073,
      "step": 2420
    },
    {
      "epoch": 0.4200208188757807,
      "grad_norm": 1.278006672859192,
      "learning_rate": 3.9984654196098825e-05,
      "loss": 0.5454,
      "step": 2421
    },
    {
      "epoch": 0.4201943095072866,
      "grad_norm": 0.9353867173194885,
      "learning_rate": 3.998438618707144e-05,
      "loss": 0.6455,
      "step": 2422
    },
    {
      "epoch": 0.4203678001387925,
      "grad_norm": 0.9579692482948303,
      "learning_rate": 3.998411585883274e-05,
      "loss": 0.5718,
      "step": 2423
    },
    {
      "epoch": 0.4205412907702984,
      "grad_norm": 0.7727288603782654,
      "learning_rate": 3.9983843211414124e-05,
      "loss": 0.6592,
      "step": 2424
    },
    {
      "epoch": 0.4207147814018043,
      "grad_norm": 0.9824633002281189,
      "learning_rate": 3.998356824484721e-05,
      "loss": 0.6136,
      "step": 2425
    },
    {
      "epoch": 0.4208882720333102,
      "grad_norm": 1.1013127565383911,
      "learning_rate": 3.9983290959163914e-05,
      "loss": 0.6833,
      "step": 2426
    },
    {
      "epoch": 0.4210617626648161,
      "grad_norm": 1.01968252658844,
      "learning_rate": 3.998301135439642e-05,
      "loss": 0.7205,
      "step": 2427
    },
    {
      "epoch": 0.421235253296322,
      "grad_norm": 1.1779837608337402,
      "learning_rate": 3.998272943057717e-05,
      "loss": 0.7167,
      "step": 2428
    },
    {
      "epoch": 0.4214087439278279,
      "grad_norm": 1.0232571363449097,
      "learning_rate": 3.9982445187738885e-05,
      "loss": 0.6255,
      "step": 2429
    },
    {
      "epoch": 0.4215822345593338,
      "grad_norm": 0.6751450896263123,
      "learning_rate": 3.998215862591455e-05,
      "loss": 0.6083,
      "step": 2430
    },
    {
      "epoch": 0.4217557251908397,
      "grad_norm": 1.1975443363189697,
      "learning_rate": 3.998186974513743e-05,
      "loss": 0.6542,
      "step": 2431
    },
    {
      "epoch": 0.4219292158223456,
      "grad_norm": 0.7380514144897461,
      "learning_rate": 3.998157854544104e-05,
      "loss": 0.7418,
      "step": 2432
    },
    {
      "epoch": 0.4221027064538515,
      "grad_norm": 0.8603578209877014,
      "learning_rate": 3.998128502685917e-05,
      "loss": 0.6229,
      "step": 2433
    },
    {
      "epoch": 0.4222761970853574,
      "grad_norm": 1.667292594909668,
      "learning_rate": 3.99809891894259e-05,
      "loss": 0.5339,
      "step": 2434
    },
    {
      "epoch": 0.4224496877168633,
      "grad_norm": 2.614480972290039,
      "learning_rate": 3.998069103317555e-05,
      "loss": 0.597,
      "step": 2435
    },
    {
      "epoch": 0.4226231783483692,
      "grad_norm": 0.8586104512214661,
      "learning_rate": 3.998039055814272e-05,
      "loss": 0.7209,
      "step": 2436
    },
    {
      "epoch": 0.4227966689798751,
      "grad_norm": 0.9705440998077393,
      "learning_rate": 3.998008776436228e-05,
      "loss": 0.6768,
      "step": 2437
    },
    {
      "epoch": 0.42297015961138096,
      "grad_norm": 0.8429592847824097,
      "learning_rate": 3.9979782651869384e-05,
      "loss": 0.6238,
      "step": 2438
    },
    {
      "epoch": 0.42314365024288686,
      "grad_norm": 1.0563840866088867,
      "learning_rate": 3.997947522069942e-05,
      "loss": 0.5486,
      "step": 2439
    },
    {
      "epoch": 0.42331714087439276,
      "grad_norm": 2.9745025634765625,
      "learning_rate": 3.997916547088808e-05,
      "loss": 0.5063,
      "step": 2440
    },
    {
      "epoch": 0.42349063150589866,
      "grad_norm": 1.5186116695404053,
      "learning_rate": 3.9978853402471306e-05,
      "loss": 0.693,
      "step": 2441
    },
    {
      "epoch": 0.42366412213740456,
      "grad_norm": 1.7089169025421143,
      "learning_rate": 3.997853901548532e-05,
      "loss": 0.5458,
      "step": 2442
    },
    {
      "epoch": 0.42383761276891047,
      "grad_norm": 1.2330875396728516,
      "learning_rate": 3.9978222309966594e-05,
      "loss": 0.6323,
      "step": 2443
    },
    {
      "epoch": 0.42401110340041637,
      "grad_norm": 1.1696192026138306,
      "learning_rate": 3.9977903285951896e-05,
      "loss": 0.5769,
      "step": 2444
    },
    {
      "epoch": 0.42418459403192227,
      "grad_norm": 1.1445473432540894,
      "learning_rate": 3.9977581943478236e-05,
      "loss": 0.5812,
      "step": 2445
    },
    {
      "epoch": 0.42435808466342817,
      "grad_norm": 1.298656702041626,
      "learning_rate": 3.9977258282582916e-05,
      "loss": 0.5652,
      "step": 2446
    },
    {
      "epoch": 0.42453157529493407,
      "grad_norm": 0.8933175206184387,
      "learning_rate": 3.99769323033035e-05,
      "loss": 0.6921,
      "step": 2447
    },
    {
      "epoch": 0.42470506592644,
      "grad_norm": 0.7664970755577087,
      "learning_rate": 3.99766040056778e-05,
      "loss": 0.7361,
      "step": 2448
    },
    {
      "epoch": 0.4248785565579459,
      "grad_norm": 0.8455670475959778,
      "learning_rate": 3.997627338974394e-05,
      "loss": 0.6073,
      "step": 2449
    },
    {
      "epoch": 0.4250520471894518,
      "grad_norm": 1.244389533996582,
      "learning_rate": 3.997594045554027e-05,
      "loss": 0.5999,
      "step": 2450
    },
    {
      "epoch": 0.4252255378209577,
      "grad_norm": 1.0283491611480713,
      "learning_rate": 3.9975605203105434e-05,
      "loss": 0.6758,
      "step": 2451
    },
    {
      "epoch": 0.4253990284524636,
      "grad_norm": 0.840735673904419,
      "learning_rate": 3.9975267632478336e-05,
      "loss": 0.6875,
      "step": 2452
    },
    {
      "epoch": 0.4255725190839695,
      "grad_norm": 0.906441867351532,
      "learning_rate": 3.997492774369816e-05,
      "loss": 0.6664,
      "step": 2453
    },
    {
      "epoch": 0.4257460097154754,
      "grad_norm": 1.0706926584243774,
      "learning_rate": 3.997458553680434e-05,
      "loss": 0.6334,
      "step": 2454
    },
    {
      "epoch": 0.4259195003469813,
      "grad_norm": 1.6295770406723022,
      "learning_rate": 3.9974241011836594e-05,
      "loss": 0.5881,
      "step": 2455
    },
    {
      "epoch": 0.4260929909784872,
      "grad_norm": 1.4740289449691772,
      "learning_rate": 3.99738941688349e-05,
      "loss": 0.73,
      "step": 2456
    },
    {
      "epoch": 0.4262664816099931,
      "grad_norm": 0.9593019485473633,
      "learning_rate": 3.997354500783952e-05,
      "loss": 0.5723,
      "step": 2457
    },
    {
      "epoch": 0.42643997224149893,
      "grad_norm": 0.8586723804473877,
      "learning_rate": 3.997319352889096e-05,
      "loss": 0.6282,
      "step": 2458
    },
    {
      "epoch": 0.42661346287300483,
      "grad_norm": 1.7914944887161255,
      "learning_rate": 3.997283973203003e-05,
      "loss": 0.6292,
      "step": 2459
    },
    {
      "epoch": 0.42678695350451074,
      "grad_norm": 1.2020705938339233,
      "learning_rate": 3.997248361729777e-05,
      "loss": 0.588,
      "step": 2460
    },
    {
      "epoch": 0.42696044413601664,
      "grad_norm": 1.2364118099212646,
      "learning_rate": 3.9972125184735505e-05,
      "loss": 0.6376,
      "step": 2461
    },
    {
      "epoch": 0.42713393476752254,
      "grad_norm": 2.9780807495117188,
      "learning_rate": 3.997176443438485e-05,
      "loss": 0.6693,
      "step": 2462
    },
    {
      "epoch": 0.42730742539902844,
      "grad_norm": 1.4414304494857788,
      "learning_rate": 3.9971401366287666e-05,
      "loss": 0.516,
      "step": 2463
    },
    {
      "epoch": 0.42748091603053434,
      "grad_norm": 1.0700013637542725,
      "learning_rate": 3.997103598048607e-05,
      "loss": 0.551,
      "step": 2464
    },
    {
      "epoch": 0.42765440666204024,
      "grad_norm": 0.8656531572341919,
      "learning_rate": 3.997066827702248e-05,
      "loss": 0.535,
      "step": 2465
    },
    {
      "epoch": 0.42782789729354614,
      "grad_norm": 1.2760169506072998,
      "learning_rate": 3.9970298255939564e-05,
      "loss": 0.5417,
      "step": 2466
    },
    {
      "epoch": 0.42800138792505205,
      "grad_norm": 1.130365252494812,
      "learning_rate": 3.9969925917280276e-05,
      "loss": 0.6696,
      "step": 2467
    },
    {
      "epoch": 0.42817487855655795,
      "grad_norm": 0.8799391984939575,
      "learning_rate": 3.9969551261087806e-05,
      "loss": 0.5598,
      "step": 2468
    },
    {
      "epoch": 0.42834836918806385,
      "grad_norm": 1.3091599941253662,
      "learning_rate": 3.996917428740565e-05,
      "loss": 0.5651,
      "step": 2469
    },
    {
      "epoch": 0.42852185981956975,
      "grad_norm": 1.28733491897583,
      "learning_rate": 3.996879499627754e-05,
      "loss": 0.5994,
      "step": 2470
    },
    {
      "epoch": 0.42869535045107565,
      "grad_norm": 1.048340916633606,
      "learning_rate": 3.996841338774751e-05,
      "loss": 0.6121,
      "step": 2471
    },
    {
      "epoch": 0.42886884108258155,
      "grad_norm": 1.209240198135376,
      "learning_rate": 3.996802946185984e-05,
      "loss": 0.7847,
      "step": 2472
    },
    {
      "epoch": 0.42904233171408745,
      "grad_norm": 1.1296306848526,
      "learning_rate": 3.996764321865907e-05,
      "loss": 0.6157,
      "step": 2473
    },
    {
      "epoch": 0.42921582234559336,
      "grad_norm": 0.9123029708862305,
      "learning_rate": 3.9967254658190055e-05,
      "loss": 0.5289,
      "step": 2474
    },
    {
      "epoch": 0.42938931297709926,
      "grad_norm": 1.675567626953125,
      "learning_rate": 3.996686378049786e-05,
      "loss": 0.6039,
      "step": 2475
    },
    {
      "epoch": 0.42956280360860516,
      "grad_norm": 0.8229936957359314,
      "learning_rate": 3.996647058562786e-05,
      "loss": 0.6963,
      "step": 2476
    },
    {
      "epoch": 0.42973629424011106,
      "grad_norm": 1.2111272811889648,
      "learning_rate": 3.9966075073625684e-05,
      "loss": 0.6133,
      "step": 2477
    },
    {
      "epoch": 0.4299097848716169,
      "grad_norm": 2.042346715927124,
      "learning_rate": 3.9965677244537226e-05,
      "loss": 0.703,
      "step": 2478
    },
    {
      "epoch": 0.4300832755031228,
      "grad_norm": 0.8567150235176086,
      "learning_rate": 3.9965277098408666e-05,
      "loss": 0.6338,
      "step": 2479
    },
    {
      "epoch": 0.4302567661346287,
      "grad_norm": 0.9839659929275513,
      "learning_rate": 3.9964874635286436e-05,
      "loss": 0.6497,
      "step": 2480
    },
    {
      "epoch": 0.4304302567661346,
      "grad_norm": 1.4143627882003784,
      "learning_rate": 3.996446985521723e-05,
      "loss": 0.5919,
      "step": 2481
    },
    {
      "epoch": 0.4306037473976405,
      "grad_norm": 1.1674302816390991,
      "learning_rate": 3.996406275824804e-05,
      "loss": 0.6061,
      "step": 2482
    },
    {
      "epoch": 0.4307772380291464,
      "grad_norm": 1.1941511631011963,
      "learning_rate": 3.996365334442611e-05,
      "loss": 0.708,
      "step": 2483
    },
    {
      "epoch": 0.4309507286606523,
      "grad_norm": 1.0614922046661377,
      "learning_rate": 3.996324161379894e-05,
      "loss": 0.4954,
      "step": 2484
    },
    {
      "epoch": 0.4311242192921582,
      "grad_norm": 1.6599159240722656,
      "learning_rate": 3.996282756641432e-05,
      "loss": 0.6565,
      "step": 2485
    },
    {
      "epoch": 0.4312977099236641,
      "grad_norm": 1.4952170848846436,
      "learning_rate": 3.9962411202320296e-05,
      "loss": 0.5383,
      "step": 2486
    },
    {
      "epoch": 0.43147120055517,
      "grad_norm": 1.8975037336349487,
      "learning_rate": 3.99619925215652e-05,
      "loss": 0.7227,
      "step": 2487
    },
    {
      "epoch": 0.4316446911866759,
      "grad_norm": 1.1745805740356445,
      "learning_rate": 3.99615715241976e-05,
      "loss": 0.6337,
      "step": 2488
    },
    {
      "epoch": 0.4318181818181818,
      "grad_norm": 2.31195068359375,
      "learning_rate": 3.996114821026638e-05,
      "loss": 0.6847,
      "step": 2489
    },
    {
      "epoch": 0.4319916724496877,
      "grad_norm": 1.2115309238433838,
      "learning_rate": 3.996072257982064e-05,
      "loss": 0.6565,
      "step": 2490
    },
    {
      "epoch": 0.4321651630811936,
      "grad_norm": 2.326763153076172,
      "learning_rate": 3.996029463290978e-05,
      "loss": 0.693,
      "step": 2491
    },
    {
      "epoch": 0.4323386537126995,
      "grad_norm": 0.8984289169311523,
      "learning_rate": 3.9959864369583485e-05,
      "loss": 0.5587,
      "step": 2492
    },
    {
      "epoch": 0.43251214434420543,
      "grad_norm": 1.6303499937057495,
      "learning_rate": 3.9959431789891665e-05,
      "loss": 0.7015,
      "step": 2493
    },
    {
      "epoch": 0.43268563497571133,
      "grad_norm": 1.4040231704711914,
      "learning_rate": 3.9958996893884525e-05,
      "loss": 0.6107,
      "step": 2494
    },
    {
      "epoch": 0.43285912560721723,
      "grad_norm": 2.0959885120391846,
      "learning_rate": 3.9958559681612544e-05,
      "loss": 0.5842,
      "step": 2495
    },
    {
      "epoch": 0.43303261623872313,
      "grad_norm": 6.375994682312012,
      "learning_rate": 3.9958120153126454e-05,
      "loss": 0.7133,
      "step": 2496
    },
    {
      "epoch": 0.43320610687022904,
      "grad_norm": 2.2398011684417725,
      "learning_rate": 3.995767830847726e-05,
      "loss": 0.627,
      "step": 2497
    },
    {
      "epoch": 0.4333795975017349,
      "grad_norm": 0.9163916707038879,
      "learning_rate": 3.995723414771625e-05,
      "loss": 0.567,
      "step": 2498
    },
    {
      "epoch": 0.4335530881332408,
      "grad_norm": 3.787116289138794,
      "learning_rate": 3.9956787670894954e-05,
      "loss": 0.5774,
      "step": 2499
    },
    {
      "epoch": 0.4337265787647467,
      "grad_norm": 0.7828592658042908,
      "learning_rate": 3.9956338878065205e-05,
      "loss": 0.6929,
      "step": 2500
    },
    {
      "epoch": 0.4339000693962526,
      "grad_norm": 0.9084985852241516,
      "learning_rate": 3.995588776927907e-05,
      "loss": 0.5309,
      "step": 2501
    },
    {
      "epoch": 0.4340735600277585,
      "grad_norm": 1.0282264947891235,
      "learning_rate": 3.99554343445889e-05,
      "loss": 0.5529,
      "step": 2502
    },
    {
      "epoch": 0.4342470506592644,
      "grad_norm": 1.0289828777313232,
      "learning_rate": 3.995497860404733e-05,
      "loss": 0.5729,
      "step": 2503
    },
    {
      "epoch": 0.4344205412907703,
      "grad_norm": 1.0444369316101074,
      "learning_rate": 3.995452054770724e-05,
      "loss": 0.5541,
      "step": 2504
    },
    {
      "epoch": 0.4345940319222762,
      "grad_norm": 0.9299532175064087,
      "learning_rate": 3.995406017562179e-05,
      "loss": 0.6864,
      "step": 2505
    },
    {
      "epoch": 0.4347675225537821,
      "grad_norm": 2.2198166847229004,
      "learning_rate": 3.99535974878444e-05,
      "loss": 0.5369,
      "step": 2506
    },
    {
      "epoch": 0.434941013185288,
      "grad_norm": 1.6497248411178589,
      "learning_rate": 3.995313248442878e-05,
      "loss": 0.5876,
      "step": 2507
    },
    {
      "epoch": 0.4351145038167939,
      "grad_norm": 1.0134512186050415,
      "learning_rate": 3.995266516542887e-05,
      "loss": 0.7045,
      "step": 2508
    },
    {
      "epoch": 0.4352879944482998,
      "grad_norm": 0.901813805103302,
      "learning_rate": 3.9952195530898926e-05,
      "loss": 0.5161,
      "step": 2509
    },
    {
      "epoch": 0.4354614850798057,
      "grad_norm": 1.057944655418396,
      "learning_rate": 3.995172358089344e-05,
      "loss": 0.5239,
      "step": 2510
    },
    {
      "epoch": 0.4356349757113116,
      "grad_norm": 0.9936065673828125,
      "learning_rate": 3.9951249315467194e-05,
      "loss": 0.6049,
      "step": 2511
    },
    {
      "epoch": 0.4358084663428175,
      "grad_norm": 1.1527905464172363,
      "learning_rate": 3.995077273467521e-05,
      "loss": 0.5753,
      "step": 2512
    },
    {
      "epoch": 0.4359819569743234,
      "grad_norm": 5.697759628295898,
      "learning_rate": 3.99502938385728e-05,
      "loss": 0.6766,
      "step": 2513
    },
    {
      "epoch": 0.4361554476058293,
      "grad_norm": 1.6759312152862549,
      "learning_rate": 3.994981262721555e-05,
      "loss": 0.5149,
      "step": 2514
    },
    {
      "epoch": 0.4363289382373352,
      "grad_norm": 1.1284745931625366,
      "learning_rate": 3.994932910065929e-05,
      "loss": 0.5564,
      "step": 2515
    },
    {
      "epoch": 0.4365024288688411,
      "grad_norm": 0.9783058762550354,
      "learning_rate": 3.9948843258960154e-05,
      "loss": 0.5891,
      "step": 2516
    },
    {
      "epoch": 0.43667591950034695,
      "grad_norm": 1.159525990486145,
      "learning_rate": 3.9948355102174503e-05,
      "loss": 0.6738,
      "step": 2517
    },
    {
      "epoch": 0.43684941013185286,
      "grad_norm": 0.8247144222259521,
      "learning_rate": 3.9947864630359005e-05,
      "loss": 0.5682,
      "step": 2518
    },
    {
      "epoch": 0.43702290076335876,
      "grad_norm": 1.1474746465682983,
      "learning_rate": 3.9947371843570565e-05,
      "loss": 0.6608,
      "step": 2519
    },
    {
      "epoch": 0.43719639139486466,
      "grad_norm": 0.994800865650177,
      "learning_rate": 3.994687674186638e-05,
      "loss": 0.5623,
      "step": 2520
    },
    {
      "epoch": 0.43736988202637056,
      "grad_norm": 1.6168469190597534,
      "learning_rate": 3.994637932530391e-05,
      "loss": 0.6833,
      "step": 2521
    },
    {
      "epoch": 0.43754337265787646,
      "grad_norm": 0.8193566203117371,
      "learning_rate": 3.9945879593940874e-05,
      "loss": 0.7202,
      "step": 2522
    },
    {
      "epoch": 0.43771686328938236,
      "grad_norm": 1.0501503944396973,
      "learning_rate": 3.994537754783527e-05,
      "loss": 0.6912,
      "step": 2523
    },
    {
      "epoch": 0.43789035392088826,
      "grad_norm": 0.9368448257446289,
      "learning_rate": 3.994487318704536e-05,
      "loss": 0.616,
      "step": 2524
    },
    {
      "epoch": 0.43806384455239417,
      "grad_norm": 0.9665560722351074,
      "learning_rate": 3.994436651162969e-05,
      "loss": 0.5624,
      "step": 2525
    },
    {
      "epoch": 0.43823733518390007,
      "grad_norm": 0.9095457196235657,
      "learning_rate": 3.994385752164703e-05,
      "loss": 0.561,
      "step": 2526
    },
    {
      "epoch": 0.43841082581540597,
      "grad_norm": 1.1988638639450073,
      "learning_rate": 3.994334621715647e-05,
      "loss": 0.5768,
      "step": 2527
    },
    {
      "epoch": 0.43858431644691187,
      "grad_norm": 0.9439307451248169,
      "learning_rate": 3.9942832598217345e-05,
      "loss": 0.6707,
      "step": 2528
    },
    {
      "epoch": 0.43875780707841777,
      "grad_norm": 1.55705726146698,
      "learning_rate": 3.9942316664889255e-05,
      "loss": 0.543,
      "step": 2529
    },
    {
      "epoch": 0.4389312977099237,
      "grad_norm": 1.1340587139129639,
      "learning_rate": 3.9941798417232084e-05,
      "loss": 0.5366,
      "step": 2530
    },
    {
      "epoch": 0.4391047883414296,
      "grad_norm": 1.848889708518982,
      "learning_rate": 3.994127785530596e-05,
      "loss": 0.5619,
      "step": 2531
    },
    {
      "epoch": 0.4392782789729355,
      "grad_norm": 0.8365806341171265,
      "learning_rate": 3.9940754979171317e-05,
      "loss": 0.6888,
      "step": 2532
    },
    {
      "epoch": 0.4394517696044414,
      "grad_norm": 0.9113152027130127,
      "learning_rate": 3.994022978888882e-05,
      "loss": 0.5945,
      "step": 2533
    },
    {
      "epoch": 0.4396252602359473,
      "grad_norm": 0.8680764436721802,
      "learning_rate": 3.9939702284519416e-05,
      "loss": 0.5983,
      "step": 2534
    },
    {
      "epoch": 0.4397987508674532,
      "grad_norm": 1.0888547897338867,
      "learning_rate": 3.993917246612433e-05,
      "loss": 0.5275,
      "step": 2535
    },
    {
      "epoch": 0.4399722414989591,
      "grad_norm": 0.8546169400215149,
      "learning_rate": 3.9938640333765046e-05,
      "loss": 0.5636,
      "step": 2536
    },
    {
      "epoch": 0.44014573213046493,
      "grad_norm": 0.8106816411018372,
      "learning_rate": 3.993810588750332e-05,
      "loss": 0.6598,
      "step": 2537
    },
    {
      "epoch": 0.44031922276197083,
      "grad_norm": 1.2467848062515259,
      "learning_rate": 3.993756912740117e-05,
      "loss": 0.7937,
      "step": 2538
    },
    {
      "epoch": 0.44049271339347673,
      "grad_norm": 0.7479244470596313,
      "learning_rate": 3.993703005352089e-05,
      "loss": 0.6787,
      "step": 2539
    },
    {
      "epoch": 0.44066620402498263,
      "grad_norm": 0.8020234107971191,
      "learning_rate": 3.9936488665925045e-05,
      "loss": 0.7542,
      "step": 2540
    },
    {
      "epoch": 0.44083969465648853,
      "grad_norm": 0.9814688563346863,
      "learning_rate": 3.993594496467646e-05,
      "loss": 0.5591,
      "step": 2541
    },
    {
      "epoch": 0.44101318528799444,
      "grad_norm": 0.9027915596961975,
      "learning_rate": 3.993539894983823e-05,
      "loss": 0.549,
      "step": 2542
    },
    {
      "epoch": 0.44118667591950034,
      "grad_norm": 1.0879294872283936,
      "learning_rate": 3.993485062147372e-05,
      "loss": 0.5062,
      "step": 2543
    },
    {
      "epoch": 0.44136016655100624,
      "grad_norm": 0.9362106323242188,
      "learning_rate": 3.993429997964657e-05,
      "loss": 0.6223,
      "step": 2544
    },
    {
      "epoch": 0.44153365718251214,
      "grad_norm": 0.9086353778839111,
      "learning_rate": 3.993374702442068e-05,
      "loss": 0.6256,
      "step": 2545
    },
    {
      "epoch": 0.44170714781401804,
      "grad_norm": 0.7001194357872009,
      "learning_rate": 3.993319175586021e-05,
      "loss": 0.663,
      "step": 2546
    },
    {
      "epoch": 0.44188063844552394,
      "grad_norm": 0.9094083905220032,
      "learning_rate": 3.993263417402962e-05,
      "loss": 0.6094,
      "step": 2547
    },
    {
      "epoch": 0.44205412907702984,
      "grad_norm": 1.1739511489868164,
      "learning_rate": 3.9932074278993604e-05,
      "loss": 0.5942,
      "step": 2548
    },
    {
      "epoch": 0.44222761970853575,
      "grad_norm": 1.1181588172912598,
      "learning_rate": 3.993151207081715e-05,
      "loss": 0.6553,
      "step": 2549
    },
    {
      "epoch": 0.44240111034004165,
      "grad_norm": 3.685606002807617,
      "learning_rate": 3.993094754956549e-05,
      "loss": 0.7366,
      "step": 2550
    },
    {
      "epoch": 0.44257460097154755,
      "grad_norm": 0.9214552044868469,
      "learning_rate": 3.9930380715304143e-05,
      "loss": 0.7626,
      "step": 2551
    },
    {
      "epoch": 0.44274809160305345,
      "grad_norm": 0.7478047013282776,
      "learning_rate": 3.992981156809889e-05,
      "loss": 0.7194,
      "step": 2552
    },
    {
      "epoch": 0.44292158223455935,
      "grad_norm": 0.7857555747032166,
      "learning_rate": 3.992924010801578e-05,
      "loss": 0.6995,
      "step": 2553
    },
    {
      "epoch": 0.44309507286606525,
      "grad_norm": 1.4917502403259277,
      "learning_rate": 3.9928666335121135e-05,
      "loss": 0.6264,
      "step": 2554
    },
    {
      "epoch": 0.44326856349757116,
      "grad_norm": 0.5582311153411865,
      "learning_rate": 3.992809024948154e-05,
      "loss": 0.7217,
      "step": 2555
    },
    {
      "epoch": 0.44344205412907706,
      "grad_norm": 1.3408156633377075,
      "learning_rate": 3.992751185116385e-05,
      "loss": 0.5981,
      "step": 2556
    },
    {
      "epoch": 0.4436155447605829,
      "grad_norm": 0.8039281964302063,
      "learning_rate": 3.992693114023519e-05,
      "loss": 0.6396,
      "step": 2557
    },
    {
      "epoch": 0.4437890353920888,
      "grad_norm": 1.0995712280273438,
      "learning_rate": 3.992634811676296e-05,
      "loss": 0.5409,
      "step": 2558
    },
    {
      "epoch": 0.4439625260235947,
      "grad_norm": 0.8444967865943909,
      "learning_rate": 3.9925762780814804e-05,
      "loss": 0.5554,
      "step": 2559
    },
    {
      "epoch": 0.4441360166551006,
      "grad_norm": 0.9820617437362671,
      "learning_rate": 3.992517513245865e-05,
      "loss": 0.5261,
      "step": 2560
    },
    {
      "epoch": 0.4443095072866065,
      "grad_norm": 1.0747051239013672,
      "learning_rate": 3.992458517176272e-05,
      "loss": 0.5458,
      "step": 2561
    },
    {
      "epoch": 0.4444829979181124,
      "grad_norm": 0.9341891407966614,
      "learning_rate": 3.992399289879546e-05,
      "loss": 0.54,
      "step": 2562
    },
    {
      "epoch": 0.4446564885496183,
      "grad_norm": 0.7590116858482361,
      "learning_rate": 3.992339831362561e-05,
      "loss": 0.6815,
      "step": 2563
    },
    {
      "epoch": 0.4448299791811242,
      "grad_norm": 1.0929893255233765,
      "learning_rate": 3.992280141632216e-05,
      "loss": 0.6704,
      "step": 2564
    },
    {
      "epoch": 0.4450034698126301,
      "grad_norm": 0.6500886678695679,
      "learning_rate": 3.9922202206954395e-05,
      "loss": 0.5415,
      "step": 2565
    },
    {
      "epoch": 0.445176960444136,
      "grad_norm": 0.9372058510780334,
      "learning_rate": 3.9921600685591856e-05,
      "loss": 0.5028,
      "step": 2566
    },
    {
      "epoch": 0.4453504510756419,
      "grad_norm": 1.0460612773895264,
      "learning_rate": 3.992099685230434e-05,
      "loss": 0.514,
      "step": 2567
    },
    {
      "epoch": 0.4455239417071478,
      "grad_norm": 0.8441975712776184,
      "learning_rate": 3.9920390707161927e-05,
      "loss": 0.6785,
      "step": 2568
    },
    {
      "epoch": 0.4456974323386537,
      "grad_norm": 0.7510903477668762,
      "learning_rate": 3.991978225023497e-05,
      "loss": 0.6934,
      "step": 2569
    },
    {
      "epoch": 0.4458709229701596,
      "grad_norm": 0.8730303645133972,
      "learning_rate": 3.9919171481594056e-05,
      "loss": 0.5898,
      "step": 2570
    },
    {
      "epoch": 0.4460444136016655,
      "grad_norm": 0.657044529914856,
      "learning_rate": 3.991855840131009e-05,
      "loss": 0.73,
      "step": 2571
    },
    {
      "epoch": 0.4462179042331714,
      "grad_norm": 1.8791756629943848,
      "learning_rate": 3.9917943009454206e-05,
      "loss": 0.5889,
      "step": 2572
    },
    {
      "epoch": 0.4463913948646773,
      "grad_norm": 0.9651908278465271,
      "learning_rate": 3.991732530609783e-05,
      "loss": 0.5627,
      "step": 2573
    },
    {
      "epoch": 0.44656488549618323,
      "grad_norm": 1.3580819368362427,
      "learning_rate": 3.9916705291312646e-05,
      "loss": 0.5333,
      "step": 2574
    },
    {
      "epoch": 0.44673837612768913,
      "grad_norm": 0.9872630834579468,
      "learning_rate": 3.9916082965170604e-05,
      "loss": 0.7241,
      "step": 2575
    },
    {
      "epoch": 0.44691186675919503,
      "grad_norm": 0.995220959186554,
      "learning_rate": 3.991545832774393e-05,
      "loss": 0.5817,
      "step": 2576
    },
    {
      "epoch": 0.4470853573907009,
      "grad_norm": 0.8517805933952332,
      "learning_rate": 3.9914831379105104e-05,
      "loss": 0.6534,
      "step": 2577
    },
    {
      "epoch": 0.4472588480222068,
      "grad_norm": 1.203552484512329,
      "learning_rate": 3.9914202119326895e-05,
      "loss": 0.6368,
      "step": 2578
    },
    {
      "epoch": 0.4474323386537127,
      "grad_norm": 0.7733704447746277,
      "learning_rate": 3.991357054848233e-05,
      "loss": 0.5702,
      "step": 2579
    },
    {
      "epoch": 0.4476058292852186,
      "grad_norm": 0.8969932794570923,
      "learning_rate": 3.991293666664469e-05,
      "loss": 0.6219,
      "step": 2580
    },
    {
      "epoch": 0.4477793199167245,
      "grad_norm": 0.8377065062522888,
      "learning_rate": 3.991230047388755e-05,
      "loss": 0.5553,
      "step": 2581
    },
    {
      "epoch": 0.4479528105482304,
      "grad_norm": 0.7633172273635864,
      "learning_rate": 3.991166197028474e-05,
      "loss": 0.4855,
      "step": 2582
    },
    {
      "epoch": 0.4481263011797363,
      "grad_norm": 0.8832114338874817,
      "learning_rate": 3.9911021155910355e-05,
      "loss": 0.594,
      "step": 2583
    },
    {
      "epoch": 0.4482997918112422,
      "grad_norm": 0.7355130314826965,
      "learning_rate": 3.9910378030838765e-05,
      "loss": 0.632,
      "step": 2584
    },
    {
      "epoch": 0.4484732824427481,
      "grad_norm": 1.0732299089431763,
      "learning_rate": 3.99097325951446e-05,
      "loss": 0.7803,
      "step": 2585
    },
    {
      "epoch": 0.448646773074254,
      "grad_norm": 1.6468517780303955,
      "learning_rate": 3.990908484890277e-05,
      "loss": 0.6067,
      "step": 2586
    },
    {
      "epoch": 0.4488202637057599,
      "grad_norm": 1.1821500062942505,
      "learning_rate": 3.9908434792188443e-05,
      "loss": 0.5513,
      "step": 2587
    },
    {
      "epoch": 0.4489937543372658,
      "grad_norm": 0.7869590520858765,
      "learning_rate": 3.990778242507707e-05,
      "loss": 0.6683,
      "step": 2588
    },
    {
      "epoch": 0.4491672449687717,
      "grad_norm": 0.6006678938865662,
      "learning_rate": 3.990712774764434e-05,
      "loss": 0.6669,
      "step": 2589
    },
    {
      "epoch": 0.4493407356002776,
      "grad_norm": 0.6672049760818481,
      "learning_rate": 3.990647075996624e-05,
      "loss": 0.6932,
      "step": 2590
    },
    {
      "epoch": 0.4495142262317835,
      "grad_norm": 1.5705081224441528,
      "learning_rate": 3.9905811462119014e-05,
      "loss": 0.6442,
      "step": 2591
    },
    {
      "epoch": 0.4496877168632894,
      "grad_norm": 0.6579003930091858,
      "learning_rate": 3.9905149854179174e-05,
      "loss": 0.6995,
      "step": 2592
    },
    {
      "epoch": 0.4498612074947953,
      "grad_norm": 0.8032071590423584,
      "learning_rate": 3.990448593622349e-05,
      "loss": 0.5646,
      "step": 2593
    },
    {
      "epoch": 0.4500346981263012,
      "grad_norm": 1.135553240776062,
      "learning_rate": 3.990381970832903e-05,
      "loss": 0.5828,
      "step": 2594
    },
    {
      "epoch": 0.4502081887578071,
      "grad_norm": 0.9185262322425842,
      "learning_rate": 3.99031511705731e-05,
      "loss": 0.7058,
      "step": 2595
    },
    {
      "epoch": 0.45038167938931295,
      "grad_norm": 0.6164620518684387,
      "learning_rate": 3.9902480323033285e-05,
      "loss": 0.7278,
      "step": 2596
    },
    {
      "epoch": 0.45055517002081885,
      "grad_norm": 0.7947415709495544,
      "learning_rate": 3.990180716578744e-05,
      "loss": 0.5625,
      "step": 2597
    },
    {
      "epoch": 0.45072866065232475,
      "grad_norm": 0.9763660430908203,
      "learning_rate": 3.990113169891367e-05,
      "loss": 0.5795,
      "step": 2598
    },
    {
      "epoch": 0.45090215128383065,
      "grad_norm": 1.2467541694641113,
      "learning_rate": 3.990045392249039e-05,
      "loss": 0.7317,
      "step": 2599
    },
    {
      "epoch": 0.45107564191533656,
      "grad_norm": 0.9953369498252869,
      "learning_rate": 3.989977383659624e-05,
      "loss": 0.5889,
      "step": 2600
    },
    {
      "epoch": 0.45124913254684246,
      "grad_norm": 1.3223135471343994,
      "learning_rate": 3.989909144131015e-05,
      "loss": 0.6748,
      "step": 2601
    },
    {
      "epoch": 0.45142262317834836,
      "grad_norm": 0.6914749145507812,
      "learning_rate": 3.989840673671131e-05,
      "loss": 0.6632,
      "step": 2602
    },
    {
      "epoch": 0.45159611380985426,
      "grad_norm": 0.9431111812591553,
      "learning_rate": 3.9897719722879176e-05,
      "loss": 0.6514,
      "step": 2603
    },
    {
      "epoch": 0.45176960444136016,
      "grad_norm": 0.8488491773605347,
      "learning_rate": 3.989703039989349e-05,
      "loss": 0.6818,
      "step": 2604
    },
    {
      "epoch": 0.45194309507286606,
      "grad_norm": 1.092769980430603,
      "learning_rate": 3.989633876783423e-05,
      "loss": 0.6993,
      "step": 2605
    },
    {
      "epoch": 0.45211658570437196,
      "grad_norm": 1.1377992630004883,
      "learning_rate": 3.989564482678168e-05,
      "loss": 0.567,
      "step": 2606
    },
    {
      "epoch": 0.45229007633587787,
      "grad_norm": 1.0154528617858887,
      "learning_rate": 3.9894948576816374e-05,
      "loss": 0.5203,
      "step": 2607
    },
    {
      "epoch": 0.45246356696738377,
      "grad_norm": 1.028075098991394,
      "learning_rate": 3.989425001801909e-05,
      "loss": 0.4795,
      "step": 2608
    },
    {
      "epoch": 0.45263705759888967,
      "grad_norm": 1.4053622484207153,
      "learning_rate": 3.9893549150470925e-05,
      "loss": 0.6066,
      "step": 2609
    },
    {
      "epoch": 0.45281054823039557,
      "grad_norm": 1.0774238109588623,
      "learning_rate": 3.9892845974253184e-05,
      "loss": 0.7133,
      "step": 2610
    },
    {
      "epoch": 0.4529840388619015,
      "grad_norm": 0.7488763928413391,
      "learning_rate": 3.989214048944749e-05,
      "loss": 0.6543,
      "step": 2611
    },
    {
      "epoch": 0.4531575294934074,
      "grad_norm": 0.9664822816848755,
      "learning_rate": 3.989143269613572e-05,
      "loss": 0.6453,
      "step": 2612
    },
    {
      "epoch": 0.4533310201249133,
      "grad_norm": 0.7693923711776733,
      "learning_rate": 3.9890722594400005e-05,
      "loss": 0.7168,
      "step": 2613
    },
    {
      "epoch": 0.4535045107564192,
      "grad_norm": 1.0447169542312622,
      "learning_rate": 3.989001018432276e-05,
      "loss": 0.5677,
      "step": 2614
    },
    {
      "epoch": 0.4536780013879251,
      "grad_norm": 1.1141068935394287,
      "learning_rate": 3.988929546598665e-05,
      "loss": 0.7236,
      "step": 2615
    },
    {
      "epoch": 0.4538514920194309,
      "grad_norm": 1.1113327741622925,
      "learning_rate": 3.988857843947463e-05,
      "loss": 0.5212,
      "step": 2616
    },
    {
      "epoch": 0.4540249826509368,
      "grad_norm": 0.832119345664978,
      "learning_rate": 3.9887859104869905e-05,
      "loss": 0.502,
      "step": 2617
    },
    {
      "epoch": 0.4541984732824427,
      "grad_norm": 0.8675135374069214,
      "learning_rate": 3.988713746225596e-05,
      "loss": 0.6472,
      "step": 2618
    },
    {
      "epoch": 0.45437196391394863,
      "grad_norm": 0.7328503727912903,
      "learning_rate": 3.988641351171653e-05,
      "loss": 0.7312,
      "step": 2619
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 1.2184706926345825,
      "learning_rate": 3.988568725333565e-05,
      "loss": 0.7324,
      "step": 2620
    },
    {
      "epoch": 0.45471894517696043,
      "grad_norm": 0.8525048494338989,
      "learning_rate": 3.988495868719759e-05,
      "loss": 0.571,
      "step": 2621
    },
    {
      "epoch": 0.45489243580846633,
      "grad_norm": 0.855289876461029,
      "learning_rate": 3.98842278133869e-05,
      "loss": 0.6549,
      "step": 2622
    },
    {
      "epoch": 0.45506592643997223,
      "grad_norm": 1.3905209302902222,
      "learning_rate": 3.988349463198841e-05,
      "loss": 0.5911,
      "step": 2623
    },
    {
      "epoch": 0.45523941707147814,
      "grad_norm": 1.1628540754318237,
      "learning_rate": 3.9882759143087194e-05,
      "loss": 0.621,
      "step": 2624
    },
    {
      "epoch": 0.45541290770298404,
      "grad_norm": 1.0131633281707764,
      "learning_rate": 3.9882021346768613e-05,
      "loss": 0.6434,
      "step": 2625
    },
    {
      "epoch": 0.45558639833448994,
      "grad_norm": 0.8525959253311157,
      "learning_rate": 3.9881281243118285e-05,
      "loss": 0.6331,
      "step": 2626
    },
    {
      "epoch": 0.45575988896599584,
      "grad_norm": 0.8659221529960632,
      "learning_rate": 3.9880538832222105e-05,
      "loss": 0.7217,
      "step": 2627
    },
    {
      "epoch": 0.45593337959750174,
      "grad_norm": 1.3382600545883179,
      "learning_rate": 3.987979411416623e-05,
      "loss": 0.5326,
      "step": 2628
    },
    {
      "epoch": 0.45610687022900764,
      "grad_norm": 2.1783976554870605,
      "learning_rate": 3.987904708903708e-05,
      "loss": 0.7019,
      "step": 2629
    },
    {
      "epoch": 0.45628036086051355,
      "grad_norm": 1.4334501028060913,
      "learning_rate": 3.987829775692135e-05,
      "loss": 0.5695,
      "step": 2630
    },
    {
      "epoch": 0.45645385149201945,
      "grad_norm": 0.6840330362319946,
      "learning_rate": 3.987754611790601e-05,
      "loss": 0.64,
      "step": 2631
    },
    {
      "epoch": 0.45662734212352535,
      "grad_norm": 1.0548428297042847,
      "learning_rate": 3.987679217207827e-05,
      "loss": 0.5468,
      "step": 2632
    },
    {
      "epoch": 0.45680083275503125,
      "grad_norm": 0.8488027453422546,
      "learning_rate": 3.9876035919525644e-05,
      "loss": 0.6248,
      "step": 2633
    },
    {
      "epoch": 0.45697432338653715,
      "grad_norm": 1.8956934213638306,
      "learning_rate": 3.987527736033589e-05,
      "loss": 0.5099,
      "step": 2634
    },
    {
      "epoch": 0.45714781401804305,
      "grad_norm": 1.124916434288025,
      "learning_rate": 3.9874516494597035e-05,
      "loss": 0.6058,
      "step": 2635
    },
    {
      "epoch": 0.4573213046495489,
      "grad_norm": 2.044217348098755,
      "learning_rate": 3.987375332239739e-05,
      "loss": 0.6123,
      "step": 2636
    },
    {
      "epoch": 0.4574947952810548,
      "grad_norm": 0.7469845414161682,
      "learning_rate": 3.9872987843825505e-05,
      "loss": 0.5177,
      "step": 2637
    },
    {
      "epoch": 0.4576682859125607,
      "grad_norm": 0.9351304173469543,
      "learning_rate": 3.9872220058970226e-05,
      "loss": 0.6732,
      "step": 2638
    },
    {
      "epoch": 0.4578417765440666,
      "grad_norm": 0.7773634791374207,
      "learning_rate": 3.9871449967920656e-05,
      "loss": 0.5753,
      "step": 2639
    },
    {
      "epoch": 0.4580152671755725,
      "grad_norm": 0.9111433029174805,
      "learning_rate": 3.9870677570766167e-05,
      "loss": 0.6011,
      "step": 2640
    },
    {
      "epoch": 0.4581887578070784,
      "grad_norm": 0.6724522113800049,
      "learning_rate": 3.986990286759639e-05,
      "loss": 0.6995,
      "step": 2641
    },
    {
      "epoch": 0.4583622484385843,
      "grad_norm": 2.940087080001831,
      "learning_rate": 3.986912585850123e-05,
      "loss": 0.6014,
      "step": 2642
    },
    {
      "epoch": 0.4585357390700902,
      "grad_norm": 1.3955363035202026,
      "learning_rate": 3.986834654357086e-05,
      "loss": 0.5264,
      "step": 2643
    },
    {
      "epoch": 0.4587092297015961,
      "grad_norm": 3.199185609817505,
      "learning_rate": 3.9867564922895724e-05,
      "loss": 0.5081,
      "step": 2644
    },
    {
      "epoch": 0.458882720333102,
      "grad_norm": 0.8836493492126465,
      "learning_rate": 3.9866780996566525e-05,
      "loss": 0.6975,
      "step": 2645
    },
    {
      "epoch": 0.4590562109646079,
      "grad_norm": 0.9459909200668335,
      "learning_rate": 3.986599476467425e-05,
      "loss": 0.7307,
      "step": 2646
    },
    {
      "epoch": 0.4592297015961138,
      "grad_norm": 0.9849326014518738,
      "learning_rate": 3.986520622731012e-05,
      "loss": 0.5869,
      "step": 2647
    },
    {
      "epoch": 0.4594031922276197,
      "grad_norm": 0.746747612953186,
      "learning_rate": 3.9864415384565675e-05,
      "loss": 0.6517,
      "step": 2648
    },
    {
      "epoch": 0.4595766828591256,
      "grad_norm": 0.7429086565971375,
      "learning_rate": 3.986362223653267e-05,
      "loss": 0.7041,
      "step": 2649
    },
    {
      "epoch": 0.4597501734906315,
      "grad_norm": 0.8128641247749329,
      "learning_rate": 3.986282678330316e-05,
      "loss": 0.7878,
      "step": 2650
    },
    {
      "epoch": 0.4599236641221374,
      "grad_norm": 0.778728187084198,
      "learning_rate": 3.9862029024969466e-05,
      "loss": 0.5446,
      "step": 2651
    },
    {
      "epoch": 0.4600971547536433,
      "grad_norm": 0.6940119862556458,
      "learning_rate": 3.9861228961624146e-05,
      "loss": 0.5607,
      "step": 2652
    },
    {
      "epoch": 0.4602706453851492,
      "grad_norm": 0.772933304309845,
      "learning_rate": 3.986042659336007e-05,
      "loss": 0.5793,
      "step": 2653
    },
    {
      "epoch": 0.4604441360166551,
      "grad_norm": 1.608131766319275,
      "learning_rate": 3.985962192027034e-05,
      "loss": 0.5588,
      "step": 2654
    },
    {
      "epoch": 0.46061762664816097,
      "grad_norm": 0.9020060300827026,
      "learning_rate": 3.985881494244835e-05,
      "loss": 0.5914,
      "step": 2655
    },
    {
      "epoch": 0.4607911172796669,
      "grad_norm": 0.9139862060546875,
      "learning_rate": 3.985800565998775e-05,
      "loss": 0.6445,
      "step": 2656
    },
    {
      "epoch": 0.4609646079111728,
      "grad_norm": 1.4942809343338013,
      "learning_rate": 3.9857194072982445e-05,
      "loss": 0.5084,
      "step": 2657
    },
    {
      "epoch": 0.4611380985426787,
      "grad_norm": 0.6657776832580566,
      "learning_rate": 3.9856380181526634e-05,
      "loss": 0.6106,
      "step": 2658
    },
    {
      "epoch": 0.4613115891741846,
      "grad_norm": 1.0785088539123535,
      "learning_rate": 3.985556398571476e-05,
      "loss": 0.5094,
      "step": 2659
    },
    {
      "epoch": 0.4614850798056905,
      "grad_norm": 0.8394649028778076,
      "learning_rate": 3.9854745485641556e-05,
      "loss": 0.5852,
      "step": 2660
    },
    {
      "epoch": 0.4616585704371964,
      "grad_norm": 2.2808496952056885,
      "learning_rate": 3.9853924681402e-05,
      "loss": 0.5568,
      "step": 2661
    },
    {
      "epoch": 0.4618320610687023,
      "grad_norm": 0.7547072768211365,
      "learning_rate": 3.985310157309135e-05,
      "loss": 0.562,
      "step": 2662
    },
    {
      "epoch": 0.4620055517002082,
      "grad_norm": 0.7576550841331482,
      "learning_rate": 3.985227616080513e-05,
      "loss": 0.5585,
      "step": 2663
    },
    {
      "epoch": 0.4621790423317141,
      "grad_norm": 1.0454773902893066,
      "learning_rate": 3.985144844463913e-05,
      "loss": 0.698,
      "step": 2664
    },
    {
      "epoch": 0.46235253296322,
      "grad_norm": 2.2017455101013184,
      "learning_rate": 3.985061842468941e-05,
      "loss": 0.5635,
      "step": 2665
    },
    {
      "epoch": 0.4625260235947259,
      "grad_norm": 1.4035037755966187,
      "learning_rate": 3.9849786101052285e-05,
      "loss": 0.5897,
      "step": 2666
    },
    {
      "epoch": 0.4626995142262318,
      "grad_norm": 0.8923456072807312,
      "learning_rate": 3.984895147382435e-05,
      "loss": 0.6019,
      "step": 2667
    },
    {
      "epoch": 0.4628730048577377,
      "grad_norm": 0.717780351638794,
      "learning_rate": 3.984811454310248e-05,
      "loss": 0.6179,
      "step": 2668
    },
    {
      "epoch": 0.4630464954892436,
      "grad_norm": 1.2886556386947632,
      "learning_rate": 3.984727530898378e-05,
      "loss": 0.5853,
      "step": 2669
    },
    {
      "epoch": 0.4632199861207495,
      "grad_norm": 0.8386170268058777,
      "learning_rate": 3.9846433771565655e-05,
      "loss": 0.6455,
      "step": 2670
    },
    {
      "epoch": 0.4633934767522554,
      "grad_norm": 1.1211575269699097,
      "learning_rate": 3.9845589930945764e-05,
      "loss": 0.6292,
      "step": 2671
    },
    {
      "epoch": 0.4635669673837613,
      "grad_norm": 0.6581203937530518,
      "learning_rate": 3.9844743787222046e-05,
      "loss": 0.6748,
      "step": 2672
    },
    {
      "epoch": 0.4637404580152672,
      "grad_norm": 0.5879610776901245,
      "learning_rate": 3.984389534049268e-05,
      "loss": 0.6105,
      "step": 2673
    },
    {
      "epoch": 0.4639139486467731,
      "grad_norm": 1.0543928146362305,
      "learning_rate": 3.984304459085614e-05,
      "loss": 0.5782,
      "step": 2674
    },
    {
      "epoch": 0.46408743927827895,
      "grad_norm": 1.3818283081054688,
      "learning_rate": 3.9842191538411145e-05,
      "loss": 0.6366,
      "step": 2675
    },
    {
      "epoch": 0.46426092990978485,
      "grad_norm": 1.5274306535720825,
      "learning_rate": 3.984133618325671e-05,
      "loss": 0.6567,
      "step": 2676
    },
    {
      "epoch": 0.46443442054129075,
      "grad_norm": 0.9548752903938293,
      "learning_rate": 3.984047852549209e-05,
      "loss": 0.5369,
      "step": 2677
    },
    {
      "epoch": 0.46460791117279665,
      "grad_norm": 0.7854692339897156,
      "learning_rate": 3.983961856521682e-05,
      "loss": 0.6729,
      "step": 2678
    },
    {
      "epoch": 0.46478140180430255,
      "grad_norm": 0.6886661648750305,
      "learning_rate": 3.983875630253069e-05,
      "loss": 0.6757,
      "step": 2679
    },
    {
      "epoch": 0.46495489243580845,
      "grad_norm": 0.6751460433006287,
      "learning_rate": 3.983789173753378e-05,
      "loss": 0.6976,
      "step": 2680
    },
    {
      "epoch": 0.46512838306731435,
      "grad_norm": 0.6498127579689026,
      "learning_rate": 3.9837024870326425e-05,
      "loss": 0.6304,
      "step": 2681
    },
    {
      "epoch": 0.46530187369882026,
      "grad_norm": 0.8613525629043579,
      "learning_rate": 3.983615570100921e-05,
      "loss": 0.556,
      "step": 2682
    },
    {
      "epoch": 0.46547536433032616,
      "grad_norm": 1.1612775325775146,
      "learning_rate": 3.983528422968301e-05,
      "loss": 0.6768,
      "step": 2683
    },
    {
      "epoch": 0.46564885496183206,
      "grad_norm": 1.0510660409927368,
      "learning_rate": 3.9834410456448966e-05,
      "loss": 0.5898,
      "step": 2684
    },
    {
      "epoch": 0.46582234559333796,
      "grad_norm": 0.819597601890564,
      "learning_rate": 3.983353438140848e-05,
      "loss": 0.6868,
      "step": 2685
    },
    {
      "epoch": 0.46599583622484386,
      "grad_norm": 1.039358139038086,
      "learning_rate": 3.983265600466321e-05,
      "loss": 0.6661,
      "step": 2686
    },
    {
      "epoch": 0.46616932685634976,
      "grad_norm": 0.6790923476219177,
      "learning_rate": 3.983177532631511e-05,
      "loss": 0.6658,
      "step": 2687
    },
    {
      "epoch": 0.46634281748785567,
      "grad_norm": 0.7647500038146973,
      "learning_rate": 3.983089234646637e-05,
      "loss": 0.6296,
      "step": 2688
    },
    {
      "epoch": 0.46651630811936157,
      "grad_norm": 0.852489709854126,
      "learning_rate": 3.9830007065219466e-05,
      "loss": 0.6422,
      "step": 2689
    },
    {
      "epoch": 0.46668979875086747,
      "grad_norm": 4.935563087463379,
      "learning_rate": 3.9829119482677144e-05,
      "loss": 0.6632,
      "step": 2690
    },
    {
      "epoch": 0.46686328938237337,
      "grad_norm": 1.0070613622665405,
      "learning_rate": 3.982822959894239e-05,
      "loss": 0.681,
      "step": 2691
    },
    {
      "epoch": 0.46703678001387927,
      "grad_norm": 0.9911800622940063,
      "learning_rate": 3.9827337414118486e-05,
      "loss": 0.6112,
      "step": 2692
    },
    {
      "epoch": 0.4672102706453852,
      "grad_norm": 0.8067095875740051,
      "learning_rate": 3.9826442928308974e-05,
      "loss": 0.5524,
      "step": 2693
    },
    {
      "epoch": 0.4673837612768911,
      "grad_norm": 0.7464631199836731,
      "learning_rate": 3.982554614161766e-05,
      "loss": 0.6768,
      "step": 2694
    },
    {
      "epoch": 0.4675572519083969,
      "grad_norm": 0.872347891330719,
      "learning_rate": 3.9824647054148614e-05,
      "loss": 0.558,
      "step": 2695
    },
    {
      "epoch": 0.4677307425399028,
      "grad_norm": 0.7208455204963684,
      "learning_rate": 3.9823745666006176e-05,
      "loss": 0.5389,
      "step": 2696
    },
    {
      "epoch": 0.4679042331714087,
      "grad_norm": 0.984927237033844,
      "learning_rate": 3.982284197729496e-05,
      "loss": 0.5515,
      "step": 2697
    },
    {
      "epoch": 0.4680777238029146,
      "grad_norm": 1.1473478078842163,
      "learning_rate": 3.982193598811983e-05,
      "loss": 0.5508,
      "step": 2698
    },
    {
      "epoch": 0.4682512144344205,
      "grad_norm": 0.9905941486358643,
      "learning_rate": 3.982102769858593e-05,
      "loss": 0.6453,
      "step": 2699
    },
    {
      "epoch": 0.46842470506592643,
      "grad_norm": 1.1189990043640137,
      "learning_rate": 3.9820117108798666e-05,
      "loss": 0.4843,
      "step": 2700
    },
    {
      "epoch": 0.46859819569743233,
      "grad_norm": 1.9971749782562256,
      "learning_rate": 3.981920421886372e-05,
      "loss": 0.4984,
      "step": 2701
    },
    {
      "epoch": 0.46877168632893823,
      "grad_norm": 0.6893688440322876,
      "learning_rate": 3.981828902888704e-05,
      "loss": 0.6013,
      "step": 2702
    },
    {
      "epoch": 0.46894517696044413,
      "grad_norm": 0.9006636142730713,
      "learning_rate": 3.981737153897481e-05,
      "loss": 0.7227,
      "step": 2703
    },
    {
      "epoch": 0.46911866759195003,
      "grad_norm": 1.91812264919281,
      "learning_rate": 3.981645174923353e-05,
      "loss": 0.5508,
      "step": 2704
    },
    {
      "epoch": 0.46929215822345594,
      "grad_norm": 1.011795997619629,
      "learning_rate": 3.981552965976993e-05,
      "loss": 0.5004,
      "step": 2705
    },
    {
      "epoch": 0.46946564885496184,
      "grad_norm": 1.296972632408142,
      "learning_rate": 3.9814605270691025e-05,
      "loss": 0.5403,
      "step": 2706
    },
    {
      "epoch": 0.46963913948646774,
      "grad_norm": 0.7811582088470459,
      "learning_rate": 3.9813678582104095e-05,
      "loss": 0.5728,
      "step": 2707
    },
    {
      "epoch": 0.46981263011797364,
      "grad_norm": 0.8036185503005981,
      "learning_rate": 3.981274959411667e-05,
      "loss": 0.6035,
      "step": 2708
    },
    {
      "epoch": 0.46998612074947954,
      "grad_norm": 0.8364619016647339,
      "learning_rate": 3.981181830683657e-05,
      "loss": 0.6265,
      "step": 2709
    },
    {
      "epoch": 0.47015961138098544,
      "grad_norm": 1.0008864402770996,
      "learning_rate": 3.9810884720371874e-05,
      "loss": 0.7524,
      "step": 2710
    },
    {
      "epoch": 0.47033310201249134,
      "grad_norm": 0.7549238801002502,
      "learning_rate": 3.9809948834830914e-05,
      "loss": 0.6472,
      "step": 2711
    },
    {
      "epoch": 0.47050659264399725,
      "grad_norm": 0.9469247460365295,
      "learning_rate": 3.980901065032232e-05,
      "loss": 0.606,
      "step": 2712
    },
    {
      "epoch": 0.47068008327550315,
      "grad_norm": 2.2984249591827393,
      "learning_rate": 3.9808070166954945e-05,
      "loss": 0.6272,
      "step": 2713
    },
    {
      "epoch": 0.47085357390700905,
      "grad_norm": 0.9015179872512817,
      "learning_rate": 3.9807127384837955e-05,
      "loss": 0.5844,
      "step": 2714
    },
    {
      "epoch": 0.4710270645385149,
      "grad_norm": 0.8548340797424316,
      "learning_rate": 3.980618230408075e-05,
      "loss": 0.5885,
      "step": 2715
    },
    {
      "epoch": 0.4712005551700208,
      "grad_norm": 0.9300568103790283,
      "learning_rate": 3.980523492479301e-05,
      "loss": 0.5674,
      "step": 2716
    },
    {
      "epoch": 0.4713740458015267,
      "grad_norm": 0.9700860977172852,
      "learning_rate": 3.9804285247084674e-05,
      "loss": 0.6692,
      "step": 2717
    },
    {
      "epoch": 0.4715475364330326,
      "grad_norm": 0.7631649374961853,
      "learning_rate": 3.980333327106596e-05,
      "loss": 0.6504,
      "step": 2718
    },
    {
      "epoch": 0.4717210270645385,
      "grad_norm": 0.7961834669113159,
      "learning_rate": 3.980237899684735e-05,
      "loss": 0.7671,
      "step": 2719
    },
    {
      "epoch": 0.4718945176960444,
      "grad_norm": 0.8147475123405457,
      "learning_rate": 3.980142242453958e-05,
      "loss": 0.6982,
      "step": 2720
    },
    {
      "epoch": 0.4720680083275503,
      "grad_norm": 0.6349263787269592,
      "learning_rate": 3.980046355425366e-05,
      "loss": 0.6766,
      "step": 2721
    },
    {
      "epoch": 0.4722414989590562,
      "grad_norm": 0.7165682911872864,
      "learning_rate": 3.979950238610088e-05,
      "loss": 0.5476,
      "step": 2722
    },
    {
      "epoch": 0.4724149895905621,
      "grad_norm": 1.1357554197311401,
      "learning_rate": 3.979853892019278e-05,
      "loss": 0.6307,
      "step": 2723
    },
    {
      "epoch": 0.472588480222068,
      "grad_norm": 1.2190383672714233,
      "learning_rate": 3.9797573156641165e-05,
      "loss": 0.5709,
      "step": 2724
    },
    {
      "epoch": 0.4727619708535739,
      "grad_norm": 0.8067264556884766,
      "learning_rate": 3.9796605095558115e-05,
      "loss": 0.5641,
      "step": 2725
    },
    {
      "epoch": 0.4729354614850798,
      "grad_norm": 1.0697906017303467,
      "learning_rate": 3.9795634737055974e-05,
      "loss": 0.6151,
      "step": 2726
    },
    {
      "epoch": 0.4731089521165857,
      "grad_norm": 1.0114651918411255,
      "learning_rate": 3.979466208124736e-05,
      "loss": 0.5023,
      "step": 2727
    },
    {
      "epoch": 0.4732824427480916,
      "grad_norm": 0.6932345032691956,
      "learning_rate": 3.979368712824514e-05,
      "loss": 0.5094,
      "step": 2728
    },
    {
      "epoch": 0.4734559333795975,
      "grad_norm": 1.4069453477859497,
      "learning_rate": 3.9792709878162466e-05,
      "loss": 0.6754,
      "step": 2729
    },
    {
      "epoch": 0.4736294240111034,
      "grad_norm": 1.1037931442260742,
      "learning_rate": 3.979173033111275e-05,
      "loss": 0.5452,
      "step": 2730
    },
    {
      "epoch": 0.4738029146426093,
      "grad_norm": 0.8605200052261353,
      "learning_rate": 3.979074848720967e-05,
      "loss": 0.6757,
      "step": 2731
    },
    {
      "epoch": 0.4739764052741152,
      "grad_norm": 0.9133217334747314,
      "learning_rate": 3.9789764346567154e-05,
      "loss": 0.6819,
      "step": 2732
    },
    {
      "epoch": 0.4741498959056211,
      "grad_norm": 1.019981861114502,
      "learning_rate": 3.978877790929944e-05,
      "loss": 0.6392,
      "step": 2733
    },
    {
      "epoch": 0.47432338653712697,
      "grad_norm": 0.7765205502510071,
      "learning_rate": 3.978778917552099e-05,
      "loss": 0.6055,
      "step": 2734
    },
    {
      "epoch": 0.47449687716863287,
      "grad_norm": 0.5978204011917114,
      "learning_rate": 3.978679814534654e-05,
      "loss": 0.7339,
      "step": 2735
    },
    {
      "epoch": 0.47467036780013877,
      "grad_norm": 0.8094828724861145,
      "learning_rate": 3.9785804818891117e-05,
      "loss": 0.5305,
      "step": 2736
    },
    {
      "epoch": 0.47484385843164467,
      "grad_norm": 1.0592631101608276,
      "learning_rate": 3.9784809196269986e-05,
      "loss": 0.5562,
      "step": 2737
    },
    {
      "epoch": 0.4750173490631506,
      "grad_norm": 0.7034501433372498,
      "learning_rate": 3.978381127759869e-05,
      "loss": 0.6902,
      "step": 2738
    },
    {
      "epoch": 0.4751908396946565,
      "grad_norm": 1.376899003982544,
      "learning_rate": 3.978281106299305e-05,
      "loss": 0.5536,
      "step": 2739
    },
    {
      "epoch": 0.4753643303261624,
      "grad_norm": 0.6912811398506165,
      "learning_rate": 3.9781808552569134e-05,
      "loss": 0.624,
      "step": 2740
    },
    {
      "epoch": 0.4755378209576683,
      "grad_norm": 0.8399140238761902,
      "learning_rate": 3.9780803746443284e-05,
      "loss": 0.6537,
      "step": 2741
    },
    {
      "epoch": 0.4757113115891742,
      "grad_norm": 0.7426314949989319,
      "learning_rate": 3.97797966447321e-05,
      "loss": 0.5227,
      "step": 2742
    },
    {
      "epoch": 0.4758848022206801,
      "grad_norm": 2.417330741882324,
      "learning_rate": 3.9778787247552484e-05,
      "loss": 0.5402,
      "step": 2743
    },
    {
      "epoch": 0.476058292852186,
      "grad_norm": 0.9911115765571594,
      "learning_rate": 3.977777555502155e-05,
      "loss": 0.5673,
      "step": 2744
    },
    {
      "epoch": 0.4762317834836919,
      "grad_norm": 0.8713995814323425,
      "learning_rate": 3.9776761567256726e-05,
      "loss": 0.6201,
      "step": 2745
    },
    {
      "epoch": 0.4764052741151978,
      "grad_norm": 1.156535029411316,
      "learning_rate": 3.977574528437567e-05,
      "loss": 0.6407,
      "step": 2746
    },
    {
      "epoch": 0.4765787647467037,
      "grad_norm": 0.5964752435684204,
      "learning_rate": 3.977472670649634e-05,
      "loss": 0.672,
      "step": 2747
    },
    {
      "epoch": 0.4767522553782096,
      "grad_norm": 0.706098735332489,
      "learning_rate": 3.977370583373692e-05,
      "loss": 0.6331,
      "step": 2748
    },
    {
      "epoch": 0.4769257460097155,
      "grad_norm": 1.0353593826293945,
      "learning_rate": 3.9772682666215906e-05,
      "loss": 0.5831,
      "step": 2749
    },
    {
      "epoch": 0.4770992366412214,
      "grad_norm": 0.7834818959236145,
      "learning_rate": 3.9771657204052026e-05,
      "loss": 0.5303,
      "step": 2750
    },
    {
      "epoch": 0.4772727272727273,
      "grad_norm": 0.5836576223373413,
      "learning_rate": 3.977062944736429e-05,
      "loss": 0.5969,
      "step": 2751
    },
    {
      "epoch": 0.4774462179042332,
      "grad_norm": 0.8149217963218689,
      "learning_rate": 3.976959939627196e-05,
      "loss": 0.6594,
      "step": 2752
    },
    {
      "epoch": 0.4776197085357391,
      "grad_norm": 1.064262866973877,
      "learning_rate": 3.97685670508946e-05,
      "loss": 0.5863,
      "step": 2753
    },
    {
      "epoch": 0.47779319916724494,
      "grad_norm": 0.7263724207878113,
      "learning_rate": 3.9767532411351985e-05,
      "loss": 0.5328,
      "step": 2754
    },
    {
      "epoch": 0.47796668979875084,
      "grad_norm": 0.7789517641067505,
      "learning_rate": 3.976649547776421e-05,
      "loss": 0.6167,
      "step": 2755
    },
    {
      "epoch": 0.47814018043025674,
      "grad_norm": 1.1462479829788208,
      "learning_rate": 3.97654562502516e-05,
      "loss": 0.605,
      "step": 2756
    },
    {
      "epoch": 0.47831367106176265,
      "grad_norm": 0.6074411869049072,
      "learning_rate": 3.9764414728934755e-05,
      "loss": 0.6172,
      "step": 2757
    },
    {
      "epoch": 0.47848716169326855,
      "grad_norm": 0.9205759167671204,
      "learning_rate": 3.9763370913934554e-05,
      "loss": 0.5411,
      "step": 2758
    },
    {
      "epoch": 0.47866065232477445,
      "grad_norm": 1.2367000579833984,
      "learning_rate": 3.976232480537213e-05,
      "loss": 0.512,
      "step": 2759
    },
    {
      "epoch": 0.47883414295628035,
      "grad_norm": 1.5458433628082275,
      "learning_rate": 3.976127640336889e-05,
      "loss": 0.6302,
      "step": 2760
    },
    {
      "epoch": 0.47900763358778625,
      "grad_norm": 0.8205773234367371,
      "learning_rate": 3.976022570804649e-05,
      "loss": 0.6331,
      "step": 2761
    },
    {
      "epoch": 0.47918112421929215,
      "grad_norm": 0.8028043508529663,
      "learning_rate": 3.9759172719526876e-05,
      "loss": 0.5426,
      "step": 2762
    },
    {
      "epoch": 0.47935461485079806,
      "grad_norm": 0.8865018486976624,
      "learning_rate": 3.975811743793224e-05,
      "loss": 0.5605,
      "step": 2763
    },
    {
      "epoch": 0.47952810548230396,
      "grad_norm": 0.722575843334198,
      "learning_rate": 3.975705986338505e-05,
      "loss": 0.6724,
      "step": 2764
    },
    {
      "epoch": 0.47970159611380986,
      "grad_norm": 1.0967686176300049,
      "learning_rate": 3.975599999600804e-05,
      "loss": 0.6622,
      "step": 2765
    },
    {
      "epoch": 0.47987508674531576,
      "grad_norm": 1.0719001293182373,
      "learning_rate": 3.9754937835924214e-05,
      "loss": 0.6163,
      "step": 2766
    },
    {
      "epoch": 0.48004857737682166,
      "grad_norm": 0.7715944051742554,
      "learning_rate": 3.975387338325684e-05,
      "loss": 0.5645,
      "step": 2767
    },
    {
      "epoch": 0.48022206800832756,
      "grad_norm": 1.0731868743896484,
      "learning_rate": 3.9752806638129435e-05,
      "loss": 0.572,
      "step": 2768
    },
    {
      "epoch": 0.48039555863983346,
      "grad_norm": 0.8332193493843079,
      "learning_rate": 3.975173760066581e-05,
      "loss": 0.7068,
      "step": 2769
    },
    {
      "epoch": 0.48056904927133937,
      "grad_norm": 0.9313003420829773,
      "learning_rate": 3.975066627099e-05,
      "loss": 0.6307,
      "step": 2770
    },
    {
      "epoch": 0.48074253990284527,
      "grad_norm": 0.9903737902641296,
      "learning_rate": 3.974959264922638e-05,
      "loss": 0.6519,
      "step": 2771
    },
    {
      "epoch": 0.48091603053435117,
      "grad_norm": 0.8947104811668396,
      "learning_rate": 3.974851673549951e-05,
      "loss": 0.5934,
      "step": 2772
    },
    {
      "epoch": 0.48108952116585707,
      "grad_norm": 0.9411084055900574,
      "learning_rate": 3.974743852993426e-05,
      "loss": 0.6036,
      "step": 2773
    },
    {
      "epoch": 0.4812630117973629,
      "grad_norm": 0.6337997913360596,
      "learning_rate": 3.9746358032655764e-05,
      "loss": 0.7129,
      "step": 2774
    },
    {
      "epoch": 0.4814365024288688,
      "grad_norm": 0.6764761209487915,
      "learning_rate": 3.9745275243789396e-05,
      "loss": 0.5551,
      "step": 2775
    },
    {
      "epoch": 0.4816099930603747,
      "grad_norm": 0.9338626265525818,
      "learning_rate": 3.974419016346084e-05,
      "loss": 0.5388,
      "step": 2776
    },
    {
      "epoch": 0.4817834836918806,
      "grad_norm": 0.6472459435462952,
      "learning_rate": 3.9743102791796005e-05,
      "loss": 0.5714,
      "step": 2777
    },
    {
      "epoch": 0.4819569743233865,
      "grad_norm": 1.0440516471862793,
      "learning_rate": 3.974201312892109e-05,
      "loss": 0.6394,
      "step": 2778
    },
    {
      "epoch": 0.4821304649548924,
      "grad_norm": 0.8014706373214722,
      "learning_rate": 3.9740921174962545e-05,
      "loss": 0.6423,
      "step": 2779
    },
    {
      "epoch": 0.4823039555863983,
      "grad_norm": 0.8197331428527832,
      "learning_rate": 3.9739826930047095e-05,
      "loss": 0.6128,
      "step": 2780
    },
    {
      "epoch": 0.4824774462179042,
      "grad_norm": 1.1526602506637573,
      "learning_rate": 3.9738730394301726e-05,
      "loss": 0.6182,
      "step": 2781
    },
    {
      "epoch": 0.48265093684941013,
      "grad_norm": 0.9170807003974915,
      "learning_rate": 3.9737631567853695e-05,
      "loss": 0.6565,
      "step": 2782
    },
    {
      "epoch": 0.48282442748091603,
      "grad_norm": 1.079443097114563,
      "learning_rate": 3.9736530450830525e-05,
      "loss": 0.6467,
      "step": 2783
    },
    {
      "epoch": 0.48299791811242193,
      "grad_norm": 0.743094265460968,
      "learning_rate": 3.9735427043359996e-05,
      "loss": 0.6473,
      "step": 2784
    },
    {
      "epoch": 0.48317140874392783,
      "grad_norm": 2.7557244300842285,
      "learning_rate": 3.9734321345570165e-05,
      "loss": 0.5278,
      "step": 2785
    },
    {
      "epoch": 0.48334489937543373,
      "grad_norm": 0.9088783860206604,
      "learning_rate": 3.973321335758934e-05,
      "loss": 0.6384,
      "step": 2786
    },
    {
      "epoch": 0.48351839000693964,
      "grad_norm": 0.7157585024833679,
      "learning_rate": 3.9732103079546125e-05,
      "loss": 0.5518,
      "step": 2787
    },
    {
      "epoch": 0.48369188063844554,
      "grad_norm": 0.6420220136642456,
      "learning_rate": 3.973099051156934e-05,
      "loss": 0.6764,
      "step": 2788
    },
    {
      "epoch": 0.48386537126995144,
      "grad_norm": 0.728564441204071,
      "learning_rate": 3.9729875653788125e-05,
      "loss": 0.6217,
      "step": 2789
    },
    {
      "epoch": 0.48403886190145734,
      "grad_norm": 0.7883340716362,
      "learning_rate": 3.972875850633184e-05,
      "loss": 0.5724,
      "step": 2790
    },
    {
      "epoch": 0.48421235253296324,
      "grad_norm": 0.7076512575149536,
      "learning_rate": 3.972763906933015e-05,
      "loss": 0.6989,
      "step": 2791
    },
    {
      "epoch": 0.48438584316446914,
      "grad_norm": 1.6119723320007324,
      "learning_rate": 3.9726517342912954e-05,
      "loss": 0.5679,
      "step": 2792
    },
    {
      "epoch": 0.48455933379597504,
      "grad_norm": 0.9080114364624023,
      "learning_rate": 3.972539332721044e-05,
      "loss": 0.5646,
      "step": 2793
    },
    {
      "epoch": 0.4847328244274809,
      "grad_norm": 0.6551885008811951,
      "learning_rate": 3.972426702235304e-05,
      "loss": 0.6938,
      "step": 2794
    },
    {
      "epoch": 0.4849063150589868,
      "grad_norm": 1.1007322072982788,
      "learning_rate": 3.972313842847148e-05,
      "loss": 0.5051,
      "step": 2795
    },
    {
      "epoch": 0.4850798056904927,
      "grad_norm": 1.7170222997665405,
      "learning_rate": 3.972200754569671e-05,
      "loss": 0.5197,
      "step": 2796
    },
    {
      "epoch": 0.4852532963219986,
      "grad_norm": 0.829792857170105,
      "learning_rate": 3.972087437415999e-05,
      "loss": 0.6917,
      "step": 2797
    },
    {
      "epoch": 0.4854267869535045,
      "grad_norm": 1.100192666053772,
      "learning_rate": 3.9719738913992815e-05,
      "loss": 0.5415,
      "step": 2798
    },
    {
      "epoch": 0.4856002775850104,
      "grad_norm": 0.8359057903289795,
      "learning_rate": 3.971860116532696e-05,
      "loss": 0.5387,
      "step": 2799
    },
    {
      "epoch": 0.4857737682165163,
      "grad_norm": 0.8219515681266785,
      "learning_rate": 3.971746112829447e-05,
      "loss": 0.6611,
      "step": 2800
    },
    {
      "epoch": 0.4859472588480222,
      "grad_norm": 0.5996765494346619,
      "learning_rate": 3.971631880302764e-05,
      "loss": 0.6564,
      "step": 2801
    },
    {
      "epoch": 0.4861207494795281,
      "grad_norm": 1.5173307657241821,
      "learning_rate": 3.9715174189659036e-05,
      "loss": 0.6659,
      "step": 2802
    },
    {
      "epoch": 0.486294240111034,
      "grad_norm": 0.9307529926300049,
      "learning_rate": 3.97140272883215e-05,
      "loss": 0.585,
      "step": 2803
    },
    {
      "epoch": 0.4864677307425399,
      "grad_norm": 0.7527129650115967,
      "learning_rate": 3.971287809914811e-05,
      "loss": 0.6942,
      "step": 2804
    },
    {
      "epoch": 0.4866412213740458,
      "grad_norm": 0.67873215675354,
      "learning_rate": 3.9711726622272266e-05,
      "loss": 0.6327,
      "step": 2805
    },
    {
      "epoch": 0.4868147120055517,
      "grad_norm": 0.8369560241699219,
      "learning_rate": 3.971057285782757e-05,
      "loss": 0.5317,
      "step": 2806
    },
    {
      "epoch": 0.4869882026370576,
      "grad_norm": 0.9180094599723816,
      "learning_rate": 3.970941680594792e-05,
      "loss": 0.6156,
      "step": 2807
    },
    {
      "epoch": 0.4871616932685635,
      "grad_norm": 1.8472323417663574,
      "learning_rate": 3.970825846676749e-05,
      "loss": 0.6224,
      "step": 2808
    },
    {
      "epoch": 0.4873351839000694,
      "grad_norm": 1.179469108581543,
      "learning_rate": 3.9707097840420706e-05,
      "loss": 0.7119,
      "step": 2809
    },
    {
      "epoch": 0.4875086745315753,
      "grad_norm": 1.036863923072815,
      "learning_rate": 3.970593492704225e-05,
      "loss": 0.5345,
      "step": 2810
    },
    {
      "epoch": 0.4876821651630812,
      "grad_norm": 1.147894263267517,
      "learning_rate": 3.970476972676708e-05,
      "loss": 0.5669,
      "step": 2811
    },
    {
      "epoch": 0.4878556557945871,
      "grad_norm": 0.7939436435699463,
      "learning_rate": 3.9703602239730425e-05,
      "loss": 0.6367,
      "step": 2812
    },
    {
      "epoch": 0.48802914642609296,
      "grad_norm": 0.6466153860092163,
      "learning_rate": 3.970243246606777e-05,
      "loss": 0.7765,
      "step": 2813
    },
    {
      "epoch": 0.48820263705759886,
      "grad_norm": 4.05562686920166,
      "learning_rate": 3.9701260405914874e-05,
      "loss": 0.6578,
      "step": 2814
    },
    {
      "epoch": 0.48837612768910477,
      "grad_norm": 0.7797324657440186,
      "learning_rate": 3.9700086059407745e-05,
      "loss": 0.6892,
      "step": 2815
    },
    {
      "epoch": 0.48854961832061067,
      "grad_norm": 1.3146294355392456,
      "learning_rate": 3.9698909426682674e-05,
      "loss": 0.5365,
      "step": 2816
    },
    {
      "epoch": 0.48872310895211657,
      "grad_norm": 0.7870042324066162,
      "learning_rate": 3.969773050787622e-05,
      "loss": 0.6057,
      "step": 2817
    },
    {
      "epoch": 0.48889659958362247,
      "grad_norm": 1.0030949115753174,
      "learning_rate": 3.9696549303125176e-05,
      "loss": 0.5239,
      "step": 2818
    },
    {
      "epoch": 0.4890700902151284,
      "grad_norm": 0.6438866257667542,
      "learning_rate": 3.9695365812566645e-05,
      "loss": 0.5275,
      "step": 2819
    },
    {
      "epoch": 0.4892435808466343,
      "grad_norm": 0.847891628742218,
      "learning_rate": 3.969418003633795e-05,
      "loss": 0.5771,
      "step": 2820
    },
    {
      "epoch": 0.4894170714781402,
      "grad_norm": 0.8128512501716614,
      "learning_rate": 3.9692991974576725e-05,
      "loss": 0.5381,
      "step": 2821
    },
    {
      "epoch": 0.4895905621096461,
      "grad_norm": 1.2491570711135864,
      "learning_rate": 3.969180162742082e-05,
      "loss": 0.5964,
      "step": 2822
    },
    {
      "epoch": 0.489764052741152,
      "grad_norm": 1.1974592208862305,
      "learning_rate": 3.96906089950084e-05,
      "loss": 0.5793,
      "step": 2823
    },
    {
      "epoch": 0.4899375433726579,
      "grad_norm": 0.808745265007019,
      "learning_rate": 3.9689414077477865e-05,
      "loss": 0.6477,
      "step": 2824
    },
    {
      "epoch": 0.4901110340041638,
      "grad_norm": 0.7065175175666809,
      "learning_rate": 3.968821687496788e-05,
      "loss": 0.6975,
      "step": 2825
    },
    {
      "epoch": 0.4902845246356697,
      "grad_norm": 0.8909404873847961,
      "learning_rate": 3.968701738761739e-05,
      "loss": 0.6786,
      "step": 2826
    },
    {
      "epoch": 0.4904580152671756,
      "grad_norm": 0.8688374757766724,
      "learning_rate": 3.968581561556558e-05,
      "loss": 0.6854,
      "step": 2827
    },
    {
      "epoch": 0.4906315058986815,
      "grad_norm": 0.8573815226554871,
      "learning_rate": 3.968461155895194e-05,
      "loss": 0.6083,
      "step": 2828
    },
    {
      "epoch": 0.4908049965301874,
      "grad_norm": 0.8470245599746704,
      "learning_rate": 3.968340521791619e-05,
      "loss": 0.6835,
      "step": 2829
    },
    {
      "epoch": 0.4909784871616933,
      "grad_norm": 0.7496466040611267,
      "learning_rate": 3.9682196592598324e-05,
      "loss": 0.562,
      "step": 2830
    },
    {
      "epoch": 0.4911519777931992,
      "grad_norm": 0.7681129574775696,
      "learning_rate": 3.968098568313862e-05,
      "loss": 0.6537,
      "step": 2831
    },
    {
      "epoch": 0.4913254684247051,
      "grad_norm": 1.5018384456634521,
      "learning_rate": 3.967977248967758e-05,
      "loss": 0.6011,
      "step": 2832
    },
    {
      "epoch": 0.49149895905621094,
      "grad_norm": 1.2373415231704712,
      "learning_rate": 3.9678557012356014e-05,
      "loss": 0.5194,
      "step": 2833
    },
    {
      "epoch": 0.49167244968771684,
      "grad_norm": 0.9008753895759583,
      "learning_rate": 3.967733925131498e-05,
      "loss": 0.5525,
      "step": 2834
    },
    {
      "epoch": 0.49184594031922274,
      "grad_norm": 0.8494754433631897,
      "learning_rate": 3.96761192066958e-05,
      "loss": 0.6589,
      "step": 2835
    },
    {
      "epoch": 0.49201943095072864,
      "grad_norm": 0.7637364864349365,
      "learning_rate": 3.9674896878640054e-05,
      "loss": 0.5907,
      "step": 2836
    },
    {
      "epoch": 0.49219292158223454,
      "grad_norm": 0.844154953956604,
      "learning_rate": 3.9673672267289604e-05,
      "loss": 0.5884,
      "step": 2837
    },
    {
      "epoch": 0.49236641221374045,
      "grad_norm": 0.9407692551612854,
      "learning_rate": 3.9672445372786565e-05,
      "loss": 0.6096,
      "step": 2838
    },
    {
      "epoch": 0.49253990284524635,
      "grad_norm": 0.8381971120834351,
      "learning_rate": 3.967121619527331e-05,
      "loss": 0.5492,
      "step": 2839
    },
    {
      "epoch": 0.49271339347675225,
      "grad_norm": 0.8256162405014038,
      "learning_rate": 3.96699847348925e-05,
      "loss": 0.6206,
      "step": 2840
    },
    {
      "epoch": 0.49288688410825815,
      "grad_norm": 0.8454298973083496,
      "learning_rate": 3.9668750991787034e-05,
      "loss": 0.7008,
      "step": 2841
    },
    {
      "epoch": 0.49306037473976405,
      "grad_norm": 0.7378323674201965,
      "learning_rate": 3.966751496610011e-05,
      "loss": 0.6375,
      "step": 2842
    },
    {
      "epoch": 0.49323386537126995,
      "grad_norm": 0.6299169063568115,
      "learning_rate": 3.9666276657975144e-05,
      "loss": 0.7336,
      "step": 2843
    },
    {
      "epoch": 0.49340735600277585,
      "grad_norm": 0.684691309928894,
      "learning_rate": 3.966503606755586e-05,
      "loss": 0.5975,
      "step": 2844
    },
    {
      "epoch": 0.49358084663428176,
      "grad_norm": 0.823893129825592,
      "learning_rate": 3.966379319498623e-05,
      "loss": 0.6161,
      "step": 2845
    },
    {
      "epoch": 0.49375433726578766,
      "grad_norm": 0.8355918526649475,
      "learning_rate": 3.9662548040410485e-05,
      "loss": 0.6804,
      "step": 2846
    },
    {
      "epoch": 0.49392782789729356,
      "grad_norm": 0.9166586399078369,
      "learning_rate": 3.966130060397312e-05,
      "loss": 0.652,
      "step": 2847
    },
    {
      "epoch": 0.49410131852879946,
      "grad_norm": 0.8756088018417358,
      "learning_rate": 3.9660050885818925e-05,
      "loss": 0.6575,
      "step": 2848
    },
    {
      "epoch": 0.49427480916030536,
      "grad_norm": 1.5226891040802002,
      "learning_rate": 3.9658798886092904e-05,
      "loss": 0.553,
      "step": 2849
    },
    {
      "epoch": 0.49444829979181126,
      "grad_norm": 0.7012729644775391,
      "learning_rate": 3.965754460494037e-05,
      "loss": 0.6741,
      "step": 2850
    },
    {
      "epoch": 0.49462179042331716,
      "grad_norm": 1.1426388025283813,
      "learning_rate": 3.965628804250688e-05,
      "loss": 0.6549,
      "step": 2851
    },
    {
      "epoch": 0.49479528105482307,
      "grad_norm": 0.9859298467636108,
      "learning_rate": 3.9655029198938256e-05,
      "loss": 0.5869,
      "step": 2852
    },
    {
      "epoch": 0.4949687716863289,
      "grad_norm": 0.8353455066680908,
      "learning_rate": 3.965376807438059e-05,
      "loss": 0.6431,
      "step": 2853
    },
    {
      "epoch": 0.4951422623178348,
      "grad_norm": 1.476958155632019,
      "learning_rate": 3.965250466898024e-05,
      "loss": 0.5901,
      "step": 2854
    },
    {
      "epoch": 0.4953157529493407,
      "grad_norm": 0.6651232838630676,
      "learning_rate": 3.9651238982883826e-05,
      "loss": 0.5836,
      "step": 2855
    },
    {
      "epoch": 0.4954892435808466,
      "grad_norm": 0.6790268421173096,
      "learning_rate": 3.964997101623823e-05,
      "loss": 0.6171,
      "step": 2856
    },
    {
      "epoch": 0.4956627342123525,
      "grad_norm": 0.8696946501731873,
      "learning_rate": 3.964870076919059e-05,
      "loss": 0.5375,
      "step": 2857
    },
    {
      "epoch": 0.4958362248438584,
      "grad_norm": 0.8823839426040649,
      "learning_rate": 3.964742824188834e-05,
      "loss": 0.6649,
      "step": 2858
    },
    {
      "epoch": 0.4960097154753643,
      "grad_norm": 0.7932507395744324,
      "learning_rate": 3.964615343447915e-05,
      "loss": 0.6865,
      "step": 2859
    },
    {
      "epoch": 0.4961832061068702,
      "grad_norm": 0.653605043888092,
      "learning_rate": 3.9644876347110956e-05,
      "loss": 0.7451,
      "step": 2860
    },
    {
      "epoch": 0.4963566967383761,
      "grad_norm": 0.6188622713088989,
      "learning_rate": 3.964359697993198e-05,
      "loss": 0.7036,
      "step": 2861
    },
    {
      "epoch": 0.496530187369882,
      "grad_norm": 1.760206937789917,
      "learning_rate": 3.964231533309067e-05,
      "loss": 0.5435,
      "step": 2862
    },
    {
      "epoch": 0.4967036780013879,
      "grad_norm": 0.7955849766731262,
      "learning_rate": 3.964103140673579e-05,
      "loss": 0.5455,
      "step": 2863
    },
    {
      "epoch": 0.49687716863289383,
      "grad_norm": 0.7729914784431458,
      "learning_rate": 3.963974520101632e-05,
      "loss": 0.6177,
      "step": 2864
    },
    {
      "epoch": 0.49705065926439973,
      "grad_norm": 1.0147491693496704,
      "learning_rate": 3.963845671608154e-05,
      "loss": 0.6644,
      "step": 2865
    },
    {
      "epoch": 0.49722414989590563,
      "grad_norm": 0.9814116954803467,
      "learning_rate": 3.963716595208098e-05,
      "loss": 0.509,
      "step": 2866
    },
    {
      "epoch": 0.49739764052741153,
      "grad_norm": 1.0014759302139282,
      "learning_rate": 3.963587290916442e-05,
      "loss": 0.6312,
      "step": 2867
    },
    {
      "epoch": 0.49757113115891743,
      "grad_norm": 0.9009021520614624,
      "learning_rate": 3.963457758748193e-05,
      "loss": 0.6477,
      "step": 2868
    },
    {
      "epoch": 0.49774462179042334,
      "grad_norm": 0.8552650213241577,
      "learning_rate": 3.963327998718385e-05,
      "loss": 0.5275,
      "step": 2869
    },
    {
      "epoch": 0.49791811242192924,
      "grad_norm": 0.9464699625968933,
      "learning_rate": 3.963198010842073e-05,
      "loss": 0.5591,
      "step": 2870
    },
    {
      "epoch": 0.49809160305343514,
      "grad_norm": 0.909501314163208,
      "learning_rate": 3.963067795134344e-05,
      "loss": 0.7686,
      "step": 2871
    },
    {
      "epoch": 0.49826509368494104,
      "grad_norm": 0.853759765625,
      "learning_rate": 3.9629373516103114e-05,
      "loss": 0.6168,
      "step": 2872
    },
    {
      "epoch": 0.4984385843164469,
      "grad_norm": 0.8711340427398682,
      "learning_rate": 3.962806680285111e-05,
      "loss": 0.5017,
      "step": 2873
    },
    {
      "epoch": 0.4986120749479528,
      "grad_norm": 0.7842757105827332,
      "learning_rate": 3.9626757811739084e-05,
      "loss": 0.5222,
      "step": 2874
    },
    {
      "epoch": 0.4987855655794587,
      "grad_norm": 0.9540364742279053,
      "learning_rate": 3.962544654291894e-05,
      "loss": 0.572,
      "step": 2875
    },
    {
      "epoch": 0.4989590562109646,
      "grad_norm": 1.2070144414901733,
      "learning_rate": 3.962413299654286e-05,
      "loss": 0.6826,
      "step": 2876
    },
    {
      "epoch": 0.4991325468424705,
      "grad_norm": 0.9257821440696716,
      "learning_rate": 3.962281717276328e-05,
      "loss": 0.5488,
      "step": 2877
    },
    {
      "epoch": 0.4993060374739764,
      "grad_norm": 1.1215496063232422,
      "learning_rate": 3.962149907173291e-05,
      "loss": 0.4994,
      "step": 2878
    },
    {
      "epoch": 0.4994795281054823,
      "grad_norm": 1.1144548654556274,
      "learning_rate": 3.9620178693604696e-05,
      "loss": 0.5066,
      "step": 2879
    },
    {
      "epoch": 0.4996530187369882,
      "grad_norm": 0.8781638741493225,
      "learning_rate": 3.961885603853189e-05,
      "loss": 0.5291,
      "step": 2880
    },
    {
      "epoch": 0.4998265093684941,
      "grad_norm": 0.7669386863708496,
      "learning_rate": 3.961753110666798e-05,
      "loss": 0.599,
      "step": 2881
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8803635835647583,
      "learning_rate": 3.9616203898166724e-05,
      "loss": 0.6178,
      "step": 2882
    },
    {
      "epoch": 0.5001734906315058,
      "grad_norm": 0.7655910849571228,
      "learning_rate": 3.9614874413182144e-05,
      "loss": 0.5726,
      "step": 2883
    },
    {
      "epoch": 0.5003469812630118,
      "grad_norm": 0.6875408291816711,
      "learning_rate": 3.961354265186854e-05,
      "loss": 0.6406,
      "step": 2884
    },
    {
      "epoch": 0.5005204718945176,
      "grad_norm": 1.0279085636138916,
      "learning_rate": 3.961220861438045e-05,
      "loss": 0.6409,
      "step": 2885
    },
    {
      "epoch": 0.5006939625260236,
      "grad_norm": 0.7553526163101196,
      "learning_rate": 3.9610872300872704e-05,
      "loss": 0.6992,
      "step": 2886
    },
    {
      "epoch": 0.5008674531575295,
      "grad_norm": 1.0226480960845947,
      "learning_rate": 3.960953371150037e-05,
      "loss": 0.5773,
      "step": 2887
    },
    {
      "epoch": 0.5010409437890354,
      "grad_norm": 0.944045901298523,
      "learning_rate": 3.96081928464188e-05,
      "loss": 0.5137,
      "step": 2888
    },
    {
      "epoch": 0.5012144344205413,
      "grad_norm": 1.0751080513000488,
      "learning_rate": 3.9606849705783606e-05,
      "loss": 0.4993,
      "step": 2889
    },
    {
      "epoch": 0.5013879250520472,
      "grad_norm": 1.8217488527297974,
      "learning_rate": 3.960550428975066e-05,
      "loss": 0.6428,
      "step": 2890
    },
    {
      "epoch": 0.5015614156835531,
      "grad_norm": 0.7758296728134155,
      "learning_rate": 3.960415659847609e-05,
      "loss": 0.5029,
      "step": 2891
    },
    {
      "epoch": 0.501734906315059,
      "grad_norm": 0.9004384875297546,
      "learning_rate": 3.9602806632116304e-05,
      "loss": 0.504,
      "step": 2892
    },
    {
      "epoch": 0.5019083969465649,
      "grad_norm": 1.2992662191390991,
      "learning_rate": 3.960145439082797e-05,
      "loss": 0.5168,
      "step": 2893
    },
    {
      "epoch": 0.5020818875780708,
      "grad_norm": 0.715490460395813,
      "learning_rate": 3.960009987476801e-05,
      "loss": 0.5402,
      "step": 2894
    },
    {
      "epoch": 0.5022553782095767,
      "grad_norm": 0.7640341520309448,
      "learning_rate": 3.959874308409362e-05,
      "loss": 0.5972,
      "step": 2895
    },
    {
      "epoch": 0.5024288688410826,
      "grad_norm": 0.6937002539634705,
      "learning_rate": 3.959738401896227e-05,
      "loss": 0.7065,
      "step": 2896
    },
    {
      "epoch": 0.5026023594725885,
      "grad_norm": 0.7549835443496704,
      "learning_rate": 3.959602267953165e-05,
      "loss": 0.54,
      "step": 2897
    },
    {
      "epoch": 0.5027758501040944,
      "grad_norm": 0.736943244934082,
      "learning_rate": 3.9594659065959774e-05,
      "loss": 0.5231,
      "step": 2898
    },
    {
      "epoch": 0.5029493407356003,
      "grad_norm": 1.0537973642349243,
      "learning_rate": 3.9593293178404885e-05,
      "loss": 0.4803,
      "step": 2899
    },
    {
      "epoch": 0.5031228313671062,
      "grad_norm": 0.8774026036262512,
      "learning_rate": 3.959192501702548e-05,
      "loss": 0.6389,
      "step": 2900
    },
    {
      "epoch": 0.5032963219986121,
      "grad_norm": 0.7064281702041626,
      "learning_rate": 3.959055458198036e-05,
      "loss": 0.5803,
      "step": 2901
    },
    {
      "epoch": 0.5034698126301179,
      "grad_norm": 0.8273089528083801,
      "learning_rate": 3.958918187342855e-05,
      "loss": 0.6331,
      "step": 2902
    },
    {
      "epoch": 0.5036433032616239,
      "grad_norm": 0.6765996813774109,
      "learning_rate": 3.9587806891529354e-05,
      "loss": 0.6755,
      "step": 2903
    },
    {
      "epoch": 0.5038167938931297,
      "grad_norm": 0.7752816677093506,
      "learning_rate": 3.9586429636442346e-05,
      "loss": 0.625,
      "step": 2904
    },
    {
      "epoch": 0.5039902845246357,
      "grad_norm": 0.7052983641624451,
      "learning_rate": 3.958505010832735e-05,
      "loss": 0.8062,
      "step": 2905
    },
    {
      "epoch": 0.5041637751561415,
      "grad_norm": 0.6991101503372192,
      "learning_rate": 3.958366830734448e-05,
      "loss": 0.7192,
      "step": 2906
    },
    {
      "epoch": 0.5043372657876475,
      "grad_norm": 0.6590566635131836,
      "learning_rate": 3.958228423365408e-05,
      "loss": 0.528,
      "step": 2907
    },
    {
      "epoch": 0.5045107564191533,
      "grad_norm": 0.7325621843338013,
      "learning_rate": 3.958089788741677e-05,
      "loss": 0.5999,
      "step": 2908
    },
    {
      "epoch": 0.5046842470506593,
      "grad_norm": 0.7469881772994995,
      "learning_rate": 3.957950926879345e-05,
      "loss": 0.627,
      "step": 2909
    },
    {
      "epoch": 0.5048577376821651,
      "grad_norm": 0.8237590193748474,
      "learning_rate": 3.957811837794526e-05,
      "loss": 0.4972,
      "step": 2910
    },
    {
      "epoch": 0.5050312283136711,
      "grad_norm": 1.2984604835510254,
      "learning_rate": 3.9576725215033624e-05,
      "loss": 0.6031,
      "step": 2911
    },
    {
      "epoch": 0.5052047189451769,
      "grad_norm": 0.7125715613365173,
      "learning_rate": 3.9575329780220215e-05,
      "loss": 0.6146,
      "step": 2912
    },
    {
      "epoch": 0.5053782095766829,
      "grad_norm": 0.7786535024642944,
      "learning_rate": 3.957393207366697e-05,
      "loss": 0.5795,
      "step": 2913
    },
    {
      "epoch": 0.5055517002081887,
      "grad_norm": 1.1961077451705933,
      "learning_rate": 3.957253209553611e-05,
      "loss": 0.645,
      "step": 2914
    },
    {
      "epoch": 0.5057251908396947,
      "grad_norm": 0.7613911032676697,
      "learning_rate": 3.9571129845990084e-05,
      "loss": 0.6475,
      "step": 2915
    },
    {
      "epoch": 0.5058986814712005,
      "grad_norm": 1.0745956897735596,
      "learning_rate": 3.956972532519164e-05,
      "loss": 0.5205,
      "step": 2916
    },
    {
      "epoch": 0.5060721721027065,
      "grad_norm": 0.8047274947166443,
      "learning_rate": 3.956831853330376e-05,
      "loss": 0.6866,
      "step": 2917
    },
    {
      "epoch": 0.5062456627342123,
      "grad_norm": 1.157810091972351,
      "learning_rate": 3.956690947048972e-05,
      "loss": 0.636,
      "step": 2918
    },
    {
      "epoch": 0.5064191533657183,
      "grad_norm": 0.8943071365356445,
      "learning_rate": 3.956549813691304e-05,
      "loss": 0.5283,
      "step": 2919
    },
    {
      "epoch": 0.5065926439972241,
      "grad_norm": 0.9355053901672363,
      "learning_rate": 3.9564084532737495e-05,
      "loss": 0.6853,
      "step": 2920
    },
    {
      "epoch": 0.5067661346287301,
      "grad_norm": 0.7883349657058716,
      "learning_rate": 3.956266865812714e-05,
      "loss": 0.5863,
      "step": 2921
    },
    {
      "epoch": 0.506939625260236,
      "grad_norm": 0.890841543674469,
      "learning_rate": 3.9561250513246306e-05,
      "loss": 0.6272,
      "step": 2922
    },
    {
      "epoch": 0.5071131158917418,
      "grad_norm": 0.7559226751327515,
      "learning_rate": 3.9559830098259544e-05,
      "loss": 0.6003,
      "step": 2923
    },
    {
      "epoch": 0.5072866065232478,
      "grad_norm": 0.8822606801986694,
      "learning_rate": 3.955840741333171e-05,
      "loss": 0.4907,
      "step": 2924
    },
    {
      "epoch": 0.5074600971547536,
      "grad_norm": 0.9082823395729065,
      "learning_rate": 3.9556982458627905e-05,
      "loss": 0.5961,
      "step": 2925
    },
    {
      "epoch": 0.5076335877862596,
      "grad_norm": 0.854621946811676,
      "learning_rate": 3.9555555234313506e-05,
      "loss": 0.5956,
      "step": 2926
    },
    {
      "epoch": 0.5078070784177654,
      "grad_norm": 0.7763813138008118,
      "learning_rate": 3.955412574055413e-05,
      "loss": 0.5427,
      "step": 2927
    },
    {
      "epoch": 0.5079805690492714,
      "grad_norm": 1.4742603302001953,
      "learning_rate": 3.9552693977515675e-05,
      "loss": 0.5181,
      "step": 2928
    },
    {
      "epoch": 0.5081540596807772,
      "grad_norm": 0.8555228114128113,
      "learning_rate": 3.95512599453643e-05,
      "loss": 0.5499,
      "step": 2929
    },
    {
      "epoch": 0.5083275503122832,
      "grad_norm": 0.7972477674484253,
      "learning_rate": 3.9549823644266434e-05,
      "loss": 0.6315,
      "step": 2930
    },
    {
      "epoch": 0.508501040943789,
      "grad_norm": 0.9488426446914673,
      "learning_rate": 3.9548385074388745e-05,
      "loss": 0.6281,
      "step": 2931
    },
    {
      "epoch": 0.508674531575295,
      "grad_norm": 0.7253492474555969,
      "learning_rate": 3.9546944235898194e-05,
      "loss": 0.7048,
      "step": 2932
    },
    {
      "epoch": 0.5088480222068008,
      "grad_norm": 0.7153453826904297,
      "learning_rate": 3.9545501128961985e-05,
      "loss": 0.6403,
      "step": 2933
    },
    {
      "epoch": 0.5090215128383068,
      "grad_norm": 0.7560308575630188,
      "learning_rate": 3.954405575374759e-05,
      "loss": 0.731,
      "step": 2934
    },
    {
      "epoch": 0.5091950034698126,
      "grad_norm": 1.0532622337341309,
      "learning_rate": 3.9542608110422764e-05,
      "loss": 0.582,
      "step": 2935
    },
    {
      "epoch": 0.5093684941013186,
      "grad_norm": 0.9665968418121338,
      "learning_rate": 3.954115819915549e-05,
      "loss": 0.5809,
      "step": 2936
    },
    {
      "epoch": 0.5095419847328244,
      "grad_norm": 2.0514838695526123,
      "learning_rate": 3.953970602011404e-05,
      "loss": 0.5996,
      "step": 2937
    },
    {
      "epoch": 0.5097154753643304,
      "grad_norm": 1.7059575319290161,
      "learning_rate": 3.9538251573466926e-05,
      "loss": 0.6071,
      "step": 2938
    },
    {
      "epoch": 0.5098889659958362,
      "grad_norm": 2.3512909412384033,
      "learning_rate": 3.9536794859382966e-05,
      "loss": 0.5767,
      "step": 2939
    },
    {
      "epoch": 0.5100624566273422,
      "grad_norm": 0.9228242635726929,
      "learning_rate": 3.9535335878031185e-05,
      "loss": 0.6882,
      "step": 2940
    },
    {
      "epoch": 0.510235947258848,
      "grad_norm": 0.7073661684989929,
      "learning_rate": 3.953387462958092e-05,
      "loss": 0.536,
      "step": 2941
    },
    {
      "epoch": 0.5104094378903539,
      "grad_norm": 0.866476833820343,
      "learning_rate": 3.953241111420174e-05,
      "loss": 0.5959,
      "step": 2942
    },
    {
      "epoch": 0.5105829285218598,
      "grad_norm": 0.7014033198356628,
      "learning_rate": 3.9530945332063486e-05,
      "loss": 0.5774,
      "step": 2943
    },
    {
      "epoch": 0.5107564191533657,
      "grad_norm": 0.7500950694084167,
      "learning_rate": 3.9529477283336274e-05,
      "loss": 0.6086,
      "step": 2944
    },
    {
      "epoch": 0.5109299097848716,
      "grad_norm": 0.8323651552200317,
      "learning_rate": 3.952800696819046e-05,
      "loss": 0.6919,
      "step": 2945
    },
    {
      "epoch": 0.5111034004163775,
      "grad_norm": 0.7332131862640381,
      "learning_rate": 3.9526534386796696e-05,
      "loss": 0.5884,
      "step": 2946
    },
    {
      "epoch": 0.5112768910478834,
      "grad_norm": 1.0053178071975708,
      "learning_rate": 3.9525059539325854e-05,
      "loss": 0.5651,
      "step": 2947
    },
    {
      "epoch": 0.5114503816793893,
      "grad_norm": 0.8901483416557312,
      "learning_rate": 3.95235824259491e-05,
      "loss": 0.5414,
      "step": 2948
    },
    {
      "epoch": 0.5116238723108952,
      "grad_norm": 0.9375765919685364,
      "learning_rate": 3.952210304683786e-05,
      "loss": 0.6737,
      "step": 2949
    },
    {
      "epoch": 0.5117973629424011,
      "grad_norm": 0.7333534359931946,
      "learning_rate": 3.952062140216381e-05,
      "loss": 0.5692,
      "step": 2950
    },
    {
      "epoch": 0.511970853573907,
      "grad_norm": 0.9000279307365417,
      "learning_rate": 3.951913749209891e-05,
      "loss": 0.644,
      "step": 2951
    },
    {
      "epoch": 0.5121443442054129,
      "grad_norm": 1.2508001327514648,
      "learning_rate": 3.951765131681535e-05,
      "loss": 0.6993,
      "step": 2952
    },
    {
      "epoch": 0.5123178348369188,
      "grad_norm": 1.0223735570907593,
      "learning_rate": 3.951616287648561e-05,
      "loss": 0.6294,
      "step": 2953
    },
    {
      "epoch": 0.5124913254684247,
      "grad_norm": 1.4913530349731445,
      "learning_rate": 3.9514672171282435e-05,
      "loss": 0.5161,
      "step": 2954
    },
    {
      "epoch": 0.5126648160999306,
      "grad_norm": 0.7307369112968445,
      "learning_rate": 3.951317920137881e-05,
      "loss": 0.6313,
      "step": 2955
    },
    {
      "epoch": 0.5128383067314365,
      "grad_norm": 2.8054757118225098,
      "learning_rate": 3.951168396694801e-05,
      "loss": 0.5863,
      "step": 2956
    },
    {
      "epoch": 0.5130117973629424,
      "grad_norm": 2.5261077880859375,
      "learning_rate": 3.951018646816354e-05,
      "loss": 0.7458,
      "step": 2957
    },
    {
      "epoch": 0.5131852879944483,
      "grad_norm": 1.7429360151290894,
      "learning_rate": 3.9508686705199196e-05,
      "loss": 0.525,
      "step": 2958
    },
    {
      "epoch": 0.5133587786259542,
      "grad_norm": 1.2919009923934937,
      "learning_rate": 3.9507184678229035e-05,
      "loss": 0.4951,
      "step": 2959
    },
    {
      "epoch": 0.5135322692574601,
      "grad_norm": 0.6654407978057861,
      "learning_rate": 3.950568038742736e-05,
      "loss": 0.7129,
      "step": 2960
    },
    {
      "epoch": 0.5137057598889659,
      "grad_norm": 0.7495904564857483,
      "learning_rate": 3.9504173832968744e-05,
      "loss": 0.6887,
      "step": 2961
    },
    {
      "epoch": 0.5138792505204719,
      "grad_norm": 0.9177446365356445,
      "learning_rate": 3.950266501502803e-05,
      "loss": 0.5347,
      "step": 2962
    },
    {
      "epoch": 0.5140527411519777,
      "grad_norm": 0.8417816758155823,
      "learning_rate": 3.9501153933780314e-05,
      "loss": 0.5763,
      "step": 2963
    },
    {
      "epoch": 0.5142262317834837,
      "grad_norm": 0.6979896426200867,
      "learning_rate": 3.9499640589400964e-05,
      "loss": 0.6661,
      "step": 2964
    },
    {
      "epoch": 0.5143997224149895,
      "grad_norm": 1.1703006029129028,
      "learning_rate": 3.94981249820656e-05,
      "loss": 0.6267,
      "step": 2965
    },
    {
      "epoch": 0.5145732130464955,
      "grad_norm": 0.7633793950080872,
      "learning_rate": 3.949660711195011e-05,
      "loss": 0.7529,
      "step": 2966
    },
    {
      "epoch": 0.5147467036780013,
      "grad_norm": 0.8751875162124634,
      "learning_rate": 3.9495086979230656e-05,
      "loss": 0.553,
      "step": 2967
    },
    {
      "epoch": 0.5149201943095073,
      "grad_norm": 0.9142911434173584,
      "learning_rate": 3.949356458408363e-05,
      "loss": 0.621,
      "step": 2968
    },
    {
      "epoch": 0.5150936849410132,
      "grad_norm": 1.1330902576446533,
      "learning_rate": 3.9492039926685724e-05,
      "loss": 0.5168,
      "step": 2969
    },
    {
      "epoch": 0.5152671755725191,
      "grad_norm": 0.8150011301040649,
      "learning_rate": 3.9490513007213874e-05,
      "loss": 0.5708,
      "step": 2970
    },
    {
      "epoch": 0.515440666204025,
      "grad_norm": 0.7211820483207703,
      "learning_rate": 3.948898382584528e-05,
      "loss": 0.5453,
      "step": 2971
    },
    {
      "epoch": 0.5156141568355309,
      "grad_norm": 1.0376445055007935,
      "learning_rate": 3.94874523827574e-05,
      "loss": 0.6359,
      "step": 2972
    },
    {
      "epoch": 0.5157876474670368,
      "grad_norm": 0.5772063136100769,
      "learning_rate": 3.9485918678127954e-05,
      "loss": 0.7893,
      "step": 2973
    },
    {
      "epoch": 0.5159611380985427,
      "grad_norm": 0.7940547466278076,
      "learning_rate": 3.9484382712134956e-05,
      "loss": 0.689,
      "step": 2974
    },
    {
      "epoch": 0.5161346287300486,
      "grad_norm": 0.7835409641265869,
      "learning_rate": 3.948284448495663e-05,
      "loss": 0.6183,
      "step": 2975
    },
    {
      "epoch": 0.5163081193615545,
      "grad_norm": 0.9172401428222656,
      "learning_rate": 3.9481303996771505e-05,
      "loss": 0.6893,
      "step": 2976
    },
    {
      "epoch": 0.5164816099930604,
      "grad_norm": 0.7964392304420471,
      "learning_rate": 3.947976124775835e-05,
      "loss": 0.657,
      "step": 2977
    },
    {
      "epoch": 0.5166551006245663,
      "grad_norm": 0.7399853467941284,
      "learning_rate": 3.9478216238096206e-05,
      "loss": 0.5171,
      "step": 2978
    },
    {
      "epoch": 0.5168285912560722,
      "grad_norm": 0.7355174422264099,
      "learning_rate": 3.947666896796436e-05,
      "loss": 0.6783,
      "step": 2979
    },
    {
      "epoch": 0.5170020818875781,
      "grad_norm": 0.6980626583099365,
      "learning_rate": 3.94751194375424e-05,
      "loss": 0.7009,
      "step": 2980
    },
    {
      "epoch": 0.517175572519084,
      "grad_norm": 0.8751781582832336,
      "learning_rate": 3.947356764701013e-05,
      "loss": 0.5306,
      "step": 2981
    },
    {
      "epoch": 0.5173490631505898,
      "grad_norm": 1.3187061548233032,
      "learning_rate": 3.9472013596547646e-05,
      "loss": 0.5841,
      "step": 2982
    },
    {
      "epoch": 0.5175225537820958,
      "grad_norm": 1.0359468460083008,
      "learning_rate": 3.947045728633529e-05,
      "loss": 0.5577,
      "step": 2983
    },
    {
      "epoch": 0.5176960444136016,
      "grad_norm": 0.7905299067497253,
      "learning_rate": 3.946889871655368e-05,
      "loss": 0.5211,
      "step": 2984
    },
    {
      "epoch": 0.5178695350451076,
      "grad_norm": 1.22454035282135,
      "learning_rate": 3.9467337887383695e-05,
      "loss": 0.5479,
      "step": 2985
    },
    {
      "epoch": 0.5180430256766134,
      "grad_norm": 0.8180719017982483,
      "learning_rate": 3.946577479900645e-05,
      "loss": 0.6879,
      "step": 2986
    },
    {
      "epoch": 0.5182165163081194,
      "grad_norm": 1.2501583099365234,
      "learning_rate": 3.9464209451603367e-05,
      "loss": 0.5841,
      "step": 2987
    },
    {
      "epoch": 0.5183900069396252,
      "grad_norm": 0.5989814400672913,
      "learning_rate": 3.9462641845356096e-05,
      "loss": 0.6218,
      "step": 2988
    },
    {
      "epoch": 0.5185634975711312,
      "grad_norm": 0.7976348996162415,
      "learning_rate": 3.946107198044656e-05,
      "loss": 0.5869,
      "step": 2989
    },
    {
      "epoch": 0.518736988202637,
      "grad_norm": 0.8901609778404236,
      "learning_rate": 3.945949985705694e-05,
      "loss": 0.6271,
      "step": 2990
    },
    {
      "epoch": 0.518910478834143,
      "grad_norm": 1.047589659690857,
      "learning_rate": 3.945792547536969e-05,
      "loss": 0.6718,
      "step": 2991
    },
    {
      "epoch": 0.5190839694656488,
      "grad_norm": 1.1930838823318481,
      "learning_rate": 3.945634883556752e-05,
      "loss": 0.5968,
      "step": 2992
    },
    {
      "epoch": 0.5192574600971548,
      "grad_norm": 0.8060930967330933,
      "learning_rate": 3.945476993783339e-05,
      "loss": 0.5712,
      "step": 2993
    },
    {
      "epoch": 0.5194309507286606,
      "grad_norm": 0.889964759349823,
      "learning_rate": 3.945318878235054e-05,
      "loss": 0.5232,
      "step": 2994
    },
    {
      "epoch": 0.5196044413601666,
      "grad_norm": 0.7452549934387207,
      "learning_rate": 3.945160536930247e-05,
      "loss": 0.6582,
      "step": 2995
    },
    {
      "epoch": 0.5197779319916724,
      "grad_norm": 0.7570145130157471,
      "learning_rate": 3.945001969887293e-05,
      "loss": 0.598,
      "step": 2996
    },
    {
      "epoch": 0.5199514226231784,
      "grad_norm": 0.6366786956787109,
      "learning_rate": 3.944843177124593e-05,
      "loss": 0.6699,
      "step": 2997
    },
    {
      "epoch": 0.5201249132546842,
      "grad_norm": 0.7185344099998474,
      "learning_rate": 3.944684158660577e-05,
      "loss": 0.5645,
      "step": 2998
    },
    {
      "epoch": 0.5202984038861902,
      "grad_norm": 0.8471353054046631,
      "learning_rate": 3.944524914513698e-05,
      "loss": 0.5682,
      "step": 2999
    },
    {
      "epoch": 0.520471894517696,
      "grad_norm": 0.7987390160560608,
      "learning_rate": 3.944365444702437e-05,
      "loss": 0.6149,
      "step": 3000
    },
    {
      "epoch": 0.5206453851492019,
      "grad_norm": 0.8394932150840759,
      "learning_rate": 3.944205749245301e-05,
      "loss": 0.6683,
      "step": 3001
    },
    {
      "epoch": 0.5208188757807078,
      "grad_norm": 0.8959985375404358,
      "learning_rate": 3.944045828160822e-05,
      "loss": 0.5833,
      "step": 3002
    },
    {
      "epoch": 0.5209923664122137,
      "grad_norm": 0.6436522603034973,
      "learning_rate": 3.943885681467559e-05,
      "loss": 0.6285,
      "step": 3003
    },
    {
      "epoch": 0.5211658570437196,
      "grad_norm": 0.7823279500007629,
      "learning_rate": 3.943725309184098e-05,
      "loss": 0.6475,
      "step": 3004
    },
    {
      "epoch": 0.5213393476752255,
      "grad_norm": 1.139787197113037,
      "learning_rate": 3.94356471132905e-05,
      "loss": 0.6207,
      "step": 3005
    },
    {
      "epoch": 0.5215128383067315,
      "grad_norm": 0.8308555483818054,
      "learning_rate": 3.943403887921052e-05,
      "loss": 0.6595,
      "step": 3006
    },
    {
      "epoch": 0.5216863289382373,
      "grad_norm": 0.9472867846488953,
      "learning_rate": 3.943242838978769e-05,
      "loss": 0.5508,
      "step": 3007
    },
    {
      "epoch": 0.5218598195697433,
      "grad_norm": 0.6742684245109558,
      "learning_rate": 3.94308156452089e-05,
      "loss": 0.6439,
      "step": 3008
    },
    {
      "epoch": 0.5220333102012491,
      "grad_norm": 0.7607699036598206,
      "learning_rate": 3.942920064566131e-05,
      "loss": 0.6178,
      "step": 3009
    },
    {
      "epoch": 0.5222068008327551,
      "grad_norm": 0.7258752584457397,
      "learning_rate": 3.9427583391332354e-05,
      "loss": 0.6803,
      "step": 3010
    },
    {
      "epoch": 0.5223802914642609,
      "grad_norm": 2.0244803428649902,
      "learning_rate": 3.94259638824097e-05,
      "loss": 0.6411,
      "step": 3011
    },
    {
      "epoch": 0.5225537820957669,
      "grad_norm": 1.4871636629104614,
      "learning_rate": 3.94243421190813e-05,
      "loss": 0.6649,
      "step": 3012
    },
    {
      "epoch": 0.5227272727272727,
      "grad_norm": 1.0151525735855103,
      "learning_rate": 3.942271810153537e-05,
      "loss": 0.5286,
      "step": 3013
    },
    {
      "epoch": 0.5229007633587787,
      "grad_norm": 0.8217700123786926,
      "learning_rate": 3.9421091829960364e-05,
      "loss": 0.562,
      "step": 3014
    },
    {
      "epoch": 0.5230742539902845,
      "grad_norm": 1.2036974430084229,
      "learning_rate": 3.941946330454503e-05,
      "loss": 0.5923,
      "step": 3015
    },
    {
      "epoch": 0.5232477446217905,
      "grad_norm": 0.8942238688468933,
      "learning_rate": 3.9417832525478344e-05,
      "loss": 0.6196,
      "step": 3016
    },
    {
      "epoch": 0.5234212352532963,
      "grad_norm": 0.6654651165008545,
      "learning_rate": 3.941619949294957e-05,
      "loss": 0.5917,
      "step": 3017
    },
    {
      "epoch": 0.5235947258848023,
      "grad_norm": 1.251190185546875,
      "learning_rate": 3.941456420714822e-05,
      "loss": 0.5902,
      "step": 3018
    },
    {
      "epoch": 0.5237682165163081,
      "grad_norm": 0.7630736827850342,
      "learning_rate": 3.941292666826408e-05,
      "loss": 0.4933,
      "step": 3019
    },
    {
      "epoch": 0.5239417071478141,
      "grad_norm": 0.8213485479354858,
      "learning_rate": 3.941128687648717e-05,
      "loss": 0.6637,
      "step": 3020
    },
    {
      "epoch": 0.5241151977793199,
      "grad_norm": 0.7555518746376038,
      "learning_rate": 3.94096448320078e-05,
      "loss": 0.6387,
      "step": 3021
    },
    {
      "epoch": 0.5242886884108258,
      "grad_norm": 1.5528206825256348,
      "learning_rate": 3.940800053501653e-05,
      "loss": 0.6968,
      "step": 3022
    },
    {
      "epoch": 0.5244621790423317,
      "grad_norm": 0.8396145105361938,
      "learning_rate": 3.940635398570418e-05,
      "loss": 0.5979,
      "step": 3023
    },
    {
      "epoch": 0.5246356696738376,
      "grad_norm": 1.0485159158706665,
      "learning_rate": 3.9404705184261846e-05,
      "loss": 0.5519,
      "step": 3024
    },
    {
      "epoch": 0.5248091603053435,
      "grad_norm": 0.7583789229393005,
      "learning_rate": 3.9403054130880864e-05,
      "loss": 0.5808,
      "step": 3025
    },
    {
      "epoch": 0.5249826509368494,
      "grad_norm": 0.5559424757957458,
      "learning_rate": 3.9401400825752835e-05,
      "loss": 0.7402,
      "step": 3026
    },
    {
      "epoch": 0.5251561415683553,
      "grad_norm": 1.0415542125701904,
      "learning_rate": 3.9399745269069636e-05,
      "loss": 0.7172,
      "step": 3027
    },
    {
      "epoch": 0.5253296321998612,
      "grad_norm": 0.8398486971855164,
      "learning_rate": 3.939808746102339e-05,
      "loss": 0.5929,
      "step": 3028
    },
    {
      "epoch": 0.5255031228313671,
      "grad_norm": 0.8926632404327393,
      "learning_rate": 3.93964274018065e-05,
      "loss": 0.5403,
      "step": 3029
    },
    {
      "epoch": 0.525676613462873,
      "grad_norm": 0.825466513633728,
      "learning_rate": 3.9394765091611596e-05,
      "loss": 0.5776,
      "step": 3030
    },
    {
      "epoch": 0.5258501040943789,
      "grad_norm": 1.4896812438964844,
      "learning_rate": 3.939310053063161e-05,
      "loss": 0.6487,
      "step": 3031
    },
    {
      "epoch": 0.5260235947258848,
      "grad_norm": 0.8119390606880188,
      "learning_rate": 3.939143371905971e-05,
      "loss": 0.7996,
      "step": 3032
    },
    {
      "epoch": 0.5261970853573907,
      "grad_norm": 0.7745625376701355,
      "learning_rate": 3.9389764657089334e-05,
      "loss": 0.6672,
      "step": 3033
    },
    {
      "epoch": 0.5263705759888966,
      "grad_norm": 0.5133965611457825,
      "learning_rate": 3.938809334491417e-05,
      "loss": 0.6941,
      "step": 3034
    },
    {
      "epoch": 0.5265440666204025,
      "grad_norm": 0.8933323621749878,
      "learning_rate": 3.938641978272819e-05,
      "loss": 0.6237,
      "step": 3035
    },
    {
      "epoch": 0.5267175572519084,
      "grad_norm": 2.33898663520813,
      "learning_rate": 3.9384743970725596e-05,
      "loss": 0.5834,
      "step": 3036
    },
    {
      "epoch": 0.5268910478834143,
      "grad_norm": 0.7867479920387268,
      "learning_rate": 3.938306590910088e-05,
      "loss": 0.6361,
      "step": 3037
    },
    {
      "epoch": 0.5270645385149202,
      "grad_norm": 1.6359503269195557,
      "learning_rate": 3.938138559804878e-05,
      "loss": 0.5477,
      "step": 3038
    },
    {
      "epoch": 0.5272380291464261,
      "grad_norm": 1.137347936630249,
      "learning_rate": 3.937970303776429e-05,
      "loss": 0.5846,
      "step": 3039
    },
    {
      "epoch": 0.527411519777932,
      "grad_norm": 0.8294497132301331,
      "learning_rate": 3.9378018228442696e-05,
      "loss": 0.6381,
      "step": 3040
    },
    {
      "epoch": 0.5275850104094378,
      "grad_norm": 0.9856337904930115,
      "learning_rate": 3.93763311702795e-05,
      "loss": 0.6945,
      "step": 3041
    },
    {
      "epoch": 0.5277585010409438,
      "grad_norm": 0.6542000770568848,
      "learning_rate": 3.937464186347049e-05,
      "loss": 0.6139,
      "step": 3042
    },
    {
      "epoch": 0.5279319916724496,
      "grad_norm": 0.8927829265594482,
      "learning_rate": 3.9372950308211715e-05,
      "loss": 0.5177,
      "step": 3043
    },
    {
      "epoch": 0.5281054823039556,
      "grad_norm": 1.0479310750961304,
      "learning_rate": 3.9371256504699486e-05,
      "loss": 0.5997,
      "step": 3044
    },
    {
      "epoch": 0.5282789729354614,
      "grad_norm": 1.0101724863052368,
      "learning_rate": 3.9369560453130366e-05,
      "loss": 0.6899,
      "step": 3045
    },
    {
      "epoch": 0.5284524635669674,
      "grad_norm": 0.889836847782135,
      "learning_rate": 3.936786215370119e-05,
      "loss": 0.571,
      "step": 3046
    },
    {
      "epoch": 0.5286259541984732,
      "grad_norm": 0.8853172659873962,
      "learning_rate": 3.9366161606609045e-05,
      "loss": 0.6084,
      "step": 3047
    },
    {
      "epoch": 0.5287994448299792,
      "grad_norm": 1.596792221069336,
      "learning_rate": 3.936445881205127e-05,
      "loss": 0.6294,
      "step": 3048
    },
    {
      "epoch": 0.528972935461485,
      "grad_norm": 0.6516552567481995,
      "learning_rate": 3.936275377022549e-05,
      "loss": 0.6698,
      "step": 3049
    },
    {
      "epoch": 0.529146426092991,
      "grad_norm": 0.9162889122962952,
      "learning_rate": 3.936104648132957e-05,
      "loss": 0.5614,
      "step": 3050
    },
    {
      "epoch": 0.5293199167244969,
      "grad_norm": 1.008646845817566,
      "learning_rate": 3.9359336945561656e-05,
      "loss": 0.6289,
      "step": 3051
    },
    {
      "epoch": 0.5294934073560028,
      "grad_norm": 0.7039345502853394,
      "learning_rate": 3.935762516312012e-05,
      "loss": 0.6151,
      "step": 3052
    },
    {
      "epoch": 0.5296668979875087,
      "grad_norm": 0.8550469875335693,
      "learning_rate": 3.9355911134203626e-05,
      "loss": 0.4678,
      "step": 3053
    },
    {
      "epoch": 0.5298403886190146,
      "grad_norm": 0.8338708877563477,
      "learning_rate": 3.9354194859011105e-05,
      "loss": 0.5522,
      "step": 3054
    },
    {
      "epoch": 0.5300138792505205,
      "grad_norm": 0.7659257650375366,
      "learning_rate": 3.935247633774171e-05,
      "loss": 0.6072,
      "step": 3055
    },
    {
      "epoch": 0.5301873698820264,
      "grad_norm": 0.6845202445983887,
      "learning_rate": 3.935075557059488e-05,
      "loss": 0.7031,
      "step": 3056
    },
    {
      "epoch": 0.5303608605135323,
      "grad_norm": 1.2755361795425415,
      "learning_rate": 3.934903255777033e-05,
      "loss": 0.6149,
      "step": 3057
    },
    {
      "epoch": 0.5305343511450382,
      "grad_norm": 0.8341980576515198,
      "learning_rate": 3.934730729946799e-05,
      "loss": 0.5929,
      "step": 3058
    },
    {
      "epoch": 0.5307078417765441,
      "grad_norm": 0.6358599066734314,
      "learning_rate": 3.934557979588811e-05,
      "loss": 0.6218,
      "step": 3059
    },
    {
      "epoch": 0.5308813324080499,
      "grad_norm": 0.808149516582489,
      "learning_rate": 3.9343850047231144e-05,
      "loss": 0.5128,
      "step": 3060
    },
    {
      "epoch": 0.5310548230395559,
      "grad_norm": 0.9138489961624146,
      "learning_rate": 3.9342118053697837e-05,
      "loss": 0.5391,
      "step": 3061
    },
    {
      "epoch": 0.5312283136710617,
      "grad_norm": 1.1662746667861938,
      "learning_rate": 3.9340383815489204e-05,
      "loss": 0.5781,
      "step": 3062
    },
    {
      "epoch": 0.5314018043025677,
      "grad_norm": 0.7738902568817139,
      "learning_rate": 3.933864733280648e-05,
      "loss": 0.5934,
      "step": 3063
    },
    {
      "epoch": 0.5315752949340735,
      "grad_norm": 0.7514986991882324,
      "learning_rate": 3.933690860585121e-05,
      "loss": 0.5417,
      "step": 3064
    },
    {
      "epoch": 0.5317487855655795,
      "grad_norm": 0.7755056023597717,
      "learning_rate": 3.933516763482516e-05,
      "loss": 0.6454,
      "step": 3065
    },
    {
      "epoch": 0.5319222761970853,
      "grad_norm": 0.8597734570503235,
      "learning_rate": 3.933342441993037e-05,
      "loss": 0.6802,
      "step": 3066
    },
    {
      "epoch": 0.5320957668285913,
      "grad_norm": 0.7433542609214783,
      "learning_rate": 3.9331678961369156e-05,
      "loss": 0.6699,
      "step": 3067
    },
    {
      "epoch": 0.5322692574600971,
      "grad_norm": 0.8031440377235413,
      "learning_rate": 3.932993125934407e-05,
      "loss": 0.5856,
      "step": 3068
    },
    {
      "epoch": 0.5324427480916031,
      "grad_norm": 0.599829912185669,
      "learning_rate": 3.932818131405794e-05,
      "loss": 0.665,
      "step": 3069
    },
    {
      "epoch": 0.5326162387231089,
      "grad_norm": 0.669053316116333,
      "learning_rate": 3.932642912571385e-05,
      "loss": 0.5923,
      "step": 3070
    },
    {
      "epoch": 0.5327897293546149,
      "grad_norm": 0.6360391974449158,
      "learning_rate": 3.9324674694515126e-05,
      "loss": 0.6462,
      "step": 3071
    },
    {
      "epoch": 0.5329632199861207,
      "grad_norm": 1.3935229778289795,
      "learning_rate": 3.932291802066539e-05,
      "loss": 0.5499,
      "step": 3072
    },
    {
      "epoch": 0.5331367106176267,
      "grad_norm": 0.7656862735748291,
      "learning_rate": 3.932115910436851e-05,
      "loss": 0.5592,
      "step": 3073
    },
    {
      "epoch": 0.5333102012491325,
      "grad_norm": 0.8608168959617615,
      "learning_rate": 3.93193979458286e-05,
      "loss": 0.5165,
      "step": 3074
    },
    {
      "epoch": 0.5334836918806385,
      "grad_norm": 0.6383200287818909,
      "learning_rate": 3.931763454525005e-05,
      "loss": 0.6392,
      "step": 3075
    },
    {
      "epoch": 0.5336571825121443,
      "grad_norm": 0.7783990502357483,
      "learning_rate": 3.93158689028375e-05,
      "loss": 0.5022,
      "step": 3076
    },
    {
      "epoch": 0.5338306731436503,
      "grad_norm": 0.9632290005683899,
      "learning_rate": 3.931410101879585e-05,
      "loss": 0.5671,
      "step": 3077
    },
    {
      "epoch": 0.5340041637751561,
      "grad_norm": 0.6838582158088684,
      "learning_rate": 3.931233089333027e-05,
      "loss": 0.5509,
      "step": 3078
    },
    {
      "epoch": 0.5341776544066621,
      "grad_norm": 0.9874531030654907,
      "learning_rate": 3.931055852664619e-05,
      "loss": 0.6674,
      "step": 3079
    },
    {
      "epoch": 0.5343511450381679,
      "grad_norm": 0.9669835567474365,
      "learning_rate": 3.9308783918949296e-05,
      "loss": 0.5638,
      "step": 3080
    },
    {
      "epoch": 0.5345246356696738,
      "grad_norm": 0.7132205963134766,
      "learning_rate": 3.930700707044552e-05,
      "loss": 0.6456,
      "step": 3081
    },
    {
      "epoch": 0.5346981263011797,
      "grad_norm": 0.7564111948013306,
      "learning_rate": 3.9305227981341085e-05,
      "loss": 0.6383,
      "step": 3082
    },
    {
      "epoch": 0.5348716169326856,
      "grad_norm": 0.8441155552864075,
      "learning_rate": 3.9303446651842444e-05,
      "loss": 0.6904,
      "step": 3083
    },
    {
      "epoch": 0.5350451075641915,
      "grad_norm": 0.8988627791404724,
      "learning_rate": 3.930166308215633e-05,
      "loss": 0.6704,
      "step": 3084
    },
    {
      "epoch": 0.5352185981956974,
      "grad_norm": 0.9254375696182251,
      "learning_rate": 3.929987727248972e-05,
      "loss": 0.629,
      "step": 3085
    },
    {
      "epoch": 0.5353920888272033,
      "grad_norm": 0.9920805096626282,
      "learning_rate": 3.929808922304987e-05,
      "loss": 0.549,
      "step": 3086
    },
    {
      "epoch": 0.5355655794587092,
      "grad_norm": 0.5927398204803467,
      "learning_rate": 3.929629893404428e-05,
      "loss": 0.59,
      "step": 3087
    },
    {
      "epoch": 0.5357390700902152,
      "grad_norm": 1.3722820281982422,
      "learning_rate": 3.92945064056807e-05,
      "loss": 0.5322,
      "step": 3088
    },
    {
      "epoch": 0.535912560721721,
      "grad_norm": 0.7934204936027527,
      "learning_rate": 3.929271163816718e-05,
      "loss": 0.6578,
      "step": 3089
    },
    {
      "epoch": 0.536086051353227,
      "grad_norm": 1.265750765800476,
      "learning_rate": 3.929091463171199e-05,
      "loss": 0.5594,
      "step": 3090
    },
    {
      "epoch": 0.5362595419847328,
      "grad_norm": 0.7382822036743164,
      "learning_rate": 3.9289115386523676e-05,
      "loss": 0.6552,
      "step": 3091
    },
    {
      "epoch": 0.5364330326162388,
      "grad_norm": 0.8243357539176941,
      "learning_rate": 3.928731390281105e-05,
      "loss": 0.6023,
      "step": 3092
    },
    {
      "epoch": 0.5366065232477446,
      "grad_norm": 0.7092339396476746,
      "learning_rate": 3.928551018078317e-05,
      "loss": 0.6342,
      "step": 3093
    },
    {
      "epoch": 0.5367800138792506,
      "grad_norm": 0.6945857405662537,
      "learning_rate": 3.928370422064936e-05,
      "loss": 0.6372,
      "step": 3094
    },
    {
      "epoch": 0.5369535045107564,
      "grad_norm": 0.6603047251701355,
      "learning_rate": 3.928189602261921e-05,
      "loss": 0.7356,
      "step": 3095
    },
    {
      "epoch": 0.5371269951422624,
      "grad_norm": 0.7895171642303467,
      "learning_rate": 3.928008558690255e-05,
      "loss": 0.6509,
      "step": 3096
    },
    {
      "epoch": 0.5373004857737682,
      "grad_norm": 1.4371894598007202,
      "learning_rate": 3.927827291370951e-05,
      "loss": 0.6134,
      "step": 3097
    },
    {
      "epoch": 0.5374739764052742,
      "grad_norm": 0.7677549719810486,
      "learning_rate": 3.927645800325041e-05,
      "loss": 0.5552,
      "step": 3098
    },
    {
      "epoch": 0.53764746703678,
      "grad_norm": 0.8762229681015015,
      "learning_rate": 3.9274640855735914e-05,
      "loss": 0.7253,
      "step": 3099
    },
    {
      "epoch": 0.5378209576682859,
      "grad_norm": 0.6970749497413635,
      "learning_rate": 3.927282147137688e-05,
      "loss": 0.707,
      "step": 3100
    },
    {
      "epoch": 0.5379944482997918,
      "grad_norm": 0.9581252932548523,
      "learning_rate": 3.927099985038446e-05,
      "loss": 0.4913,
      "step": 3101
    },
    {
      "epoch": 0.5381679389312977,
      "grad_norm": 0.9242482781410217,
      "learning_rate": 3.9269175992970055e-05,
      "loss": 0.5629,
      "step": 3102
    },
    {
      "epoch": 0.5383414295628036,
      "grad_norm": 1.2358778715133667,
      "learning_rate": 3.926734989934532e-05,
      "loss": 0.5692,
      "step": 3103
    },
    {
      "epoch": 0.5385149201943095,
      "grad_norm": 0.9537628889083862,
      "learning_rate": 3.9265521569722176e-05,
      "loss": 0.626,
      "step": 3104
    },
    {
      "epoch": 0.5386884108258154,
      "grad_norm": 0.7534976005554199,
      "learning_rate": 3.9263691004312804e-05,
      "loss": 0.7104,
      "step": 3105
    },
    {
      "epoch": 0.5388619014573213,
      "grad_norm": 0.6508181095123291,
      "learning_rate": 3.926185820332965e-05,
      "loss": 0.5377,
      "step": 3106
    },
    {
      "epoch": 0.5390353920888272,
      "grad_norm": 0.6256936192512512,
      "learning_rate": 3.9260023166985407e-05,
      "loss": 0.5938,
      "step": 3107
    },
    {
      "epoch": 0.5392088827203331,
      "grad_norm": 0.8080528378486633,
      "learning_rate": 3.9258185895493026e-05,
      "loss": 0.6345,
      "step": 3108
    },
    {
      "epoch": 0.539382373351839,
      "grad_norm": 0.6470572352409363,
      "learning_rate": 3.925634638906574e-05,
      "loss": 0.592,
      "step": 3109
    },
    {
      "epoch": 0.5395558639833449,
      "grad_norm": 0.9834403395652771,
      "learning_rate": 3.925450464791701e-05,
      "loss": 0.5455,
      "step": 3110
    },
    {
      "epoch": 0.5397293546148508,
      "grad_norm": 0.9988872408866882,
      "learning_rate": 3.925266067226058e-05,
      "loss": 0.5189,
      "step": 3111
    },
    {
      "epoch": 0.5399028452463567,
      "grad_norm": 0.608731210231781,
      "learning_rate": 3.925081446231045e-05,
      "loss": 0.5674,
      "step": 3112
    },
    {
      "epoch": 0.5400763358778626,
      "grad_norm": 0.7310196757316589,
      "learning_rate": 3.924896601828087e-05,
      "loss": 0.5524,
      "step": 3113
    },
    {
      "epoch": 0.5402498265093685,
      "grad_norm": 0.6233857274055481,
      "learning_rate": 3.924711534038635e-05,
      "loss": 0.5348,
      "step": 3114
    },
    {
      "epoch": 0.5404233171408744,
      "grad_norm": 0.6901100873947144,
      "learning_rate": 3.924526242884167e-05,
      "loss": 0.736,
      "step": 3115
    },
    {
      "epoch": 0.5405968077723803,
      "grad_norm": 0.9781090617179871,
      "learning_rate": 3.9243407283861866e-05,
      "loss": 0.6023,
      "step": 3116
    },
    {
      "epoch": 0.5407702984038862,
      "grad_norm": 0.9208560585975647,
      "learning_rate": 3.924154990566222e-05,
      "loss": 0.6281,
      "step": 3117
    },
    {
      "epoch": 0.5409437890353921,
      "grad_norm": 0.7432438731193542,
      "learning_rate": 3.923969029445828e-05,
      "loss": 0.7271,
      "step": 3118
    },
    {
      "epoch": 0.5411172796668979,
      "grad_norm": 0.6769471168518066,
      "learning_rate": 3.9237828450465866e-05,
      "loss": 0.6505,
      "step": 3119
    },
    {
      "epoch": 0.5412907702984039,
      "grad_norm": 0.7981407642364502,
      "learning_rate": 3.923596437390105e-05,
      "loss": 0.5497,
      "step": 3120
    },
    {
      "epoch": 0.5414642609299097,
      "grad_norm": 1.166545033454895,
      "learning_rate": 3.9234098064980145e-05,
      "loss": 0.6946,
      "step": 3121
    },
    {
      "epoch": 0.5416377515614157,
      "grad_norm": 0.9433131814002991,
      "learning_rate": 3.923222952391975e-05,
      "loss": 0.6458,
      "step": 3122
    },
    {
      "epoch": 0.5418112421929215,
      "grad_norm": 1.5492539405822754,
      "learning_rate": 3.923035875093671e-05,
      "loss": 0.6902,
      "step": 3123
    },
    {
      "epoch": 0.5419847328244275,
      "grad_norm": 0.9652863144874573,
      "learning_rate": 3.9228485746248134e-05,
      "loss": 0.5554,
      "step": 3124
    },
    {
      "epoch": 0.5421582234559333,
      "grad_norm": 0.9015517830848694,
      "learning_rate": 3.922661051007137e-05,
      "loss": 0.5894,
      "step": 3125
    },
    {
      "epoch": 0.5423317140874393,
      "grad_norm": 1.0942059755325317,
      "learning_rate": 3.922473304262406e-05,
      "loss": 0.5625,
      "step": 3126
    },
    {
      "epoch": 0.5425052047189451,
      "grad_norm": 0.9288303256034851,
      "learning_rate": 3.922285334412408e-05,
      "loss": 0.5653,
      "step": 3127
    },
    {
      "epoch": 0.5426786953504511,
      "grad_norm": 0.7825427055358887,
      "learning_rate": 3.922097141478957e-05,
      "loss": 0.691,
      "step": 3128
    },
    {
      "epoch": 0.542852185981957,
      "grad_norm": 0.8568305969238281,
      "learning_rate": 3.9219087254838925e-05,
      "loss": 0.6082,
      "step": 3129
    },
    {
      "epoch": 0.5430256766134629,
      "grad_norm": 1.0168123245239258,
      "learning_rate": 3.921720086449082e-05,
      "loss": 0.6301,
      "step": 3130
    },
    {
      "epoch": 0.5431991672449688,
      "grad_norm": 1.1532608270645142,
      "learning_rate": 3.921531224396415e-05,
      "loss": 0.5955,
      "step": 3131
    },
    {
      "epoch": 0.5433726578764747,
      "grad_norm": 0.7430927753448486,
      "learning_rate": 3.921342139347811e-05,
      "loss": 0.5428,
      "step": 3132
    },
    {
      "epoch": 0.5435461485079806,
      "grad_norm": 1.0280706882476807,
      "learning_rate": 3.921152831325213e-05,
      "loss": 0.5558,
      "step": 3133
    },
    {
      "epoch": 0.5437196391394865,
      "grad_norm": 0.6628148555755615,
      "learning_rate": 3.92096330035059e-05,
      "loss": 0.6188,
      "step": 3134
    },
    {
      "epoch": 0.5438931297709924,
      "grad_norm": 1.6297130584716797,
      "learning_rate": 3.920773546445938e-05,
      "loss": 0.6769,
      "step": 3135
    },
    {
      "epoch": 0.5440666204024983,
      "grad_norm": 0.6971448659896851,
      "learning_rate": 3.9205835696332775e-05,
      "loss": 0.585,
      "step": 3136
    },
    {
      "epoch": 0.5442401110340042,
      "grad_norm": 0.9060026407241821,
      "learning_rate": 3.9203933699346555e-05,
      "loss": 0.5519,
      "step": 3137
    },
    {
      "epoch": 0.5444136016655101,
      "grad_norm": 0.7070338129997253,
      "learning_rate": 3.920202947372146e-05,
      "loss": 0.5883,
      "step": 3138
    },
    {
      "epoch": 0.544587092297016,
      "grad_norm": 0.9702456593513489,
      "learning_rate": 3.9200123019678467e-05,
      "loss": 0.5353,
      "step": 3139
    },
    {
      "epoch": 0.5447605829285218,
      "grad_norm": 1.4283995628356934,
      "learning_rate": 3.919821433743882e-05,
      "loss": 0.7267,
      "step": 3140
    },
    {
      "epoch": 0.5449340735600278,
      "grad_norm": 0.7251282930374146,
      "learning_rate": 3.9196303427224036e-05,
      "loss": 0.6918,
      "step": 3141
    },
    {
      "epoch": 0.5451075641915336,
      "grad_norm": 0.8550883531570435,
      "learning_rate": 3.919439028925587e-05,
      "loss": 0.6157,
      "step": 3142
    },
    {
      "epoch": 0.5452810548230396,
      "grad_norm": 1.5072968006134033,
      "learning_rate": 3.919247492375634e-05,
      "loss": 0.6156,
      "step": 3143
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.7578117251396179,
      "learning_rate": 3.919055733094774e-05,
      "loss": 0.5886,
      "step": 3144
    },
    {
      "epoch": 0.5456280360860514,
      "grad_norm": 0.8776355981826782,
      "learning_rate": 3.918863751105259e-05,
      "loss": 0.5455,
      "step": 3145
    },
    {
      "epoch": 0.5458015267175572,
      "grad_norm": 0.7595939636230469,
      "learning_rate": 3.91867154642937e-05,
      "loss": 0.6096,
      "step": 3146
    },
    {
      "epoch": 0.5459750173490632,
      "grad_norm": 1.0227677822113037,
      "learning_rate": 3.918479119089413e-05,
      "loss": 0.5476,
      "step": 3147
    },
    {
      "epoch": 0.546148507980569,
      "grad_norm": 0.7157589197158813,
      "learning_rate": 3.918286469107718e-05,
      "loss": 0.6128,
      "step": 3148
    },
    {
      "epoch": 0.546321998612075,
      "grad_norm": 0.995482861995697,
      "learning_rate": 3.918093596506643e-05,
      "loss": 0.5902,
      "step": 3149
    },
    {
      "epoch": 0.5464954892435808,
      "grad_norm": 0.7816616892814636,
      "learning_rate": 3.917900501308572e-05,
      "loss": 0.6029,
      "step": 3150
    },
    {
      "epoch": 0.5466689798750868,
      "grad_norm": 1.0965698957443237,
      "learning_rate": 3.917707183535913e-05,
      "loss": 0.6873,
      "step": 3151
    },
    {
      "epoch": 0.5468424705065926,
      "grad_norm": 0.5522381663322449,
      "learning_rate": 3.9175136432111e-05,
      "loss": 0.6611,
      "step": 3152
    },
    {
      "epoch": 0.5470159611380986,
      "grad_norm": 1.5331714153289795,
      "learning_rate": 3.917319880356594e-05,
      "loss": 0.6505,
      "step": 3153
    },
    {
      "epoch": 0.5471894517696044,
      "grad_norm": 1.0777934789657593,
      "learning_rate": 3.9171258949948827e-05,
      "loss": 0.5635,
      "step": 3154
    },
    {
      "epoch": 0.5473629424011104,
      "grad_norm": 0.8994247913360596,
      "learning_rate": 3.916931687148477e-05,
      "loss": 0.5776,
      "step": 3155
    },
    {
      "epoch": 0.5475364330326162,
      "grad_norm": 1.6511632204055786,
      "learning_rate": 3.916737256839916e-05,
      "loss": 0.5968,
      "step": 3156
    },
    {
      "epoch": 0.5477099236641222,
      "grad_norm": 0.8579990267753601,
      "learning_rate": 3.916542604091762e-05,
      "loss": 0.5574,
      "step": 3157
    },
    {
      "epoch": 0.547883414295628,
      "grad_norm": 1.2080751657485962,
      "learning_rate": 3.916347728926606e-05,
      "loss": 0.7275,
      "step": 3158
    },
    {
      "epoch": 0.5480569049271339,
      "grad_norm": 0.7340068221092224,
      "learning_rate": 3.916152631367063e-05,
      "loss": 0.5321,
      "step": 3159
    },
    {
      "epoch": 0.5482303955586398,
      "grad_norm": 0.6173186302185059,
      "learning_rate": 3.915957311435774e-05,
      "loss": 0.5669,
      "step": 3160
    },
    {
      "epoch": 0.5484038861901457,
      "grad_norm": 0.7848614454269409,
      "learning_rate": 3.915761769155407e-05,
      "loss": 0.5255,
      "step": 3161
    },
    {
      "epoch": 0.5485773768216516,
      "grad_norm": 0.5917115807533264,
      "learning_rate": 3.915566004548654e-05,
      "loss": 0.5249,
      "step": 3162
    },
    {
      "epoch": 0.5487508674531575,
      "grad_norm": 1.5309841632843018,
      "learning_rate": 3.9153700176382344e-05,
      "loss": 0.6642,
      "step": 3163
    },
    {
      "epoch": 0.5489243580846634,
      "grad_norm": 0.7895891666412354,
      "learning_rate": 3.915173808446892e-05,
      "loss": 0.6454,
      "step": 3164
    },
    {
      "epoch": 0.5490978487161693,
      "grad_norm": 0.8298311233520508,
      "learning_rate": 3.9149773769973985e-05,
      "loss": 0.4957,
      "step": 3165
    },
    {
      "epoch": 0.5492713393476752,
      "grad_norm": 1.134933590888977,
      "learning_rate": 3.914780723312548e-05,
      "loss": 0.5394,
      "step": 3166
    },
    {
      "epoch": 0.5494448299791811,
      "grad_norm": 1.13173246383667,
      "learning_rate": 3.9145838474151633e-05,
      "loss": 0.5302,
      "step": 3167
    },
    {
      "epoch": 0.549618320610687,
      "grad_norm": 1.4297860860824585,
      "learning_rate": 3.914386749328093e-05,
      "loss": 0.6407,
      "step": 3168
    },
    {
      "epoch": 0.5497918112421929,
      "grad_norm": 0.7096014618873596,
      "learning_rate": 3.914189429074209e-05,
      "loss": 0.6606,
      "step": 3169
    },
    {
      "epoch": 0.5499653018736989,
      "grad_norm": 1.2400507926940918,
      "learning_rate": 3.913991886676412e-05,
      "loss": 0.6764,
      "step": 3170
    },
    {
      "epoch": 0.5501387925052047,
      "grad_norm": 0.7265576720237732,
      "learning_rate": 3.913794122157626e-05,
      "loss": 0.6105,
      "step": 3171
    },
    {
      "epoch": 0.5503122831367107,
      "grad_norm": 0.8316229581832886,
      "learning_rate": 3.9135961355408024e-05,
      "loss": 0.564,
      "step": 3172
    },
    {
      "epoch": 0.5504857737682165,
      "grad_norm": 1.0690675973892212,
      "learning_rate": 3.913397926848917e-05,
      "loss": 0.5505,
      "step": 3173
    },
    {
      "epoch": 0.5506592643997225,
      "grad_norm": 0.6928693056106567,
      "learning_rate": 3.913199496104972e-05,
      "loss": 0.6057,
      "step": 3174
    },
    {
      "epoch": 0.5508327550312283,
      "grad_norm": 0.5716657638549805,
      "learning_rate": 3.9130008433319974e-05,
      "loss": 0.7322,
      "step": 3175
    },
    {
      "epoch": 0.5510062456627343,
      "grad_norm": 0.6816802024841309,
      "learning_rate": 3.912801968553045e-05,
      "loss": 0.6383,
      "step": 3176
    },
    {
      "epoch": 0.5511797362942401,
      "grad_norm": 0.6757462024688721,
      "learning_rate": 3.912602871791196e-05,
      "loss": 0.6283,
      "step": 3177
    },
    {
      "epoch": 0.5513532269257461,
      "grad_norm": 0.7462841272354126,
      "learning_rate": 3.9124035530695546e-05,
      "loss": 0.6243,
      "step": 3178
    },
    {
      "epoch": 0.5515267175572519,
      "grad_norm": 0.875676155090332,
      "learning_rate": 3.912204012411253e-05,
      "loss": 0.5619,
      "step": 3179
    },
    {
      "epoch": 0.5517002081887578,
      "grad_norm": 0.673236072063446,
      "learning_rate": 3.912004249839447e-05,
      "loss": 0.5929,
      "step": 3180
    },
    {
      "epoch": 0.5518736988202637,
      "grad_norm": 0.9303373694419861,
      "learning_rate": 3.9118042653773194e-05,
      "loss": 0.6926,
      "step": 3181
    },
    {
      "epoch": 0.5520471894517696,
      "grad_norm": 0.684323787689209,
      "learning_rate": 3.91160405904808e-05,
      "loss": 0.6083,
      "step": 3182
    },
    {
      "epoch": 0.5522206800832755,
      "grad_norm": 0.7034314274787903,
      "learning_rate": 3.9114036308749625e-05,
      "loss": 0.5876,
      "step": 3183
    },
    {
      "epoch": 0.5523941707147814,
      "grad_norm": 2.226392984390259,
      "learning_rate": 3.911202980881226e-05,
      "loss": 0.5818,
      "step": 3184
    },
    {
      "epoch": 0.5525676613462873,
      "grad_norm": 0.9508465528488159,
      "learning_rate": 3.911002109090156e-05,
      "loss": 0.6475,
      "step": 3185
    },
    {
      "epoch": 0.5527411519777932,
      "grad_norm": 0.8926861882209778,
      "learning_rate": 3.910801015525064e-05,
      "loss": 0.613,
      "step": 3186
    },
    {
      "epoch": 0.5529146426092991,
      "grad_norm": 0.8146591782569885,
      "learning_rate": 3.9105997002092896e-05,
      "loss": 0.6594,
      "step": 3187
    },
    {
      "epoch": 0.553088133240805,
      "grad_norm": 1.1827658414840698,
      "learning_rate": 3.910398163166192e-05,
      "loss": 0.6274,
      "step": 3188
    },
    {
      "epoch": 0.5532616238723109,
      "grad_norm": 1.186378836631775,
      "learning_rate": 3.910196404419163e-05,
      "loss": 0.5598,
      "step": 3189
    },
    {
      "epoch": 0.5534351145038168,
      "grad_norm": 1.9686994552612305,
      "learning_rate": 3.909994423991614e-05,
      "loss": 0.5443,
      "step": 3190
    },
    {
      "epoch": 0.5536086051353227,
      "grad_norm": 0.8391484618186951,
      "learning_rate": 3.909792221906987e-05,
      "loss": 0.5592,
      "step": 3191
    },
    {
      "epoch": 0.5537820957668286,
      "grad_norm": 0.9526604413986206,
      "learning_rate": 3.909589798188747e-05,
      "loss": 0.614,
      "step": 3192
    },
    {
      "epoch": 0.5539555863983345,
      "grad_norm": 1.4479925632476807,
      "learning_rate": 3.909387152860386e-05,
      "loss": 0.5209,
      "step": 3193
    },
    {
      "epoch": 0.5541290770298404,
      "grad_norm": 2.3782713413238525,
      "learning_rate": 3.909184285945421e-05,
      "loss": 0.5903,
      "step": 3194
    },
    {
      "epoch": 0.5543025676613463,
      "grad_norm": 1.63461434841156,
      "learning_rate": 3.908981197467396e-05,
      "loss": 0.5917,
      "step": 3195
    },
    {
      "epoch": 0.5544760582928522,
      "grad_norm": 0.9873148202896118,
      "learning_rate": 3.908777887449877e-05,
      "loss": 0.7415,
      "step": 3196
    },
    {
      "epoch": 0.5546495489243581,
      "grad_norm": 0.669289231300354,
      "learning_rate": 3.908574355916461e-05,
      "loss": 0.6958,
      "step": 3197
    },
    {
      "epoch": 0.554823039555864,
      "grad_norm": 0.6251704096794128,
      "learning_rate": 3.9083706028907665e-05,
      "loss": 0.6174,
      "step": 3198
    },
    {
      "epoch": 0.5549965301873698,
      "grad_norm": 0.7469822764396667,
      "learning_rate": 3.90816662839644e-05,
      "loss": 0.6473,
      "step": 3199
    },
    {
      "epoch": 0.5551700208188758,
      "grad_norm": 0.8963797092437744,
      "learning_rate": 3.9079624324571536e-05,
      "loss": 0.5769,
      "step": 3200
    },
    {
      "epoch": 0.5553435114503816,
      "grad_norm": 0.5527542233467102,
      "learning_rate": 3.907758015096603e-05,
      "loss": 0.6652,
      "step": 3201
    },
    {
      "epoch": 0.5555170020818876,
      "grad_norm": 0.6194865107536316,
      "learning_rate": 3.9075533763385116e-05,
      "loss": 0.7543,
      "step": 3202
    },
    {
      "epoch": 0.5556904927133934,
      "grad_norm": 2.4148449897766113,
      "learning_rate": 3.907348516206629e-05,
      "loss": 0.6011,
      "step": 3203
    },
    {
      "epoch": 0.5558639833448994,
      "grad_norm": 0.6149037480354309,
      "learning_rate": 3.9071434347247275e-05,
      "loss": 0.6194,
      "step": 3204
    },
    {
      "epoch": 0.5560374739764052,
      "grad_norm": 0.7264496684074402,
      "learning_rate": 3.906938131916609e-05,
      "loss": 0.5947,
      "step": 3205
    },
    {
      "epoch": 0.5562109646079112,
      "grad_norm": 1.004270076751709,
      "learning_rate": 3.906732607806098e-05,
      "loss": 0.5686,
      "step": 3206
    },
    {
      "epoch": 0.556384455239417,
      "grad_norm": 0.5807811617851257,
      "learning_rate": 3.906526862417046e-05,
      "loss": 0.6287,
      "step": 3207
    },
    {
      "epoch": 0.556557945870923,
      "grad_norm": 0.8036131262779236,
      "learning_rate": 3.906320895773329e-05,
      "loss": 0.6539,
      "step": 3208
    },
    {
      "epoch": 0.5567314365024288,
      "grad_norm": 0.7775804996490479,
      "learning_rate": 3.9061147078988526e-05,
      "loss": 0.6222,
      "step": 3209
    },
    {
      "epoch": 0.5569049271339348,
      "grad_norm": 0.962177038192749,
      "learning_rate": 3.905908298817543e-05,
      "loss": 0.6736,
      "step": 3210
    },
    {
      "epoch": 0.5570784177654406,
      "grad_norm": 1.0258029699325562,
      "learning_rate": 3.905701668553353e-05,
      "loss": 0.6505,
      "step": 3211
    },
    {
      "epoch": 0.5572519083969466,
      "grad_norm": 0.8902332186698914,
      "learning_rate": 3.905494817130265e-05,
      "loss": 0.6418,
      "step": 3212
    },
    {
      "epoch": 0.5574253990284525,
      "grad_norm": 0.803525984287262,
      "learning_rate": 3.905287744572283e-05,
      "loss": 0.5302,
      "step": 3213
    },
    {
      "epoch": 0.5575988896599584,
      "grad_norm": 0.9576848745346069,
      "learning_rate": 3.9050804509034383e-05,
      "loss": 0.5084,
      "step": 3214
    },
    {
      "epoch": 0.5577723802914643,
      "grad_norm": 2.1501924991607666,
      "learning_rate": 3.904872936147787e-05,
      "loss": 0.5497,
      "step": 3215
    },
    {
      "epoch": 0.5579458709229702,
      "grad_norm": 1.1034975051879883,
      "learning_rate": 3.904665200329411e-05,
      "loss": 0.5637,
      "step": 3216
    },
    {
      "epoch": 0.5581193615544761,
      "grad_norm": 0.7428118586540222,
      "learning_rate": 3.904457243472421e-05,
      "loss": 0.613,
      "step": 3217
    },
    {
      "epoch": 0.5582928521859819,
      "grad_norm": 1.164508581161499,
      "learning_rate": 3.904249065600948e-05,
      "loss": 0.6355,
      "step": 3218
    },
    {
      "epoch": 0.5584663428174879,
      "grad_norm": 0.8807821869850159,
      "learning_rate": 3.904040666739151e-05,
      "loss": 0.6489,
      "step": 3219
    },
    {
      "epoch": 0.5586398334489937,
      "grad_norm": 0.7938810586929321,
      "learning_rate": 3.903832046911218e-05,
      "loss": 0.6605,
      "step": 3220
    },
    {
      "epoch": 0.5588133240804997,
      "grad_norm": 0.7629036903381348,
      "learning_rate": 3.903623206141356e-05,
      "loss": 0.5359,
      "step": 3221
    },
    {
      "epoch": 0.5589868147120055,
      "grad_norm": 0.7561957836151123,
      "learning_rate": 3.9034141444538034e-05,
      "loss": 0.6113,
      "step": 3222
    },
    {
      "epoch": 0.5591603053435115,
      "grad_norm": 1.1009613275527954,
      "learning_rate": 3.903204861872821e-05,
      "loss": 0.6989,
      "step": 3223
    },
    {
      "epoch": 0.5593337959750173,
      "grad_norm": 0.6956751942634583,
      "learning_rate": 3.902995358422697e-05,
      "loss": 0.5776,
      "step": 3224
    },
    {
      "epoch": 0.5595072866065233,
      "grad_norm": 0.6899406909942627,
      "learning_rate": 3.902785634127744e-05,
      "loss": 0.7062,
      "step": 3225
    },
    {
      "epoch": 0.5596807772380291,
      "grad_norm": 0.9534558057785034,
      "learning_rate": 3.902575689012301e-05,
      "loss": 0.6958,
      "step": 3226
    },
    {
      "epoch": 0.5598542678695351,
      "grad_norm": 1.1844556331634521,
      "learning_rate": 3.9023655231007325e-05,
      "loss": 0.5811,
      "step": 3227
    },
    {
      "epoch": 0.5600277585010409,
      "grad_norm": 0.6327782869338989,
      "learning_rate": 3.9021551364174286e-05,
      "loss": 0.5557,
      "step": 3228
    },
    {
      "epoch": 0.5602012491325469,
      "grad_norm": 0.8286023139953613,
      "learning_rate": 3.901944528986804e-05,
      "loss": 0.6006,
      "step": 3229
    },
    {
      "epoch": 0.5603747397640527,
      "grad_norm": 0.9115365743637085,
      "learning_rate": 3.901733700833301e-05,
      "loss": 0.6746,
      "step": 3230
    },
    {
      "epoch": 0.5605482303955587,
      "grad_norm": 0.7401015162467957,
      "learning_rate": 3.9015226519813864e-05,
      "loss": 0.6975,
      "step": 3231
    },
    {
      "epoch": 0.5607217210270645,
      "grad_norm": 1.462897539138794,
      "learning_rate": 3.9013113824555515e-05,
      "loss": 0.7756,
      "step": 3232
    },
    {
      "epoch": 0.5608952116585705,
      "grad_norm": 1.277316689491272,
      "learning_rate": 3.901099892280316e-05,
      "loss": 0.6306,
      "step": 3233
    },
    {
      "epoch": 0.5610687022900763,
      "grad_norm": 0.779032826423645,
      "learning_rate": 3.9008881814802225e-05,
      "loss": 0.6066,
      "step": 3234
    },
    {
      "epoch": 0.5612421929215823,
      "grad_norm": 0.7906771898269653,
      "learning_rate": 3.900676250079841e-05,
      "loss": 0.6125,
      "step": 3235
    },
    {
      "epoch": 0.5614156835530881,
      "grad_norm": 0.7157325148582458,
      "learning_rate": 3.900464098103765e-05,
      "loss": 0.6409,
      "step": 3236
    },
    {
      "epoch": 0.5615891741845941,
      "grad_norm": 0.7426790595054626,
      "learning_rate": 3.9002517255766163e-05,
      "loss": 0.5331,
      "step": 3237
    },
    {
      "epoch": 0.5617626648160999,
      "grad_norm": 0.8521499633789062,
      "learning_rate": 3.9000391325230405e-05,
      "loss": 0.5262,
      "step": 3238
    },
    {
      "epoch": 0.5619361554476058,
      "grad_norm": 0.8248336911201477,
      "learning_rate": 3.89982631896771e-05,
      "loss": 0.6337,
      "step": 3239
    },
    {
      "epoch": 0.5621096460791117,
      "grad_norm": 0.874073326587677,
      "learning_rate": 3.899613284935321e-05,
      "loss": 0.5125,
      "step": 3240
    },
    {
      "epoch": 0.5622831367106176,
      "grad_norm": 0.6740594506263733,
      "learning_rate": 3.899400030450597e-05,
      "loss": 0.5568,
      "step": 3241
    },
    {
      "epoch": 0.5624566273421235,
      "grad_norm": 0.867178201675415,
      "learning_rate": 3.899186555538286e-05,
      "loss": 0.6821,
      "step": 3242
    },
    {
      "epoch": 0.5626301179736294,
      "grad_norm": 0.7415544986724854,
      "learning_rate": 3.8989728602231623e-05,
      "loss": 0.6562,
      "step": 3243
    },
    {
      "epoch": 0.5628036086051353,
      "grad_norm": 1.0366650819778442,
      "learning_rate": 3.898758944530025e-05,
      "loss": 0.616,
      "step": 3244
    },
    {
      "epoch": 0.5629770992366412,
      "grad_norm": 0.976502001285553,
      "learning_rate": 3.8985448084837e-05,
      "loss": 0.5437,
      "step": 3245
    },
    {
      "epoch": 0.5631505898681471,
      "grad_norm": 0.8038862943649292,
      "learning_rate": 3.898330452109038e-05,
      "loss": 0.5486,
      "step": 3246
    },
    {
      "epoch": 0.563324080499653,
      "grad_norm": 0.7974483370780945,
      "learning_rate": 3.8981158754309156e-05,
      "loss": 0.5627,
      "step": 3247
    },
    {
      "epoch": 0.563497571131159,
      "grad_norm": 0.7054277062416077,
      "learning_rate": 3.897901078474233e-05,
      "loss": 0.6125,
      "step": 3248
    },
    {
      "epoch": 0.5636710617626648,
      "grad_norm": 0.6825677752494812,
      "learning_rate": 3.897686061263919e-05,
      "loss": 0.5743,
      "step": 3249
    },
    {
      "epoch": 0.5638445523941708,
      "grad_norm": 1.8333244323730469,
      "learning_rate": 3.897470823824927e-05,
      "loss": 0.551,
      "step": 3250
    },
    {
      "epoch": 0.5640180430256766,
      "grad_norm": 0.9035657644271851,
      "learning_rate": 3.8972553661822334e-05,
      "loss": 0.6897,
      "step": 3251
    },
    {
      "epoch": 0.5641915336571826,
      "grad_norm": 1.166638731956482,
      "learning_rate": 3.897039688360845e-05,
      "loss": 0.6208,
      "step": 3252
    },
    {
      "epoch": 0.5643650242886884,
      "grad_norm": 1.1381759643554688,
      "learning_rate": 3.8968237903857906e-05,
      "loss": 0.5958,
      "step": 3253
    },
    {
      "epoch": 0.5645385149201944,
      "grad_norm": 0.7991777658462524,
      "learning_rate": 3.8966076722821245e-05,
      "loss": 0.6111,
      "step": 3254
    },
    {
      "epoch": 0.5647120055517002,
      "grad_norm": 0.9115409851074219,
      "learning_rate": 3.896391334074928e-05,
      "loss": 0.5956,
      "step": 3255
    },
    {
      "epoch": 0.5648854961832062,
      "grad_norm": 1.2931652069091797,
      "learning_rate": 3.8961747757893075e-05,
      "loss": 0.5609,
      "step": 3256
    },
    {
      "epoch": 0.565058986814712,
      "grad_norm": 1.0862016677856445,
      "learning_rate": 3.8959579974503947e-05,
      "loss": 0.5239,
      "step": 3257
    },
    {
      "epoch": 0.5652324774462179,
      "grad_norm": 0.8347268104553223,
      "learning_rate": 3.895740999083347e-05,
      "loss": 0.5426,
      "step": 3258
    },
    {
      "epoch": 0.5654059680777238,
      "grad_norm": 1.2553099393844604,
      "learning_rate": 3.8955237807133485e-05,
      "loss": 0.5894,
      "step": 3259
    },
    {
      "epoch": 0.5655794587092297,
      "grad_norm": 0.8644529581069946,
      "learning_rate": 3.8953063423656055e-05,
      "loss": 0.615,
      "step": 3260
    },
    {
      "epoch": 0.5657529493407356,
      "grad_norm": 0.7972590923309326,
      "learning_rate": 3.8950886840653524e-05,
      "loss": 0.5249,
      "step": 3261
    },
    {
      "epoch": 0.5659264399722415,
      "grad_norm": 0.6203532814979553,
      "learning_rate": 3.8948708058378504e-05,
      "loss": 0.7021,
      "step": 3262
    },
    {
      "epoch": 0.5660999306037474,
      "grad_norm": 0.6847710013389587,
      "learning_rate": 3.894652707708383e-05,
      "loss": 0.6949,
      "step": 3263
    },
    {
      "epoch": 0.5662734212352533,
      "grad_norm": 1.1506133079528809,
      "learning_rate": 3.894434389702261e-05,
      "loss": 0.5867,
      "step": 3264
    },
    {
      "epoch": 0.5664469118667592,
      "grad_norm": 1.1352781057357788,
      "learning_rate": 3.894215851844821e-05,
      "loss": 0.5782,
      "step": 3265
    },
    {
      "epoch": 0.5666204024982651,
      "grad_norm": 0.7782225012779236,
      "learning_rate": 3.8939970941614247e-05,
      "loss": 0.5499,
      "step": 3266
    },
    {
      "epoch": 0.566793893129771,
      "grad_norm": 2.2187085151672363,
      "learning_rate": 3.893778116677457e-05,
      "loss": 0.5884,
      "step": 3267
    },
    {
      "epoch": 0.5669673837612769,
      "grad_norm": 1.034565806388855,
      "learning_rate": 3.893558919418334e-05,
      "loss": 0.5935,
      "step": 3268
    },
    {
      "epoch": 0.5671408743927828,
      "grad_norm": 0.6328162550926208,
      "learning_rate": 3.8933395024094904e-05,
      "loss": 0.6771,
      "step": 3269
    },
    {
      "epoch": 0.5673143650242887,
      "grad_norm": 1.196004033088684,
      "learning_rate": 3.893119865676393e-05,
      "loss": 0.5558,
      "step": 3270
    },
    {
      "epoch": 0.5674878556557946,
      "grad_norm": 1.0808980464935303,
      "learning_rate": 3.892900009244528e-05,
      "loss": 0.6027,
      "step": 3271
    },
    {
      "epoch": 0.5676613462873005,
      "grad_norm": 0.7612327933311462,
      "learning_rate": 3.892679933139412e-05,
      "loss": 0.5385,
      "step": 3272
    },
    {
      "epoch": 0.5678348369188064,
      "grad_norm": 0.8031080365180969,
      "learning_rate": 3.8924596373865834e-05,
      "loss": 0.6525,
      "step": 3273
    },
    {
      "epoch": 0.5680083275503123,
      "grad_norm": 0.6683812737464905,
      "learning_rate": 3.8922391220116094e-05,
      "loss": 0.5999,
      "step": 3274
    },
    {
      "epoch": 0.5681818181818182,
      "grad_norm": 0.7505687475204468,
      "learning_rate": 3.89201838704008e-05,
      "loss": 0.6857,
      "step": 3275
    },
    {
      "epoch": 0.5683553088133241,
      "grad_norm": 0.8488569855690002,
      "learning_rate": 3.891797432497613e-05,
      "loss": 0.4913,
      "step": 3276
    },
    {
      "epoch": 0.5685287994448299,
      "grad_norm": 0.9531639814376831,
      "learning_rate": 3.8915762584098484e-05,
      "loss": 0.4996,
      "step": 3277
    },
    {
      "epoch": 0.5687022900763359,
      "grad_norm": 1.6543718576431274,
      "learning_rate": 3.891354864802455e-05,
      "loss": 0.6038,
      "step": 3278
    },
    {
      "epoch": 0.5688757807078417,
      "grad_norm": 0.8249922394752502,
      "learning_rate": 3.891133251701127e-05,
      "loss": 0.6492,
      "step": 3279
    },
    {
      "epoch": 0.5690492713393477,
      "grad_norm": 0.7629801630973816,
      "learning_rate": 3.89091141913158e-05,
      "loss": 0.5842,
      "step": 3280
    },
    {
      "epoch": 0.5692227619708535,
      "grad_norm": 1.0736356973648071,
      "learning_rate": 3.890689367119561e-05,
      "loss": 0.5034,
      "step": 3281
    },
    {
      "epoch": 0.5693962526023595,
      "grad_norm": 0.8037618398666382,
      "learning_rate": 3.890467095690837e-05,
      "loss": 0.6332,
      "step": 3282
    },
    {
      "epoch": 0.5695697432338653,
      "grad_norm": 1.05054771900177,
      "learning_rate": 3.890244604871203e-05,
      "loss": 0.6654,
      "step": 3283
    },
    {
      "epoch": 0.5697432338653713,
      "grad_norm": 0.8441582322120667,
      "learning_rate": 3.890021894686481e-05,
      "loss": 0.651,
      "step": 3284
    },
    {
      "epoch": 0.5699167244968771,
      "grad_norm": 0.895268440246582,
      "learning_rate": 3.889798965162516e-05,
      "loss": 0.7266,
      "step": 3285
    },
    {
      "epoch": 0.5700902151283831,
      "grad_norm": 0.8688643574714661,
      "learning_rate": 3.8895758163251783e-05,
      "loss": 0.652,
      "step": 3286
    },
    {
      "epoch": 0.5702637057598889,
      "grad_norm": 0.9622326493263245,
      "learning_rate": 3.889352448200366e-05,
      "loss": 0.5823,
      "step": 3287
    },
    {
      "epoch": 0.5704371963913949,
      "grad_norm": 0.8369941115379333,
      "learning_rate": 3.889128860814e-05,
      "loss": 0.5424,
      "step": 3288
    },
    {
      "epoch": 0.5706106870229007,
      "grad_norm": 1.4068269729614258,
      "learning_rate": 3.8889050541920285e-05,
      "loss": 0.5499,
      "step": 3289
    },
    {
      "epoch": 0.5707841776544067,
      "grad_norm": 0.8182569146156311,
      "learning_rate": 3.8886810283604245e-05,
      "loss": 0.5638,
      "step": 3290
    },
    {
      "epoch": 0.5709576682859125,
      "grad_norm": 1.9418103694915771,
      "learning_rate": 3.888456783345187e-05,
      "loss": 0.5587,
      "step": 3291
    },
    {
      "epoch": 0.5711311589174185,
      "grad_norm": 1.0349828004837036,
      "learning_rate": 3.888232319172338e-05,
      "loss": 0.6218,
      "step": 3292
    },
    {
      "epoch": 0.5713046495489243,
      "grad_norm": 0.7936072945594788,
      "learning_rate": 3.8880076358679295e-05,
      "loss": 0.6178,
      "step": 3293
    },
    {
      "epoch": 0.5714781401804303,
      "grad_norm": 0.9973400831222534,
      "learning_rate": 3.887782733458034e-05,
      "loss": 0.5144,
      "step": 3294
    },
    {
      "epoch": 0.5716516308119362,
      "grad_norm": 0.9298374056816101,
      "learning_rate": 3.887557611968753e-05,
      "loss": 0.5629,
      "step": 3295
    },
    {
      "epoch": 0.5718251214434421,
      "grad_norm": 0.887874960899353,
      "learning_rate": 3.887332271426211e-05,
      "loss": 0.6729,
      "step": 3296
    },
    {
      "epoch": 0.571998612074948,
      "grad_norm": 1.6880004405975342,
      "learning_rate": 3.887106711856561e-05,
      "loss": 0.5826,
      "step": 3297
    },
    {
      "epoch": 0.5721721027064538,
      "grad_norm": 0.9786666035652161,
      "learning_rate": 3.886880933285977e-05,
      "loss": 0.6461,
      "step": 3298
    },
    {
      "epoch": 0.5723455933379598,
      "grad_norm": 1.2034778594970703,
      "learning_rate": 3.886654935740663e-05,
      "loss": 0.6392,
      "step": 3299
    },
    {
      "epoch": 0.5725190839694656,
      "grad_norm": 0.7050906419754028,
      "learning_rate": 3.886428719246845e-05,
      "loss": 0.6788,
      "step": 3300
    },
    {
      "epoch": 0.5726925746009716,
      "grad_norm": 0.8306242823600769,
      "learning_rate": 3.886202283830776e-05,
      "loss": 0.6774,
      "step": 3301
    },
    {
      "epoch": 0.5728660652324774,
      "grad_norm": 0.8787949085235596,
      "learning_rate": 3.885975629518734e-05,
      "loss": 0.5353,
      "step": 3302
    },
    {
      "epoch": 0.5730395558639834,
      "grad_norm": 1.0400038957595825,
      "learning_rate": 3.885748756337022e-05,
      "loss": 0.5448,
      "step": 3303
    },
    {
      "epoch": 0.5732130464954892,
      "grad_norm": 0.7915439605712891,
      "learning_rate": 3.8855216643119697e-05,
      "loss": 0.657,
      "step": 3304
    },
    {
      "epoch": 0.5733865371269952,
      "grad_norm": 0.8591414093971252,
      "learning_rate": 3.885294353469932e-05,
      "loss": 0.5073,
      "step": 3305
    },
    {
      "epoch": 0.573560027758501,
      "grad_norm": 1.1922606229782104,
      "learning_rate": 3.885066823837287e-05,
      "loss": 0.6675,
      "step": 3306
    },
    {
      "epoch": 0.573733518390007,
      "grad_norm": 0.7343970537185669,
      "learning_rate": 3.884839075440441e-05,
      "loss": 0.5431,
      "step": 3307
    },
    {
      "epoch": 0.5739070090215128,
      "grad_norm": 0.8767739534378052,
      "learning_rate": 3.884611108305824e-05,
      "loss": 0.6619,
      "step": 3308
    },
    {
      "epoch": 0.5740804996530188,
      "grad_norm": 1.115496039390564,
      "learning_rate": 3.884382922459891e-05,
      "loss": 0.6133,
      "step": 3309
    },
    {
      "epoch": 0.5742539902845246,
      "grad_norm": 0.813336193561554,
      "learning_rate": 3.8841545179291254e-05,
      "loss": 0.6106,
      "step": 3310
    },
    {
      "epoch": 0.5744274809160306,
      "grad_norm": 0.7772932052612305,
      "learning_rate": 3.8839258947400325e-05,
      "loss": 0.6854,
      "step": 3311
    },
    {
      "epoch": 0.5746009715475364,
      "grad_norm": 0.823560357093811,
      "learning_rate": 3.883697052919143e-05,
      "loss": 0.5686,
      "step": 3312
    },
    {
      "epoch": 0.5747744621790424,
      "grad_norm": 0.7215649485588074,
      "learning_rate": 3.883467992493017e-05,
      "loss": 0.6415,
      "step": 3313
    },
    {
      "epoch": 0.5749479528105482,
      "grad_norm": 0.746321976184845,
      "learning_rate": 3.883238713488235e-05,
      "loss": 0.5688,
      "step": 3314
    },
    {
      "epoch": 0.5751214434420542,
      "grad_norm": 0.6839783787727356,
      "learning_rate": 3.883009215931406e-05,
      "loss": 0.6318,
      "step": 3315
    },
    {
      "epoch": 0.57529493407356,
      "grad_norm": 0.7651140689849854,
      "learning_rate": 3.882779499849163e-05,
      "loss": 0.6566,
      "step": 3316
    },
    {
      "epoch": 0.5754684247050659,
      "grad_norm": 1.2375156879425049,
      "learning_rate": 3.8825495652681653e-05,
      "loss": 0.6702,
      "step": 3317
    },
    {
      "epoch": 0.5756419153365718,
      "grad_norm": 0.7743232250213623,
      "learning_rate": 3.8823194122150975e-05,
      "loss": 0.5146,
      "step": 3318
    },
    {
      "epoch": 0.5758154059680777,
      "grad_norm": 1.2336572408676147,
      "learning_rate": 3.8820890407166683e-05,
      "loss": 0.7351,
      "step": 3319
    },
    {
      "epoch": 0.5759888965995836,
      "grad_norm": 0.9283602237701416,
      "learning_rate": 3.881858450799612e-05,
      "loss": 0.6333,
      "step": 3320
    },
    {
      "epoch": 0.5761623872310895,
      "grad_norm": 0.7599928379058838,
      "learning_rate": 3.881627642490691e-05,
      "loss": 0.6134,
      "step": 3321
    },
    {
      "epoch": 0.5763358778625954,
      "grad_norm": 1.3574687242507935,
      "learning_rate": 3.8813966158166894e-05,
      "loss": 0.6586,
      "step": 3322
    },
    {
      "epoch": 0.5765093684941013,
      "grad_norm": 1.052731990814209,
      "learning_rate": 3.8811653708044176e-05,
      "loss": 0.5859,
      "step": 3323
    },
    {
      "epoch": 0.5766828591256072,
      "grad_norm": 0.8059406280517578,
      "learning_rate": 3.8809339074807125e-05,
      "loss": 0.4701,
      "step": 3324
    },
    {
      "epoch": 0.5768563497571131,
      "grad_norm": 0.7636902332305908,
      "learning_rate": 3.880702225872437e-05,
      "loss": 0.5542,
      "step": 3325
    },
    {
      "epoch": 0.577029840388619,
      "grad_norm": 0.8611479997634888,
      "learning_rate": 3.8804703260064756e-05,
      "loss": 0.5389,
      "step": 3326
    },
    {
      "epoch": 0.5772033310201249,
      "grad_norm": 0.8257524371147156,
      "learning_rate": 3.880238207909742e-05,
      "loss": 0.4996,
      "step": 3327
    },
    {
      "epoch": 0.5773768216516308,
      "grad_norm": 1.1965727806091309,
      "learning_rate": 3.880005871609173e-05,
      "loss": 0.5571,
      "step": 3328
    },
    {
      "epoch": 0.5775503122831367,
      "grad_norm": 0.7814649343490601,
      "learning_rate": 3.879773317131732e-05,
      "loss": 0.4926,
      "step": 3329
    },
    {
      "epoch": 0.5777238029146426,
      "grad_norm": 0.9013784527778625,
      "learning_rate": 3.879540544504408e-05,
      "loss": 0.5669,
      "step": 3330
    },
    {
      "epoch": 0.5778972935461485,
      "grad_norm": 0.834071695804596,
      "learning_rate": 3.879307553754213e-05,
      "loss": 0.7269,
      "step": 3331
    },
    {
      "epoch": 0.5780707841776545,
      "grad_norm": 0.7171045541763306,
      "learning_rate": 3.879074344908187e-05,
      "loss": 0.7321,
      "step": 3332
    },
    {
      "epoch": 0.5782442748091603,
      "grad_norm": 0.5801519751548767,
      "learning_rate": 3.878840917993393e-05,
      "loss": 0.6796,
      "step": 3333
    },
    {
      "epoch": 0.5784177654406663,
      "grad_norm": 1.0420644283294678,
      "learning_rate": 3.878607273036922e-05,
      "loss": 0.6821,
      "step": 3334
    },
    {
      "epoch": 0.5785912560721721,
      "grad_norm": 0.7662199139595032,
      "learning_rate": 3.8783734100658874e-05,
      "loss": 0.5862,
      "step": 3335
    },
    {
      "epoch": 0.5787647467036781,
      "grad_norm": 0.8534917831420898,
      "learning_rate": 3.8781393291074296e-05,
      "loss": 0.6056,
      "step": 3336
    },
    {
      "epoch": 0.5789382373351839,
      "grad_norm": 0.7719308733940125,
      "learning_rate": 3.877905030188715e-05,
      "loss": 0.5311,
      "step": 3337
    },
    {
      "epoch": 0.5791117279666897,
      "grad_norm": 0.8380523920059204,
      "learning_rate": 3.8776705133369333e-05,
      "loss": 0.5789,
      "step": 3338
    },
    {
      "epoch": 0.5792852185981957,
      "grad_norm": 0.7678638100624084,
      "learning_rate": 3.8774357785793e-05,
      "loss": 0.6484,
      "step": 3339
    },
    {
      "epoch": 0.5794587092297016,
      "grad_norm": 0.788212776184082,
      "learning_rate": 3.8772008259430575e-05,
      "loss": 0.6382,
      "step": 3340
    },
    {
      "epoch": 0.5796321998612075,
      "grad_norm": 0.738366961479187,
      "learning_rate": 3.8769656554554716e-05,
      "loss": 0.6045,
      "step": 3341
    },
    {
      "epoch": 0.5798056904927134,
      "grad_norm": 0.9916058778762817,
      "learning_rate": 3.876730267143834e-05,
      "loss": 0.6257,
      "step": 3342
    },
    {
      "epoch": 0.5799791811242193,
      "grad_norm": 1.277799129486084,
      "learning_rate": 3.8764946610354626e-05,
      "loss": 0.5104,
      "step": 3343
    },
    {
      "epoch": 0.5801526717557252,
      "grad_norm": 0.9335923194885254,
      "learning_rate": 3.876258837157699e-05,
      "loss": 0.5049,
      "step": 3344
    },
    {
      "epoch": 0.5803261623872311,
      "grad_norm": 0.7122304439544678,
      "learning_rate": 3.876022795537911e-05,
      "loss": 0.5026,
      "step": 3345
    },
    {
      "epoch": 0.580499653018737,
      "grad_norm": 1.5594797134399414,
      "learning_rate": 3.8757865362034914e-05,
      "loss": 0.5314,
      "step": 3346
    },
    {
      "epoch": 0.5806731436502429,
      "grad_norm": 1.2827972173690796,
      "learning_rate": 3.875550059181859e-05,
      "loss": 0.6298,
      "step": 3347
    },
    {
      "epoch": 0.5808466342817488,
      "grad_norm": 1.3870747089385986,
      "learning_rate": 3.875313364500456e-05,
      "loss": 0.6448,
      "step": 3348
    },
    {
      "epoch": 0.5810201249132547,
      "grad_norm": 1.5545334815979004,
      "learning_rate": 3.8750764521867526e-05,
      "loss": 0.5092,
      "step": 3349
    },
    {
      "epoch": 0.5811936155447606,
      "grad_norm": 0.780291736125946,
      "learning_rate": 3.8748393222682425e-05,
      "loss": 0.5072,
      "step": 3350
    },
    {
      "epoch": 0.5813671061762665,
      "grad_norm": 1.0323545932769775,
      "learning_rate": 3.8746019747724436e-05,
      "loss": 0.5936,
      "step": 3351
    },
    {
      "epoch": 0.5815405968077724,
      "grad_norm": 0.9434840679168701,
      "learning_rate": 3.874364409726901e-05,
      "loss": 0.651,
      "step": 3352
    },
    {
      "epoch": 0.5817140874392783,
      "grad_norm": 1.0047508478164673,
      "learning_rate": 3.8741266271591846e-05,
      "loss": 0.6742,
      "step": 3353
    },
    {
      "epoch": 0.5818875780707842,
      "grad_norm": 1.0800899267196655,
      "learning_rate": 3.87388862709689e-05,
      "loss": 0.5039,
      "step": 3354
    },
    {
      "epoch": 0.5820610687022901,
      "grad_norm": 0.9108862280845642,
      "learning_rate": 3.8736504095676364e-05,
      "loss": 0.5164,
      "step": 3355
    },
    {
      "epoch": 0.582234559333796,
      "grad_norm": 0.7775710821151733,
      "learning_rate": 3.8734119745990696e-05,
      "loss": 0.6127,
      "step": 3356
    },
    {
      "epoch": 0.5824080499653018,
      "grad_norm": 1.023120641708374,
      "learning_rate": 3.8731733222188605e-05,
      "loss": 0.641,
      "step": 3357
    },
    {
      "epoch": 0.5825815405968078,
      "grad_norm": 0.6763226985931396,
      "learning_rate": 3.872934452454704e-05,
      "loss": 0.5679,
      "step": 3358
    },
    {
      "epoch": 0.5827550312283136,
      "grad_norm": 0.5067607164382935,
      "learning_rate": 3.8726953653343226e-05,
      "loss": 0.6472,
      "step": 3359
    },
    {
      "epoch": 0.5829285218598196,
      "grad_norm": 0.7462106347084045,
      "learning_rate": 3.872456060885461e-05,
      "loss": 0.594,
      "step": 3360
    },
    {
      "epoch": 0.5831020124913254,
      "grad_norm": 0.8248808979988098,
      "learning_rate": 3.8722165391358926e-05,
      "loss": 0.4955,
      "step": 3361
    },
    {
      "epoch": 0.5832755031228314,
      "grad_norm": 0.7597683072090149,
      "learning_rate": 3.8719768001134124e-05,
      "loss": 0.6084,
      "step": 3362
    },
    {
      "epoch": 0.5834489937543372,
      "grad_norm": 1.114116907119751,
      "learning_rate": 3.8717368438458435e-05,
      "loss": 0.5118,
      "step": 3363
    },
    {
      "epoch": 0.5836224843858432,
      "grad_norm": 1.0715887546539307,
      "learning_rate": 3.871496670361033e-05,
      "loss": 0.6173,
      "step": 3364
    },
    {
      "epoch": 0.583795975017349,
      "grad_norm": 0.9598300457000732,
      "learning_rate": 3.871256279686854e-05,
      "loss": 0.6216,
      "step": 3365
    },
    {
      "epoch": 0.583969465648855,
      "grad_norm": 1.053706407546997,
      "learning_rate": 3.871015671851202e-05,
      "loss": 0.5491,
      "step": 3366
    },
    {
      "epoch": 0.5841429562803608,
      "grad_norm": 1.0202147960662842,
      "learning_rate": 3.8707748468820024e-05,
      "loss": 0.6046,
      "step": 3367
    },
    {
      "epoch": 0.5843164469118668,
      "grad_norm": 0.9622864127159119,
      "learning_rate": 3.870533804807201e-05,
      "loss": 0.5989,
      "step": 3368
    },
    {
      "epoch": 0.5844899375433726,
      "grad_norm": 0.8145611882209778,
      "learning_rate": 3.870292545654772e-05,
      "loss": 0.5956,
      "step": 3369
    },
    {
      "epoch": 0.5846634281748786,
      "grad_norm": 0.9293409585952759,
      "learning_rate": 3.870051069452714e-05,
      "loss": 0.5316,
      "step": 3370
    },
    {
      "epoch": 0.5848369188063844,
      "grad_norm": 0.7547447681427002,
      "learning_rate": 3.86980937622905e-05,
      "loss": 0.6106,
      "step": 3371
    },
    {
      "epoch": 0.5850104094378904,
      "grad_norm": 0.7704901099205017,
      "learning_rate": 3.8695674660118294e-05,
      "loss": 0.7321,
      "step": 3372
    },
    {
      "epoch": 0.5851839000693962,
      "grad_norm": 1.2197743654251099,
      "learning_rate": 3.8693253388291256e-05,
      "loss": 0.7271,
      "step": 3373
    },
    {
      "epoch": 0.5853573907009022,
      "grad_norm": 0.7334229946136475,
      "learning_rate": 3.8690829947090386e-05,
      "loss": 0.5839,
      "step": 3374
    },
    {
      "epoch": 0.585530881332408,
      "grad_norm": 0.7082890272140503,
      "learning_rate": 3.868840433679692e-05,
      "loss": 0.6974,
      "step": 3375
    },
    {
      "epoch": 0.5857043719639139,
      "grad_norm": 0.7663260102272034,
      "learning_rate": 3.868597655769235e-05,
      "loss": 0.6614,
      "step": 3376
    },
    {
      "epoch": 0.5858778625954199,
      "grad_norm": 0.7415555715560913,
      "learning_rate": 3.8683546610058434e-05,
      "loss": 0.5551,
      "step": 3377
    },
    {
      "epoch": 0.5860513532269257,
      "grad_norm": 0.8075070381164551,
      "learning_rate": 3.868111449417716e-05,
      "loss": 0.6012,
      "step": 3378
    },
    {
      "epoch": 0.5862248438584317,
      "grad_norm": 0.704871416091919,
      "learning_rate": 3.867868021033078e-05,
      "loss": 0.6234,
      "step": 3379
    },
    {
      "epoch": 0.5863983344899375,
      "grad_norm": 0.6264182925224304,
      "learning_rate": 3.867624375880179e-05,
      "loss": 0.6178,
      "step": 3380
    },
    {
      "epoch": 0.5865718251214435,
      "grad_norm": 1.559768557548523,
      "learning_rate": 3.8673805139872966e-05,
      "loss": 0.502,
      "step": 3381
    },
    {
      "epoch": 0.5867453157529493,
      "grad_norm": 0.8009780645370483,
      "learning_rate": 3.8671364353827284e-05,
      "loss": 0.6245,
      "step": 3382
    },
    {
      "epoch": 0.5869188063844553,
      "grad_norm": 0.746608555316925,
      "learning_rate": 3.8668921400948015e-05,
      "loss": 0.5422,
      "step": 3383
    },
    {
      "epoch": 0.5870922970159611,
      "grad_norm": 0.8048222064971924,
      "learning_rate": 3.8666476281518665e-05,
      "loss": 0.5168,
      "step": 3384
    },
    {
      "epoch": 0.5872657876474671,
      "grad_norm": 0.7640017867088318,
      "learning_rate": 3.866402899582299e-05,
      "loss": 0.7377,
      "step": 3385
    },
    {
      "epoch": 0.5874392782789729,
      "grad_norm": 0.8351792693138123,
      "learning_rate": 3.8661579544145e-05,
      "loss": 0.6089,
      "step": 3386
    },
    {
      "epoch": 0.5876127689104789,
      "grad_norm": 0.6824077367782593,
      "learning_rate": 3.865912792676897e-05,
      "loss": 0.5981,
      "step": 3387
    },
    {
      "epoch": 0.5877862595419847,
      "grad_norm": 0.9101720452308655,
      "learning_rate": 3.8656674143979386e-05,
      "loss": 0.5658,
      "step": 3388
    },
    {
      "epoch": 0.5879597501734907,
      "grad_norm": 0.753296434879303,
      "learning_rate": 3.865421819606104e-05,
      "loss": 0.6121,
      "step": 3389
    },
    {
      "epoch": 0.5881332408049965,
      "grad_norm": 0.7643285989761353,
      "learning_rate": 3.8651760083298926e-05,
      "loss": 0.5245,
      "step": 3390
    },
    {
      "epoch": 0.5883067314365025,
      "grad_norm": 0.6890398859977722,
      "learning_rate": 3.8649299805978324e-05,
      "loss": 0.6505,
      "step": 3391
    },
    {
      "epoch": 0.5884802220680083,
      "grad_norm": 0.638952374458313,
      "learning_rate": 3.864683736438475e-05,
      "loss": 0.6943,
      "step": 3392
    },
    {
      "epoch": 0.5886537126995143,
      "grad_norm": 0.8157287240028381,
      "learning_rate": 3.864437275880398e-05,
      "loss": 0.5846,
      "step": 3393
    },
    {
      "epoch": 0.5888272033310201,
      "grad_norm": 0.7618674039840698,
      "learning_rate": 3.8641905989522016e-05,
      "loss": 0.6025,
      "step": 3394
    },
    {
      "epoch": 0.5890006939625261,
      "grad_norm": 1.0098158121109009,
      "learning_rate": 3.8639437056825146e-05,
      "loss": 0.5103,
      "step": 3395
    },
    {
      "epoch": 0.5891741845940319,
      "grad_norm": 0.9587876796722412,
      "learning_rate": 3.863696596099988e-05,
      "loss": 0.5607,
      "step": 3396
    },
    {
      "epoch": 0.5893476752255378,
      "grad_norm": 0.5911865830421448,
      "learning_rate": 3.8634492702333e-05,
      "loss": 0.6744,
      "step": 3397
    },
    {
      "epoch": 0.5895211658570437,
      "grad_norm": 0.850542426109314,
      "learning_rate": 3.863201728111153e-05,
      "loss": 0.6318,
      "step": 3398
    },
    {
      "epoch": 0.5896946564885496,
      "grad_norm": 0.9997515678405762,
      "learning_rate": 3.8629539697622746e-05,
      "loss": 0.7252,
      "step": 3399
    },
    {
      "epoch": 0.5898681471200555,
      "grad_norm": 0.9618603587150574,
      "learning_rate": 3.862705995215417e-05,
      "loss": 0.5009,
      "step": 3400
    },
    {
      "epoch": 0.5900416377515614,
      "grad_norm": 0.8880718350410461,
      "learning_rate": 3.862457804499358e-05,
      "loss": 0.562,
      "step": 3401
    },
    {
      "epoch": 0.5902151283830673,
      "grad_norm": 0.7900727987289429,
      "learning_rate": 3.862209397642901e-05,
      "loss": 0.6401,
      "step": 3402
    },
    {
      "epoch": 0.5903886190145732,
      "grad_norm": 0.847778856754303,
      "learning_rate": 3.861960774674874e-05,
      "loss": 0.4969,
      "step": 3403
    },
    {
      "epoch": 0.5905621096460791,
      "grad_norm": 0.9456307291984558,
      "learning_rate": 3.861711935624129e-05,
      "loss": 0.5255,
      "step": 3404
    },
    {
      "epoch": 0.590735600277585,
      "grad_norm": 1.0695728063583374,
      "learning_rate": 3.861462880519543e-05,
      "loss": 0.6191,
      "step": 3405
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 0.7595635056495667,
      "learning_rate": 3.8612136093900224e-05,
      "loss": 0.5824,
      "step": 3406
    },
    {
      "epoch": 0.5910825815405968,
      "grad_norm": 11.357579231262207,
      "learning_rate": 3.860964122264493e-05,
      "loss": 0.691,
      "step": 3407
    },
    {
      "epoch": 0.5912560721721027,
      "grad_norm": 1.208006739616394,
      "learning_rate": 3.860714419171909e-05,
      "loss": 0.5909,
      "step": 3408
    },
    {
      "epoch": 0.5914295628036086,
      "grad_norm": 0.839406430721283,
      "learning_rate": 3.860464500141249e-05,
      "loss": 0.572,
      "step": 3409
    },
    {
      "epoch": 0.5916030534351145,
      "grad_norm": 1.164560317993164,
      "learning_rate": 3.860214365201515e-05,
      "loss": 0.6687,
      "step": 3410
    },
    {
      "epoch": 0.5917765440666204,
      "grad_norm": 1.1396112442016602,
      "learning_rate": 3.859964014381737e-05,
      "loss": 0.6771,
      "step": 3411
    },
    {
      "epoch": 0.5919500346981263,
      "grad_norm": 0.8461340665817261,
      "learning_rate": 3.8597134477109674e-05,
      "loss": 0.7394,
      "step": 3412
    },
    {
      "epoch": 0.5921235253296322,
      "grad_norm": 0.7491380572319031,
      "learning_rate": 3.859462665218286e-05,
      "loss": 0.6487,
      "step": 3413
    },
    {
      "epoch": 0.5922970159611382,
      "grad_norm": 0.7270049452781677,
      "learning_rate": 3.8592116669327945e-05,
      "loss": 0.5641,
      "step": 3414
    },
    {
      "epoch": 0.592470506592644,
      "grad_norm": 0.7229993343353271,
      "learning_rate": 3.858960452883623e-05,
      "loss": 0.6256,
      "step": 3415
    },
    {
      "epoch": 0.5926439972241498,
      "grad_norm": 0.7910577058792114,
      "learning_rate": 3.858709023099925e-05,
      "loss": 0.5709,
      "step": 3416
    },
    {
      "epoch": 0.5928174878556558,
      "grad_norm": 0.7861661911010742,
      "learning_rate": 3.8584573776108794e-05,
      "loss": 0.578,
      "step": 3417
    },
    {
      "epoch": 0.5929909784871616,
      "grad_norm": 0.7965797781944275,
      "learning_rate": 3.858205516445689e-05,
      "loss": 0.5209,
      "step": 3418
    },
    {
      "epoch": 0.5931644691186676,
      "grad_norm": 1.3844408988952637,
      "learning_rate": 3.8579534396335835e-05,
      "loss": 0.6051,
      "step": 3419
    },
    {
      "epoch": 0.5933379597501734,
      "grad_norm": 1.988527536392212,
      "learning_rate": 3.857701147203816e-05,
      "loss": 0.5623,
      "step": 3420
    },
    {
      "epoch": 0.5935114503816794,
      "grad_norm": 0.8353315591812134,
      "learning_rate": 3.8574486391856655e-05,
      "loss": 0.4869,
      "step": 3421
    },
    {
      "epoch": 0.5936849410131853,
      "grad_norm": 0.7921696305274963,
      "learning_rate": 3.857195915608437e-05,
      "loss": 0.6094,
      "step": 3422
    },
    {
      "epoch": 0.5938584316446912,
      "grad_norm": 1.1405901908874512,
      "learning_rate": 3.856942976501458e-05,
      "loss": 0.6217,
      "step": 3423
    },
    {
      "epoch": 0.594031922276197,
      "grad_norm": 0.9423086643218994,
      "learning_rate": 3.8566898218940825e-05,
      "loss": 0.5925,
      "step": 3424
    },
    {
      "epoch": 0.594205412907703,
      "grad_norm": 0.6545419096946716,
      "learning_rate": 3.85643645181569e-05,
      "loss": 0.5238,
      "step": 3425
    },
    {
      "epoch": 0.5943789035392089,
      "grad_norm": 0.9272714853286743,
      "learning_rate": 3.856182866295684e-05,
      "loss": 0.5372,
      "step": 3426
    },
    {
      "epoch": 0.5945523941707148,
      "grad_norm": 0.8032915592193604,
      "learning_rate": 3.855929065363493e-05,
      "loss": 0.6144,
      "step": 3427
    },
    {
      "epoch": 0.5947258848022207,
      "grad_norm": 0.9311384558677673,
      "learning_rate": 3.8556750490485724e-05,
      "loss": 0.6423,
      "step": 3428
    },
    {
      "epoch": 0.5948993754337266,
      "grad_norm": 0.9234505295753479,
      "learning_rate": 3.8554208173804e-05,
      "loss": 0.5463,
      "step": 3429
    },
    {
      "epoch": 0.5950728660652325,
      "grad_norm": 0.9996792674064636,
      "learning_rate": 3.855166370388479e-05,
      "loss": 0.6635,
      "step": 3430
    },
    {
      "epoch": 0.5952463566967384,
      "grad_norm": 0.7507815361022949,
      "learning_rate": 3.854911708102339e-05,
      "loss": 0.6295,
      "step": 3431
    },
    {
      "epoch": 0.5954198473282443,
      "grad_norm": 0.8840140104293823,
      "learning_rate": 3.8546568305515345e-05,
      "loss": 0.5497,
      "step": 3432
    },
    {
      "epoch": 0.5955933379597502,
      "grad_norm": 1.1633116006851196,
      "learning_rate": 3.854401737765644e-05,
      "loss": 0.5869,
      "step": 3433
    },
    {
      "epoch": 0.5957668285912561,
      "grad_norm": 0.8514946103096008,
      "learning_rate": 3.85414642977427e-05,
      "loss": 0.6907,
      "step": 3434
    },
    {
      "epoch": 0.5959403192227619,
      "grad_norm": 1.305191159248352,
      "learning_rate": 3.853890906607043e-05,
      "loss": 0.5542,
      "step": 3435
    },
    {
      "epoch": 0.5961138098542679,
      "grad_norm": 0.9533877968788147,
      "learning_rate": 3.8536351682936155e-05,
      "loss": 0.6359,
      "step": 3436
    },
    {
      "epoch": 0.5962873004857737,
      "grad_norm": 0.7394745945930481,
      "learning_rate": 3.853379214863667e-05,
      "loss": 0.58,
      "step": 3437
    },
    {
      "epoch": 0.5964607911172797,
      "grad_norm": 0.6518771648406982,
      "learning_rate": 3.8531230463469015e-05,
      "loss": 0.6638,
      "step": 3438
    },
    {
      "epoch": 0.5966342817487855,
      "grad_norm": 0.9110044240951538,
      "learning_rate": 3.852866662773047e-05,
      "loss": 0.7,
      "step": 3439
    },
    {
      "epoch": 0.5968077723802915,
      "grad_norm": 1.4987447261810303,
      "learning_rate": 3.852610064171857e-05,
      "loss": 0.728,
      "step": 3440
    },
    {
      "epoch": 0.5969812630117973,
      "grad_norm": 0.9062539935112,
      "learning_rate": 3.85235325057311e-05,
      "loss": 0.6445,
      "step": 3441
    },
    {
      "epoch": 0.5971547536433033,
      "grad_norm": 1.3974779844284058,
      "learning_rate": 3.85209622200661e-05,
      "loss": 0.6837,
      "step": 3442
    },
    {
      "epoch": 0.5973282442748091,
      "grad_norm": 0.6922855973243713,
      "learning_rate": 3.851838978502186e-05,
      "loss": 0.6296,
      "step": 3443
    },
    {
      "epoch": 0.5975017349063151,
      "grad_norm": 0.7842308878898621,
      "learning_rate": 3.8515815200896905e-05,
      "loss": 0.5752,
      "step": 3444
    },
    {
      "epoch": 0.5976752255378209,
      "grad_norm": 0.7984820604324341,
      "learning_rate": 3.851323846799002e-05,
      "loss": 0.6096,
      "step": 3445
    },
    {
      "epoch": 0.5978487161693269,
      "grad_norm": 1.3559335470199585,
      "learning_rate": 3.851065958660023e-05,
      "loss": 0.6342,
      "step": 3446
    },
    {
      "epoch": 0.5980222068008327,
      "grad_norm": 0.9452993273735046,
      "learning_rate": 3.8508078557026835e-05,
      "loss": 0.7297,
      "step": 3447
    },
    {
      "epoch": 0.5981956974323387,
      "grad_norm": 0.634530246257782,
      "learning_rate": 3.8505495379569354e-05,
      "loss": 0.5994,
      "step": 3448
    },
    {
      "epoch": 0.5983691880638445,
      "grad_norm": 0.8228620290756226,
      "learning_rate": 3.850291005452757e-05,
      "loss": 0.5238,
      "step": 3449
    },
    {
      "epoch": 0.5985426786953505,
      "grad_norm": 1.0455272197723389,
      "learning_rate": 3.850032258220152e-05,
      "loss": 0.5449,
      "step": 3450
    },
    {
      "epoch": 0.5987161693268563,
      "grad_norm": 0.7928491234779358,
      "learning_rate": 3.849773296289147e-05,
      "loss": 0.6968,
      "step": 3451
    },
    {
      "epoch": 0.5988896599583623,
      "grad_norm": 1.4088414907455444,
      "learning_rate": 3.849514119689796e-05,
      "loss": 0.5996,
      "step": 3452
    },
    {
      "epoch": 0.5990631505898681,
      "grad_norm": 0.9783432483673096,
      "learning_rate": 3.849254728452176e-05,
      "loss": 0.5813,
      "step": 3453
    },
    {
      "epoch": 0.5992366412213741,
      "grad_norm": 0.7994115352630615,
      "learning_rate": 3.84899512260639e-05,
      "loss": 0.6154,
      "step": 3454
    },
    {
      "epoch": 0.59941013185288,
      "grad_norm": 0.754689633846283,
      "learning_rate": 3.848735302182566e-05,
      "loss": 0.584,
      "step": 3455
    },
    {
      "epoch": 0.5995836224843858,
      "grad_norm": 2.09782075881958,
      "learning_rate": 3.848475267210856e-05,
      "loss": 0.5867,
      "step": 3456
    },
    {
      "epoch": 0.5997571131158917,
      "grad_norm": 1.2456940412521362,
      "learning_rate": 3.848215017721437e-05,
      "loss": 0.5066,
      "step": 3457
    },
    {
      "epoch": 0.5999306037473976,
      "grad_norm": 0.7088345289230347,
      "learning_rate": 3.8479545537445115e-05,
      "loss": 0.6174,
      "step": 3458
    },
    {
      "epoch": 0.6001040943789036,
      "grad_norm": 2.3128747940063477,
      "learning_rate": 3.8476938753103066e-05,
      "loss": 0.47,
      "step": 3459
    },
    {
      "epoch": 0.6002775850104094,
      "grad_norm": 1.0873178243637085,
      "learning_rate": 3.847432982449075e-05,
      "loss": 0.6378,
      "step": 3460
    },
    {
      "epoch": 0.6004510756419154,
      "grad_norm": 0.6794579029083252,
      "learning_rate": 3.8471718751910926e-05,
      "loss": 0.661,
      "step": 3461
    },
    {
      "epoch": 0.6006245662734212,
      "grad_norm": 0.8252881765365601,
      "learning_rate": 3.846910553566662e-05,
      "loss": 0.5981,
      "step": 3462
    },
    {
      "epoch": 0.6007980569049272,
      "grad_norm": 0.7919279336929321,
      "learning_rate": 3.846649017606109e-05,
      "loss": 0.6964,
      "step": 3463
    },
    {
      "epoch": 0.600971547536433,
      "grad_norm": 0.8595472574234009,
      "learning_rate": 3.846387267339787e-05,
      "loss": 0.6184,
      "step": 3464
    },
    {
      "epoch": 0.601145038167939,
      "grad_norm": 0.7740964889526367,
      "learning_rate": 3.84612530279807e-05,
      "loss": 0.6564,
      "step": 3465
    },
    {
      "epoch": 0.6013185287994448,
      "grad_norm": 0.8735344409942627,
      "learning_rate": 3.845863124011361e-05,
      "loss": 0.6528,
      "step": 3466
    },
    {
      "epoch": 0.6014920194309508,
      "grad_norm": 1.1475926637649536,
      "learning_rate": 3.845600731010085e-05,
      "loss": 0.5629,
      "step": 3467
    },
    {
      "epoch": 0.6016655100624566,
      "grad_norm": 0.6880063414573669,
      "learning_rate": 3.845338123824694e-05,
      "loss": 0.6332,
      "step": 3468
    },
    {
      "epoch": 0.6018390006939626,
      "grad_norm": 2.3320505619049072,
      "learning_rate": 3.845075302485664e-05,
      "loss": 0.6035,
      "step": 3469
    },
    {
      "epoch": 0.6020124913254684,
      "grad_norm": 0.8664422631263733,
      "learning_rate": 3.844812267023495e-05,
      "loss": 0.6415,
      "step": 3470
    },
    {
      "epoch": 0.6021859819569744,
      "grad_norm": 0.9907106161117554,
      "learning_rate": 3.844549017468712e-05,
      "loss": 0.563,
      "step": 3471
    },
    {
      "epoch": 0.6023594725884802,
      "grad_norm": 0.7867692112922668,
      "learning_rate": 3.8442855538518667e-05,
      "loss": 0.5328,
      "step": 3472
    },
    {
      "epoch": 0.6025329632199862,
      "grad_norm": 1.5940982103347778,
      "learning_rate": 3.844021876203534e-05,
      "loss": 0.516,
      "step": 3473
    },
    {
      "epoch": 0.602706453851492,
      "grad_norm": 0.8625582456588745,
      "learning_rate": 3.8437579845543133e-05,
      "loss": 0.67,
      "step": 3474
    },
    {
      "epoch": 0.6028799444829979,
      "grad_norm": 1.81771719455719,
      "learning_rate": 3.843493878934831e-05,
      "loss": 0.6008,
      "step": 3475
    },
    {
      "epoch": 0.6030534351145038,
      "grad_norm": 1.882793664932251,
      "learning_rate": 3.843229559375735e-05,
      "loss": 0.6074,
      "step": 3476
    },
    {
      "epoch": 0.6032269257460097,
      "grad_norm": 0.9315357804298401,
      "learning_rate": 3.842965025907701e-05,
      "loss": 0.5636,
      "step": 3477
    },
    {
      "epoch": 0.6034004163775156,
      "grad_norm": 1.3701409101486206,
      "learning_rate": 3.842700278561429e-05,
      "loss": 0.4924,
      "step": 3478
    },
    {
      "epoch": 0.6035739070090215,
      "grad_norm": 1.117668867111206,
      "learning_rate": 3.842435317367642e-05,
      "loss": 0.6133,
      "step": 3479
    },
    {
      "epoch": 0.6037473976405274,
      "grad_norm": 1.1667253971099854,
      "learning_rate": 3.8421701423570895e-05,
      "loss": 0.601,
      "step": 3480
    },
    {
      "epoch": 0.6039208882720333,
      "grad_norm": 1.159469485282898,
      "learning_rate": 3.8419047535605456e-05,
      "loss": 0.724,
      "step": 3481
    },
    {
      "epoch": 0.6040943789035392,
      "grad_norm": 1.304949402809143,
      "learning_rate": 3.841639151008809e-05,
      "loss": 0.5074,
      "step": 3482
    },
    {
      "epoch": 0.6042678695350451,
      "grad_norm": 1.1901766061782837,
      "learning_rate": 3.8413733347327024e-05,
      "loss": 0.5042,
      "step": 3483
    },
    {
      "epoch": 0.604441360166551,
      "grad_norm": 1.2651426792144775,
      "learning_rate": 3.8411073047630745e-05,
      "loss": 0.5394,
      "step": 3484
    },
    {
      "epoch": 0.6046148507980569,
      "grad_norm": 0.7895896434783936,
      "learning_rate": 3.8408410611308e-05,
      "loss": 0.5227,
      "step": 3485
    },
    {
      "epoch": 0.6047883414295628,
      "grad_norm": 1.146105408668518,
      "learning_rate": 3.840574603866774e-05,
      "loss": 0.5277,
      "step": 3486
    },
    {
      "epoch": 0.6049618320610687,
      "grad_norm": 1.4869883060455322,
      "learning_rate": 3.840307933001921e-05,
      "loss": 0.5737,
      "step": 3487
    },
    {
      "epoch": 0.6051353226925746,
      "grad_norm": 1.337009072303772,
      "learning_rate": 3.840041048567188e-05,
      "loss": 0.5547,
      "step": 3488
    },
    {
      "epoch": 0.6053088133240805,
      "grad_norm": 1.0347092151641846,
      "learning_rate": 3.839773950593547e-05,
      "loss": 0.5471,
      "step": 3489
    },
    {
      "epoch": 0.6054823039555864,
      "grad_norm": 0.7365402579307556,
      "learning_rate": 3.839506639111996e-05,
      "loss": 0.6643,
      "step": 3490
    },
    {
      "epoch": 0.6056557945870923,
      "grad_norm": 0.7664414644241333,
      "learning_rate": 3.839239114153555e-05,
      "loss": 0.5186,
      "step": 3491
    },
    {
      "epoch": 0.6058292852185982,
      "grad_norm": 0.8788231611251831,
      "learning_rate": 3.838971375749272e-05,
      "loss": 0.496,
      "step": 3492
    },
    {
      "epoch": 0.6060027758501041,
      "grad_norm": 1.0685210227966309,
      "learning_rate": 3.838703423930218e-05,
      "loss": 0.6027,
      "step": 3493
    },
    {
      "epoch": 0.6061762664816099,
      "grad_norm": 1.3240433931350708,
      "learning_rate": 3.83843525872749e-05,
      "loss": 0.687,
      "step": 3494
    },
    {
      "epoch": 0.6063497571131159,
      "grad_norm": 0.7077116370201111,
      "learning_rate": 3.838166880172207e-05,
      "loss": 0.5708,
      "step": 3495
    },
    {
      "epoch": 0.6065232477446217,
      "grad_norm": 0.778710663318634,
      "learning_rate": 3.837898288295516e-05,
      "loss": 0.572,
      "step": 3496
    },
    {
      "epoch": 0.6066967383761277,
      "grad_norm": 1.6770925521850586,
      "learning_rate": 3.837629483128587e-05,
      "loss": 0.6476,
      "step": 3497
    },
    {
      "epoch": 0.6068702290076335,
      "grad_norm": 0.8148396015167236,
      "learning_rate": 3.837360464702616e-05,
      "loss": 0.589,
      "step": 3498
    },
    {
      "epoch": 0.6070437196391395,
      "grad_norm": 0.9807456731796265,
      "learning_rate": 3.837091233048821e-05,
      "loss": 0.5621,
      "step": 3499
    },
    {
      "epoch": 0.6072172102706453,
      "grad_norm": 1.5750974416732788,
      "learning_rate": 3.8368217881984484e-05,
      "loss": 0.515,
      "step": 3500
    },
    {
      "epoch": 0.6073907009021513,
      "grad_norm": 0.8710517883300781,
      "learning_rate": 3.836552130182766e-05,
      "loss": 0.5878,
      "step": 3501
    },
    {
      "epoch": 0.6075641915336571,
      "grad_norm": 0.8788391351699829,
      "learning_rate": 3.83628225903307e-05,
      "loss": 0.5024,
      "step": 3502
    },
    {
      "epoch": 0.6077376821651631,
      "grad_norm": 0.8667015433311462,
      "learning_rate": 3.836012174780678e-05,
      "loss": 0.5999,
      "step": 3503
    },
    {
      "epoch": 0.607911172796669,
      "grad_norm": 0.7130902409553528,
      "learning_rate": 3.8357418774569335e-05,
      "loss": 0.6808,
      "step": 3504
    },
    {
      "epoch": 0.6080846634281749,
      "grad_norm": 1.0192363262176514,
      "learning_rate": 3.835471367093205e-05,
      "loss": 0.5347,
      "step": 3505
    },
    {
      "epoch": 0.6082581540596808,
      "grad_norm": 2.0837340354919434,
      "learning_rate": 3.835200643720886e-05,
      "loss": 0.5211,
      "step": 3506
    },
    {
      "epoch": 0.6084316446911867,
      "grad_norm": 0.8674420714378357,
      "learning_rate": 3.834929707371394e-05,
      "loss": 0.6078,
      "step": 3507
    },
    {
      "epoch": 0.6086051353226926,
      "grad_norm": 0.8269285559654236,
      "learning_rate": 3.8346585580761705e-05,
      "loss": 0.6461,
      "step": 3508
    },
    {
      "epoch": 0.6087786259541985,
      "grad_norm": 0.7027774453163147,
      "learning_rate": 3.834387195866684e-05,
      "loss": 0.6669,
      "step": 3509
    },
    {
      "epoch": 0.6089521165857044,
      "grad_norm": 1.2846192121505737,
      "learning_rate": 3.8341156207744254e-05,
      "loss": 0.5911,
      "step": 3510
    },
    {
      "epoch": 0.6091256072172103,
      "grad_norm": 0.6821696162223816,
      "learning_rate": 3.8338438328309126e-05,
      "loss": 0.542,
      "step": 3511
    },
    {
      "epoch": 0.6092990978487162,
      "grad_norm": 0.8131824135780334,
      "learning_rate": 3.833571832067685e-05,
      "loss": 0.5776,
      "step": 3512
    },
    {
      "epoch": 0.6094725884802221,
      "grad_norm": 0.6710551977157593,
      "learning_rate": 3.833299618516311e-05,
      "loss": 0.5839,
      "step": 3513
    },
    {
      "epoch": 0.609646079111728,
      "grad_norm": 1.05852210521698,
      "learning_rate": 3.8330271922083795e-05,
      "loss": 0.6022,
      "step": 3514
    },
    {
      "epoch": 0.6098195697432338,
      "grad_norm": 0.6794431209564209,
      "learning_rate": 3.832754553175507e-05,
      "loss": 0.536,
      "step": 3515
    },
    {
      "epoch": 0.6099930603747398,
      "grad_norm": 4.972042083740234,
      "learning_rate": 3.8324817014493326e-05,
      "loss": 0.5684,
      "step": 3516
    },
    {
      "epoch": 0.6101665510062456,
      "grad_norm": 0.8742214441299438,
      "learning_rate": 3.832208637061522e-05,
      "loss": 0.5333,
      "step": 3517
    },
    {
      "epoch": 0.6103400416377516,
      "grad_norm": 0.9846548438072205,
      "learning_rate": 3.831935360043763e-05,
      "loss": 0.6179,
      "step": 3518
    },
    {
      "epoch": 0.6105135322692574,
      "grad_norm": 1.2224737405776978,
      "learning_rate": 3.8316618704277715e-05,
      "loss": 0.6616,
      "step": 3519
    },
    {
      "epoch": 0.6106870229007634,
      "grad_norm": 0.6444560885429382,
      "learning_rate": 3.8313881682452854e-05,
      "loss": 0.5886,
      "step": 3520
    },
    {
      "epoch": 0.6108605135322692,
      "grad_norm": 1.0943591594696045,
      "learning_rate": 3.8311142535280684e-05,
      "loss": 0.5546,
      "step": 3521
    },
    {
      "epoch": 0.6110340041637752,
      "grad_norm": 0.7637893557548523,
      "learning_rate": 3.830840126307909e-05,
      "loss": 0.575,
      "step": 3522
    },
    {
      "epoch": 0.611207494795281,
      "grad_norm": 0.822937548160553,
      "learning_rate": 3.830565786616619e-05,
      "loss": 0.5853,
      "step": 3523
    },
    {
      "epoch": 0.611380985426787,
      "grad_norm": 0.9483650326728821,
      "learning_rate": 3.830291234486037e-05,
      "loss": 0.5626,
      "step": 3524
    },
    {
      "epoch": 0.6115544760582928,
      "grad_norm": 0.7164247632026672,
      "learning_rate": 3.8300164699480246e-05,
      "loss": 0.5519,
      "step": 3525
    },
    {
      "epoch": 0.6117279666897988,
      "grad_norm": 0.7991425395011902,
      "learning_rate": 3.8297414930344684e-05,
      "loss": 0.6359,
      "step": 3526
    },
    {
      "epoch": 0.6119014573213046,
      "grad_norm": 0.918340802192688,
      "learning_rate": 3.8294663037772794e-05,
      "loss": 0.6689,
      "step": 3527
    },
    {
      "epoch": 0.6120749479528106,
      "grad_norm": 1.0987683534622192,
      "learning_rate": 3.829190902208394e-05,
      "loss": 0.5338,
      "step": 3528
    },
    {
      "epoch": 0.6122484385843164,
      "grad_norm": 1.5816664695739746,
      "learning_rate": 3.828915288359774e-05,
      "loss": 0.5103,
      "step": 3529
    },
    {
      "epoch": 0.6124219292158224,
      "grad_norm": 0.9146783351898193,
      "learning_rate": 3.828639462263403e-05,
      "loss": 0.525,
      "step": 3530
    },
    {
      "epoch": 0.6125954198473282,
      "grad_norm": 1.0526752471923828,
      "learning_rate": 3.828363423951291e-05,
      "loss": 0.6726,
      "step": 3531
    },
    {
      "epoch": 0.6127689104788342,
      "grad_norm": 0.8325849175453186,
      "learning_rate": 3.8280871734554746e-05,
      "loss": 0.5735,
      "step": 3532
    },
    {
      "epoch": 0.61294240111034,
      "grad_norm": 1.0969488620758057,
      "learning_rate": 3.8278107108080104e-05,
      "loss": 0.6194,
      "step": 3533
    },
    {
      "epoch": 0.6131158917418459,
      "grad_norm": 0.6949189305305481,
      "learning_rate": 3.827534036040984e-05,
      "loss": 0.7388,
      "step": 3534
    },
    {
      "epoch": 0.6132893823733518,
      "grad_norm": 0.9849112629890442,
      "learning_rate": 3.827257149186502e-05,
      "loss": 0.6366,
      "step": 3535
    },
    {
      "epoch": 0.6134628730048577,
      "grad_norm": 0.8824640512466431,
      "learning_rate": 3.8269800502767e-05,
      "loss": 0.4998,
      "step": 3536
    },
    {
      "epoch": 0.6136363636363636,
      "grad_norm": 1.1439348459243774,
      "learning_rate": 3.826702739343734e-05,
      "loss": 0.5786,
      "step": 3537
    },
    {
      "epoch": 0.6138098542678695,
      "grad_norm": 0.8941345810890198,
      "learning_rate": 3.8264252164197866e-05,
      "loss": 0.599,
      "step": 3538
    },
    {
      "epoch": 0.6139833448993754,
      "grad_norm": 0.8163082003593445,
      "learning_rate": 3.826147481537065e-05,
      "loss": 0.5389,
      "step": 3539
    },
    {
      "epoch": 0.6141568355308813,
      "grad_norm": 1.1131253242492676,
      "learning_rate": 3.825869534727799e-05,
      "loss": 0.5281,
      "step": 3540
    },
    {
      "epoch": 0.6143303261623873,
      "grad_norm": 1.7549223899841309,
      "learning_rate": 3.825591376024247e-05,
      "loss": 0.5359,
      "step": 3541
    },
    {
      "epoch": 0.6145038167938931,
      "grad_norm": 0.6971312761306763,
      "learning_rate": 3.8253130054586886e-05,
      "loss": 0.5726,
      "step": 3542
    },
    {
      "epoch": 0.614677307425399,
      "grad_norm": 0.8808236122131348,
      "learning_rate": 3.825034423063429e-05,
      "loss": 0.5706,
      "step": 3543
    },
    {
      "epoch": 0.6148507980569049,
      "grad_norm": 0.9417666792869568,
      "learning_rate": 3.824755628870797e-05,
      "loss": 0.6189,
      "step": 3544
    },
    {
      "epoch": 0.6150242886884109,
      "grad_norm": 1.137608289718628,
      "learning_rate": 3.824476622913149e-05,
      "loss": 0.5914,
      "step": 3545
    },
    {
      "epoch": 0.6151977793199167,
      "grad_norm": 0.6162335276603699,
      "learning_rate": 3.824197405222863e-05,
      "loss": 0.7371,
      "step": 3546
    },
    {
      "epoch": 0.6153712699514227,
      "grad_norm": 0.6862666606903076,
      "learning_rate": 3.8239179758323424e-05,
      "loss": 0.666,
      "step": 3547
    },
    {
      "epoch": 0.6155447605829285,
      "grad_norm": 0.8268684148788452,
      "learning_rate": 3.8236383347740146e-05,
      "loss": 0.6798,
      "step": 3548
    },
    {
      "epoch": 0.6157182512144345,
      "grad_norm": 0.7263082265853882,
      "learning_rate": 3.823358482080334e-05,
      "loss": 0.6742,
      "step": 3549
    },
    {
      "epoch": 0.6158917418459403,
      "grad_norm": 0.8222211003303528,
      "learning_rate": 3.823078417783777e-05,
      "loss": 0.6381,
      "step": 3550
    },
    {
      "epoch": 0.6160652324774463,
      "grad_norm": 0.8517816662788391,
      "learning_rate": 3.8227981419168445e-05,
      "loss": 0.5504,
      "step": 3551
    },
    {
      "epoch": 0.6162387231089521,
      "grad_norm": 1.240667700767517,
      "learning_rate": 3.8225176545120646e-05,
      "loss": 0.5581,
      "step": 3552
    },
    {
      "epoch": 0.6164122137404581,
      "grad_norm": 0.7769044041633606,
      "learning_rate": 3.822236955601987e-05,
      "loss": 0.6517,
      "step": 3553
    },
    {
      "epoch": 0.6165857043719639,
      "grad_norm": 1.2511776685714722,
      "learning_rate": 3.821956045219186e-05,
      "loss": 0.6025,
      "step": 3554
    },
    {
      "epoch": 0.6167591950034698,
      "grad_norm": 0.6716343760490417,
      "learning_rate": 3.821674923396265e-05,
      "loss": 0.5631,
      "step": 3555
    },
    {
      "epoch": 0.6169326856349757,
      "grad_norm": 1.228094458580017,
      "learning_rate": 3.821393590165845e-05,
      "loss": 0.5713,
      "step": 3556
    },
    {
      "epoch": 0.6171061762664816,
      "grad_norm": 0.6799286603927612,
      "learning_rate": 3.8211120455605774e-05,
      "loss": 0.6575,
      "step": 3557
    },
    {
      "epoch": 0.6172796668979875,
      "grad_norm": 0.5899439454078674,
      "learning_rate": 3.8208302896131344e-05,
      "loss": 0.6927,
      "step": 3558
    },
    {
      "epoch": 0.6174531575294934,
      "grad_norm": 0.7722693085670471,
      "learning_rate": 3.820548322356215e-05,
      "loss": 0.6003,
      "step": 3559
    },
    {
      "epoch": 0.6176266481609993,
      "grad_norm": 0.8632433414459229,
      "learning_rate": 3.820266143822541e-05,
      "loss": 0.4967,
      "step": 3560
    },
    {
      "epoch": 0.6178001387925052,
      "grad_norm": 0.756759762763977,
      "learning_rate": 3.81998375404486e-05,
      "loss": 0.6062,
      "step": 3561
    },
    {
      "epoch": 0.6179736294240111,
      "grad_norm": 0.6955695748329163,
      "learning_rate": 3.819701153055944e-05,
      "loss": 0.5519,
      "step": 3562
    },
    {
      "epoch": 0.618147120055517,
      "grad_norm": 0.6327463984489441,
      "learning_rate": 3.8194183408885885e-05,
      "loss": 0.657,
      "step": 3563
    },
    {
      "epoch": 0.6183206106870229,
      "grad_norm": 0.7696917057037354,
      "learning_rate": 3.8191353175756145e-05,
      "loss": 0.6415,
      "step": 3564
    },
    {
      "epoch": 0.6184941013185288,
      "grad_norm": 0.7328816056251526,
      "learning_rate": 3.818852083149867e-05,
      "loss": 0.6356,
      "step": 3565
    },
    {
      "epoch": 0.6186675919500347,
      "grad_norm": 1.2973989248275757,
      "learning_rate": 3.818568637644217e-05,
      "loss": 0.6426,
      "step": 3566
    },
    {
      "epoch": 0.6188410825815406,
      "grad_norm": 1.6103951930999756,
      "learning_rate": 3.818284981091556e-05,
      "loss": 0.5831,
      "step": 3567
    },
    {
      "epoch": 0.6190145732130465,
      "grad_norm": 1.0352448225021362,
      "learning_rate": 3.8180011135248055e-05,
      "loss": 0.6628,
      "step": 3568
    },
    {
      "epoch": 0.6191880638445524,
      "grad_norm": 0.9566216468811035,
      "learning_rate": 3.8177170349769064e-05,
      "loss": 0.561,
      "step": 3569
    },
    {
      "epoch": 0.6193615544760583,
      "grad_norm": 0.8397760987281799,
      "learning_rate": 3.8174327454808275e-05,
      "loss": 0.6062,
      "step": 3570
    },
    {
      "epoch": 0.6195350451075642,
      "grad_norm": 0.9678451418876648,
      "learning_rate": 3.8171482450695616e-05,
      "loss": 0.5275,
      "step": 3571
    },
    {
      "epoch": 0.6197085357390701,
      "grad_norm": 1.1903417110443115,
      "learning_rate": 3.816863533776124e-05,
      "loss": 0.6281,
      "step": 3572
    },
    {
      "epoch": 0.619882026370576,
      "grad_norm": 0.8403670191764832,
      "learning_rate": 3.816578611633556e-05,
      "loss": 0.707,
      "step": 3573
    },
    {
      "epoch": 0.6200555170020818,
      "grad_norm": 0.7110308408737183,
      "learning_rate": 3.816293478674923e-05,
      "loss": 0.5867,
      "step": 3574
    },
    {
      "epoch": 0.6202290076335878,
      "grad_norm": 0.7667453289031982,
      "learning_rate": 3.816008134933317e-05,
      "loss": 0.4829,
      "step": 3575
    },
    {
      "epoch": 0.6204024982650936,
      "grad_norm": 1.0670323371887207,
      "learning_rate": 3.815722580441849e-05,
      "loss": 0.5897,
      "step": 3576
    },
    {
      "epoch": 0.6205759888965996,
      "grad_norm": 0.7509621381759644,
      "learning_rate": 3.8154368152336606e-05,
      "loss": 0.6483,
      "step": 3577
    },
    {
      "epoch": 0.6207494795281054,
      "grad_norm": 0.9665395617485046,
      "learning_rate": 3.815150839341915e-05,
      "loss": 0.6711,
      "step": 3578
    },
    {
      "epoch": 0.6209229701596114,
      "grad_norm": 0.6299415826797485,
      "learning_rate": 3.814864652799798e-05,
      "loss": 0.7004,
      "step": 3579
    },
    {
      "epoch": 0.6210964607911172,
      "grad_norm": 0.620185375213623,
      "learning_rate": 3.8145782556405244e-05,
      "loss": 0.7126,
      "step": 3580
    },
    {
      "epoch": 0.6212699514226232,
      "grad_norm": 0.8498899340629578,
      "learning_rate": 3.81429164789733e-05,
      "loss": 0.5984,
      "step": 3581
    },
    {
      "epoch": 0.621443442054129,
      "grad_norm": 0.8451429605484009,
      "learning_rate": 3.814004829603475e-05,
      "loss": 0.5302,
      "step": 3582
    },
    {
      "epoch": 0.621616932685635,
      "grad_norm": 0.6915079355239868,
      "learning_rate": 3.813717800792246e-05,
      "loss": 0.6641,
      "step": 3583
    },
    {
      "epoch": 0.6217904233171409,
      "grad_norm": 1.105806589126587,
      "learning_rate": 3.813430561496953e-05,
      "loss": 0.6299,
      "step": 3584
    },
    {
      "epoch": 0.6219639139486468,
      "grad_norm": 0.5665948390960693,
      "learning_rate": 3.81314311175093e-05,
      "loss": 0.6212,
      "step": 3585
    },
    {
      "epoch": 0.6221374045801527,
      "grad_norm": 0.7293710708618164,
      "learning_rate": 3.812855451587537e-05,
      "loss": 0.541,
      "step": 3586
    },
    {
      "epoch": 0.6223108952116586,
      "grad_norm": 0.6838366985321045,
      "learning_rate": 3.812567581040155e-05,
      "loss": 0.6431,
      "step": 3587
    },
    {
      "epoch": 0.6224843858431645,
      "grad_norm": 0.8503236174583435,
      "learning_rate": 3.812279500142194e-05,
      "loss": 0.6522,
      "step": 3588
    },
    {
      "epoch": 0.6226578764746704,
      "grad_norm": 0.882392406463623,
      "learning_rate": 3.811991208927085e-05,
      "loss": 0.5879,
      "step": 3589
    },
    {
      "epoch": 0.6228313671061763,
      "grad_norm": 0.8290567994117737,
      "learning_rate": 3.811702707428285e-05,
      "loss": 0.6743,
      "step": 3590
    },
    {
      "epoch": 0.6230048577376822,
      "grad_norm": 0.9682883620262146,
      "learning_rate": 3.811413995679275e-05,
      "loss": 0.5339,
      "step": 3591
    },
    {
      "epoch": 0.6231783483691881,
      "grad_norm": 0.7133485078811646,
      "learning_rate": 3.81112507371356e-05,
      "loss": 0.5846,
      "step": 3592
    },
    {
      "epoch": 0.6233518390006939,
      "grad_norm": 0.6190349459648132,
      "learning_rate": 3.8108359415646694e-05,
      "loss": 0.6305,
      "step": 3593
    },
    {
      "epoch": 0.6235253296321999,
      "grad_norm": 0.5973116755485535,
      "learning_rate": 3.810546599266158e-05,
      "loss": 0.5449,
      "step": 3594
    },
    {
      "epoch": 0.6236988202637057,
      "grad_norm": 0.7279934287071228,
      "learning_rate": 3.810257046851604e-05,
      "loss": 0.4998,
      "step": 3595
    },
    {
      "epoch": 0.6238723108952117,
      "grad_norm": 0.8145825862884521,
      "learning_rate": 3.8099672843546106e-05,
      "loss": 0.5234,
      "step": 3596
    },
    {
      "epoch": 0.6240458015267175,
      "grad_norm": 0.8200207352638245,
      "learning_rate": 3.8096773118088045e-05,
      "loss": 0.5153,
      "step": 3597
    },
    {
      "epoch": 0.6242192921582235,
      "grad_norm": 0.8757025599479675,
      "learning_rate": 3.809387129247838e-05,
      "loss": 0.5671,
      "step": 3598
    },
    {
      "epoch": 0.6243927827897293,
      "grad_norm": 0.7421393394470215,
      "learning_rate": 3.809096736705387e-05,
      "loss": 0.6555,
      "step": 3599
    },
    {
      "epoch": 0.6245662734212353,
      "grad_norm": 0.8077309131622314,
      "learning_rate": 3.808806134215151e-05,
      "loss": 0.6855,
      "step": 3600
    },
    {
      "epoch": 0.6247397640527411,
      "grad_norm": 0.7182208299636841,
      "learning_rate": 3.8085153218108555e-05,
      "loss": 0.6501,
      "step": 3601
    },
    {
      "epoch": 0.6249132546842471,
      "grad_norm": 0.8924763202667236,
      "learning_rate": 3.80822429952625e-05,
      "loss": 0.5684,
      "step": 3602
    },
    {
      "epoch": 0.6250867453157529,
      "grad_norm": 0.6226588487625122,
      "learning_rate": 3.807933067395108e-05,
      "loss": 0.6013,
      "step": 3603
    },
    {
      "epoch": 0.6252602359472589,
      "grad_norm": 0.7973635196685791,
      "learning_rate": 3.8076416254512256e-05,
      "loss": 0.701,
      "step": 3604
    },
    {
      "epoch": 0.6254337265787647,
      "grad_norm": 0.5976009368896484,
      "learning_rate": 3.807349973728427e-05,
      "loss": 0.5522,
      "step": 3605
    },
    {
      "epoch": 0.6256072172102707,
      "grad_norm": 0.7112151384353638,
      "learning_rate": 3.807058112260558e-05,
      "loss": 0.6797,
      "step": 3606
    },
    {
      "epoch": 0.6257807078417765,
      "grad_norm": 0.8459869623184204,
      "learning_rate": 3.8067660410814895e-05,
      "loss": 0.6395,
      "step": 3607
    },
    {
      "epoch": 0.6259541984732825,
      "grad_norm": 0.9730987548828125,
      "learning_rate": 3.8064737602251155e-05,
      "loss": 0.6216,
      "step": 3608
    },
    {
      "epoch": 0.6261276891047883,
      "grad_norm": 0.7733995318412781,
      "learning_rate": 3.8061812697253576e-05,
      "loss": 0.6182,
      "step": 3609
    },
    {
      "epoch": 0.6263011797362943,
      "grad_norm": 0.6619139313697815,
      "learning_rate": 3.8058885696161595e-05,
      "loss": 0.5614,
      "step": 3610
    },
    {
      "epoch": 0.6264746703678001,
      "grad_norm": 0.7139784693717957,
      "learning_rate": 3.805595659931487e-05,
      "loss": 0.6194,
      "step": 3611
    },
    {
      "epoch": 0.6266481609993061,
      "grad_norm": 0.7814990282058716,
      "learning_rate": 3.805302540705335e-05,
      "loss": 0.6265,
      "step": 3612
    },
    {
      "epoch": 0.6268216516308119,
      "grad_norm": 0.7852691411972046,
      "learning_rate": 3.80500921197172e-05,
      "loss": 0.5624,
      "step": 3613
    },
    {
      "epoch": 0.6269951422623178,
      "grad_norm": 1.1444884538650513,
      "learning_rate": 3.8047156737646825e-05,
      "loss": 0.6121,
      "step": 3614
    },
    {
      "epoch": 0.6271686328938237,
      "grad_norm": 0.6964013576507568,
      "learning_rate": 3.8044219261182876e-05,
      "loss": 0.6388,
      "step": 3615
    },
    {
      "epoch": 0.6273421235253296,
      "grad_norm": 0.7965458035469055,
      "learning_rate": 3.8041279690666254e-05,
      "loss": 0.5717,
      "step": 3616
    },
    {
      "epoch": 0.6275156141568355,
      "grad_norm": 0.8811926245689392,
      "learning_rate": 3.8038338026438116e-05,
      "loss": 0.5613,
      "step": 3617
    },
    {
      "epoch": 0.6276891047883414,
      "grad_norm": 0.8348861932754517,
      "learning_rate": 3.803539426883982e-05,
      "loss": 0.6429,
      "step": 3618
    },
    {
      "epoch": 0.6278625954198473,
      "grad_norm": 0.8193329572677612,
      "learning_rate": 3.803244841821301e-05,
      "loss": 0.571,
      "step": 3619
    },
    {
      "epoch": 0.6280360860513532,
      "grad_norm": 0.7115561366081238,
      "learning_rate": 3.8029500474899544e-05,
      "loss": 0.5219,
      "step": 3620
    },
    {
      "epoch": 0.6282095766828591,
      "grad_norm": 0.7684637904167175,
      "learning_rate": 3.8026550439241535e-05,
      "loss": 0.6139,
      "step": 3621
    },
    {
      "epoch": 0.628383067314365,
      "grad_norm": 0.6806290745735168,
      "learning_rate": 3.802359831158135e-05,
      "loss": 0.5044,
      "step": 3622
    },
    {
      "epoch": 0.628556557945871,
      "grad_norm": 0.7397187352180481,
      "learning_rate": 3.802064409226158e-05,
      "loss": 0.5122,
      "step": 3623
    },
    {
      "epoch": 0.6287300485773768,
      "grad_norm": 2.4987173080444336,
      "learning_rate": 3.801768778162506e-05,
      "loss": 0.5508,
      "step": 3624
    },
    {
      "epoch": 0.6289035392088828,
      "grad_norm": 0.8442872762680054,
      "learning_rate": 3.801472938001488e-05,
      "loss": 0.5547,
      "step": 3625
    },
    {
      "epoch": 0.6290770298403886,
      "grad_norm": 0.7826525568962097,
      "learning_rate": 3.8011768887774365e-05,
      "loss": 0.7209,
      "step": 3626
    },
    {
      "epoch": 0.6292505204718946,
      "grad_norm": 0.8531131744384766,
      "learning_rate": 3.8008806305247083e-05,
      "loss": 0.5614,
      "step": 3627
    },
    {
      "epoch": 0.6294240111034004,
      "grad_norm": 1.044183373451233,
      "learning_rate": 3.800584163277684e-05,
      "loss": 0.6444,
      "step": 3628
    },
    {
      "epoch": 0.6295975017349064,
      "grad_norm": 0.8884148001670837,
      "learning_rate": 3.80028748707077e-05,
      "loss": 0.5234,
      "step": 3629
    },
    {
      "epoch": 0.6297709923664122,
      "grad_norm": 0.9937842488288879,
      "learning_rate": 3.7999906019383954e-05,
      "loss": 0.5386,
      "step": 3630
    },
    {
      "epoch": 0.6299444829979182,
      "grad_norm": 0.8778430819511414,
      "learning_rate": 3.799693507915014e-05,
      "loss": 0.6859,
      "step": 3631
    },
    {
      "epoch": 0.630117973629424,
      "grad_norm": 0.9054518938064575,
      "learning_rate": 3.799396205035104e-05,
      "loss": 0.4907,
      "step": 3632
    },
    {
      "epoch": 0.6302914642609299,
      "grad_norm": 0.9029878377914429,
      "learning_rate": 3.799098693333167e-05,
      "loss": 0.526,
      "step": 3633
    },
    {
      "epoch": 0.6304649548924358,
      "grad_norm": 1.7739448547363281,
      "learning_rate": 3.7988009728437304e-05,
      "loss": 0.4902,
      "step": 3634
    },
    {
      "epoch": 0.6306384455239417,
      "grad_norm": 0.6787098050117493,
      "learning_rate": 3.7985030436013454e-05,
      "loss": 0.5785,
      "step": 3635
    },
    {
      "epoch": 0.6308119361554476,
      "grad_norm": 0.8755621314048767,
      "learning_rate": 3.7982049056405866e-05,
      "loss": 0.5568,
      "step": 3636
    },
    {
      "epoch": 0.6309854267869535,
      "grad_norm": 0.662980318069458,
      "learning_rate": 3.797906558996053e-05,
      "loss": 0.5785,
      "step": 3637
    },
    {
      "epoch": 0.6311589174184594,
      "grad_norm": 0.9010483622550964,
      "learning_rate": 3.797608003702368e-05,
      "loss": 0.5588,
      "step": 3638
    },
    {
      "epoch": 0.6313324080499653,
      "grad_norm": 1.4660987854003906,
      "learning_rate": 3.79730923979418e-05,
      "loss": 0.515,
      "step": 3639
    },
    {
      "epoch": 0.6315058986814712,
      "grad_norm": 0.8304885625839233,
      "learning_rate": 3.79701026730616e-05,
      "loss": 0.574,
      "step": 3640
    },
    {
      "epoch": 0.6316793893129771,
      "grad_norm": 0.8260068893432617,
      "learning_rate": 3.7967110862730045e-05,
      "loss": 0.5349,
      "step": 3641
    },
    {
      "epoch": 0.631852879944483,
      "grad_norm": 0.65542072057724,
      "learning_rate": 3.796411696729434e-05,
      "loss": 0.6261,
      "step": 3642
    },
    {
      "epoch": 0.6320263705759889,
      "grad_norm": 0.7199093699455261,
      "learning_rate": 3.7961120987101933e-05,
      "loss": 0.7023,
      "step": 3643
    },
    {
      "epoch": 0.6321998612074948,
      "grad_norm": 0.9273139238357544,
      "learning_rate": 3.79581229225005e-05,
      "loss": 0.5454,
      "step": 3644
    },
    {
      "epoch": 0.6323733518390007,
      "grad_norm": 0.7068122029304504,
      "learning_rate": 3.795512277383798e-05,
      "loss": 0.6456,
      "step": 3645
    },
    {
      "epoch": 0.6325468424705066,
      "grad_norm": 0.6215282678604126,
      "learning_rate": 3.795212054146254e-05,
      "loss": 0.6597,
      "step": 3646
    },
    {
      "epoch": 0.6327203331020125,
      "grad_norm": 1.4631701707839966,
      "learning_rate": 3.794911622572259e-05,
      "loss": 0.6924,
      "step": 3647
    },
    {
      "epoch": 0.6328938237335184,
      "grad_norm": 1.1630256175994873,
      "learning_rate": 3.794610982696679e-05,
      "loss": 0.4807,
      "step": 3648
    },
    {
      "epoch": 0.6330673143650243,
      "grad_norm": 0.5245219469070435,
      "learning_rate": 3.794310134554403e-05,
      "loss": 0.6442,
      "step": 3649
    },
    {
      "epoch": 0.6332408049965302,
      "grad_norm": 0.7798471450805664,
      "learning_rate": 3.7940090781803454e-05,
      "loss": 0.5171,
      "step": 3650
    },
    {
      "epoch": 0.6334142956280361,
      "grad_norm": 1.0232371091842651,
      "learning_rate": 3.793707813609444e-05,
      "loss": 0.5538,
      "step": 3651
    },
    {
      "epoch": 0.6335877862595419,
      "grad_norm": 1.0044523477554321,
      "learning_rate": 3.7934063408766606e-05,
      "loss": 0.6938,
      "step": 3652
    },
    {
      "epoch": 0.6337612768910479,
      "grad_norm": 0.8687499165534973,
      "learning_rate": 3.7931046600169815e-05,
      "loss": 0.6128,
      "step": 3653
    },
    {
      "epoch": 0.6339347675225537,
      "grad_norm": 0.7496787309646606,
      "learning_rate": 3.792802771065417e-05,
      "loss": 0.5076,
      "step": 3654
    },
    {
      "epoch": 0.6341082581540597,
      "grad_norm": 0.714073657989502,
      "learning_rate": 3.792500674057002e-05,
      "loss": 0.6096,
      "step": 3655
    },
    {
      "epoch": 0.6342817487855655,
      "grad_norm": 0.9160600304603577,
      "learning_rate": 3.792198369026796e-05,
      "loss": 0.5463,
      "step": 3656
    },
    {
      "epoch": 0.6344552394170715,
      "grad_norm": 1.1145319938659668,
      "learning_rate": 3.79189585600988e-05,
      "loss": 0.63,
      "step": 3657
    },
    {
      "epoch": 0.6346287300485773,
      "grad_norm": 1.1254013776779175,
      "learning_rate": 3.791593135041362e-05,
      "loss": 0.7301,
      "step": 3658
    },
    {
      "epoch": 0.6348022206800833,
      "grad_norm": 0.9876613020896912,
      "learning_rate": 3.791290206156373e-05,
      "loss": 0.6985,
      "step": 3659
    },
    {
      "epoch": 0.6349757113115891,
      "grad_norm": 0.6043176651000977,
      "learning_rate": 3.790987069390069e-05,
      "loss": 0.6089,
      "step": 3660
    },
    {
      "epoch": 0.6351492019430951,
      "grad_norm": 0.5808443427085876,
      "learning_rate": 3.790683724777628e-05,
      "loss": 0.6603,
      "step": 3661
    },
    {
      "epoch": 0.635322692574601,
      "grad_norm": 0.9439184665679932,
      "learning_rate": 3.790380172354255e-05,
      "loss": 0.5812,
      "step": 3662
    },
    {
      "epoch": 0.6354961832061069,
      "grad_norm": 0.6527266502380371,
      "learning_rate": 3.790076412155176e-05,
      "loss": 0.6115,
      "step": 3663
    },
    {
      "epoch": 0.6356696738376127,
      "grad_norm": 0.7660706639289856,
      "learning_rate": 3.789772444215644e-05,
      "loss": 0.6599,
      "step": 3664
    },
    {
      "epoch": 0.6358431644691187,
      "grad_norm": 0.9501388072967529,
      "learning_rate": 3.7894682685709335e-05,
      "loss": 0.749,
      "step": 3665
    },
    {
      "epoch": 0.6360166551006246,
      "grad_norm": 1.4506312608718872,
      "learning_rate": 3.7891638852563455e-05,
      "loss": 0.494,
      "step": 3666
    },
    {
      "epoch": 0.6361901457321305,
      "grad_norm": 0.711625337600708,
      "learning_rate": 3.788859294307204e-05,
      "loss": 0.6405,
      "step": 3667
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.6802700161933899,
      "learning_rate": 3.788554495758858e-05,
      "loss": 0.5615,
      "step": 3668
    },
    {
      "epoch": 0.6365371269951423,
      "grad_norm": 0.7881873846054077,
      "learning_rate": 3.788249489646677e-05,
      "loss": 0.5312,
      "step": 3669
    },
    {
      "epoch": 0.6367106176266482,
      "grad_norm": 0.7416797876358032,
      "learning_rate": 3.7879442760060604e-05,
      "loss": 0.5347,
      "step": 3670
    },
    {
      "epoch": 0.6368841082581541,
      "grad_norm": 0.6100966334342957,
      "learning_rate": 3.7876388548724256e-05,
      "loss": 0.6012,
      "step": 3671
    },
    {
      "epoch": 0.63705759888966,
      "grad_norm": 0.918250560760498,
      "learning_rate": 3.78733322628122e-05,
      "loss": 0.5327,
      "step": 3672
    },
    {
      "epoch": 0.6372310895211658,
      "grad_norm": 0.8421688079833984,
      "learning_rate": 3.787027390267911e-05,
      "loss": 0.6249,
      "step": 3673
    },
    {
      "epoch": 0.6374045801526718,
      "grad_norm": 0.6886511445045471,
      "learning_rate": 3.786721346867991e-05,
      "loss": 0.6,
      "step": 3674
    },
    {
      "epoch": 0.6375780707841776,
      "grad_norm": 0.7019809484481812,
      "learning_rate": 3.786415096116976e-05,
      "loss": 0.6454,
      "step": 3675
    },
    {
      "epoch": 0.6377515614156836,
      "grad_norm": 0.7325215339660645,
      "learning_rate": 3.786108638050408e-05,
      "loss": 0.6024,
      "step": 3676
    },
    {
      "epoch": 0.6379250520471894,
      "grad_norm": 0.6856330037117004,
      "learning_rate": 3.785801972703851e-05,
      "loss": 0.6418,
      "step": 3677
    },
    {
      "epoch": 0.6380985426786954,
      "grad_norm": 0.5166242122650146,
      "learning_rate": 3.785495100112894e-05,
      "loss": 0.7063,
      "step": 3678
    },
    {
      "epoch": 0.6382720333102012,
      "grad_norm": 0.6169303059577942,
      "learning_rate": 3.7851880203131506e-05,
      "loss": 0.7209,
      "step": 3679
    },
    {
      "epoch": 0.6384455239417072,
      "grad_norm": 0.7382290959358215,
      "learning_rate": 3.784880733340257e-05,
      "loss": 0.7397,
      "step": 3680
    },
    {
      "epoch": 0.638619014573213,
      "grad_norm": 0.6344757080078125,
      "learning_rate": 3.7845732392298746e-05,
      "loss": 0.6554,
      "step": 3681
    },
    {
      "epoch": 0.638792505204719,
      "grad_norm": 1.1019749641418457,
      "learning_rate": 3.784265538017689e-05,
      "loss": 0.5975,
      "step": 3682
    },
    {
      "epoch": 0.6389659958362248,
      "grad_norm": 0.8122680187225342,
      "learning_rate": 3.783957629739408e-05,
      "loss": 0.6364,
      "step": 3683
    },
    {
      "epoch": 0.6391394864677308,
      "grad_norm": 0.7165302038192749,
      "learning_rate": 3.7836495144307644e-05,
      "loss": 0.5143,
      "step": 3684
    },
    {
      "epoch": 0.6393129770992366,
      "grad_norm": 0.6541818976402283,
      "learning_rate": 3.783341192127518e-05,
      "loss": 0.5791,
      "step": 3685
    },
    {
      "epoch": 0.6394864677307426,
      "grad_norm": 0.5491472482681274,
      "learning_rate": 3.783032662865447e-05,
      "loss": 0.5684,
      "step": 3686
    },
    {
      "epoch": 0.6396599583622484,
      "grad_norm": 2.529473066329956,
      "learning_rate": 3.7827239266803584e-05,
      "loss": 0.5333,
      "step": 3687
    },
    {
      "epoch": 0.6398334489937544,
      "grad_norm": 0.7565951943397522,
      "learning_rate": 3.782414983608081e-05,
      "loss": 0.5792,
      "step": 3688
    },
    {
      "epoch": 0.6400069396252602,
      "grad_norm": 0.8166025280952454,
      "learning_rate": 3.7821058336844676e-05,
      "loss": 0.5707,
      "step": 3689
    },
    {
      "epoch": 0.6401804302567662,
      "grad_norm": 0.6751120090484619,
      "learning_rate": 3.7817964769453956e-05,
      "loss": 0.6344,
      "step": 3690
    },
    {
      "epoch": 0.640353920888272,
      "grad_norm": 0.6678987145423889,
      "learning_rate": 3.781486913426766e-05,
      "loss": 0.6392,
      "step": 3691
    },
    {
      "epoch": 0.6405274115197779,
      "grad_norm": 0.7985641360282898,
      "learning_rate": 3.781177143164505e-05,
      "loss": 0.5919,
      "step": 3692
    },
    {
      "epoch": 0.6407009021512838,
      "grad_norm": 0.740614116191864,
      "learning_rate": 3.7808671661945606e-05,
      "loss": 0.5917,
      "step": 3693
    },
    {
      "epoch": 0.6408743927827897,
      "grad_norm": 0.5903328657150269,
      "learning_rate": 3.7805569825529055e-05,
      "loss": 0.6633,
      "step": 3694
    },
    {
      "epoch": 0.6410478834142956,
      "grad_norm": 0.6017022132873535,
      "learning_rate": 3.780246592275539e-05,
      "loss": 0.7225,
      "step": 3695
    },
    {
      "epoch": 0.6412213740458015,
      "grad_norm": 0.919784426689148,
      "learning_rate": 3.779935995398481e-05,
      "loss": 0.5295,
      "step": 3696
    },
    {
      "epoch": 0.6413948646773074,
      "grad_norm": 0.5991265773773193,
      "learning_rate": 3.7796251919577764e-05,
      "loss": 0.6859,
      "step": 3697
    },
    {
      "epoch": 0.6415683553088133,
      "grad_norm": 0.6057894825935364,
      "learning_rate": 3.7793141819894955e-05,
      "loss": 0.6577,
      "step": 3698
    },
    {
      "epoch": 0.6417418459403192,
      "grad_norm": 0.7780317664146423,
      "learning_rate": 3.779002965529729e-05,
      "loss": 0.5516,
      "step": 3699
    },
    {
      "epoch": 0.6419153365718251,
      "grad_norm": 1.1641170978546143,
      "learning_rate": 3.778691542614596e-05,
      "loss": 0.6179,
      "step": 3700
    },
    {
      "epoch": 0.642088827203331,
      "grad_norm": 0.7804569602012634,
      "learning_rate": 3.7783799132802365e-05,
      "loss": 0.6097,
      "step": 3701
    },
    {
      "epoch": 0.6422623178348369,
      "grad_norm": 0.7090769410133362,
      "learning_rate": 3.778068077562817e-05,
      "loss": 0.6624,
      "step": 3702
    },
    {
      "epoch": 0.6424358084663429,
      "grad_norm": 0.6512264609336853,
      "learning_rate": 3.7777560354985246e-05,
      "loss": 0.6213,
      "step": 3703
    },
    {
      "epoch": 0.6426092990978487,
      "grad_norm": 0.7044447660446167,
      "learning_rate": 3.7774437871235724e-05,
      "loss": 0.5912,
      "step": 3704
    },
    {
      "epoch": 0.6427827897293547,
      "grad_norm": 0.7511945366859436,
      "learning_rate": 3.7771313324741974e-05,
      "loss": 0.5898,
      "step": 3705
    },
    {
      "epoch": 0.6429562803608605,
      "grad_norm": 1.123663306236267,
      "learning_rate": 3.776818671586662e-05,
      "loss": 0.568,
      "step": 3706
    },
    {
      "epoch": 0.6431297709923665,
      "grad_norm": 0.8128359913825989,
      "learning_rate": 3.776505804497248e-05,
      "loss": 0.5587,
      "step": 3707
    },
    {
      "epoch": 0.6433032616238723,
      "grad_norm": 0.7907178997993469,
      "learning_rate": 3.776192731242265e-05,
      "loss": 0.6127,
      "step": 3708
    },
    {
      "epoch": 0.6434767522553783,
      "grad_norm": 0.7276610136032104,
      "learning_rate": 3.775879451858047e-05,
      "loss": 0.5898,
      "step": 3709
    },
    {
      "epoch": 0.6436502428868841,
      "grad_norm": 0.6831201910972595,
      "learning_rate": 3.775565966380949e-05,
      "loss": 0.547,
      "step": 3710
    },
    {
      "epoch": 0.6438237335183901,
      "grad_norm": 0.752452552318573,
      "learning_rate": 3.7752522748473517e-05,
      "loss": 0.6072,
      "step": 3711
    },
    {
      "epoch": 0.6439972241498959,
      "grad_norm": 0.9028571248054504,
      "learning_rate": 3.774938377293659e-05,
      "loss": 0.6353,
      "step": 3712
    },
    {
      "epoch": 0.6441707147814018,
      "grad_norm": 0.7592943906784058,
      "learning_rate": 3.774624273756299e-05,
      "loss": 0.5831,
      "step": 3713
    },
    {
      "epoch": 0.6443442054129077,
      "grad_norm": 2.0297415256500244,
      "learning_rate": 3.774309964271725e-05,
      "loss": 0.6412,
      "step": 3714
    },
    {
      "epoch": 0.6445176960444136,
      "grad_norm": 0.8096716403961182,
      "learning_rate": 3.773995448876412e-05,
      "loss": 0.6069,
      "step": 3715
    },
    {
      "epoch": 0.6446911866759195,
      "grad_norm": 0.7666990160942078,
      "learning_rate": 3.7736807276068604e-05,
      "loss": 0.6991,
      "step": 3716
    },
    {
      "epoch": 0.6448646773074254,
      "grad_norm": 0.6890394687652588,
      "learning_rate": 3.773365800499592e-05,
      "loss": 0.6167,
      "step": 3717
    },
    {
      "epoch": 0.6450381679389313,
      "grad_norm": 0.6246633529663086,
      "learning_rate": 3.773050667591158e-05,
      "loss": 0.5603,
      "step": 3718
    },
    {
      "epoch": 0.6452116585704372,
      "grad_norm": 0.950532853603363,
      "learning_rate": 3.772735328918127e-05,
      "loss": 0.5162,
      "step": 3719
    },
    {
      "epoch": 0.6453851492019431,
      "grad_norm": 0.8982919454574585,
      "learning_rate": 3.772419784517095e-05,
      "loss": 0.6329,
      "step": 3720
    },
    {
      "epoch": 0.645558639833449,
      "grad_norm": 1.2791414260864258,
      "learning_rate": 3.7721040344246824e-05,
      "loss": 0.543,
      "step": 3721
    },
    {
      "epoch": 0.6457321304649549,
      "grad_norm": 0.6306223273277283,
      "learning_rate": 3.771788078677532e-05,
      "loss": 0.5756,
      "step": 3722
    },
    {
      "epoch": 0.6459056210964608,
      "grad_norm": 0.8798757195472717,
      "learning_rate": 3.7714719173123104e-05,
      "loss": 0.4753,
      "step": 3723
    },
    {
      "epoch": 0.6460791117279667,
      "grad_norm": 0.6941823959350586,
      "learning_rate": 3.771155550365708e-05,
      "loss": 0.5959,
      "step": 3724
    },
    {
      "epoch": 0.6462526023594726,
      "grad_norm": 0.70530766248703,
      "learning_rate": 3.770838977874441e-05,
      "loss": 0.6239,
      "step": 3725
    },
    {
      "epoch": 0.6464260929909785,
      "grad_norm": 0.8281103372573853,
      "learning_rate": 3.770522199875247e-05,
      "loss": 0.5128,
      "step": 3726
    },
    {
      "epoch": 0.6465995836224844,
      "grad_norm": 1.0173561573028564,
      "learning_rate": 3.770205216404888e-05,
      "loss": 0.6163,
      "step": 3727
    },
    {
      "epoch": 0.6467730742539903,
      "grad_norm": 0.9786710143089294,
      "learning_rate": 3.7698880275001516e-05,
      "loss": 0.4941,
      "step": 3728
    },
    {
      "epoch": 0.6469465648854962,
      "grad_norm": 0.6810334920883179,
      "learning_rate": 3.7695706331978465e-05,
      "loss": 0.6027,
      "step": 3729
    },
    {
      "epoch": 0.6471200555170021,
      "grad_norm": 0.8536402583122253,
      "learning_rate": 3.769253033534808e-05,
      "loss": 0.6367,
      "step": 3730
    },
    {
      "epoch": 0.647293546148508,
      "grad_norm": 0.8468443751335144,
      "learning_rate": 3.768935228547894e-05,
      "loss": 0.625,
      "step": 3731
    },
    {
      "epoch": 0.6474670367800138,
      "grad_norm": 0.7573752403259277,
      "learning_rate": 3.7686172182739845e-05,
      "loss": 0.5385,
      "step": 3732
    },
    {
      "epoch": 0.6476405274115198,
      "grad_norm": 0.7154396176338196,
      "learning_rate": 3.7682990027499864e-05,
      "loss": 0.6071,
      "step": 3733
    },
    {
      "epoch": 0.6478140180430256,
      "grad_norm": 1.3461333513259888,
      "learning_rate": 3.767980582012828e-05,
      "loss": 0.6571,
      "step": 3734
    },
    {
      "epoch": 0.6479875086745316,
      "grad_norm": 0.6264582276344299,
      "learning_rate": 3.767661956099464e-05,
      "loss": 0.6262,
      "step": 3735
    },
    {
      "epoch": 0.6481609993060374,
      "grad_norm": 0.6777341961860657,
      "learning_rate": 3.7673431250468695e-05,
      "loss": 0.6062,
      "step": 3736
    },
    {
      "epoch": 0.6483344899375434,
      "grad_norm": 1.1197324991226196,
      "learning_rate": 3.767024088892046e-05,
      "loss": 0.5386,
      "step": 3737
    },
    {
      "epoch": 0.6485079805690492,
      "grad_norm": 0.6859471797943115,
      "learning_rate": 3.766704847672018e-05,
      "loss": 0.6283,
      "step": 3738
    },
    {
      "epoch": 0.6486814712005552,
      "grad_norm": 0.9123736619949341,
      "learning_rate": 3.7663854014238344e-05,
      "loss": 0.5356,
      "step": 3739
    },
    {
      "epoch": 0.648854961832061,
      "grad_norm": 1.0681298971176147,
      "learning_rate": 3.766065750184566e-05,
      "loss": 0.5234,
      "step": 3740
    },
    {
      "epoch": 0.649028452463567,
      "grad_norm": 0.7163435220718384,
      "learning_rate": 3.765745893991309e-05,
      "loss": 0.4725,
      "step": 3741
    },
    {
      "epoch": 0.6492019430950728,
      "grad_norm": 1.132663369178772,
      "learning_rate": 3.7654258328811856e-05,
      "loss": 0.5359,
      "step": 3742
    },
    {
      "epoch": 0.6493754337265788,
      "grad_norm": 1.1401888132095337,
      "learning_rate": 3.765105566891335e-05,
      "loss": 0.516,
      "step": 3743
    },
    {
      "epoch": 0.6495489243580846,
      "grad_norm": 0.5803067088127136,
      "learning_rate": 3.764785096058927e-05,
      "loss": 0.5691,
      "step": 3744
    },
    {
      "epoch": 0.6497224149895906,
      "grad_norm": 2.6133241653442383,
      "learning_rate": 3.764464420421153e-05,
      "loss": 0.6187,
      "step": 3745
    },
    {
      "epoch": 0.6498959056210964,
      "grad_norm": 0.6367717385292053,
      "learning_rate": 3.764143540015227e-05,
      "loss": 0.6381,
      "step": 3746
    },
    {
      "epoch": 0.6500693962526024,
      "grad_norm": 0.5755173563957214,
      "learning_rate": 3.763822454878387e-05,
      "loss": 0.6499,
      "step": 3747
    },
    {
      "epoch": 0.6502428868841083,
      "grad_norm": 0.6816663146018982,
      "learning_rate": 3.763501165047896e-05,
      "loss": 0.6312,
      "step": 3748
    },
    {
      "epoch": 0.6504163775156142,
      "grad_norm": 0.7217597961425781,
      "learning_rate": 3.76317967056104e-05,
      "loss": 0.488,
      "step": 3749
    },
    {
      "epoch": 0.65058986814712,
      "grad_norm": 0.7203596234321594,
      "learning_rate": 3.7628579714551285e-05,
      "loss": 0.6619,
      "step": 3750
    },
    {
      "epoch": 0.6507633587786259,
      "grad_norm": 0.9814811944961548,
      "learning_rate": 3.762536067767495e-05,
      "loss": 0.7832,
      "step": 3751
    },
    {
      "epoch": 0.6509368494101319,
      "grad_norm": 0.719393253326416,
      "learning_rate": 3.7622139595354976e-05,
      "loss": 0.6853,
      "step": 3752
    },
    {
      "epoch": 0.6511103400416377,
      "grad_norm": 0.8610318303108215,
      "learning_rate": 3.761891646796517e-05,
      "loss": 0.5215,
      "step": 3753
    },
    {
      "epoch": 0.6512838306731437,
      "grad_norm": 1.0605509281158447,
      "learning_rate": 3.7615691295879574e-05,
      "loss": 0.5273,
      "step": 3754
    },
    {
      "epoch": 0.6514573213046495,
      "grad_norm": 0.6611588597297668,
      "learning_rate": 3.7612464079472474e-05,
      "loss": 0.6794,
      "step": 3755
    },
    {
      "epoch": 0.6516308119361555,
      "grad_norm": 0.7887797355651855,
      "learning_rate": 3.76092348191184e-05,
      "loss": 0.6007,
      "step": 3756
    },
    {
      "epoch": 0.6518043025676613,
      "grad_norm": 0.7849190831184387,
      "learning_rate": 3.7606003515192103e-05,
      "loss": 0.6798,
      "step": 3757
    },
    {
      "epoch": 0.6519777931991673,
      "grad_norm": 0.8868094086647034,
      "learning_rate": 3.7602770168068586e-05,
      "loss": 0.6073,
      "step": 3758
    },
    {
      "epoch": 0.6521512838306731,
      "grad_norm": 0.6700664162635803,
      "learning_rate": 3.7599534778123074e-05,
      "loss": 0.5991,
      "step": 3759
    },
    {
      "epoch": 0.6523247744621791,
      "grad_norm": 0.7111955285072327,
      "learning_rate": 3.759629734573105e-05,
      "loss": 0.6134,
      "step": 3760
    },
    {
      "epoch": 0.6524982650936849,
      "grad_norm": 0.5644627809524536,
      "learning_rate": 3.759305787126821e-05,
      "loss": 0.6559,
      "step": 3761
    },
    {
      "epoch": 0.6526717557251909,
      "grad_norm": 0.6465210914611816,
      "learning_rate": 3.758981635511051e-05,
      "loss": 0.6991,
      "step": 3762
    },
    {
      "epoch": 0.6528452463566967,
      "grad_norm": 0.8938121795654297,
      "learning_rate": 3.758657279763412e-05,
      "loss": 0.7344,
      "step": 3763
    },
    {
      "epoch": 0.6530187369882027,
      "grad_norm": 0.8159437775611877,
      "learning_rate": 3.758332719921547e-05,
      "loss": 0.6278,
      "step": 3764
    },
    {
      "epoch": 0.6531922276197085,
      "grad_norm": 0.661339282989502,
      "learning_rate": 3.758007956023121e-05,
      "loss": 0.5828,
      "step": 3765
    },
    {
      "epoch": 0.6533657182512145,
      "grad_norm": 0.5961443185806274,
      "learning_rate": 3.757682988105823e-05,
      "loss": 0.6519,
      "step": 3766
    },
    {
      "epoch": 0.6535392088827203,
      "grad_norm": 0.7474210262298584,
      "learning_rate": 3.757357816207366e-05,
      "loss": 0.5505,
      "step": 3767
    },
    {
      "epoch": 0.6537126995142263,
      "grad_norm": 0.5833430290222168,
      "learning_rate": 3.7570324403654866e-05,
      "loss": 0.5767,
      "step": 3768
    },
    {
      "epoch": 0.6538861901457321,
      "grad_norm": 0.8346602320671082,
      "learning_rate": 3.756706860617945e-05,
      "loss": 0.6614,
      "step": 3769
    },
    {
      "epoch": 0.6540596807772381,
      "grad_norm": 0.6843047738075256,
      "learning_rate": 3.756381077002526e-05,
      "loss": 0.6467,
      "step": 3770
    },
    {
      "epoch": 0.6542331714087439,
      "grad_norm": 0.7140724062919617,
      "learning_rate": 3.756055089557036e-05,
      "loss": 0.5745,
      "step": 3771
    },
    {
      "epoch": 0.6544066620402498,
      "grad_norm": 0.7032603621482849,
      "learning_rate": 3.755728898319306e-05,
      "loss": 0.6626,
      "step": 3772
    },
    {
      "epoch": 0.6545801526717557,
      "grad_norm": 1.0892679691314697,
      "learning_rate": 3.7554025033271923e-05,
      "loss": 0.5334,
      "step": 3773
    },
    {
      "epoch": 0.6547536433032616,
      "grad_norm": 0.6795757412910461,
      "learning_rate": 3.7550759046185726e-05,
      "loss": 0.5492,
      "step": 3774
    },
    {
      "epoch": 0.6549271339347675,
      "grad_norm": 0.7899928092956543,
      "learning_rate": 3.754749102231349e-05,
      "loss": 0.5258,
      "step": 3775
    },
    {
      "epoch": 0.6551006245662734,
      "grad_norm": 0.9099841117858887,
      "learning_rate": 3.7544220962034475e-05,
      "loss": 0.5477,
      "step": 3776
    },
    {
      "epoch": 0.6552741151977793,
      "grad_norm": 0.7497731447219849,
      "learning_rate": 3.7540948865728174e-05,
      "loss": 0.6212,
      "step": 3777
    },
    {
      "epoch": 0.6554476058292852,
      "grad_norm": 0.724162220954895,
      "learning_rate": 3.7537674733774315e-05,
      "loss": 0.6899,
      "step": 3778
    },
    {
      "epoch": 0.6556210964607911,
      "grad_norm": 0.7789028882980347,
      "learning_rate": 3.7534398566552866e-05,
      "loss": 0.6855,
      "step": 3779
    },
    {
      "epoch": 0.655794587092297,
      "grad_norm": 0.6694778800010681,
      "learning_rate": 3.753112036444404e-05,
      "loss": 0.5677,
      "step": 3780
    },
    {
      "epoch": 0.655968077723803,
      "grad_norm": 0.671506404876709,
      "learning_rate": 3.7527840127828256e-05,
      "loss": 0.5542,
      "step": 3781
    },
    {
      "epoch": 0.6561415683553088,
      "grad_norm": 0.9021669626235962,
      "learning_rate": 3.752455785708622e-05,
      "loss": 0.7089,
      "step": 3782
    },
    {
      "epoch": 0.6563150589868147,
      "grad_norm": 0.925146222114563,
      "learning_rate": 3.752127355259881e-05,
      "loss": 0.5023,
      "step": 3783
    },
    {
      "epoch": 0.6564885496183206,
      "grad_norm": 0.7789936065673828,
      "learning_rate": 3.7517987214747186e-05,
      "loss": 0.4883,
      "step": 3784
    },
    {
      "epoch": 0.6566620402498266,
      "grad_norm": 2.8964085578918457,
      "learning_rate": 3.751469884391274e-05,
      "loss": 0.5497,
      "step": 3785
    },
    {
      "epoch": 0.6568355308813324,
      "grad_norm": 0.6731738448143005,
      "learning_rate": 3.751140844047708e-05,
      "loss": 0.5436,
      "step": 3786
    },
    {
      "epoch": 0.6570090215128384,
      "grad_norm": 0.6144465804100037,
      "learning_rate": 3.750811600482207e-05,
      "loss": 0.6362,
      "step": 3787
    },
    {
      "epoch": 0.6571825121443442,
      "grad_norm": 1.0024034976959229,
      "learning_rate": 3.7504821537329795e-05,
      "loss": 0.6335,
      "step": 3788
    },
    {
      "epoch": 0.6573560027758502,
      "grad_norm": 0.6064333915710449,
      "learning_rate": 3.750152503838258e-05,
      "loss": 0.6765,
      "step": 3789
    },
    {
      "epoch": 0.657529493407356,
      "grad_norm": 0.8006238341331482,
      "learning_rate": 3.7498226508362996e-05,
      "loss": 0.5896,
      "step": 3790
    },
    {
      "epoch": 0.6577029840388618,
      "grad_norm": 0.9184333682060242,
      "learning_rate": 3.7494925947653835e-05,
      "loss": 0.5553,
      "step": 3791
    },
    {
      "epoch": 0.6578764746703678,
      "grad_norm": 0.774874746799469,
      "learning_rate": 3.749162335663813e-05,
      "loss": 0.7189,
      "step": 3792
    },
    {
      "epoch": 0.6580499653018737,
      "grad_norm": 0.68658846616745,
      "learning_rate": 3.7488318735699154e-05,
      "loss": 0.7041,
      "step": 3793
    },
    {
      "epoch": 0.6582234559333796,
      "grad_norm": 1.0522260665893555,
      "learning_rate": 3.7485012085220416e-05,
      "loss": 0.6862,
      "step": 3794
    },
    {
      "epoch": 0.6583969465648855,
      "grad_norm": 0.6765912175178528,
      "learning_rate": 3.7481703405585646e-05,
      "loss": 0.5156,
      "step": 3795
    },
    {
      "epoch": 0.6585704371963914,
      "grad_norm": 0.63466876745224,
      "learning_rate": 3.747839269717882e-05,
      "loss": 0.6035,
      "step": 3796
    },
    {
      "epoch": 0.6587439278278973,
      "grad_norm": 0.736077070236206,
      "learning_rate": 3.747507996038416e-05,
      "loss": 0.554,
      "step": 3797
    },
    {
      "epoch": 0.6589174184594032,
      "grad_norm": 2.358635663986206,
      "learning_rate": 3.7471765195586115e-05,
      "loss": 0.5072,
      "step": 3798
    },
    {
      "epoch": 0.6590909090909091,
      "grad_norm": 0.7252426743507385,
      "learning_rate": 3.746844840316935e-05,
      "loss": 0.5627,
      "step": 3799
    },
    {
      "epoch": 0.659264399722415,
      "grad_norm": 0.9332752227783203,
      "learning_rate": 3.74651295835188e-05,
      "loss": 0.5043,
      "step": 3800
    },
    {
      "epoch": 0.6594378903539209,
      "grad_norm": 0.8140218257904053,
      "learning_rate": 3.7461808737019606e-05,
      "loss": 0.5044,
      "step": 3801
    },
    {
      "epoch": 0.6596113809854268,
      "grad_norm": 0.7537535429000854,
      "learning_rate": 3.745848586405717e-05,
      "loss": 0.6608,
      "step": 3802
    },
    {
      "epoch": 0.6597848716169327,
      "grad_norm": 0.8436901569366455,
      "learning_rate": 3.745516096501709e-05,
      "loss": 0.5093,
      "step": 3803
    },
    {
      "epoch": 0.6599583622484386,
      "grad_norm": 0.9159920811653137,
      "learning_rate": 3.745183404028525e-05,
      "loss": 0.4623,
      "step": 3804
    },
    {
      "epoch": 0.6601318528799445,
      "grad_norm": 1.2931699752807617,
      "learning_rate": 3.744850509024774e-05,
      "loss": 0.5604,
      "step": 3805
    },
    {
      "epoch": 0.6603053435114504,
      "grad_norm": 0.6823956370353699,
      "learning_rate": 3.7445174115290875e-05,
      "loss": 0.5564,
      "step": 3806
    },
    {
      "epoch": 0.6604788341429563,
      "grad_norm": 0.720169186592102,
      "learning_rate": 3.744184111580123e-05,
      "loss": 0.5358,
      "step": 3807
    },
    {
      "epoch": 0.6606523247744622,
      "grad_norm": 0.9941638708114624,
      "learning_rate": 3.74385060921656e-05,
      "loss": 0.5621,
      "step": 3808
    },
    {
      "epoch": 0.6608258154059681,
      "grad_norm": 0.7703692317008972,
      "learning_rate": 3.7435169044771016e-05,
      "loss": 0.58,
      "step": 3809
    },
    {
      "epoch": 0.6609993060374739,
      "grad_norm": 0.8619455695152283,
      "learning_rate": 3.743182997400475e-05,
      "loss": 0.593,
      "step": 3810
    },
    {
      "epoch": 0.6611727966689799,
      "grad_norm": 1.0943324565887451,
      "learning_rate": 3.74284888802543e-05,
      "loss": 0.4762,
      "step": 3811
    },
    {
      "epoch": 0.6613462873004857,
      "grad_norm": 1.0455622673034668,
      "learning_rate": 3.742514576390741e-05,
      "loss": 0.5531,
      "step": 3812
    },
    {
      "epoch": 0.6615197779319917,
      "grad_norm": 4.09170389175415,
      "learning_rate": 3.742180062535205e-05,
      "loss": 0.6555,
      "step": 3813
    },
    {
      "epoch": 0.6616932685634975,
      "grad_norm": 0.5392979383468628,
      "learning_rate": 3.741845346497643e-05,
      "loss": 0.7412,
      "step": 3814
    },
    {
      "epoch": 0.6618667591950035,
      "grad_norm": 0.5717268586158752,
      "learning_rate": 3.741510428316898e-05,
      "loss": 0.5106,
      "step": 3815
    },
    {
      "epoch": 0.6620402498265093,
      "grad_norm": 0.8426308035850525,
      "learning_rate": 3.741175308031839e-05,
      "loss": 0.5457,
      "step": 3816
    },
    {
      "epoch": 0.6622137404580153,
      "grad_norm": 1.7442169189453125,
      "learning_rate": 3.7408399856813565e-05,
      "loss": 0.6315,
      "step": 3817
    },
    {
      "epoch": 0.6623872310895211,
      "grad_norm": 0.6281899213790894,
      "learning_rate": 3.740504461304366e-05,
      "loss": 0.7258,
      "step": 3818
    },
    {
      "epoch": 0.6625607217210271,
      "grad_norm": 0.624485194683075,
      "learning_rate": 3.7401687349398044e-05,
      "loss": 0.5144,
      "step": 3819
    },
    {
      "epoch": 0.6627342123525329,
      "grad_norm": 0.7151901125907898,
      "learning_rate": 3.739832806626632e-05,
      "loss": 0.4746,
      "step": 3820
    },
    {
      "epoch": 0.6629077029840389,
      "grad_norm": 0.48618757724761963,
      "learning_rate": 3.7394966764038366e-05,
      "loss": 0.7042,
      "step": 3821
    },
    {
      "epoch": 0.6630811936155447,
      "grad_norm": 0.9319300055503845,
      "learning_rate": 3.7391603443104244e-05,
      "loss": 0.6274,
      "step": 3822
    },
    {
      "epoch": 0.6632546842470507,
      "grad_norm": 0.5655573010444641,
      "learning_rate": 3.738823810385428e-05,
      "loss": 0.629,
      "step": 3823
    },
    {
      "epoch": 0.6634281748785565,
      "grad_norm": 0.5381420850753784,
      "learning_rate": 3.738487074667902e-05,
      "loss": 0.6809,
      "step": 3824
    },
    {
      "epoch": 0.6636016655100625,
      "grad_norm": 0.5153032541275024,
      "learning_rate": 3.738150137196925e-05,
      "loss": 0.7103,
      "step": 3825
    },
    {
      "epoch": 0.6637751561415683,
      "grad_norm": 0.8185628056526184,
      "learning_rate": 3.7378129980116e-05,
      "loss": 0.6635,
      "step": 3826
    },
    {
      "epoch": 0.6639486467730743,
      "grad_norm": 0.705909013748169,
      "learning_rate": 3.737475657151051e-05,
      "loss": 0.6787,
      "step": 3827
    },
    {
      "epoch": 0.6641221374045801,
      "grad_norm": 0.6243531703948975,
      "learning_rate": 3.7371381146544276e-05,
      "loss": 0.6552,
      "step": 3828
    },
    {
      "epoch": 0.6642956280360861,
      "grad_norm": 0.7824167013168335,
      "learning_rate": 3.736800370560902e-05,
      "loss": 0.6104,
      "step": 3829
    },
    {
      "epoch": 0.664469118667592,
      "grad_norm": 1.479079008102417,
      "learning_rate": 3.736462424909669e-05,
      "loss": 0.5413,
      "step": 3830
    },
    {
      "epoch": 0.6646426092990978,
      "grad_norm": 1.029030442237854,
      "learning_rate": 3.736124277739949e-05,
      "loss": 0.6205,
      "step": 3831
    },
    {
      "epoch": 0.6648160999306038,
      "grad_norm": 0.8108117580413818,
      "learning_rate": 3.735785929090983e-05,
      "loss": 0.5165,
      "step": 3832
    },
    {
      "epoch": 0.6649895905621096,
      "grad_norm": 0.5535688996315002,
      "learning_rate": 3.7354473790020375e-05,
      "loss": 0.666,
      "step": 3833
    },
    {
      "epoch": 0.6651630811936156,
      "grad_norm": 0.6154541373252869,
      "learning_rate": 3.7351086275124023e-05,
      "loss": 0.6436,
      "step": 3834
    },
    {
      "epoch": 0.6653365718251214,
      "grad_norm": 0.6031832695007324,
      "learning_rate": 3.734769674661388e-05,
      "loss": 0.7224,
      "step": 3835
    },
    {
      "epoch": 0.6655100624566274,
      "grad_norm": 1.4116318225860596,
      "learning_rate": 3.7344305204883326e-05,
      "loss": 0.5701,
      "step": 3836
    },
    {
      "epoch": 0.6656835530881332,
      "grad_norm": 0.553779125213623,
      "learning_rate": 3.7340911650325934e-05,
      "loss": 0.6707,
      "step": 3837
    },
    {
      "epoch": 0.6658570437196392,
      "grad_norm": 2.773503303527832,
      "learning_rate": 3.7337516083335536e-05,
      "loss": 0.6321,
      "step": 3838
    },
    {
      "epoch": 0.666030534351145,
      "grad_norm": 1.2878464460372925,
      "learning_rate": 3.73341185043062e-05,
      "loss": 0.6041,
      "step": 3839
    },
    {
      "epoch": 0.666204024982651,
      "grad_norm": 0.9436421394348145,
      "learning_rate": 3.7330718913632215e-05,
      "loss": 0.6665,
      "step": 3840
    },
    {
      "epoch": 0.6663775156141568,
      "grad_norm": 0.6700578927993774,
      "learning_rate": 3.7327317311708095e-05,
      "loss": 0.6097,
      "step": 3841
    },
    {
      "epoch": 0.6665510062456628,
      "grad_norm": 0.658366858959198,
      "learning_rate": 3.732391369892862e-05,
      "loss": 0.5806,
      "step": 3842
    },
    {
      "epoch": 0.6667244968771686,
      "grad_norm": 0.688478410243988,
      "learning_rate": 3.732050807568878e-05,
      "loss": 0.589,
      "step": 3843
    },
    {
      "epoch": 0.6668979875086746,
      "grad_norm": 0.7162792086601257,
      "learning_rate": 3.731710044238378e-05,
      "loss": 0.5082,
      "step": 3844
    },
    {
      "epoch": 0.6670714781401804,
      "grad_norm": 0.749352216720581,
      "learning_rate": 3.731369079940911e-05,
      "loss": 0.495,
      "step": 3845
    },
    {
      "epoch": 0.6672449687716864,
      "grad_norm": 0.7571706771850586,
      "learning_rate": 3.731027914716044e-05,
      "loss": 0.5282,
      "step": 3846
    },
    {
      "epoch": 0.6674184594031922,
      "grad_norm": 1.0114901065826416,
      "learning_rate": 3.73068654860337e-05,
      "loss": 0.5433,
      "step": 3847
    },
    {
      "epoch": 0.6675919500346982,
      "grad_norm": 0.8244972229003906,
      "learning_rate": 3.7303449816425066e-05,
      "loss": 0.5514,
      "step": 3848
    },
    {
      "epoch": 0.667765440666204,
      "grad_norm": 0.9451310634613037,
      "learning_rate": 3.730003213873091e-05,
      "loss": 0.6599,
      "step": 3849
    },
    {
      "epoch": 0.6679389312977099,
      "grad_norm": 0.7476672530174255,
      "learning_rate": 3.729661245334787e-05,
      "loss": 0.5265,
      "step": 3850
    },
    {
      "epoch": 0.6681124219292158,
      "grad_norm": 1.3498326539993286,
      "learning_rate": 3.72931907606728e-05,
      "loss": 0.6234,
      "step": 3851
    },
    {
      "epoch": 0.6682859125607217,
      "grad_norm": 0.7990555763244629,
      "learning_rate": 3.728976706110278e-05,
      "loss": 0.6112,
      "step": 3852
    },
    {
      "epoch": 0.6684594031922276,
      "grad_norm": 0.702285647392273,
      "learning_rate": 3.7286341355035155e-05,
      "loss": 0.6128,
      "step": 3853
    },
    {
      "epoch": 0.6686328938237335,
      "grad_norm": 1.068147897720337,
      "learning_rate": 3.7282913642867484e-05,
      "loss": 0.5525,
      "step": 3854
    },
    {
      "epoch": 0.6688063844552394,
      "grad_norm": 0.9744014143943787,
      "learning_rate": 3.7279483924997534e-05,
      "loss": 0.6964,
      "step": 3855
    },
    {
      "epoch": 0.6689798750867453,
      "grad_norm": 0.6208134293556213,
      "learning_rate": 3.727605220182334e-05,
      "loss": 0.5535,
      "step": 3856
    },
    {
      "epoch": 0.6691533657182512,
      "grad_norm": 0.6973011493682861,
      "learning_rate": 3.727261847374316e-05,
      "loss": 0.5404,
      "step": 3857
    },
    {
      "epoch": 0.6693268563497571,
      "grad_norm": 0.6904463171958923,
      "learning_rate": 3.726918274115548e-05,
      "loss": 0.5416,
      "step": 3858
    },
    {
      "epoch": 0.669500346981263,
      "grad_norm": 0.59099280834198,
      "learning_rate": 3.726574500445902e-05,
      "loss": 0.6119,
      "step": 3859
    },
    {
      "epoch": 0.6696738376127689,
      "grad_norm": 0.644543468952179,
      "learning_rate": 3.726230526405273e-05,
      "loss": 0.6818,
      "step": 3860
    },
    {
      "epoch": 0.6698473282442748,
      "grad_norm": 0.7885875701904297,
      "learning_rate": 3.7258863520335804e-05,
      "loss": 0.5179,
      "step": 3861
    },
    {
      "epoch": 0.6700208188757807,
      "grad_norm": 0.8245733380317688,
      "learning_rate": 3.725541977370765e-05,
      "loss": 0.5963,
      "step": 3862
    },
    {
      "epoch": 0.6701943095072866,
      "grad_norm": 0.8566779494285583,
      "learning_rate": 3.725197402456793e-05,
      "loss": 0.493,
      "step": 3863
    },
    {
      "epoch": 0.6703678001387925,
      "grad_norm": 0.8461937308311462,
      "learning_rate": 3.7248526273316524e-05,
      "loss": 0.5311,
      "step": 3864
    },
    {
      "epoch": 0.6705412907702984,
      "grad_norm": 0.5525814890861511,
      "learning_rate": 3.724507652035354e-05,
      "loss": 0.5729,
      "step": 3865
    },
    {
      "epoch": 0.6707147814018043,
      "grad_norm": 0.7559514045715332,
      "learning_rate": 3.724162476607933e-05,
      "loss": 0.5732,
      "step": 3866
    },
    {
      "epoch": 0.6708882720333103,
      "grad_norm": 1.441557765007019,
      "learning_rate": 3.723817101089448e-05,
      "loss": 0.697,
      "step": 3867
    },
    {
      "epoch": 0.6710617626648161,
      "grad_norm": 0.8079794049263,
      "learning_rate": 3.72347152551998e-05,
      "loss": 0.594,
      "step": 3868
    },
    {
      "epoch": 0.6712352532963219,
      "grad_norm": 0.6825093030929565,
      "learning_rate": 3.723125749939633e-05,
      "loss": 0.6075,
      "step": 3869
    },
    {
      "epoch": 0.6714087439278279,
      "grad_norm": 1.3218646049499512,
      "learning_rate": 3.722779774388535e-05,
      "loss": 0.6593,
      "step": 3870
    },
    {
      "epoch": 0.6715822345593337,
      "grad_norm": 1.1525205373764038,
      "learning_rate": 3.722433598906836e-05,
      "loss": 0.6132,
      "step": 3871
    },
    {
      "epoch": 0.6717557251908397,
      "grad_norm": 1.3912229537963867,
      "learning_rate": 3.722087223534711e-05,
      "loss": 0.4697,
      "step": 3872
    },
    {
      "epoch": 0.6719292158223455,
      "grad_norm": 1.0367560386657715,
      "learning_rate": 3.7217406483123575e-05,
      "loss": 0.6156,
      "step": 3873
    },
    {
      "epoch": 0.6721027064538515,
      "grad_norm": 0.9154960513114929,
      "learning_rate": 3.721393873279996e-05,
      "loss": 0.5574,
      "step": 3874
    },
    {
      "epoch": 0.6722761970853574,
      "grad_norm": 0.7309114336967468,
      "learning_rate": 3.721046898477869e-05,
      "loss": 0.6003,
      "step": 3875
    },
    {
      "epoch": 0.6724496877168633,
      "grad_norm": 0.6787078976631165,
      "learning_rate": 3.720699723946244e-05,
      "loss": 0.6046,
      "step": 3876
    },
    {
      "epoch": 0.6726231783483692,
      "grad_norm": 0.6763220429420471,
      "learning_rate": 3.720352349725411e-05,
      "loss": 0.6923,
      "step": 3877
    },
    {
      "epoch": 0.6727966689798751,
      "grad_norm": 0.8732578158378601,
      "learning_rate": 3.720004775855684e-05,
      "loss": 0.6654,
      "step": 3878
    },
    {
      "epoch": 0.672970159611381,
      "grad_norm": 0.6277561783790588,
      "learning_rate": 3.719657002377397e-05,
      "loss": 0.5978,
      "step": 3879
    },
    {
      "epoch": 0.6731436502428869,
      "grad_norm": 1.2320107221603394,
      "learning_rate": 3.719309029330912e-05,
      "loss": 0.4968,
      "step": 3880
    },
    {
      "epoch": 0.6733171408743928,
      "grad_norm": 0.7353063225746155,
      "learning_rate": 3.718960856756611e-05,
      "loss": 0.58,
      "step": 3881
    },
    {
      "epoch": 0.6734906315058987,
      "grad_norm": 0.8003371357917786,
      "learning_rate": 3.7186124846948995e-05,
      "loss": 0.63,
      "step": 3882
    },
    {
      "epoch": 0.6736641221374046,
      "grad_norm": 0.6182820796966553,
      "learning_rate": 3.718263913186206e-05,
      "loss": 0.6464,
      "step": 3883
    },
    {
      "epoch": 0.6738376127689105,
      "grad_norm": 0.9211560487747192,
      "learning_rate": 3.7179151422709845e-05,
      "loss": 0.6189,
      "step": 3884
    },
    {
      "epoch": 0.6740111034004164,
      "grad_norm": 0.8812063336372375,
      "learning_rate": 3.717566171989708e-05,
      "loss": 0.7413,
      "step": 3885
    },
    {
      "epoch": 0.6741845940319223,
      "grad_norm": 0.9910178780555725,
      "learning_rate": 3.717217002382875e-05,
      "loss": 0.5789,
      "step": 3886
    },
    {
      "epoch": 0.6743580846634282,
      "grad_norm": 0.730434238910675,
      "learning_rate": 3.716867633491009e-05,
      "loss": 0.5742,
      "step": 3887
    },
    {
      "epoch": 0.6745315752949341,
      "grad_norm": 0.7126477956771851,
      "learning_rate": 3.716518065354654e-05,
      "loss": 0.6165,
      "step": 3888
    },
    {
      "epoch": 0.67470506592644,
      "grad_norm": 0.9688850045204163,
      "learning_rate": 3.7161682980143766e-05,
      "loss": 0.5385,
      "step": 3889
    },
    {
      "epoch": 0.6748785565579458,
      "grad_norm": 0.7977421283721924,
      "learning_rate": 3.715818331510769e-05,
      "loss": 0.6541,
      "step": 3890
    },
    {
      "epoch": 0.6750520471894518,
      "grad_norm": 0.782105028629303,
      "learning_rate": 3.715468165884444e-05,
      "loss": 0.5095,
      "step": 3891
    },
    {
      "epoch": 0.6752255378209576,
      "grad_norm": 1.3656920194625854,
      "learning_rate": 3.71511780117604e-05,
      "loss": 0.6842,
      "step": 3892
    },
    {
      "epoch": 0.6753990284524636,
      "grad_norm": 0.7901886701583862,
      "learning_rate": 3.7147672374262165e-05,
      "loss": 0.5476,
      "step": 3893
    },
    {
      "epoch": 0.6755725190839694,
      "grad_norm": 0.7951087355613708,
      "learning_rate": 3.714416474675657e-05,
      "loss": 0.4933,
      "step": 3894
    },
    {
      "epoch": 0.6757460097154754,
      "grad_norm": 0.8243618011474609,
      "learning_rate": 3.7140655129650676e-05,
      "loss": 0.4677,
      "step": 3895
    },
    {
      "epoch": 0.6759195003469812,
      "grad_norm": 0.8517226576805115,
      "learning_rate": 3.7137143523351787e-05,
      "loss": 0.639,
      "step": 3896
    },
    {
      "epoch": 0.6760929909784872,
      "grad_norm": 1.2318670749664307,
      "learning_rate": 3.713362992826742e-05,
      "loss": 0.6829,
      "step": 3897
    },
    {
      "epoch": 0.676266481609993,
      "grad_norm": 0.7173767685890198,
      "learning_rate": 3.713011434480534e-05,
      "loss": 0.6592,
      "step": 3898
    },
    {
      "epoch": 0.676439972241499,
      "grad_norm": 0.7821794152259827,
      "learning_rate": 3.712659677337352e-05,
      "loss": 0.6282,
      "step": 3899
    },
    {
      "epoch": 0.6766134628730048,
      "grad_norm": 0.7882617712020874,
      "learning_rate": 3.71230772143802e-05,
      "loss": 0.5424,
      "step": 3900
    },
    {
      "epoch": 0.6767869535045108,
      "grad_norm": 0.8059515953063965,
      "learning_rate": 3.711955566823381e-05,
      "loss": 0.5006,
      "step": 3901
    },
    {
      "epoch": 0.6769604441360166,
      "grad_norm": 0.7536359429359436,
      "learning_rate": 3.711603213534303e-05,
      "loss": 0.5542,
      "step": 3902
    },
    {
      "epoch": 0.6771339347675226,
      "grad_norm": 0.7734538912773132,
      "learning_rate": 3.7112506616116794e-05,
      "loss": 0.6211,
      "step": 3903
    },
    {
      "epoch": 0.6773074253990284,
      "grad_norm": 0.6895074248313904,
      "learning_rate": 3.710897911096421e-05,
      "loss": 0.5533,
      "step": 3904
    },
    {
      "epoch": 0.6774809160305344,
      "grad_norm": 1.085900902748108,
      "learning_rate": 3.710544962029467e-05,
      "loss": 0.4974,
      "step": 3905
    },
    {
      "epoch": 0.6776544066620402,
      "grad_norm": 0.7151042222976685,
      "learning_rate": 3.710191814451777e-05,
      "loss": 0.5927,
      "step": 3906
    },
    {
      "epoch": 0.6778278972935462,
      "grad_norm": 0.8727800846099854,
      "learning_rate": 3.709838468404334e-05,
      "loss": 0.6307,
      "step": 3907
    },
    {
      "epoch": 0.678001387925052,
      "grad_norm": 1.0264941453933716,
      "learning_rate": 3.7094849239281444e-05,
      "loss": 0.5277,
      "step": 3908
    },
    {
      "epoch": 0.6781748785565579,
      "grad_norm": 1.1419225931167603,
      "learning_rate": 3.709131181064238e-05,
      "loss": 0.4739,
      "step": 3909
    },
    {
      "epoch": 0.6783483691880638,
      "grad_norm": 0.7475083470344543,
      "learning_rate": 3.7087772398536656e-05,
      "loss": 0.5382,
      "step": 3910
    },
    {
      "epoch": 0.6785218598195697,
      "grad_norm": 0.8183844089508057,
      "learning_rate": 3.708423100337504e-05,
      "loss": 0.4733,
      "step": 3911
    },
    {
      "epoch": 0.6786953504510757,
      "grad_norm": 0.8566032648086548,
      "learning_rate": 3.70806876255685e-05,
      "loss": 0.5155,
      "step": 3912
    },
    {
      "epoch": 0.6788688410825815,
      "grad_norm": 0.6601811051368713,
      "learning_rate": 3.707714226552827e-05,
      "loss": 0.6111,
      "step": 3913
    },
    {
      "epoch": 0.6790423317140875,
      "grad_norm": 0.8909444212913513,
      "learning_rate": 3.7073594923665774e-05,
      "loss": 0.6725,
      "step": 3914
    },
    {
      "epoch": 0.6792158223455933,
      "grad_norm": 0.5450756549835205,
      "learning_rate": 3.70700456003927e-05,
      "loss": 0.5869,
      "step": 3915
    },
    {
      "epoch": 0.6793893129770993,
      "grad_norm": 0.8035801649093628,
      "learning_rate": 3.7066494296120935e-05,
      "loss": 0.512,
      "step": 3916
    },
    {
      "epoch": 0.6795628036086051,
      "grad_norm": 0.6066131591796875,
      "learning_rate": 3.7062941011262624e-05,
      "loss": 0.6162,
      "step": 3917
    },
    {
      "epoch": 0.6797362942401111,
      "grad_norm": 0.7695337533950806,
      "learning_rate": 3.705938574623012e-05,
      "loss": 0.5311,
      "step": 3918
    },
    {
      "epoch": 0.6799097848716169,
      "grad_norm": 0.7239415645599365,
      "learning_rate": 3.705582850143603e-05,
      "loss": 0.5833,
      "step": 3919
    },
    {
      "epoch": 0.6800832755031229,
      "grad_norm": 0.5847352147102356,
      "learning_rate": 3.705226927729317e-05,
      "loss": 0.5315,
      "step": 3920
    },
    {
      "epoch": 0.6802567661346287,
      "grad_norm": 1.0682518482208252,
      "learning_rate": 3.7048708074214586e-05,
      "loss": 0.6537,
      "step": 3921
    },
    {
      "epoch": 0.6804302567661347,
      "grad_norm": 0.690758466720581,
      "learning_rate": 3.704514489261357e-05,
      "loss": 0.637,
      "step": 3922
    },
    {
      "epoch": 0.6806037473976405,
      "grad_norm": 0.7355716824531555,
      "learning_rate": 3.7041579732903617e-05,
      "loss": 0.5459,
      "step": 3923
    },
    {
      "epoch": 0.6807772380291465,
      "grad_norm": 1.3008263111114502,
      "learning_rate": 3.703801259549848e-05,
      "loss": 0.647,
      "step": 3924
    },
    {
      "epoch": 0.6809507286606523,
      "grad_norm": 0.6505176424980164,
      "learning_rate": 3.7034443480812144e-05,
      "loss": 0.5614,
      "step": 3925
    },
    {
      "epoch": 0.6811242192921583,
      "grad_norm": 0.6328716278076172,
      "learning_rate": 3.7030872389258777e-05,
      "loss": 0.5817,
      "step": 3926
    },
    {
      "epoch": 0.6812977099236641,
      "grad_norm": 0.5805076956748962,
      "learning_rate": 3.7027299321252825e-05,
      "loss": 0.6594,
      "step": 3927
    },
    {
      "epoch": 0.6814712005551701,
      "grad_norm": 1.107621192932129,
      "learning_rate": 3.702372427720895e-05,
      "loss": 0.5508,
      "step": 3928
    },
    {
      "epoch": 0.6816446911866759,
      "grad_norm": 0.7718678712844849,
      "learning_rate": 3.702014725754204e-05,
      "loss": 0.5469,
      "step": 3929
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.722978949546814,
      "learning_rate": 3.701656826266721e-05,
      "loss": 0.5425,
      "step": 3930
    },
    {
      "epoch": 0.6819916724496877,
      "grad_norm": 0.7684771418571472,
      "learning_rate": 3.701298729299979e-05,
      "loss": 0.6063,
      "step": 3931
    },
    {
      "epoch": 0.6821651630811936,
      "grad_norm": 0.6785157322883606,
      "learning_rate": 3.7009404348955385e-05,
      "loss": 0.6631,
      "step": 3932
    },
    {
      "epoch": 0.6823386537126995,
      "grad_norm": 0.8750243782997131,
      "learning_rate": 3.700581943094978e-05,
      "loss": 0.5497,
      "step": 3933
    },
    {
      "epoch": 0.6825121443442054,
      "grad_norm": 0.6999068856239319,
      "learning_rate": 3.7002232539399014e-05,
      "loss": 0.7249,
      "step": 3934
    },
    {
      "epoch": 0.6826856349757113,
      "grad_norm": 0.7761996984481812,
      "learning_rate": 3.699864367471935e-05,
      "loss": 0.6375,
      "step": 3935
    },
    {
      "epoch": 0.6828591256072172,
      "grad_norm": 0.7740322351455688,
      "learning_rate": 3.6995052837327274e-05,
      "loss": 0.5499,
      "step": 3936
    },
    {
      "epoch": 0.6830326162387231,
      "grad_norm": 0.7279134392738342,
      "learning_rate": 3.699146002763953e-05,
      "loss": 0.7603,
      "step": 3937
    },
    {
      "epoch": 0.683206106870229,
      "grad_norm": 1.0422098636627197,
      "learning_rate": 3.6987865246073035e-05,
      "loss": 0.5532,
      "step": 3938
    },
    {
      "epoch": 0.6833795975017349,
      "grad_norm": 0.8805818557739258,
      "learning_rate": 3.6984268493044994e-05,
      "loss": 0.5599,
      "step": 3939
    },
    {
      "epoch": 0.6835530881332408,
      "grad_norm": 0.8069884181022644,
      "learning_rate": 3.6980669768972795e-05,
      "loss": 0.6333,
      "step": 3940
    },
    {
      "epoch": 0.6837265787647467,
      "grad_norm": 0.7256645560264587,
      "learning_rate": 3.697706907427409e-05,
      "loss": 0.7054,
      "step": 3941
    },
    {
      "epoch": 0.6839000693962526,
      "grad_norm": 0.6287798285484314,
      "learning_rate": 3.6973466409366735e-05,
      "loss": 0.7194,
      "step": 3942
    },
    {
      "epoch": 0.6840735600277585,
      "grad_norm": 1.039019227027893,
      "learning_rate": 3.696986177466882e-05,
      "loss": 0.5857,
      "step": 3943
    },
    {
      "epoch": 0.6842470506592644,
      "grad_norm": 1.0137076377868652,
      "learning_rate": 3.696625517059868e-05,
      "loss": 0.5593,
      "step": 3944
    },
    {
      "epoch": 0.6844205412907703,
      "grad_norm": 0.8316372036933899,
      "learning_rate": 3.696264659757485e-05,
      "loss": 0.605,
      "step": 3945
    },
    {
      "epoch": 0.6845940319222762,
      "grad_norm": 0.7433950901031494,
      "learning_rate": 3.695903605601612e-05,
      "loss": 0.6311,
      "step": 3946
    },
    {
      "epoch": 0.6847675225537821,
      "grad_norm": 0.7588471174240112,
      "learning_rate": 3.6955423546341494e-05,
      "loss": 0.5829,
      "step": 3947
    },
    {
      "epoch": 0.684941013185288,
      "grad_norm": 0.7198483943939209,
      "learning_rate": 3.695180906897021e-05,
      "loss": 0.5868,
      "step": 3948
    },
    {
      "epoch": 0.6851145038167938,
      "grad_norm": 0.7661027908325195,
      "learning_rate": 3.694819262432173e-05,
      "loss": 0.4917,
      "step": 3949
    },
    {
      "epoch": 0.6852879944482998,
      "grad_norm": 0.6911413669586182,
      "learning_rate": 3.694457421281575e-05,
      "loss": 0.6302,
      "step": 3950
    },
    {
      "epoch": 0.6854614850798056,
      "grad_norm": 0.6780367493629456,
      "learning_rate": 3.694095383487219e-05,
      "loss": 0.6379,
      "step": 3951
    },
    {
      "epoch": 0.6856349757113116,
      "grad_norm": 0.6340448260307312,
      "learning_rate": 3.693733149091119e-05,
      "loss": 0.6678,
      "step": 3952
    },
    {
      "epoch": 0.6858084663428174,
      "grad_norm": 0.6367129683494568,
      "learning_rate": 3.693370718135314e-05,
      "loss": 0.6665,
      "step": 3953
    },
    {
      "epoch": 0.6859819569743234,
      "grad_norm": 0.9811844825744629,
      "learning_rate": 3.693008090661864e-05,
      "loss": 0.5535,
      "step": 3954
    },
    {
      "epoch": 0.6861554476058292,
      "grad_norm": 1.003396987915039,
      "learning_rate": 3.692645266712852e-05,
      "loss": 0.7156,
      "step": 3955
    },
    {
      "epoch": 0.6863289382373352,
      "grad_norm": 1.047627568244934,
      "learning_rate": 3.6922822463303846e-05,
      "loss": 0.4995,
      "step": 3956
    },
    {
      "epoch": 0.686502428868841,
      "grad_norm": 0.689240038394928,
      "learning_rate": 3.691919029556591e-05,
      "loss": 0.5833,
      "step": 3957
    },
    {
      "epoch": 0.686675919500347,
      "grad_norm": 0.8044683933258057,
      "learning_rate": 3.691555616433622e-05,
      "loss": 0.6375,
      "step": 3958
    },
    {
      "epoch": 0.6868494101318529,
      "grad_norm": 0.7325993776321411,
      "learning_rate": 3.691192007003652e-05,
      "loss": 0.6768,
      "step": 3959
    },
    {
      "epoch": 0.6870229007633588,
      "grad_norm": 1.0790187120437622,
      "learning_rate": 3.69082820130888e-05,
      "loss": 0.6622,
      "step": 3960
    },
    {
      "epoch": 0.6871963913948647,
      "grad_norm": 0.856217622756958,
      "learning_rate": 3.690464199391525e-05,
      "loss": 0.5355,
      "step": 3961
    },
    {
      "epoch": 0.6873698820263706,
      "grad_norm": 0.6712712049484253,
      "learning_rate": 3.69010000129383e-05,
      "loss": 0.5717,
      "step": 3962
    },
    {
      "epoch": 0.6875433726578765,
      "grad_norm": 0.5234697461128235,
      "learning_rate": 3.6897356070580596e-05,
      "loss": 0.642,
      "step": 3963
    },
    {
      "epoch": 0.6877168632893824,
      "grad_norm": 0.7027401924133301,
      "learning_rate": 3.689371016726504e-05,
      "loss": 0.5718,
      "step": 3964
    },
    {
      "epoch": 0.6878903539208883,
      "grad_norm": 0.6622681617736816,
      "learning_rate": 3.6890062303414734e-05,
      "loss": 0.5706,
      "step": 3965
    },
    {
      "epoch": 0.6880638445523942,
      "grad_norm": 1.1115381717681885,
      "learning_rate": 3.6886412479453004e-05,
      "loss": 0.6348,
      "step": 3966
    },
    {
      "epoch": 0.6882373351839001,
      "grad_norm": 0.6777711510658264,
      "learning_rate": 3.6882760695803444e-05,
      "loss": 0.5442,
      "step": 3967
    },
    {
      "epoch": 0.6884108258154059,
      "grad_norm": 0.8778414130210876,
      "learning_rate": 3.6879106952889826e-05,
      "loss": 0.5205,
      "step": 3968
    },
    {
      "epoch": 0.6885843164469119,
      "grad_norm": 0.753093957901001,
      "learning_rate": 3.687545125113618e-05,
      "loss": 0.6105,
      "step": 3969
    },
    {
      "epoch": 0.6887578070784177,
      "grad_norm": 0.8035931587219238,
      "learning_rate": 3.687179359096675e-05,
      "loss": 0.568,
      "step": 3970
    },
    {
      "epoch": 0.6889312977099237,
      "grad_norm": 0.6357128620147705,
      "learning_rate": 3.686813397280602e-05,
      "loss": 0.548,
      "step": 3971
    },
    {
      "epoch": 0.6891047883414295,
      "grad_norm": 0.9069519639015198,
      "learning_rate": 3.686447239707868e-05,
      "loss": 0.6504,
      "step": 3972
    },
    {
      "epoch": 0.6892782789729355,
      "grad_norm": 0.6657199859619141,
      "learning_rate": 3.686080886420968e-05,
      "loss": 0.5139,
      "step": 3973
    },
    {
      "epoch": 0.6894517696044413,
      "grad_norm": 0.7213208675384521,
      "learning_rate": 3.685714337462415e-05,
      "loss": 0.5786,
      "step": 3974
    },
    {
      "epoch": 0.6896252602359473,
      "grad_norm": 0.6772657036781311,
      "learning_rate": 3.685347592874749e-05,
      "loss": 0.6503,
      "step": 3975
    },
    {
      "epoch": 0.6897987508674531,
      "grad_norm": 0.7594093084335327,
      "learning_rate": 3.6849806527005316e-05,
      "loss": 0.708,
      "step": 3976
    },
    {
      "epoch": 0.6899722414989591,
      "grad_norm": 1.1025997400283813,
      "learning_rate": 3.684613516982346e-05,
      "loss": 0.5925,
      "step": 3977
    },
    {
      "epoch": 0.6901457321304649,
      "grad_norm": 0.7050821781158447,
      "learning_rate": 3.6842461857627986e-05,
      "loss": 0.525,
      "step": 3978
    },
    {
      "epoch": 0.6903192227619709,
      "grad_norm": 0.8308770060539246,
      "learning_rate": 3.683878659084519e-05,
      "loss": 0.5809,
      "step": 3979
    },
    {
      "epoch": 0.6904927133934767,
      "grad_norm": 0.612371563911438,
      "learning_rate": 3.6835109369901586e-05,
      "loss": 0.5927,
      "step": 3980
    },
    {
      "epoch": 0.6906662040249827,
      "grad_norm": 0.6579895615577698,
      "learning_rate": 3.6831430195223927e-05,
      "loss": 0.6783,
      "step": 3981
    },
    {
      "epoch": 0.6908396946564885,
      "grad_norm": 0.8142983317375183,
      "learning_rate": 3.682774906723918e-05,
      "loss": 0.4998,
      "step": 3982
    },
    {
      "epoch": 0.6910131852879945,
      "grad_norm": 0.7529022097587585,
      "learning_rate": 3.6824065986374546e-05,
      "loss": 0.6178,
      "step": 3983
    },
    {
      "epoch": 0.6911866759195003,
      "grad_norm": 1.1721090078353882,
      "learning_rate": 3.6820380953057446e-05,
      "loss": 0.536,
      "step": 3984
    },
    {
      "epoch": 0.6913601665510063,
      "grad_norm": 0.7267649173736572,
      "learning_rate": 3.681669396771554e-05,
      "loss": 0.5693,
      "step": 3985
    },
    {
      "epoch": 0.6915336571825121,
      "grad_norm": 0.8769500851631165,
      "learning_rate": 3.681300503077671e-05,
      "loss": 0.5223,
      "step": 3986
    },
    {
      "epoch": 0.6917071478140181,
      "grad_norm": 0.7207356691360474,
      "learning_rate": 3.6809314142669044e-05,
      "loss": 0.6375,
      "step": 3987
    },
    {
      "epoch": 0.6918806384455239,
      "grad_norm": 0.6993891000747681,
      "learning_rate": 3.680562130382089e-05,
      "loss": 0.6946,
      "step": 3988
    },
    {
      "epoch": 0.6920541290770298,
      "grad_norm": 0.5964895486831665,
      "learning_rate": 3.68019265146608e-05,
      "loss": 0.6268,
      "step": 3989
    },
    {
      "epoch": 0.6922276197085357,
      "grad_norm": 0.887861967086792,
      "learning_rate": 3.679822977561756e-05,
      "loss": 0.6449,
      "step": 3990
    },
    {
      "epoch": 0.6924011103400416,
      "grad_norm": 0.4853559732437134,
      "learning_rate": 3.679453108712018e-05,
      "loss": 0.6294,
      "step": 3991
    },
    {
      "epoch": 0.6925746009715475,
      "grad_norm": 1.1081832647323608,
      "learning_rate": 3.67908304495979e-05,
      "loss": 0.5466,
      "step": 3992
    },
    {
      "epoch": 0.6927480916030534,
      "grad_norm": 0.8581452369689941,
      "learning_rate": 3.678712786348018e-05,
      "loss": 0.5555,
      "step": 3993
    },
    {
      "epoch": 0.6929215822345594,
      "grad_norm": 0.7335177659988403,
      "learning_rate": 3.678342332919671e-05,
      "loss": 0.6768,
      "step": 3994
    },
    {
      "epoch": 0.6930950728660652,
      "grad_norm": 0.6366673707962036,
      "learning_rate": 3.677971684717741e-05,
      "loss": 0.7219,
      "step": 3995
    },
    {
      "epoch": 0.6932685634975712,
      "grad_norm": 0.6510780453681946,
      "learning_rate": 3.6776008417852415e-05,
      "loss": 0.6423,
      "step": 3996
    },
    {
      "epoch": 0.693442054129077,
      "grad_norm": 0.7125526070594788,
      "learning_rate": 3.6772298041652095e-05,
      "loss": 0.6307,
      "step": 3997
    },
    {
      "epoch": 0.693615544760583,
      "grad_norm": 0.920978844165802,
      "learning_rate": 3.676858571900704e-05,
      "loss": 0.6475,
      "step": 3998
    },
    {
      "epoch": 0.6937890353920888,
      "grad_norm": 0.6296781897544861,
      "learning_rate": 3.676487145034808e-05,
      "loss": 0.621,
      "step": 3999
    },
    {
      "epoch": 0.6939625260235948,
      "grad_norm": 0.8165046572685242,
      "learning_rate": 3.6761155236106246e-05,
      "loss": 0.6057,
      "step": 4000
    },
    {
      "epoch": 0.6941360166551006,
      "grad_norm": 0.5971564650535583,
      "learning_rate": 3.675743707671282e-05,
      "loss": 0.6636,
      "step": 4001
    },
    {
      "epoch": 0.6943095072866066,
      "grad_norm": 0.5804475545883179,
      "learning_rate": 3.67537169725993e-05,
      "loss": 0.6046,
      "step": 4002
    },
    {
      "epoch": 0.6944829979181124,
      "grad_norm": 0.8879976868629456,
      "learning_rate": 3.6749994924197394e-05,
      "loss": 0.5135,
      "step": 4003
    },
    {
      "epoch": 0.6946564885496184,
      "grad_norm": 0.6747565865516663,
      "learning_rate": 3.6746270931939064e-05,
      "loss": 0.5944,
      "step": 4004
    },
    {
      "epoch": 0.6948299791811242,
      "grad_norm": 0.8641384243965149,
      "learning_rate": 3.674254499625648e-05,
      "loss": 0.54,
      "step": 4005
    },
    {
      "epoch": 0.6950034698126302,
      "grad_norm": 0.7296652793884277,
      "learning_rate": 3.6738817117582045e-05,
      "loss": 0.4667,
      "step": 4006
    },
    {
      "epoch": 0.695176960444136,
      "grad_norm": 0.9644712209701538,
      "learning_rate": 3.6735087296348366e-05,
      "loss": 0.6002,
      "step": 4007
    },
    {
      "epoch": 0.6953504510756419,
      "grad_norm": 1.2004969120025635,
      "learning_rate": 3.6731355532988315e-05,
      "loss": 0.537,
      "step": 4008
    },
    {
      "epoch": 0.6955239417071478,
      "grad_norm": 0.5749496817588806,
      "learning_rate": 3.672762182793496e-05,
      "loss": 0.6667,
      "step": 4009
    },
    {
      "epoch": 0.6956974323386537,
      "grad_norm": 0.6510900855064392,
      "learning_rate": 3.6723886181621595e-05,
      "loss": 0.6166,
      "step": 4010
    },
    {
      "epoch": 0.6958709229701596,
      "grad_norm": 0.723423182964325,
      "learning_rate": 3.672014859448175e-05,
      "loss": 0.6903,
      "step": 4011
    },
    {
      "epoch": 0.6960444136016655,
      "grad_norm": 1.0655949115753174,
      "learning_rate": 3.6716409066949184e-05,
      "loss": 0.5122,
      "step": 4012
    },
    {
      "epoch": 0.6962179042331714,
      "grad_norm": 0.749419629573822,
      "learning_rate": 3.671266759945786e-05,
      "loss": 0.6561,
      "step": 4013
    },
    {
      "epoch": 0.6963913948646773,
      "grad_norm": 0.7480120658874512,
      "learning_rate": 3.670892419244199e-05,
      "loss": 0.5352,
      "step": 4014
    },
    {
      "epoch": 0.6965648854961832,
      "grad_norm": 0.6904357671737671,
      "learning_rate": 3.6705178846336004e-05,
      "loss": 0.606,
      "step": 4015
    },
    {
      "epoch": 0.6967383761276891,
      "grad_norm": 0.6453255414962769,
      "learning_rate": 3.670143156157454e-05,
      "loss": 0.6123,
      "step": 4016
    },
    {
      "epoch": 0.696911866759195,
      "grad_norm": 0.8253284096717834,
      "learning_rate": 3.669768233859249e-05,
      "loss": 0.5403,
      "step": 4017
    },
    {
      "epoch": 0.6970853573907009,
      "grad_norm": 0.5644474029541016,
      "learning_rate": 3.6693931177824934e-05,
      "loss": 0.5294,
      "step": 4018
    },
    {
      "epoch": 0.6972588480222068,
      "grad_norm": 0.6671839952468872,
      "learning_rate": 3.6690178079707226e-05,
      "loss": 0.5482,
      "step": 4019
    },
    {
      "epoch": 0.6974323386537127,
      "grad_norm": 0.7043505907058716,
      "learning_rate": 3.66864230446749e-05,
      "loss": 0.5681,
      "step": 4020
    },
    {
      "epoch": 0.6976058292852186,
      "grad_norm": 0.6432799100875854,
      "learning_rate": 3.668266607316373e-05,
      "loss": 0.6671,
      "step": 4021
    },
    {
      "epoch": 0.6977793199167245,
      "grad_norm": 0.7257577180862427,
      "learning_rate": 3.667890716560973e-05,
      "loss": 0.6791,
      "step": 4022
    },
    {
      "epoch": 0.6979528105482304,
      "grad_norm": 0.685150682926178,
      "learning_rate": 3.667514632244912e-05,
      "loss": 0.6329,
      "step": 4023
    },
    {
      "epoch": 0.6981263011797363,
      "grad_norm": 0.584290087223053,
      "learning_rate": 3.667138354411834e-05,
      "loss": 0.6116,
      "step": 4024
    },
    {
      "epoch": 0.6982997918112422,
      "grad_norm": 1.7903361320495605,
      "learning_rate": 3.666761883105408e-05,
      "loss": 0.7131,
      "step": 4025
    },
    {
      "epoch": 0.6984732824427481,
      "grad_norm": 1.1780725717544556,
      "learning_rate": 3.666385218369324e-05,
      "loss": 0.575,
      "step": 4026
    },
    {
      "epoch": 0.6986467730742539,
      "grad_norm": 0.793237030506134,
      "learning_rate": 3.6660083602472924e-05,
      "loss": 0.6355,
      "step": 4027
    },
    {
      "epoch": 0.6988202637057599,
      "grad_norm": 0.8836570978164673,
      "learning_rate": 3.6656313087830505e-05,
      "loss": 0.5863,
      "step": 4028
    },
    {
      "epoch": 0.6989937543372657,
      "grad_norm": 0.68492192029953,
      "learning_rate": 3.665254064020353e-05,
      "loss": 0.4689,
      "step": 4029
    },
    {
      "epoch": 0.6991672449687717,
      "grad_norm": 0.7318260669708252,
      "learning_rate": 3.664876626002982e-05,
      "loss": 0.717,
      "step": 4030
    },
    {
      "epoch": 0.6993407356002775,
      "grad_norm": 0.6521219611167908,
      "learning_rate": 3.664498994774738e-05,
      "loss": 0.6057,
      "step": 4031
    },
    {
      "epoch": 0.6995142262317835,
      "grad_norm": 0.8591718673706055,
      "learning_rate": 3.6641211703794466e-05,
      "loss": 0.6348,
      "step": 4032
    },
    {
      "epoch": 0.6996877168632893,
      "grad_norm": 0.47968804836273193,
      "learning_rate": 3.663743152860954e-05,
      "loss": 0.728,
      "step": 4033
    },
    {
      "epoch": 0.6998612074947953,
      "grad_norm": 0.9473422765731812,
      "learning_rate": 3.66336494226313e-05,
      "loss": 0.6333,
      "step": 4034
    },
    {
      "epoch": 0.7000346981263011,
      "grad_norm": 0.6778231263160706,
      "learning_rate": 3.662986538629866e-05,
      "loss": 0.5881,
      "step": 4035
    },
    {
      "epoch": 0.7002081887578071,
      "grad_norm": 1.6221238374710083,
      "learning_rate": 3.662607942005077e-05,
      "loss": 0.6375,
      "step": 4036
    },
    {
      "epoch": 0.700381679389313,
      "grad_norm": 0.6425724029541016,
      "learning_rate": 3.6622291524326986e-05,
      "loss": 0.5854,
      "step": 4037
    },
    {
      "epoch": 0.7005551700208189,
      "grad_norm": 0.6844794750213623,
      "learning_rate": 3.66185016995669e-05,
      "loss": 0.5521,
      "step": 4038
    },
    {
      "epoch": 0.7007286606523248,
      "grad_norm": 0.902301013469696,
      "learning_rate": 3.661470994621033e-05,
      "loss": 0.5486,
      "step": 4039
    },
    {
      "epoch": 0.7009021512838307,
      "grad_norm": 0.6516269445419312,
      "learning_rate": 3.661091626469731e-05,
      "loss": 0.582,
      "step": 4040
    },
    {
      "epoch": 0.7010756419153366,
      "grad_norm": 0.7208936214447021,
      "learning_rate": 3.66071206554681e-05,
      "loss": 0.7527,
      "step": 4041
    },
    {
      "epoch": 0.7012491325468425,
      "grad_norm": 0.48527052998542786,
      "learning_rate": 3.6603323118963194e-05,
      "loss": 0.6077,
      "step": 4042
    },
    {
      "epoch": 0.7014226231783484,
      "grad_norm": 0.9502211809158325,
      "learning_rate": 3.6599523655623285e-05,
      "loss": 0.6543,
      "step": 4043
    },
    {
      "epoch": 0.7015961138098543,
      "grad_norm": 0.622065544128418,
      "learning_rate": 3.659572226588932e-05,
      "loss": 0.5192,
      "step": 4044
    },
    {
      "epoch": 0.7017696044413602,
      "grad_norm": 0.6672574281692505,
      "learning_rate": 3.659191895020244e-05,
      "loss": 0.5074,
      "step": 4045
    },
    {
      "epoch": 0.7019430950728661,
      "grad_norm": 0.6417238712310791,
      "learning_rate": 3.658811370900404e-05,
      "loss": 0.6324,
      "step": 4046
    },
    {
      "epoch": 0.702116585704372,
      "grad_norm": 0.7055219411849976,
      "learning_rate": 3.6584306542735715e-05,
      "loss": 0.6863,
      "step": 4047
    },
    {
      "epoch": 0.7022900763358778,
      "grad_norm": 0.7298030853271484,
      "learning_rate": 3.658049745183928e-05,
      "loss": 0.5736,
      "step": 4048
    },
    {
      "epoch": 0.7024635669673838,
      "grad_norm": 0.6878877282142639,
      "learning_rate": 3.657668643675681e-05,
      "loss": 0.5702,
      "step": 4049
    },
    {
      "epoch": 0.7026370575988896,
      "grad_norm": 0.5812045335769653,
      "learning_rate": 3.657287349793056e-05,
      "loss": 0.6903,
      "step": 4050
    },
    {
      "epoch": 0.7028105482303956,
      "grad_norm": 0.7609200477600098,
      "learning_rate": 3.656905863580302e-05,
      "loss": 0.5286,
      "step": 4051
    },
    {
      "epoch": 0.7029840388619014,
      "grad_norm": 0.7479817271232605,
      "learning_rate": 3.656524185081693e-05,
      "loss": 0.5653,
      "step": 4052
    },
    {
      "epoch": 0.7031575294934074,
      "grad_norm": 0.6225118041038513,
      "learning_rate": 3.6561423143415216e-05,
      "loss": 0.6211,
      "step": 4053
    },
    {
      "epoch": 0.7033310201249132,
      "grad_norm": 0.6613042950630188,
      "learning_rate": 3.655760251404105e-05,
      "loss": 0.6046,
      "step": 4054
    },
    {
      "epoch": 0.7035045107564192,
      "grad_norm": 0.8066621422767639,
      "learning_rate": 3.655377996313782e-05,
      "loss": 0.5394,
      "step": 4055
    },
    {
      "epoch": 0.703678001387925,
      "grad_norm": 0.7806961536407471,
      "learning_rate": 3.654995549114913e-05,
      "loss": 0.5581,
      "step": 4056
    },
    {
      "epoch": 0.703851492019431,
      "grad_norm": 0.803566038608551,
      "learning_rate": 3.654612909851882e-05,
      "loss": 0.4973,
      "step": 4057
    },
    {
      "epoch": 0.7040249826509368,
      "grad_norm": 0.998016893863678,
      "learning_rate": 3.6542300785690954e-05,
      "loss": 0.5757,
      "step": 4058
    },
    {
      "epoch": 0.7041984732824428,
      "grad_norm": 0.49902257323265076,
      "learning_rate": 3.653847055310981e-05,
      "loss": 0.6404,
      "step": 4059
    },
    {
      "epoch": 0.7043719639139486,
      "grad_norm": 1.0214604139328003,
      "learning_rate": 3.6534638401219874e-05,
      "loss": 0.5183,
      "step": 4060
    },
    {
      "epoch": 0.7045454545454546,
      "grad_norm": 1.037064552307129,
      "learning_rate": 3.653080433046589e-05,
      "loss": 0.4678,
      "step": 4061
    },
    {
      "epoch": 0.7047189451769604,
      "grad_norm": 0.6829334497451782,
      "learning_rate": 3.652696834129281e-05,
      "loss": 0.5582,
      "step": 4062
    },
    {
      "epoch": 0.7048924358084664,
      "grad_norm": 1.581860065460205,
      "learning_rate": 3.652313043414579e-05,
      "loss": 0.5304,
      "step": 4063
    },
    {
      "epoch": 0.7050659264399722,
      "grad_norm": 0.5808870792388916,
      "learning_rate": 3.6519290609470225e-05,
      "loss": 0.6938,
      "step": 4064
    },
    {
      "epoch": 0.7052394170714782,
      "grad_norm": 0.6855258941650391,
      "learning_rate": 3.651544886771174e-05,
      "loss": 0.5574,
      "step": 4065
    },
    {
      "epoch": 0.705412907702984,
      "grad_norm": 0.970683217048645,
      "learning_rate": 3.651160520931617e-05,
      "loss": 0.5873,
      "step": 4066
    },
    {
      "epoch": 0.7055863983344899,
      "grad_norm": 0.6593789458274841,
      "learning_rate": 3.650775963472958e-05,
      "loss": 0.5842,
      "step": 4067
    },
    {
      "epoch": 0.7057598889659958,
      "grad_norm": 0.8516277074813843,
      "learning_rate": 3.650391214439825e-05,
      "loss": 0.5527,
      "step": 4068
    },
    {
      "epoch": 0.7059333795975017,
      "grad_norm": 0.7127820253372192,
      "learning_rate": 3.6500062738768675e-05,
      "loss": 0.598,
      "step": 4069
    },
    {
      "epoch": 0.7061068702290076,
      "grad_norm": 0.6315264105796814,
      "learning_rate": 3.64962114182876e-05,
      "loss": 0.563,
      "step": 4070
    },
    {
      "epoch": 0.7062803608605135,
      "grad_norm": 0.6952970623970032,
      "learning_rate": 3.649235818340197e-05,
      "loss": 0.5381,
      "step": 4071
    },
    {
      "epoch": 0.7064538514920194,
      "grad_norm": 0.9195808172225952,
      "learning_rate": 3.648850303455895e-05,
      "loss": 0.5012,
      "step": 4072
    },
    {
      "epoch": 0.7066273421235253,
      "grad_norm": 0.7608962655067444,
      "learning_rate": 3.648464597220594e-05,
      "loss": 0.6113,
      "step": 4073
    },
    {
      "epoch": 0.7068008327550312,
      "grad_norm": 0.8608875274658203,
      "learning_rate": 3.6480786996790554e-05,
      "loss": 0.4757,
      "step": 4074
    },
    {
      "epoch": 0.7069743233865371,
      "grad_norm": 0.5480713844299316,
      "learning_rate": 3.647692610876064e-05,
      "loss": 0.5889,
      "step": 4075
    },
    {
      "epoch": 0.707147814018043,
      "grad_norm": 0.7620058655738831,
      "learning_rate": 3.647306330856425e-05,
      "loss": 0.5334,
      "step": 4076
    },
    {
      "epoch": 0.7073213046495489,
      "grad_norm": 0.7351611256599426,
      "learning_rate": 3.6469198596649663e-05,
      "loss": 0.6351,
      "step": 4077
    },
    {
      "epoch": 0.7074947952810549,
      "grad_norm": 0.877390444278717,
      "learning_rate": 3.646533197346539e-05,
      "loss": 0.6279,
      "step": 4078
    },
    {
      "epoch": 0.7076682859125607,
      "grad_norm": 0.5145160555839539,
      "learning_rate": 3.6461463439460156e-05,
      "loss": 0.5768,
      "step": 4079
    },
    {
      "epoch": 0.7078417765440667,
      "grad_norm": 0.6730548739433289,
      "learning_rate": 3.6457592995082915e-05,
      "loss": 0.6096,
      "step": 4080
    },
    {
      "epoch": 0.7080152671755725,
      "grad_norm": 0.6766643524169922,
      "learning_rate": 3.645372064078282e-05,
      "loss": 0.5292,
      "step": 4081
    },
    {
      "epoch": 0.7081887578070785,
      "grad_norm": 0.7172795534133911,
      "learning_rate": 3.644984637700928e-05,
      "loss": 0.6473,
      "step": 4082
    },
    {
      "epoch": 0.7083622484385843,
      "grad_norm": 0.6874622702598572,
      "learning_rate": 3.644597020421189e-05,
      "loss": 0.5645,
      "step": 4083
    },
    {
      "epoch": 0.7085357390700903,
      "grad_norm": 0.8783832788467407,
      "learning_rate": 3.6442092122840505e-05,
      "loss": 0.5518,
      "step": 4084
    },
    {
      "epoch": 0.7087092297015961,
      "grad_norm": 0.7163074016571045,
      "learning_rate": 3.6438212133345164e-05,
      "loss": 0.5182,
      "step": 4085
    },
    {
      "epoch": 0.7088827203331021,
      "grad_norm": 0.7176882028579712,
      "learning_rate": 3.643433023617616e-05,
      "loss": 0.6757,
      "step": 4086
    },
    {
      "epoch": 0.7090562109646079,
      "grad_norm": 0.6373384594917297,
      "learning_rate": 3.643044643178397e-05,
      "loss": 0.598,
      "step": 4087
    },
    {
      "epoch": 0.7092297015961138,
      "grad_norm": 0.6766403317451477,
      "learning_rate": 3.642656072061933e-05,
      "loss": 0.6661,
      "step": 4088
    },
    {
      "epoch": 0.7094031922276197,
      "grad_norm": 0.9148829579353333,
      "learning_rate": 3.6422673103133186e-05,
      "loss": 0.5396,
      "step": 4089
    },
    {
      "epoch": 0.7095766828591256,
      "grad_norm": 0.7896472811698914,
      "learning_rate": 3.641878357977668e-05,
      "loss": 0.5364,
      "step": 4090
    },
    {
      "epoch": 0.7097501734906315,
      "grad_norm": 0.7675262689590454,
      "learning_rate": 3.641489215100122e-05,
      "loss": 0.5026,
      "step": 4091
    },
    {
      "epoch": 0.7099236641221374,
      "grad_norm": 0.7292678952217102,
      "learning_rate": 3.641099881725839e-05,
      "loss": 0.6086,
      "step": 4092
    },
    {
      "epoch": 0.7100971547536433,
      "grad_norm": 0.9237269759178162,
      "learning_rate": 3.6407103579000024e-05,
      "loss": 0.5127,
      "step": 4093
    },
    {
      "epoch": 0.7102706453851492,
      "grad_norm": 0.7068149447441101,
      "learning_rate": 3.6403206436678173e-05,
      "loss": 0.4916,
      "step": 4094
    },
    {
      "epoch": 0.7104441360166551,
      "grad_norm": 1.187597393989563,
      "learning_rate": 3.63993073907451e-05,
      "loss": 0.4594,
      "step": 4095
    },
    {
      "epoch": 0.710617626648161,
      "grad_norm": 1.7699358463287354,
      "learning_rate": 3.63954064416533e-05,
      "loss": 0.5876,
      "step": 4096
    },
    {
      "epoch": 0.7107911172796669,
      "grad_norm": 0.8028684258460999,
      "learning_rate": 3.639150358985547e-05,
      "loss": 0.5413,
      "step": 4097
    },
    {
      "epoch": 0.7109646079111728,
      "grad_norm": 0.7019853591918945,
      "learning_rate": 3.6387598835804555e-05,
      "loss": 0.6101,
      "step": 4098
    },
    {
      "epoch": 0.7111380985426787,
      "grad_norm": 0.8166154623031616,
      "learning_rate": 3.63836921799537e-05,
      "loss": 0.588,
      "step": 4099
    },
    {
      "epoch": 0.7113115891741846,
      "grad_norm": 1.1425904035568237,
      "learning_rate": 3.6379783622756275e-05,
      "loss": 0.611,
      "step": 4100
    },
    {
      "epoch": 0.7114850798056905,
      "grad_norm": 0.6073325276374817,
      "learning_rate": 3.637587316466587e-05,
      "loss": 0.5802,
      "step": 4101
    },
    {
      "epoch": 0.7116585704371964,
      "grad_norm": 0.864340603351593,
      "learning_rate": 3.6371960806136313e-05,
      "loss": 0.5504,
      "step": 4102
    },
    {
      "epoch": 0.7118320610687023,
      "grad_norm": 0.8540159463882446,
      "learning_rate": 3.636804654762162e-05,
      "loss": 0.5891,
      "step": 4103
    },
    {
      "epoch": 0.7120055517002082,
      "grad_norm": 0.9849295616149902,
      "learning_rate": 3.636413038957605e-05,
      "loss": 0.5096,
      "step": 4104
    },
    {
      "epoch": 0.7121790423317141,
      "grad_norm": 0.8594913482666016,
      "learning_rate": 3.6360212332454087e-05,
      "loss": 0.5494,
      "step": 4105
    },
    {
      "epoch": 0.71235253296322,
      "grad_norm": 0.7813236713409424,
      "learning_rate": 3.635629237671041e-05,
      "loss": 0.5718,
      "step": 4106
    },
    {
      "epoch": 0.7125260235947258,
      "grad_norm": 0.8800257444381714,
      "learning_rate": 3.6352370522799956e-05,
      "loss": 0.5593,
      "step": 4107
    },
    {
      "epoch": 0.7126995142262318,
      "grad_norm": 0.680556058883667,
      "learning_rate": 3.634844677117784e-05,
      "loss": 0.5435,
      "step": 4108
    },
    {
      "epoch": 0.7128730048577376,
      "grad_norm": 0.6794553399085999,
      "learning_rate": 3.634452112229942e-05,
      "loss": 0.5635,
      "step": 4109
    },
    {
      "epoch": 0.7130464954892436,
      "grad_norm": 0.762883722782135,
      "learning_rate": 3.63405935766203e-05,
      "loss": 0.547,
      "step": 4110
    },
    {
      "epoch": 0.7132199861207494,
      "grad_norm": 0.7484919428825378,
      "learning_rate": 3.633666413459624e-05,
      "loss": 0.5802,
      "step": 4111
    },
    {
      "epoch": 0.7133934767522554,
      "grad_norm": 0.7314563393592834,
      "learning_rate": 3.633273279668327e-05,
      "loss": 0.5463,
      "step": 4112
    },
    {
      "epoch": 0.7135669673837612,
      "grad_norm": 0.7025768160820007,
      "learning_rate": 3.632879956333763e-05,
      "loss": 0.5608,
      "step": 4113
    },
    {
      "epoch": 0.7137404580152672,
      "grad_norm": 0.6107587218284607,
      "learning_rate": 3.632486443501578e-05,
      "loss": 0.5657,
      "step": 4114
    },
    {
      "epoch": 0.713913948646773,
      "grad_norm": 0.6207438111305237,
      "learning_rate": 3.632092741217438e-05,
      "loss": 0.6782,
      "step": 4115
    },
    {
      "epoch": 0.714087439278279,
      "grad_norm": 0.7434753775596619,
      "learning_rate": 3.631698849527034e-05,
      "loss": 0.6816,
      "step": 4116
    },
    {
      "epoch": 0.7142609299097848,
      "grad_norm": 0.7651646733283997,
      "learning_rate": 3.631304768476078e-05,
      "loss": 0.5726,
      "step": 4117
    },
    {
      "epoch": 0.7144344205412908,
      "grad_norm": 0.6007829904556274,
      "learning_rate": 3.630910498110302e-05,
      "loss": 0.5927,
      "step": 4118
    },
    {
      "epoch": 0.7146079111727967,
      "grad_norm": 0.6312075853347778,
      "learning_rate": 3.630516038475462e-05,
      "loss": 0.6816,
      "step": 4119
    },
    {
      "epoch": 0.7147814018043026,
      "grad_norm": 0.9027377963066101,
      "learning_rate": 3.630121389617336e-05,
      "loss": 0.6041,
      "step": 4120
    },
    {
      "epoch": 0.7149548924358085,
      "grad_norm": 0.628914475440979,
      "learning_rate": 3.6297265515817234e-05,
      "loss": 0.6572,
      "step": 4121
    },
    {
      "epoch": 0.7151283830673144,
      "grad_norm": 0.7190338969230652,
      "learning_rate": 3.629331524414446e-05,
      "loss": 0.5398,
      "step": 4122
    },
    {
      "epoch": 0.7153018736988203,
      "grad_norm": 0.9281529784202576,
      "learning_rate": 3.628936308161346e-05,
      "loss": 0.6353,
      "step": 4123
    },
    {
      "epoch": 0.7154753643303262,
      "grad_norm": 0.8841820955276489,
      "learning_rate": 3.6285409028682895e-05,
      "loss": 0.6016,
      "step": 4124
    },
    {
      "epoch": 0.7156488549618321,
      "grad_norm": 0.7699342370033264,
      "learning_rate": 3.6281453085811634e-05,
      "loss": 0.5547,
      "step": 4125
    },
    {
      "epoch": 0.7158223455933379,
      "grad_norm": 0.8891981244087219,
      "learning_rate": 3.627749525345878e-05,
      "loss": 0.5593,
      "step": 4126
    },
    {
      "epoch": 0.7159958362248439,
      "grad_norm": 0.8777641654014587,
      "learning_rate": 3.627353553208362e-05,
      "loss": 0.5215,
      "step": 4127
    },
    {
      "epoch": 0.7161693268563497,
      "grad_norm": 0.6126023530960083,
      "learning_rate": 3.626957392214571e-05,
      "loss": 0.5089,
      "step": 4128
    },
    {
      "epoch": 0.7163428174878557,
      "grad_norm": 1.155997395515442,
      "learning_rate": 3.626561042410479e-05,
      "loss": 0.5682,
      "step": 4129
    },
    {
      "epoch": 0.7165163081193615,
      "grad_norm": 0.8990711569786072,
      "learning_rate": 3.626164503842082e-05,
      "loss": 0.4742,
      "step": 4130
    },
    {
      "epoch": 0.7166897987508675,
      "grad_norm": 0.8453020453453064,
      "learning_rate": 3.6257677765553996e-05,
      "loss": 0.5734,
      "step": 4131
    },
    {
      "epoch": 0.7168632893823733,
      "grad_norm": 0.8402464389801025,
      "learning_rate": 3.6253708605964724e-05,
      "loss": 0.6011,
      "step": 4132
    },
    {
      "epoch": 0.7170367800138793,
      "grad_norm": 0.7821096777915955,
      "learning_rate": 3.624973756011363e-05,
      "loss": 0.5906,
      "step": 4133
    },
    {
      "epoch": 0.7172102706453851,
      "grad_norm": 0.5626051425933838,
      "learning_rate": 3.6245764628461556e-05,
      "loss": 0.6475,
      "step": 4134
    },
    {
      "epoch": 0.7173837612768911,
      "grad_norm": 0.6475934982299805,
      "learning_rate": 3.624178981146956e-05,
      "loss": 0.5848,
      "step": 4135
    },
    {
      "epoch": 0.7175572519083969,
      "grad_norm": 0.88841313123703,
      "learning_rate": 3.6237813109598944e-05,
      "loss": 0.7047,
      "step": 4136
    },
    {
      "epoch": 0.7177307425399029,
      "grad_norm": 0.6983802914619446,
      "learning_rate": 3.623383452331119e-05,
      "loss": 0.5806,
      "step": 4137
    },
    {
      "epoch": 0.7179042331714087,
      "grad_norm": 0.7312623858451843,
      "learning_rate": 3.622985405306803e-05,
      "loss": 0.5026,
      "step": 4138
    },
    {
      "epoch": 0.7180777238029147,
      "grad_norm": 1.1107406616210938,
      "learning_rate": 3.622587169933138e-05,
      "loss": 0.5413,
      "step": 4139
    },
    {
      "epoch": 0.7182512144344205,
      "grad_norm": 0.672258198261261,
      "learning_rate": 3.622188746256343e-05,
      "loss": 0.5979,
      "step": 4140
    },
    {
      "epoch": 0.7184247050659265,
      "grad_norm": 0.5452670454978943,
      "learning_rate": 3.6217901343226526e-05,
      "loss": 0.6443,
      "step": 4141
    },
    {
      "epoch": 0.7185981956974323,
      "grad_norm": 0.6215023398399353,
      "learning_rate": 3.621391334178328e-05,
      "loss": 0.6107,
      "step": 4142
    },
    {
      "epoch": 0.7187716863289383,
      "grad_norm": 0.8545822501182556,
      "learning_rate": 3.620992345869649e-05,
      "loss": 0.6942,
      "step": 4143
    },
    {
      "epoch": 0.7189451769604441,
      "grad_norm": 1.0239531993865967,
      "learning_rate": 3.62059316944292e-05,
      "loss": 0.6053,
      "step": 4144
    },
    {
      "epoch": 0.7191186675919501,
      "grad_norm": 0.798069953918457,
      "learning_rate": 3.6201938049444654e-05,
      "loss": 0.5001,
      "step": 4145
    },
    {
      "epoch": 0.7192921582234559,
      "grad_norm": 0.8938401937484741,
      "learning_rate": 3.619794252420632e-05,
      "loss": 0.5348,
      "step": 4146
    },
    {
      "epoch": 0.7194656488549618,
      "grad_norm": 0.8637401461601257,
      "learning_rate": 3.619394511917788e-05,
      "loss": 0.6536,
      "step": 4147
    },
    {
      "epoch": 0.7196391394864677,
      "grad_norm": 0.9881258010864258,
      "learning_rate": 3.618994583482323e-05,
      "loss": 0.5515,
      "step": 4148
    },
    {
      "epoch": 0.7198126301179736,
      "grad_norm": 0.9566797018051147,
      "learning_rate": 3.618594467160651e-05,
      "loss": 0.6193,
      "step": 4149
    },
    {
      "epoch": 0.7199861207494795,
      "grad_norm": 1.0019598007202148,
      "learning_rate": 3.618194162999205e-05,
      "loss": 0.7379,
      "step": 4150
    },
    {
      "epoch": 0.7201596113809854,
      "grad_norm": 1.4422179460525513,
      "learning_rate": 3.617793671044441e-05,
      "loss": 0.5209,
      "step": 4151
    },
    {
      "epoch": 0.7203331020124913,
      "grad_norm": 0.6101250648498535,
      "learning_rate": 3.617392991342836e-05,
      "loss": 0.5112,
      "step": 4152
    },
    {
      "epoch": 0.7205065926439972,
      "grad_norm": 0.7908729910850525,
      "learning_rate": 3.6169921239408894e-05,
      "loss": 0.5563,
      "step": 4153
    },
    {
      "epoch": 0.7206800832755031,
      "grad_norm": 0.8282957673072815,
      "learning_rate": 3.616591068885123e-05,
      "loss": 0.5977,
      "step": 4154
    },
    {
      "epoch": 0.720853573907009,
      "grad_norm": 0.7281785011291504,
      "learning_rate": 3.61618982622208e-05,
      "loss": 0.5371,
      "step": 4155
    },
    {
      "epoch": 0.721027064538515,
      "grad_norm": 0.5883955955505371,
      "learning_rate": 3.6157883959983234e-05,
      "loss": 0.6997,
      "step": 4156
    },
    {
      "epoch": 0.7212005551700208,
      "grad_norm": 0.8235392570495605,
      "learning_rate": 3.615386778260441e-05,
      "loss": 0.582,
      "step": 4157
    },
    {
      "epoch": 0.7213740458015268,
      "grad_norm": 0.762932538986206,
      "learning_rate": 3.614984973055041e-05,
      "loss": 0.5698,
      "step": 4158
    },
    {
      "epoch": 0.7215475364330326,
      "grad_norm": 0.9984418153762817,
      "learning_rate": 3.6145829804287526e-05,
      "loss": 0.5438,
      "step": 4159
    },
    {
      "epoch": 0.7217210270645386,
      "grad_norm": 0.5380730032920837,
      "learning_rate": 3.614180800428228e-05,
      "loss": 0.6814,
      "step": 4160
    },
    {
      "epoch": 0.7218945176960444,
      "grad_norm": 0.65350341796875,
      "learning_rate": 3.613778433100141e-05,
      "loss": 0.6564,
      "step": 4161
    },
    {
      "epoch": 0.7220680083275504,
      "grad_norm": 0.7141516804695129,
      "learning_rate": 3.6133758784911864e-05,
      "loss": 0.5327,
      "step": 4162
    },
    {
      "epoch": 0.7222414989590562,
      "grad_norm": 0.8622221350669861,
      "learning_rate": 3.612973136648081e-05,
      "loss": 0.6542,
      "step": 4163
    },
    {
      "epoch": 0.7224149895905622,
      "grad_norm": 0.6311085224151611,
      "learning_rate": 3.6125702076175636e-05,
      "loss": 0.6011,
      "step": 4164
    },
    {
      "epoch": 0.722588480222068,
      "grad_norm": 0.929018497467041,
      "learning_rate": 3.612167091446394e-05,
      "loss": 0.6543,
      "step": 4165
    },
    {
      "epoch": 0.7227619708535739,
      "grad_norm": 0.7394356727600098,
      "learning_rate": 3.611763788181356e-05,
      "loss": 0.5372,
      "step": 4166
    },
    {
      "epoch": 0.7229354614850798,
      "grad_norm": 0.8344247937202454,
      "learning_rate": 3.6113602978692514e-05,
      "loss": 0.5393,
      "step": 4167
    },
    {
      "epoch": 0.7231089521165857,
      "grad_norm": 0.6699580550193787,
      "learning_rate": 3.610956620556907e-05,
      "loss": 0.491,
      "step": 4168
    },
    {
      "epoch": 0.7232824427480916,
      "grad_norm": 1.0480327606201172,
      "learning_rate": 3.61055275629117e-05,
      "loss": 0.5978,
      "step": 4169
    },
    {
      "epoch": 0.7234559333795975,
      "grad_norm": 0.6671690940856934,
      "learning_rate": 3.610148705118908e-05,
      "loss": 0.6597,
      "step": 4170
    },
    {
      "epoch": 0.7236294240111034,
      "grad_norm": 0.7958476543426514,
      "learning_rate": 3.6097444670870136e-05,
      "loss": 0.635,
      "step": 4171
    },
    {
      "epoch": 0.7238029146426093,
      "grad_norm": 0.6575822830200195,
      "learning_rate": 3.609340042242397e-05,
      "loss": 0.5558,
      "step": 4172
    },
    {
      "epoch": 0.7239764052741152,
      "grad_norm": 0.7265298366546631,
      "learning_rate": 3.608935430631994e-05,
      "loss": 0.5389,
      "step": 4173
    },
    {
      "epoch": 0.7241498959056211,
      "grad_norm": 0.7857984900474548,
      "learning_rate": 3.6085306323027596e-05,
      "loss": 0.5946,
      "step": 4174
    },
    {
      "epoch": 0.724323386537127,
      "grad_norm": 0.7398806214332581,
      "learning_rate": 3.608125647301671e-05,
      "loss": 0.53,
      "step": 4175
    },
    {
      "epoch": 0.7244968771686329,
      "grad_norm": 0.5658463835716248,
      "learning_rate": 3.607720475675727e-05,
      "loss": 0.5669,
      "step": 4176
    },
    {
      "epoch": 0.7246703678001388,
      "grad_norm": 0.6145365834236145,
      "learning_rate": 3.607315117471948e-05,
      "loss": 0.647,
      "step": 4177
    },
    {
      "epoch": 0.7248438584316447,
      "grad_norm": 0.8134177923202515,
      "learning_rate": 3.606909572737378e-05,
      "loss": 0.5913,
      "step": 4178
    },
    {
      "epoch": 0.7250173490631506,
      "grad_norm": 0.8853358626365662,
      "learning_rate": 3.6065038415190775e-05,
      "loss": 0.6432,
      "step": 4179
    },
    {
      "epoch": 0.7251908396946565,
      "grad_norm": 0.5950773358345032,
      "learning_rate": 3.6060979238641363e-05,
      "loss": 0.5945,
      "step": 4180
    },
    {
      "epoch": 0.7253643303261624,
      "grad_norm": 0.6152825951576233,
      "learning_rate": 3.605691819819659e-05,
      "loss": 0.5674,
      "step": 4181
    },
    {
      "epoch": 0.7255378209576683,
      "grad_norm": 0.8700108528137207,
      "learning_rate": 3.6052855294327746e-05,
      "loss": 0.6201,
      "step": 4182
    },
    {
      "epoch": 0.7257113115891742,
      "grad_norm": 0.7685799598693848,
      "learning_rate": 3.604879052750634e-05,
      "loss": 0.5131,
      "step": 4183
    },
    {
      "epoch": 0.7258848022206801,
      "grad_norm": 1.0418882369995117,
      "learning_rate": 3.604472389820409e-05,
      "loss": 0.5507,
      "step": 4184
    },
    {
      "epoch": 0.7260582928521859,
      "grad_norm": 0.6010865569114685,
      "learning_rate": 3.604065540689295e-05,
      "loss": 0.5939,
      "step": 4185
    },
    {
      "epoch": 0.7262317834836919,
      "grad_norm": 0.9228698015213013,
      "learning_rate": 3.6036585054045044e-05,
      "loss": 0.5513,
      "step": 4186
    },
    {
      "epoch": 0.7264052741151977,
      "grad_norm": 1.3044053316116333,
      "learning_rate": 3.603251284013276e-05,
      "loss": 0.5715,
      "step": 4187
    },
    {
      "epoch": 0.7265787647467037,
      "grad_norm": 0.7584335207939148,
      "learning_rate": 3.602843876562868e-05,
      "loss": 0.4867,
      "step": 4188
    },
    {
      "epoch": 0.7267522553782095,
      "grad_norm": 0.6108629703521729,
      "learning_rate": 3.602436283100561e-05,
      "loss": 0.6052,
      "step": 4189
    },
    {
      "epoch": 0.7269257460097155,
      "grad_norm": 0.6309239864349365,
      "learning_rate": 3.6020285036736554e-05,
      "loss": 0.6798,
      "step": 4190
    },
    {
      "epoch": 0.7270992366412213,
      "grad_norm": 0.7907925844192505,
      "learning_rate": 3.601620538329476e-05,
      "loss": 0.5789,
      "step": 4191
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.6239023208618164,
      "learning_rate": 3.601212387115366e-05,
      "loss": 0.5896,
      "step": 4192
    },
    {
      "epoch": 0.7274462179042331,
      "grad_norm": 0.6699426770210266,
      "learning_rate": 3.6008040500786926e-05,
      "loss": 0.5713,
      "step": 4193
    },
    {
      "epoch": 0.7276197085357391,
      "grad_norm": 1.3908271789550781,
      "learning_rate": 3.6003955272668444e-05,
      "loss": 0.6373,
      "step": 4194
    },
    {
      "epoch": 0.7277931991672449,
      "grad_norm": 0.7028026580810547,
      "learning_rate": 3.599986818727231e-05,
      "loss": 0.5668,
      "step": 4195
    },
    {
      "epoch": 0.7279666897987509,
      "grad_norm": 0.5071716904640198,
      "learning_rate": 3.5995779245072816e-05,
      "loss": 0.589,
      "step": 4196
    },
    {
      "epoch": 0.7281401804302567,
      "grad_norm": 0.5421550869941711,
      "learning_rate": 3.599168844654451e-05,
      "loss": 0.7155,
      "step": 4197
    },
    {
      "epoch": 0.7283136710617627,
      "grad_norm": 0.5536943078041077,
      "learning_rate": 3.5987595792162126e-05,
      "loss": 0.5199,
      "step": 4198
    },
    {
      "epoch": 0.7284871616932685,
      "grad_norm": 0.7120516300201416,
      "learning_rate": 3.5983501282400617e-05,
      "loss": 0.6636,
      "step": 4199
    },
    {
      "epoch": 0.7286606523247745,
      "grad_norm": 1.147731900215149,
      "learning_rate": 3.597940491773516e-05,
      "loss": 0.5612,
      "step": 4200
    },
    {
      "epoch": 0.7288341429562804,
      "grad_norm": 0.6481512188911438,
      "learning_rate": 3.597530669864115e-05,
      "loss": 0.6503,
      "step": 4201
    },
    {
      "epoch": 0.7290076335877863,
      "grad_norm": 0.8503965139389038,
      "learning_rate": 3.5971206625594176e-05,
      "loss": 0.5345,
      "step": 4202
    },
    {
      "epoch": 0.7291811242192922,
      "grad_norm": 0.49129992723464966,
      "learning_rate": 3.596710469907006e-05,
      "loss": 0.6738,
      "step": 4203
    },
    {
      "epoch": 0.7293546148507981,
      "grad_norm": 0.6439663171768188,
      "learning_rate": 3.5963000919544844e-05,
      "loss": 0.6027,
      "step": 4204
    },
    {
      "epoch": 0.729528105482304,
      "grad_norm": 0.7014073729515076,
      "learning_rate": 3.595889528749477e-05,
      "loss": 0.6725,
      "step": 4205
    },
    {
      "epoch": 0.7297015961138098,
      "grad_norm": 0.6441563963890076,
      "learning_rate": 3.59547878033963e-05,
      "loss": 0.5385,
      "step": 4206
    },
    {
      "epoch": 0.7298750867453158,
      "grad_norm": 0.6568038463592529,
      "learning_rate": 3.595067846772612e-05,
      "loss": 0.5377,
      "step": 4207
    },
    {
      "epoch": 0.7300485773768216,
      "grad_norm": 0.7476823925971985,
      "learning_rate": 3.594656728096111e-05,
      "loss": 0.5637,
      "step": 4208
    },
    {
      "epoch": 0.7302220680083276,
      "grad_norm": 0.7711260914802551,
      "learning_rate": 3.594245424357839e-05,
      "loss": 0.5879,
      "step": 4209
    },
    {
      "epoch": 0.7303955586398334,
      "grad_norm": 0.6500536203384399,
      "learning_rate": 3.5938339356055274e-05,
      "loss": 0.6674,
      "step": 4210
    },
    {
      "epoch": 0.7305690492713394,
      "grad_norm": 0.6398093700408936,
      "learning_rate": 3.593422261886931e-05,
      "loss": 0.5714,
      "step": 4211
    },
    {
      "epoch": 0.7307425399028452,
      "grad_norm": 0.9162209630012512,
      "learning_rate": 3.593010403249824e-05,
      "loss": 0.6091,
      "step": 4212
    },
    {
      "epoch": 0.7309160305343512,
      "grad_norm": 0.7451730370521545,
      "learning_rate": 3.592598359742004e-05,
      "loss": 0.549,
      "step": 4213
    },
    {
      "epoch": 0.731089521165857,
      "grad_norm": 0.6620838642120361,
      "learning_rate": 3.592186131411288e-05,
      "loss": 0.5551,
      "step": 4214
    },
    {
      "epoch": 0.731263011797363,
      "grad_norm": 0.603359043598175,
      "learning_rate": 3.591773718305517e-05,
      "loss": 0.7273,
      "step": 4215
    },
    {
      "epoch": 0.7314365024288688,
      "grad_norm": 0.637779176235199,
      "learning_rate": 3.5913611204725496e-05,
      "loss": 0.6625,
      "step": 4216
    },
    {
      "epoch": 0.7316099930603748,
      "grad_norm": 0.8620924949645996,
      "learning_rate": 3.590948337960271e-05,
      "loss": 0.5035,
      "step": 4217
    },
    {
      "epoch": 0.7317834836918806,
      "grad_norm": 0.6196661591529846,
      "learning_rate": 3.590535370816584e-05,
      "loss": 0.4924,
      "step": 4218
    },
    {
      "epoch": 0.7319569743233866,
      "grad_norm": 0.7694399356842041,
      "learning_rate": 3.5901222190894136e-05,
      "loss": 0.6069,
      "step": 4219
    },
    {
      "epoch": 0.7321304649548924,
      "grad_norm": 0.7322472333908081,
      "learning_rate": 3.589708882826707e-05,
      "loss": 0.6136,
      "step": 4220
    },
    {
      "epoch": 0.7323039555863984,
      "grad_norm": 0.9868308901786804,
      "learning_rate": 3.589295362076432e-05,
      "loss": 0.5369,
      "step": 4221
    },
    {
      "epoch": 0.7324774462179042,
      "grad_norm": 0.8929972648620605,
      "learning_rate": 3.588881656886578e-05,
      "loss": 0.5729,
      "step": 4222
    },
    {
      "epoch": 0.7326509368494102,
      "grad_norm": 0.9018761515617371,
      "learning_rate": 3.588467767305157e-05,
      "loss": 0.5753,
      "step": 4223
    },
    {
      "epoch": 0.732824427480916,
      "grad_norm": 0.6200088262557983,
      "learning_rate": 3.5880536933802e-05,
      "loss": 0.6248,
      "step": 4224
    },
    {
      "epoch": 0.7329979181124219,
      "grad_norm": 0.5684300065040588,
      "learning_rate": 3.587639435159762e-05,
      "loss": 0.6528,
      "step": 4225
    },
    {
      "epoch": 0.7331714087439278,
      "grad_norm": 0.8975379467010498,
      "learning_rate": 3.587224992691917e-05,
      "loss": 0.5836,
      "step": 4226
    },
    {
      "epoch": 0.7333448993754337,
      "grad_norm": 0.6046488285064697,
      "learning_rate": 3.586810366024763e-05,
      "loss": 0.5603,
      "step": 4227
    },
    {
      "epoch": 0.7335183900069396,
      "grad_norm": 0.9245457649230957,
      "learning_rate": 3.586395555206417e-05,
      "loss": 0.5519,
      "step": 4228
    },
    {
      "epoch": 0.7336918806384455,
      "grad_norm": 0.681410014629364,
      "learning_rate": 3.585980560285017e-05,
      "loss": 0.5819,
      "step": 4229
    },
    {
      "epoch": 0.7338653712699514,
      "grad_norm": 0.8801109194755554,
      "learning_rate": 3.585565381308726e-05,
      "loss": 0.5975,
      "step": 4230
    },
    {
      "epoch": 0.7340388619014573,
      "grad_norm": 0.6511202454566956,
      "learning_rate": 3.5851500183257246e-05,
      "loss": 0.5861,
      "step": 4231
    },
    {
      "epoch": 0.7342123525329632,
      "grad_norm": 0.8441863656044006,
      "learning_rate": 3.584734471384217e-05,
      "loss": 0.5149,
      "step": 4232
    },
    {
      "epoch": 0.7343858431644691,
      "grad_norm": 0.7297527194023132,
      "learning_rate": 3.5843187405324266e-05,
      "loss": 0.5925,
      "step": 4233
    },
    {
      "epoch": 0.734559333795975,
      "grad_norm": 0.7790377140045166,
      "learning_rate": 3.5839028258186014e-05,
      "loss": 0.5204,
      "step": 4234
    },
    {
      "epoch": 0.7347328244274809,
      "grad_norm": 0.6037988662719727,
      "learning_rate": 3.583486727291007e-05,
      "loss": 0.6438,
      "step": 4235
    },
    {
      "epoch": 0.7349063150589868,
      "grad_norm": 0.9271818399429321,
      "learning_rate": 3.583070444997932e-05,
      "loss": 0.4866,
      "step": 4236
    },
    {
      "epoch": 0.7350798056904927,
      "grad_norm": 0.5512932538986206,
      "learning_rate": 3.5826539789876885e-05,
      "loss": 0.6483,
      "step": 4237
    },
    {
      "epoch": 0.7352532963219987,
      "grad_norm": 1.7557094097137451,
      "learning_rate": 3.5822373293086055e-05,
      "loss": 0.6453,
      "step": 4238
    },
    {
      "epoch": 0.7354267869535045,
      "grad_norm": 0.8460390567779541,
      "learning_rate": 3.581820496009038e-05,
      "loss": 0.559,
      "step": 4239
    },
    {
      "epoch": 0.7356002775850105,
      "grad_norm": 0.6969012022018433,
      "learning_rate": 3.581403479137358e-05,
      "loss": 0.5341,
      "step": 4240
    },
    {
      "epoch": 0.7357737682165163,
      "grad_norm": 0.6169527769088745,
      "learning_rate": 3.580986278741961e-05,
      "loss": 0.709,
      "step": 4241
    },
    {
      "epoch": 0.7359472588480223,
      "grad_norm": 0.6563184261322021,
      "learning_rate": 3.580568894871265e-05,
      "loss": 0.7075,
      "step": 4242
    },
    {
      "epoch": 0.7361207494795281,
      "grad_norm": 0.6531060338020325,
      "learning_rate": 3.580151327573707e-05,
      "loss": 0.6005,
      "step": 4243
    },
    {
      "epoch": 0.7362942401110341,
      "grad_norm": 0.7275852560997009,
      "learning_rate": 3.579733576897746e-05,
      "loss": 0.5295,
      "step": 4244
    },
    {
      "epoch": 0.7364677307425399,
      "grad_norm": 0.6438325643539429,
      "learning_rate": 3.579315642891862e-05,
      "loss": 0.681,
      "step": 4245
    },
    {
      "epoch": 0.7366412213740458,
      "grad_norm": 0.6312209963798523,
      "learning_rate": 3.578897525604558e-05,
      "loss": 0.7383,
      "step": 4246
    },
    {
      "epoch": 0.7368147120055517,
      "grad_norm": 1.2075726985931396,
      "learning_rate": 3.5784792250843564e-05,
      "loss": 0.5753,
      "step": 4247
    },
    {
      "epoch": 0.7369882026370576,
      "grad_norm": 0.7225402593612671,
      "learning_rate": 3.578060741379801e-05,
      "loss": 0.6145,
      "step": 4248
    },
    {
      "epoch": 0.7371616932685635,
      "grad_norm": 0.685030996799469,
      "learning_rate": 3.5776420745394584e-05,
      "loss": 0.5864,
      "step": 4249
    },
    {
      "epoch": 0.7373351839000694,
      "grad_norm": 0.668433666229248,
      "learning_rate": 3.577223224611915e-05,
      "loss": 0.6794,
      "step": 4250
    },
    {
      "epoch": 0.7375086745315753,
      "grad_norm": 0.6833786964416504,
      "learning_rate": 3.576804191645778e-05,
      "loss": 0.6241,
      "step": 4251
    },
    {
      "epoch": 0.7376821651630812,
      "grad_norm": 0.529434323310852,
      "learning_rate": 3.576384975689677e-05,
      "loss": 0.6364,
      "step": 4252
    },
    {
      "epoch": 0.7378556557945871,
      "grad_norm": 1.049576997756958,
      "learning_rate": 3.5759655767922624e-05,
      "loss": 0.5231,
      "step": 4253
    },
    {
      "epoch": 0.738029146426093,
      "grad_norm": 1.0075207948684692,
      "learning_rate": 3.575545995002207e-05,
      "loss": 0.4973,
      "step": 4254
    },
    {
      "epoch": 0.7382026370575989,
      "grad_norm": 0.8046914935112,
      "learning_rate": 3.5751262303682034e-05,
      "loss": 0.5264,
      "step": 4255
    },
    {
      "epoch": 0.7383761276891048,
      "grad_norm": 0.8025881052017212,
      "learning_rate": 3.574706282938964e-05,
      "loss": 0.6394,
      "step": 4256
    },
    {
      "epoch": 0.7385496183206107,
      "grad_norm": 0.5996862649917603,
      "learning_rate": 3.574286152763226e-05,
      "loss": 0.636,
      "step": 4257
    },
    {
      "epoch": 0.7387231089521166,
      "grad_norm": 0.8055362105369568,
      "learning_rate": 3.573865839889746e-05,
      "loss": 0.541,
      "step": 4258
    },
    {
      "epoch": 0.7388965995836225,
      "grad_norm": 0.7214866280555725,
      "learning_rate": 3.573445344367302e-05,
      "loss": 0.6044,
      "step": 4259
    },
    {
      "epoch": 0.7390700902151284,
      "grad_norm": 0.5948684811592102,
      "learning_rate": 3.5730246662446916e-05,
      "loss": 0.6464,
      "step": 4260
    },
    {
      "epoch": 0.7392435808466343,
      "grad_norm": 0.8364885449409485,
      "learning_rate": 3.572603805570736e-05,
      "loss": 0.6919,
      "step": 4261
    },
    {
      "epoch": 0.7394170714781402,
      "grad_norm": 0.5774275064468384,
      "learning_rate": 3.572182762394276e-05,
      "loss": 0.5867,
      "step": 4262
    },
    {
      "epoch": 0.7395905621096461,
      "grad_norm": 0.7644405364990234,
      "learning_rate": 3.571761536764174e-05,
      "loss": 0.6997,
      "step": 4263
    },
    {
      "epoch": 0.739764052741152,
      "grad_norm": 1.6703912019729614,
      "learning_rate": 3.571340128729315e-05,
      "loss": 0.593,
      "step": 4264
    },
    {
      "epoch": 0.7399375433726578,
      "grad_norm": 0.6243732571601868,
      "learning_rate": 3.5709185383386024e-05,
      "loss": 0.6183,
      "step": 4265
    },
    {
      "epoch": 0.7401110340041638,
      "grad_norm": 0.7216456532478333,
      "learning_rate": 3.570496765640964e-05,
      "loss": 0.5013,
      "step": 4266
    },
    {
      "epoch": 0.7402845246356696,
      "grad_norm": 0.7056600451469421,
      "learning_rate": 3.570074810685345e-05,
      "loss": 0.4917,
      "step": 4267
    },
    {
      "epoch": 0.7404580152671756,
      "grad_norm": 0.8297005295753479,
      "learning_rate": 3.569652673520715e-05,
      "loss": 0.5697,
      "step": 4268
    },
    {
      "epoch": 0.7406315058986814,
      "grad_norm": 0.7679122686386108,
      "learning_rate": 3.569230354196063e-05,
      "loss": 0.5293,
      "step": 4269
    },
    {
      "epoch": 0.7408049965301874,
      "grad_norm": 0.7023241519927979,
      "learning_rate": 3.5688078527604e-05,
      "loss": 0.6606,
      "step": 4270
    },
    {
      "epoch": 0.7409784871616932,
      "grad_norm": 0.6871032118797302,
      "learning_rate": 3.568385169262758e-05,
      "loss": 0.6316,
      "step": 4271
    },
    {
      "epoch": 0.7411519777931992,
      "grad_norm": 1.1948171854019165,
      "learning_rate": 3.56796230375219e-05,
      "loss": 0.495,
      "step": 4272
    },
    {
      "epoch": 0.741325468424705,
      "grad_norm": 0.5721275806427002,
      "learning_rate": 3.567539256277769e-05,
      "loss": 0.5731,
      "step": 4273
    },
    {
      "epoch": 0.741498959056211,
      "grad_norm": 1.6316579580307007,
      "learning_rate": 3.567116026888591e-05,
      "loss": 0.6626,
      "step": 4274
    },
    {
      "epoch": 0.7416724496877168,
      "grad_norm": 0.6861583590507507,
      "learning_rate": 3.566692615633771e-05,
      "loss": 0.6213,
      "step": 4275
    },
    {
      "epoch": 0.7418459403192228,
      "grad_norm": 0.932975709438324,
      "learning_rate": 3.5662690225624484e-05,
      "loss": 0.5396,
      "step": 4276
    },
    {
      "epoch": 0.7420194309507286,
      "grad_norm": 0.8106069564819336,
      "learning_rate": 3.56584524772378e-05,
      "loss": 0.4543,
      "step": 4277
    },
    {
      "epoch": 0.7421929215822346,
      "grad_norm": 1.3101338148117065,
      "learning_rate": 3.565421291166946e-05,
      "loss": 0.4977,
      "step": 4278
    },
    {
      "epoch": 0.7423664122137404,
      "grad_norm": 0.7285441756248474,
      "learning_rate": 3.564997152941148e-05,
      "loss": 0.605,
      "step": 4279
    },
    {
      "epoch": 0.7425399028452464,
      "grad_norm": 0.619378924369812,
      "learning_rate": 3.5645728330956074e-05,
      "loss": 0.7219,
      "step": 4280
    },
    {
      "epoch": 0.7427133934767522,
      "grad_norm": 1.1070685386657715,
      "learning_rate": 3.564148331679565e-05,
      "loss": 0.5995,
      "step": 4281
    },
    {
      "epoch": 0.7428868841082582,
      "grad_norm": 0.656654417514801,
      "learning_rate": 3.563723648742286e-05,
      "loss": 0.6273,
      "step": 4282
    },
    {
      "epoch": 0.743060374739764,
      "grad_norm": 0.8392003774642944,
      "learning_rate": 3.563298784333056e-05,
      "loss": 0.5298,
      "step": 4283
    },
    {
      "epoch": 0.7432338653712699,
      "grad_norm": 0.889798641204834,
      "learning_rate": 3.5628737385011814e-05,
      "loss": 0.5413,
      "step": 4284
    },
    {
      "epoch": 0.7434073560027759,
      "grad_norm": 0.8371496796607971,
      "learning_rate": 3.562448511295987e-05,
      "loss": 0.6377,
      "step": 4285
    },
    {
      "epoch": 0.7435808466342817,
      "grad_norm": 0.5523166656494141,
      "learning_rate": 3.562023102766822e-05,
      "loss": 0.6639,
      "step": 4286
    },
    {
      "epoch": 0.7437543372657877,
      "grad_norm": 0.9057916402816772,
      "learning_rate": 3.561597512963057e-05,
      "loss": 0.5361,
      "step": 4287
    },
    {
      "epoch": 0.7439278278972935,
      "grad_norm": 0.7643842697143555,
      "learning_rate": 3.561171741934081e-05,
      "loss": 0.4995,
      "step": 4288
    },
    {
      "epoch": 0.7441013185287995,
      "grad_norm": 0.7651995420455933,
      "learning_rate": 3.560745789729304e-05,
      "loss": 0.5859,
      "step": 4289
    },
    {
      "epoch": 0.7442748091603053,
      "grad_norm": 0.7631704807281494,
      "learning_rate": 3.56031965639816e-05,
      "loss": 0.5461,
      "step": 4290
    },
    {
      "epoch": 0.7444482997918113,
      "grad_norm": 0.928712010383606,
      "learning_rate": 3.559893341990102e-05,
      "loss": 0.6956,
      "step": 4291
    },
    {
      "epoch": 0.7446217904233171,
      "grad_norm": 0.5718164443969727,
      "learning_rate": 3.559466846554604e-05,
      "loss": 0.6771,
      "step": 4292
    },
    {
      "epoch": 0.7447952810548231,
      "grad_norm": 0.7099377512931824,
      "learning_rate": 3.559040170141161e-05,
      "loss": 0.6006,
      "step": 4293
    },
    {
      "epoch": 0.7449687716863289,
      "grad_norm": 0.7230462431907654,
      "learning_rate": 3.5586133127992904e-05,
      "loss": 0.6356,
      "step": 4294
    },
    {
      "epoch": 0.7451422623178349,
      "grad_norm": 0.7968019843101501,
      "learning_rate": 3.558186274578527e-05,
      "loss": 0.5314,
      "step": 4295
    },
    {
      "epoch": 0.7453157529493407,
      "grad_norm": 0.7601330280303955,
      "learning_rate": 3.557759055528433e-05,
      "loss": 0.5615,
      "step": 4296
    },
    {
      "epoch": 0.7454892435808467,
      "grad_norm": 0.7466877698898315,
      "learning_rate": 3.5573316556985845e-05,
      "loss": 0.6106,
      "step": 4297
    },
    {
      "epoch": 0.7456627342123525,
      "grad_norm": 0.6433128118515015,
      "learning_rate": 3.5569040751385825e-05,
      "loss": 0.619,
      "step": 4298
    },
    {
      "epoch": 0.7458362248438585,
      "grad_norm": 0.7198663353919983,
      "learning_rate": 3.556476313898048e-05,
      "loss": 0.5938,
      "step": 4299
    },
    {
      "epoch": 0.7460097154753643,
      "grad_norm": 0.8491593599319458,
      "learning_rate": 3.556048372026625e-05,
      "loss": 0.6785,
      "step": 4300
    },
    {
      "epoch": 0.7461832061068703,
      "grad_norm": 0.7524433732032776,
      "learning_rate": 3.5556202495739736e-05,
      "loss": 0.5038,
      "step": 4301
    },
    {
      "epoch": 0.7463566967383761,
      "grad_norm": 0.4702468514442444,
      "learning_rate": 3.555191946589781e-05,
      "loss": 0.5743,
      "step": 4302
    },
    {
      "epoch": 0.7465301873698821,
      "grad_norm": 0.6478613018989563,
      "learning_rate": 3.55476346312375e-05,
      "loss": 0.5717,
      "step": 4303
    },
    {
      "epoch": 0.7467036780013879,
      "grad_norm": 0.6415278911590576,
      "learning_rate": 3.554334799225608e-05,
      "loss": 0.4797,
      "step": 4304
    },
    {
      "epoch": 0.7468771686328938,
      "grad_norm": 0.6717489361763,
      "learning_rate": 3.5539059549451e-05,
      "loss": 0.5934,
      "step": 4305
    },
    {
      "epoch": 0.7470506592643997,
      "grad_norm": 0.7758067846298218,
      "learning_rate": 3.553476930331996e-05,
      "loss": 0.6525,
      "step": 4306
    },
    {
      "epoch": 0.7472241498959056,
      "grad_norm": 0.6344873905181885,
      "learning_rate": 3.553047725436085e-05,
      "loss": 0.547,
      "step": 4307
    },
    {
      "epoch": 0.7473976405274115,
      "grad_norm": 0.6888913512229919,
      "learning_rate": 3.5526183403071754e-05,
      "loss": 0.5333,
      "step": 4308
    },
    {
      "epoch": 0.7475711311589174,
      "grad_norm": 0.871826708316803,
      "learning_rate": 3.552188774995098e-05,
      "loss": 0.6178,
      "step": 4309
    },
    {
      "epoch": 0.7477446217904233,
      "grad_norm": 0.9191585183143616,
      "learning_rate": 3.551759029549705e-05,
      "loss": 0.6038,
      "step": 4310
    },
    {
      "epoch": 0.7479181124219292,
      "grad_norm": 0.7116936445236206,
      "learning_rate": 3.5513291040208674e-05,
      "loss": 0.675,
      "step": 4311
    },
    {
      "epoch": 0.7480916030534351,
      "grad_norm": 0.7276498675346375,
      "learning_rate": 3.550898998458481e-05,
      "loss": 0.6112,
      "step": 4312
    },
    {
      "epoch": 0.748265093684941,
      "grad_norm": 0.7814944982528687,
      "learning_rate": 3.550468712912458e-05,
      "loss": 0.5112,
      "step": 4313
    },
    {
      "epoch": 0.7484385843164469,
      "grad_norm": 0.5744490623474121,
      "learning_rate": 3.550038247432734e-05,
      "loss": 0.6537,
      "step": 4314
    },
    {
      "epoch": 0.7486120749479528,
      "grad_norm": 0.6232500076293945,
      "learning_rate": 3.549607602069265e-05,
      "loss": 0.5288,
      "step": 4315
    },
    {
      "epoch": 0.7487855655794587,
      "grad_norm": 1.3707828521728516,
      "learning_rate": 3.549176776872029e-05,
      "loss": 0.5879,
      "step": 4316
    },
    {
      "epoch": 0.7489590562109646,
      "grad_norm": 0.8000961542129517,
      "learning_rate": 3.5487457718910226e-05,
      "loss": 0.5605,
      "step": 4317
    },
    {
      "epoch": 0.7491325468424705,
      "grad_norm": 0.8479990363121033,
      "learning_rate": 3.5483145871762646e-05,
      "loss": 0.5392,
      "step": 4318
    },
    {
      "epoch": 0.7493060374739764,
      "grad_norm": 0.7323888540267944,
      "learning_rate": 3.5478832227777945e-05,
      "loss": 0.562,
      "step": 4319
    },
    {
      "epoch": 0.7494795281054824,
      "grad_norm": 0.7212268710136414,
      "learning_rate": 3.547451678745673e-05,
      "loss": 0.5671,
      "step": 4320
    },
    {
      "epoch": 0.7496530187369882,
      "grad_norm": 1.0141124725341797,
      "learning_rate": 3.547019955129981e-05,
      "loss": 0.5107,
      "step": 4321
    },
    {
      "epoch": 0.7498265093684942,
      "grad_norm": 0.853689968585968,
      "learning_rate": 3.54658805198082e-05,
      "loss": 0.5624,
      "step": 4322
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6390842199325562,
      "learning_rate": 3.546155969348315e-05,
      "loss": 0.5446,
      "step": 4323
    },
    {
      "epoch": 0.7501734906315058,
      "grad_norm": 0.7222588658332825,
      "learning_rate": 3.545723707282606e-05,
      "loss": 0.6731,
      "step": 4324
    },
    {
      "epoch": 0.7503469812630118,
      "grad_norm": 0.7966939210891724,
      "learning_rate": 3.5452912658338605e-05,
      "loss": 0.5018,
      "step": 4325
    },
    {
      "epoch": 0.7505204718945176,
      "grad_norm": 0.6748339533805847,
      "learning_rate": 3.5448586450522635e-05,
      "loss": 0.6138,
      "step": 4326
    },
    {
      "epoch": 0.7506939625260236,
      "grad_norm": 0.7996774911880493,
      "learning_rate": 3.5444258449880205e-05,
      "loss": 0.4681,
      "step": 4327
    },
    {
      "epoch": 0.7508674531575295,
      "grad_norm": 0.8768160939216614,
      "learning_rate": 3.5439928656913586e-05,
      "loss": 0.4862,
      "step": 4328
    },
    {
      "epoch": 0.7510409437890354,
      "grad_norm": 0.957241415977478,
      "learning_rate": 3.543559707212525e-05,
      "loss": 0.5496,
      "step": 4329
    },
    {
      "epoch": 0.7512144344205413,
      "grad_norm": 0.8147217631340027,
      "learning_rate": 3.543126369601789e-05,
      "loss": 0.6323,
      "step": 4330
    },
    {
      "epoch": 0.7513879250520472,
      "grad_norm": 0.7651244401931763,
      "learning_rate": 3.54269285290944e-05,
      "loss": 0.6366,
      "step": 4331
    },
    {
      "epoch": 0.7515614156835531,
      "grad_norm": 0.7931744456291199,
      "learning_rate": 3.542259157185787e-05,
      "loss": 0.6262,
      "step": 4332
    },
    {
      "epoch": 0.751734906315059,
      "grad_norm": 0.9671867489814758,
      "learning_rate": 3.541825282481162e-05,
      "loss": 0.5729,
      "step": 4333
    },
    {
      "epoch": 0.7519083969465649,
      "grad_norm": 0.8910104632377625,
      "learning_rate": 3.5413912288459174e-05,
      "loss": 0.4968,
      "step": 4334
    },
    {
      "epoch": 0.7520818875780708,
      "grad_norm": 0.6812747716903687,
      "learning_rate": 3.540956996330424e-05,
      "loss": 0.5334,
      "step": 4335
    },
    {
      "epoch": 0.7522553782095767,
      "grad_norm": 0.6249709129333496,
      "learning_rate": 3.5405225849850754e-05,
      "loss": 0.5815,
      "step": 4336
    },
    {
      "epoch": 0.7524288688410826,
      "grad_norm": 0.6938480138778687,
      "learning_rate": 3.5400879948602854e-05,
      "loss": 0.5894,
      "step": 4337
    },
    {
      "epoch": 0.7526023594725885,
      "grad_norm": 0.8001736402511597,
      "learning_rate": 3.53965322600649e-05,
      "loss": 0.5944,
      "step": 4338
    },
    {
      "epoch": 0.7527758501040944,
      "grad_norm": 0.6530340313911438,
      "learning_rate": 3.539218278474143e-05,
      "loss": 0.641,
      "step": 4339
    },
    {
      "epoch": 0.7529493407356003,
      "grad_norm": 0.736674964427948,
      "learning_rate": 3.5387831523137216e-05,
      "loss": 0.6057,
      "step": 4340
    },
    {
      "epoch": 0.7531228313671062,
      "grad_norm": 0.7169548869132996,
      "learning_rate": 3.538347847575722e-05,
      "loss": 0.578,
      "step": 4341
    },
    {
      "epoch": 0.7532963219986121,
      "grad_norm": 0.9289293885231018,
      "learning_rate": 3.5379123643106625e-05,
      "loss": 0.5378,
      "step": 4342
    },
    {
      "epoch": 0.7534698126301179,
      "grad_norm": 0.6985144019126892,
      "learning_rate": 3.537476702569081e-05,
      "loss": 0.5327,
      "step": 4343
    },
    {
      "epoch": 0.7536433032616239,
      "grad_norm": 0.8143550753593445,
      "learning_rate": 3.5370408624015364e-05,
      "loss": 0.5986,
      "step": 4344
    },
    {
      "epoch": 0.7538167938931297,
      "grad_norm": 0.563184916973114,
      "learning_rate": 3.536604843858609e-05,
      "loss": 0.6035,
      "step": 4345
    },
    {
      "epoch": 0.7539902845246357,
      "grad_norm": 0.8249099254608154,
      "learning_rate": 3.536168646990899e-05,
      "loss": 0.6566,
      "step": 4346
    },
    {
      "epoch": 0.7541637751561415,
      "grad_norm": 0.8906046152114868,
      "learning_rate": 3.535732271849028e-05,
      "loss": 0.573,
      "step": 4347
    },
    {
      "epoch": 0.7543372657876475,
      "grad_norm": 0.7494427561759949,
      "learning_rate": 3.535295718483636e-05,
      "loss": 0.4839,
      "step": 4348
    },
    {
      "epoch": 0.7545107564191533,
      "grad_norm": 0.6854152679443359,
      "learning_rate": 3.5348589869453874e-05,
      "loss": 0.5288,
      "step": 4349
    },
    {
      "epoch": 0.7546842470506593,
      "grad_norm": 0.7171089053153992,
      "learning_rate": 3.5344220772849654e-05,
      "loss": 0.6676,
      "step": 4350
    },
    {
      "epoch": 0.7548577376821651,
      "grad_norm": 0.6131109595298767,
      "learning_rate": 3.533984989553073e-05,
      "loss": 0.6404,
      "step": 4351
    },
    {
      "epoch": 0.7550312283136711,
      "grad_norm": 0.6121817827224731,
      "learning_rate": 3.533547723800435e-05,
      "loss": 0.6564,
      "step": 4352
    },
    {
      "epoch": 0.7552047189451769,
      "grad_norm": 0.6851575970649719,
      "learning_rate": 3.533110280077797e-05,
      "loss": 0.4674,
      "step": 4353
    },
    {
      "epoch": 0.7553782095766829,
      "grad_norm": 0.8325938582420349,
      "learning_rate": 3.532672658435925e-05,
      "loss": 0.525,
      "step": 4354
    },
    {
      "epoch": 0.7555517002081887,
      "grad_norm": 0.8352726697921753,
      "learning_rate": 3.5322348589256044e-05,
      "loss": 0.5544,
      "step": 4355
    },
    {
      "epoch": 0.7557251908396947,
      "grad_norm": 0.8470213413238525,
      "learning_rate": 3.531796881597643e-05,
      "loss": 0.5129,
      "step": 4356
    },
    {
      "epoch": 0.7558986814712005,
      "grad_norm": 0.8853175044059753,
      "learning_rate": 3.5313587265028686e-05,
      "loss": 0.4796,
      "step": 4357
    },
    {
      "epoch": 0.7560721721027065,
      "grad_norm": 0.6890913844108582,
      "learning_rate": 3.53092039369213e-05,
      "loss": 0.527,
      "step": 4358
    },
    {
      "epoch": 0.7562456627342123,
      "grad_norm": 0.6278190016746521,
      "learning_rate": 3.5304818832162956e-05,
      "loss": 0.5302,
      "step": 4359
    },
    {
      "epoch": 0.7564191533657183,
      "grad_norm": 0.7474961280822754,
      "learning_rate": 3.530043195126255e-05,
      "loss": 0.5416,
      "step": 4360
    },
    {
      "epoch": 0.7565926439972241,
      "grad_norm": 0.6712853908538818,
      "learning_rate": 3.529604329472919e-05,
      "loss": 0.5948,
      "step": 4361
    },
    {
      "epoch": 0.7567661346287301,
      "grad_norm": 0.6338025331497192,
      "learning_rate": 3.529165286307219e-05,
      "loss": 0.6205,
      "step": 4362
    },
    {
      "epoch": 0.756939625260236,
      "grad_norm": 0.6051141619682312,
      "learning_rate": 3.5287260656801044e-05,
      "loss": 0.6592,
      "step": 4363
    },
    {
      "epoch": 0.7571131158917418,
      "grad_norm": 0.6509369611740112,
      "learning_rate": 3.528286667642549e-05,
      "loss": 0.5748,
      "step": 4364
    },
    {
      "epoch": 0.7572866065232478,
      "grad_norm": 1.2799158096313477,
      "learning_rate": 3.5278470922455453e-05,
      "loss": 0.5452,
      "step": 4365
    },
    {
      "epoch": 0.7574600971547536,
      "grad_norm": 1.0939301252365112,
      "learning_rate": 3.527407339540106e-05,
      "loss": 0.729,
      "step": 4366
    },
    {
      "epoch": 0.7576335877862596,
      "grad_norm": 0.759294331073761,
      "learning_rate": 3.5269674095772654e-05,
      "loss": 0.5856,
      "step": 4367
    },
    {
      "epoch": 0.7578070784177654,
      "grad_norm": 0.782491147518158,
      "learning_rate": 3.5265273024080776e-05,
      "loss": 0.5333,
      "step": 4368
    },
    {
      "epoch": 0.7579805690492714,
      "grad_norm": 0.6015889644622803,
      "learning_rate": 3.526087018083617e-05,
      "loss": 0.6262,
      "step": 4369
    },
    {
      "epoch": 0.7581540596807772,
      "grad_norm": 0.5154745578765869,
      "learning_rate": 3.52564655665498e-05,
      "loss": 0.7069,
      "step": 4370
    },
    {
      "epoch": 0.7583275503122832,
      "grad_norm": 0.7049304842948914,
      "learning_rate": 3.525205918173283e-05,
      "loss": 0.5781,
      "step": 4371
    },
    {
      "epoch": 0.758501040943789,
      "grad_norm": 0.7382335662841797,
      "learning_rate": 3.524765102689662e-05,
      "loss": 0.5284,
      "step": 4372
    },
    {
      "epoch": 0.758674531575295,
      "grad_norm": 0.8790127038955688,
      "learning_rate": 3.524324110255273e-05,
      "loss": 0.5142,
      "step": 4373
    },
    {
      "epoch": 0.7588480222068008,
      "grad_norm": 0.9203531742095947,
      "learning_rate": 3.523882940921296e-05,
      "loss": 0.4869,
      "step": 4374
    },
    {
      "epoch": 0.7590215128383068,
      "grad_norm": 1.163057565689087,
      "learning_rate": 3.523441594738927e-05,
      "loss": 0.5157,
      "step": 4375
    },
    {
      "epoch": 0.7591950034698126,
      "grad_norm": 0.6108386516571045,
      "learning_rate": 3.523000071759387e-05,
      "loss": 0.5833,
      "step": 4376
    },
    {
      "epoch": 0.7593684941013186,
      "grad_norm": 0.5940057039260864,
      "learning_rate": 3.522558372033912e-05,
      "loss": 0.6936,
      "step": 4377
    },
    {
      "epoch": 0.7595419847328244,
      "grad_norm": 0.6746407151222229,
      "learning_rate": 3.522116495613766e-05,
      "loss": 0.5966,
      "step": 4378
    },
    {
      "epoch": 0.7597154753643304,
      "grad_norm": 0.8779783248901367,
      "learning_rate": 3.521674442550226e-05,
      "loss": 0.5787,
      "step": 4379
    },
    {
      "epoch": 0.7598889659958362,
      "grad_norm": 0.8782005906105042,
      "learning_rate": 3.521232212894594e-05,
      "loss": 0.5933,
      "step": 4380
    },
    {
      "epoch": 0.7600624566273422,
      "grad_norm": 0.5405778884887695,
      "learning_rate": 3.520789806698191e-05,
      "loss": 0.6982,
      "step": 4381
    },
    {
      "epoch": 0.760235947258848,
      "grad_norm": 0.5692259669303894,
      "learning_rate": 3.5203472240123594e-05,
      "loss": 0.6979,
      "step": 4382
    },
    {
      "epoch": 0.7604094378903539,
      "grad_norm": 1.0693467855453491,
      "learning_rate": 3.5199044648884605e-05,
      "loss": 0.5969,
      "step": 4383
    },
    {
      "epoch": 0.7605829285218598,
      "grad_norm": 1.04296875,
      "learning_rate": 3.519461529377877e-05,
      "loss": 0.5542,
      "step": 4384
    },
    {
      "epoch": 0.7607564191533657,
      "grad_norm": 0.725059449672699,
      "learning_rate": 3.519018417532013e-05,
      "loss": 0.6002,
      "step": 4385
    },
    {
      "epoch": 0.7609299097848716,
      "grad_norm": 0.8393305540084839,
      "learning_rate": 3.5185751294022914e-05,
      "loss": 0.5996,
      "step": 4386
    },
    {
      "epoch": 0.7611034004163775,
      "grad_norm": 0.745979905128479,
      "learning_rate": 3.518131665040157e-05,
      "loss": 0.5409,
      "step": 4387
    },
    {
      "epoch": 0.7612768910478834,
      "grad_norm": 0.6680980920791626,
      "learning_rate": 3.5176880244970735e-05,
      "loss": 0.5209,
      "step": 4388
    },
    {
      "epoch": 0.7614503816793893,
      "grad_norm": 1.4826394319534302,
      "learning_rate": 3.517244207824526e-05,
      "loss": 0.6648,
      "step": 4389
    },
    {
      "epoch": 0.7616238723108952,
      "grad_norm": 0.6956479549407959,
      "learning_rate": 3.516800215074021e-05,
      "loss": 0.5179,
      "step": 4390
    },
    {
      "epoch": 0.7617973629424011,
      "grad_norm": 0.7217515110969543,
      "learning_rate": 3.516356046297083e-05,
      "loss": 0.6899,
      "step": 4391
    },
    {
      "epoch": 0.761970853573907,
      "grad_norm": 0.7801789045333862,
      "learning_rate": 3.515911701545259e-05,
      "loss": 0.5912,
      "step": 4392
    },
    {
      "epoch": 0.7621443442054129,
      "grad_norm": 0.902409553527832,
      "learning_rate": 3.515467180870116e-05,
      "loss": 0.5863,
      "step": 4393
    },
    {
      "epoch": 0.7623178348369188,
      "grad_norm": 0.6760614514350891,
      "learning_rate": 3.5150224843232405e-05,
      "loss": 0.5063,
      "step": 4394
    },
    {
      "epoch": 0.7624913254684247,
      "grad_norm": 0.8342067003250122,
      "learning_rate": 3.51457761195624e-05,
      "loss": 0.5487,
      "step": 4395
    },
    {
      "epoch": 0.7626648160999306,
      "grad_norm": 0.7199931144714355,
      "learning_rate": 3.514132563820744e-05,
      "loss": 0.5853,
      "step": 4396
    },
    {
      "epoch": 0.7628383067314365,
      "grad_norm": 0.9212760329246521,
      "learning_rate": 3.513687339968399e-05,
      "loss": 0.6139,
      "step": 4397
    },
    {
      "epoch": 0.7630117973629424,
      "grad_norm": 0.6572199463844299,
      "learning_rate": 3.513241940450874e-05,
      "loss": 0.5898,
      "step": 4398
    },
    {
      "epoch": 0.7631852879944483,
      "grad_norm": 0.9196416735649109,
      "learning_rate": 3.5127963653198583e-05,
      "loss": 0.6492,
      "step": 4399
    },
    {
      "epoch": 0.7633587786259542,
      "grad_norm": 0.66299968957901,
      "learning_rate": 3.512350614627062e-05,
      "loss": 0.5554,
      "step": 4400
    },
    {
      "epoch": 0.7635322692574601,
      "grad_norm": 0.7973070740699768,
      "learning_rate": 3.511904688424215e-05,
      "loss": 0.6779,
      "step": 4401
    },
    {
      "epoch": 0.7637057598889659,
      "grad_norm": 0.8950045704841614,
      "learning_rate": 3.511458586763067e-05,
      "loss": 0.499,
      "step": 4402
    },
    {
      "epoch": 0.7638792505204719,
      "grad_norm": 0.8955358862876892,
      "learning_rate": 3.511012309695389e-05,
      "loss": 0.5558,
      "step": 4403
    },
    {
      "epoch": 0.7640527411519777,
      "grad_norm": 1.0587538480758667,
      "learning_rate": 3.510565857272972e-05,
      "loss": 0.5178,
      "step": 4404
    },
    {
      "epoch": 0.7642262317834837,
      "grad_norm": 0.8319945335388184,
      "learning_rate": 3.510119229547626e-05,
      "loss": 0.5959,
      "step": 4405
    },
    {
      "epoch": 0.7643997224149895,
      "grad_norm": 0.9041110277175903,
      "learning_rate": 3.509672426571185e-05,
      "loss": 0.5969,
      "step": 4406
    },
    {
      "epoch": 0.7645732130464955,
      "grad_norm": 0.8114092350006104,
      "learning_rate": 3.509225448395499e-05,
      "loss": 0.532,
      "step": 4407
    },
    {
      "epoch": 0.7647467036780013,
      "grad_norm": 0.7527754902839661,
      "learning_rate": 3.508778295072441e-05,
      "loss": 0.5211,
      "step": 4408
    },
    {
      "epoch": 0.7649201943095073,
      "grad_norm": 0.9807412624359131,
      "learning_rate": 3.5083309666539043e-05,
      "loss": 0.5741,
      "step": 4409
    },
    {
      "epoch": 0.7650936849410132,
      "grad_norm": 0.7295682430267334,
      "learning_rate": 3.5078834631918014e-05,
      "loss": 0.6608,
      "step": 4410
    },
    {
      "epoch": 0.7652671755725191,
      "grad_norm": 0.8115662336349487,
      "learning_rate": 3.507435784738065e-05,
      "loss": 0.5958,
      "step": 4411
    },
    {
      "epoch": 0.765440666204025,
      "grad_norm": 0.6505253314971924,
      "learning_rate": 3.506987931344649e-05,
      "loss": 0.6276,
      "step": 4412
    },
    {
      "epoch": 0.7656141568355309,
      "grad_norm": 0.6952764391899109,
      "learning_rate": 3.5065399030635286e-05,
      "loss": 0.5837,
      "step": 4413
    },
    {
      "epoch": 0.7657876474670368,
      "grad_norm": 0.9617361426353455,
      "learning_rate": 3.506091699946697e-05,
      "loss": 0.5822,
      "step": 4414
    },
    {
      "epoch": 0.7659611380985427,
      "grad_norm": 0.6262912154197693,
      "learning_rate": 3.505643322046168e-05,
      "loss": 0.5688,
      "step": 4415
    },
    {
      "epoch": 0.7661346287300486,
      "grad_norm": 0.8222533464431763,
      "learning_rate": 3.505194769413977e-05,
      "loss": 0.4937,
      "step": 4416
    },
    {
      "epoch": 0.7663081193615545,
      "grad_norm": 0.7947388887405396,
      "learning_rate": 3.5047460421021796e-05,
      "loss": 0.5691,
      "step": 4417
    },
    {
      "epoch": 0.7664816099930604,
      "grad_norm": 0.9605175852775574,
      "learning_rate": 3.504297140162851e-05,
      "loss": 0.5865,
      "step": 4418
    },
    {
      "epoch": 0.7666551006245663,
      "grad_norm": 0.8637752532958984,
      "learning_rate": 3.503848063648086e-05,
      "loss": 0.637,
      "step": 4419
    },
    {
      "epoch": 0.7668285912560722,
      "grad_norm": 0.7540271878242493,
      "learning_rate": 3.50339881261e-05,
      "loss": 0.5406,
      "step": 4420
    },
    {
      "epoch": 0.7670020818875781,
      "grad_norm": 0.6558127403259277,
      "learning_rate": 3.502949387100731e-05,
      "loss": 0.6517,
      "step": 4421
    },
    {
      "epoch": 0.767175572519084,
      "grad_norm": 0.6077901124954224,
      "learning_rate": 3.502499787172434e-05,
      "loss": 0.5991,
      "step": 4422
    },
    {
      "epoch": 0.7673490631505898,
      "grad_norm": 1.032698154449463,
      "learning_rate": 3.502050012877286e-05,
      "loss": 0.5167,
      "step": 4423
    },
    {
      "epoch": 0.7675225537820958,
      "grad_norm": 0.7792657613754272,
      "learning_rate": 3.5016000642674836e-05,
      "loss": 0.6167,
      "step": 4424
    },
    {
      "epoch": 0.7676960444136016,
      "grad_norm": 0.6926698684692383,
      "learning_rate": 3.501149941395245e-05,
      "loss": 0.6359,
      "step": 4425
    },
    {
      "epoch": 0.7678695350451076,
      "grad_norm": 0.7828586101531982,
      "learning_rate": 3.500699644312805e-05,
      "loss": 0.6838,
      "step": 4426
    },
    {
      "epoch": 0.7680430256766134,
      "grad_norm": 0.7258651256561279,
      "learning_rate": 3.5002491730724235e-05,
      "loss": 0.6031,
      "step": 4427
    },
    {
      "epoch": 0.7682165163081194,
      "grad_norm": 0.7690781354904175,
      "learning_rate": 3.4997985277263765e-05,
      "loss": 0.5748,
      "step": 4428
    },
    {
      "epoch": 0.7683900069396252,
      "grad_norm": 0.5877462029457092,
      "learning_rate": 3.499347708326964e-05,
      "loss": 0.5717,
      "step": 4429
    },
    {
      "epoch": 0.7685634975711312,
      "grad_norm": 0.5916820168495178,
      "learning_rate": 3.498896714926502e-05,
      "loss": 0.5902,
      "step": 4430
    },
    {
      "epoch": 0.768736988202637,
      "grad_norm": 0.7683703303337097,
      "learning_rate": 3.4984455475773304e-05,
      "loss": 0.6587,
      "step": 4431
    },
    {
      "epoch": 0.768910478834143,
      "grad_norm": 0.7843685746192932,
      "learning_rate": 3.4979942063318066e-05,
      "loss": 0.5276,
      "step": 4432
    },
    {
      "epoch": 0.7690839694656488,
      "grad_norm": 0.620903491973877,
      "learning_rate": 3.497542691242309e-05,
      "loss": 0.6189,
      "step": 4433
    },
    {
      "epoch": 0.7692574600971548,
      "grad_norm": 0.5340049862861633,
      "learning_rate": 3.497091002361238e-05,
      "loss": 0.6627,
      "step": 4434
    },
    {
      "epoch": 0.7694309507286606,
      "grad_norm": 1.0713303089141846,
      "learning_rate": 3.496639139741011e-05,
      "loss": 0.5232,
      "step": 4435
    },
    {
      "epoch": 0.7696044413601666,
      "grad_norm": 0.6281771063804626,
      "learning_rate": 3.496187103434069e-05,
      "loss": 0.4862,
      "step": 4436
    },
    {
      "epoch": 0.7697779319916724,
      "grad_norm": 0.7040949463844299,
      "learning_rate": 3.495734893492869e-05,
      "loss": 0.5281,
      "step": 4437
    },
    {
      "epoch": 0.7699514226231784,
      "grad_norm": 0.6082890033721924,
      "learning_rate": 3.4952825099698926e-05,
      "loss": 0.6719,
      "step": 4438
    },
    {
      "epoch": 0.7701249132546842,
      "grad_norm": 0.6842986941337585,
      "learning_rate": 3.494829952917638e-05,
      "loss": 0.7261,
      "step": 4439
    },
    {
      "epoch": 0.7702984038861902,
      "grad_norm": 1.3153560161590576,
      "learning_rate": 3.4943772223886264e-05,
      "loss": 0.5721,
      "step": 4440
    },
    {
      "epoch": 0.770471894517696,
      "grad_norm": 0.6036491990089417,
      "learning_rate": 3.493924318435395e-05,
      "loss": 0.6052,
      "step": 4441
    },
    {
      "epoch": 0.7706453851492019,
      "grad_norm": 0.7688340544700623,
      "learning_rate": 3.493471241110507e-05,
      "loss": 0.5905,
      "step": 4442
    },
    {
      "epoch": 0.7708188757807078,
      "grad_norm": 0.8035319447517395,
      "learning_rate": 3.493017990466542e-05,
      "loss": 0.6389,
      "step": 4443
    },
    {
      "epoch": 0.7709923664122137,
      "grad_norm": 0.7157716751098633,
      "learning_rate": 3.492564566556098e-05,
      "loss": 0.5776,
      "step": 4444
    },
    {
      "epoch": 0.7711658570437196,
      "grad_norm": 0.9449665546417236,
      "learning_rate": 3.4921109694317974e-05,
      "loss": 0.5808,
      "step": 4445
    },
    {
      "epoch": 0.7713393476752255,
      "grad_norm": 1.098851203918457,
      "learning_rate": 3.491657199146281e-05,
      "loss": 0.6249,
      "step": 4446
    },
    {
      "epoch": 0.7715128383067315,
      "grad_norm": 1.1828197240829468,
      "learning_rate": 3.4912032557522075e-05,
      "loss": 0.6592,
      "step": 4447
    },
    {
      "epoch": 0.7716863289382373,
      "grad_norm": 0.6309101581573486,
      "learning_rate": 3.490749139302258e-05,
      "loss": 0.5276,
      "step": 4448
    },
    {
      "epoch": 0.7718598195697433,
      "grad_norm": 0.7277196645736694,
      "learning_rate": 3.4902948498491357e-05,
      "loss": 0.5864,
      "step": 4449
    },
    {
      "epoch": 0.7720333102012491,
      "grad_norm": 0.8082743287086487,
      "learning_rate": 3.4898403874455584e-05,
      "loss": 0.6727,
      "step": 4450
    },
    {
      "epoch": 0.7722068008327551,
      "grad_norm": 1.0545870065689087,
      "learning_rate": 3.489385752144268e-05,
      "loss": 0.6199,
      "step": 4451
    },
    {
      "epoch": 0.7723802914642609,
      "grad_norm": 0.7641955018043518,
      "learning_rate": 3.4889309439980256e-05,
      "loss": 0.6483,
      "step": 4452
    },
    {
      "epoch": 0.7725537820957669,
      "grad_norm": 0.6104955077171326,
      "learning_rate": 3.4884759630596124e-05,
      "loss": 0.6729,
      "step": 4453
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 0.9830421209335327,
      "learning_rate": 3.488020809381829e-05,
      "loss": 0.5653,
      "step": 4454
    },
    {
      "epoch": 0.7729007633587787,
      "grad_norm": 0.8413013815879822,
      "learning_rate": 3.4875654830174975e-05,
      "loss": 0.5801,
      "step": 4455
    },
    {
      "epoch": 0.7730742539902845,
      "grad_norm": 0.956383228302002,
      "learning_rate": 3.4871099840194575e-05,
      "loss": 0.585,
      "step": 4456
    },
    {
      "epoch": 0.7732477446217905,
      "grad_norm": 0.6991445422172546,
      "learning_rate": 3.4866543124405714e-05,
      "loss": 0.611,
      "step": 4457
    },
    {
      "epoch": 0.7734212352532963,
      "grad_norm": 0.7721594572067261,
      "learning_rate": 3.4861984683337205e-05,
      "loss": 0.5787,
      "step": 4458
    },
    {
      "epoch": 0.7735947258848023,
      "grad_norm": 0.7705695629119873,
      "learning_rate": 3.485742451751805e-05,
      "loss": 0.5342,
      "step": 4459
    },
    {
      "epoch": 0.7737682165163081,
      "grad_norm": 0.7413557767868042,
      "learning_rate": 3.485286262747747e-05,
      "loss": 0.6382,
      "step": 4460
    },
    {
      "epoch": 0.7739417071478141,
      "grad_norm": 0.7928658127784729,
      "learning_rate": 3.484829901374487e-05,
      "loss": 0.5946,
      "step": 4461
    },
    {
      "epoch": 0.7741151977793199,
      "grad_norm": 0.81223464012146,
      "learning_rate": 3.4843733676849876e-05,
      "loss": 0.6542,
      "step": 4462
    },
    {
      "epoch": 0.7742886884108258,
      "grad_norm": 1.0212039947509766,
      "learning_rate": 3.4839166617322285e-05,
      "loss": 0.6143,
      "step": 4463
    },
    {
      "epoch": 0.7744621790423317,
      "grad_norm": 0.9145832657814026,
      "learning_rate": 3.4834597835692117e-05,
      "loss": 0.5707,
      "step": 4464
    },
    {
      "epoch": 0.7746356696738376,
      "grad_norm": 0.7010418176651001,
      "learning_rate": 3.483002733248959e-05,
      "loss": 0.6993,
      "step": 4465
    },
    {
      "epoch": 0.7748091603053435,
      "grad_norm": 1.256196141242981,
      "learning_rate": 3.482545510824511e-05,
      "loss": 0.5703,
      "step": 4466
    },
    {
      "epoch": 0.7749826509368494,
      "grad_norm": 0.8264151811599731,
      "learning_rate": 3.4820881163489284e-05,
      "loss": 0.5764,
      "step": 4467
    },
    {
      "epoch": 0.7751561415683553,
      "grad_norm": 0.9197185039520264,
      "learning_rate": 3.481630549875293e-05,
      "loss": 0.6288,
      "step": 4468
    },
    {
      "epoch": 0.7753296321998612,
      "grad_norm": 0.9843990206718445,
      "learning_rate": 3.481172811456707e-05,
      "loss": 0.5416,
      "step": 4469
    },
    {
      "epoch": 0.7755031228313671,
      "grad_norm": 1.1497993469238281,
      "learning_rate": 3.480714901146289e-05,
      "loss": 0.5309,
      "step": 4470
    },
    {
      "epoch": 0.775676613462873,
      "grad_norm": 1.3240598440170288,
      "learning_rate": 3.4802568189971814e-05,
      "loss": 0.6338,
      "step": 4471
    },
    {
      "epoch": 0.7758501040943789,
      "grad_norm": 0.811797559261322,
      "learning_rate": 3.479798565062546e-05,
      "loss": 0.5841,
      "step": 4472
    },
    {
      "epoch": 0.7760235947258848,
      "grad_norm": 0.6281471848487854,
      "learning_rate": 3.479340139395562e-05,
      "loss": 0.7012,
      "step": 4473
    },
    {
      "epoch": 0.7761970853573907,
      "grad_norm": 0.5829970836639404,
      "learning_rate": 3.478881542049432e-05,
      "loss": 0.5142,
      "step": 4474
    },
    {
      "epoch": 0.7763705759888966,
      "grad_norm": 0.6744431257247925,
      "learning_rate": 3.478422773077375e-05,
      "loss": 0.6349,
      "step": 4475
    },
    {
      "epoch": 0.7765440666204025,
      "grad_norm": 1.3758318424224854,
      "learning_rate": 3.4779638325326326e-05,
      "loss": 0.5693,
      "step": 4476
    },
    {
      "epoch": 0.7767175572519084,
      "grad_norm": 0.8429315090179443,
      "learning_rate": 3.477504720468465e-05,
      "loss": 0.5776,
      "step": 4477
    },
    {
      "epoch": 0.7768910478834143,
      "grad_norm": 1.0065104961395264,
      "learning_rate": 3.477045436938154e-05,
      "loss": 0.5262,
      "step": 4478
    },
    {
      "epoch": 0.7770645385149202,
      "grad_norm": 0.9056709408760071,
      "learning_rate": 3.4765859819949977e-05,
      "loss": 0.545,
      "step": 4479
    },
    {
      "epoch": 0.7772380291464261,
      "grad_norm": 0.7382376790046692,
      "learning_rate": 3.476126355692318e-05,
      "loss": 0.7358,
      "step": 4480
    },
    {
      "epoch": 0.777411519777932,
      "grad_norm": 0.6907861828804016,
      "learning_rate": 3.475666558083455e-05,
      "loss": 0.6331,
      "step": 4481
    },
    {
      "epoch": 0.7775850104094378,
      "grad_norm": 0.852795422077179,
      "learning_rate": 3.475206589221768e-05,
      "loss": 0.4722,
      "step": 4482
    },
    {
      "epoch": 0.7777585010409438,
      "grad_norm": 0.8108128309249878,
      "learning_rate": 3.4747464491606376e-05,
      "loss": 0.5543,
      "step": 4483
    },
    {
      "epoch": 0.7779319916724496,
      "grad_norm": 0.6749537587165833,
      "learning_rate": 3.4742861379534636e-05,
      "loss": 0.6477,
      "step": 4484
    },
    {
      "epoch": 0.7781054823039556,
      "grad_norm": 0.8126578330993652,
      "learning_rate": 3.4738256556536654e-05,
      "loss": 0.6495,
      "step": 4485
    },
    {
      "epoch": 0.7782789729354614,
      "grad_norm": 1.5315734148025513,
      "learning_rate": 3.473365002314682e-05,
      "loss": 0.5951,
      "step": 4486
    },
    {
      "epoch": 0.7784524635669674,
      "grad_norm": 0.9686574935913086,
      "learning_rate": 3.4729041779899736e-05,
      "loss": 0.4756,
      "step": 4487
    },
    {
      "epoch": 0.7786259541984732,
      "grad_norm": 1.302496075630188,
      "learning_rate": 3.4724431827330196e-05,
      "loss": 0.4712,
      "step": 4488
    },
    {
      "epoch": 0.7787994448299792,
      "grad_norm": 1.0299724340438843,
      "learning_rate": 3.471982016597317e-05,
      "loss": 0.5264,
      "step": 4489
    },
    {
      "epoch": 0.778972935461485,
      "grad_norm": 0.5932894945144653,
      "learning_rate": 3.4715206796363876e-05,
      "loss": 0.6855,
      "step": 4490
    },
    {
      "epoch": 0.779146426092991,
      "grad_norm": 0.752143144607544,
      "learning_rate": 3.4710591719037685e-05,
      "loss": 0.6594,
      "step": 4491
    },
    {
      "epoch": 0.7793199167244969,
      "grad_norm": 2.0506062507629395,
      "learning_rate": 3.470597493453018e-05,
      "loss": 0.6189,
      "step": 4492
    },
    {
      "epoch": 0.7794934073560028,
      "grad_norm": 2.3081183433532715,
      "learning_rate": 3.470135644337715e-05,
      "loss": 0.6396,
      "step": 4493
    },
    {
      "epoch": 0.7796668979875087,
      "grad_norm": 0.7178435325622559,
      "learning_rate": 3.469673624611457e-05,
      "loss": 0.5553,
      "step": 4494
    },
    {
      "epoch": 0.7798403886190146,
      "grad_norm": 0.7959488034248352,
      "learning_rate": 3.4692114343278626e-05,
      "loss": 0.5393,
      "step": 4495
    },
    {
      "epoch": 0.7800138792505205,
      "grad_norm": 1.1394597291946411,
      "learning_rate": 3.4687490735405696e-05,
      "loss": 0.4987,
      "step": 4496
    },
    {
      "epoch": 0.7801873698820264,
      "grad_norm": 0.7853170037269592,
      "learning_rate": 3.468286542303235e-05,
      "loss": 0.4734,
      "step": 4497
    },
    {
      "epoch": 0.7803608605135323,
      "grad_norm": 0.6878113150596619,
      "learning_rate": 3.467823840669536e-05,
      "loss": 0.439,
      "step": 4498
    },
    {
      "epoch": 0.7805343511450382,
      "grad_norm": 0.646817147731781,
      "learning_rate": 3.4673609686931697e-05,
      "loss": 0.6442,
      "step": 4499
    },
    {
      "epoch": 0.7807078417765441,
      "grad_norm": 0.7293394804000854,
      "learning_rate": 3.466897926427854e-05,
      "loss": 0.5532,
      "step": 4500
    },
    {
      "epoch": 0.7808813324080499,
      "grad_norm": 0.8035613298416138,
      "learning_rate": 3.4664347139273245e-05,
      "loss": 0.6101,
      "step": 4501
    },
    {
      "epoch": 0.7810548230395559,
      "grad_norm": 0.8770453333854675,
      "learning_rate": 3.465971331245337e-05,
      "loss": 0.525,
      "step": 4502
    },
    {
      "epoch": 0.7812283136710617,
      "grad_norm": 0.7347700595855713,
      "learning_rate": 3.465507778435669e-05,
      "loss": 0.4902,
      "step": 4503
    },
    {
      "epoch": 0.7814018043025677,
      "grad_norm": 0.5748605132102966,
      "learning_rate": 3.465044055552116e-05,
      "loss": 0.6661,
      "step": 4504
    },
    {
      "epoch": 0.7815752949340735,
      "grad_norm": 0.5695153474807739,
      "learning_rate": 3.464580162648492e-05,
      "loss": 0.6606,
      "step": 4505
    },
    {
      "epoch": 0.7817487855655795,
      "grad_norm": 0.7740654349327087,
      "learning_rate": 3.464116099778634e-05,
      "loss": 0.5966,
      "step": 4506
    },
    {
      "epoch": 0.7819222761970853,
      "grad_norm": 3.157895088195801,
      "learning_rate": 3.463651866996397e-05,
      "loss": 0.5482,
      "step": 4507
    },
    {
      "epoch": 0.7820957668285913,
      "grad_norm": 0.9121180176734924,
      "learning_rate": 3.463187464355655e-05,
      "loss": 0.676,
      "step": 4508
    },
    {
      "epoch": 0.7822692574600971,
      "grad_norm": 0.6632499694824219,
      "learning_rate": 3.462722891910303e-05,
      "loss": 0.6271,
      "step": 4509
    },
    {
      "epoch": 0.7824427480916031,
      "grad_norm": 0.696527361869812,
      "learning_rate": 3.462258149714255e-05,
      "loss": 0.6219,
      "step": 4510
    },
    {
      "epoch": 0.7826162387231089,
      "grad_norm": 1.442875623703003,
      "learning_rate": 3.461793237821445e-05,
      "loss": 0.6343,
      "step": 4511
    },
    {
      "epoch": 0.7827897293546149,
      "grad_norm": 0.8152379393577576,
      "learning_rate": 3.461328156285826e-05,
      "loss": 0.7207,
      "step": 4512
    },
    {
      "epoch": 0.7829632199861207,
      "grad_norm": 1.4653974771499634,
      "learning_rate": 3.460862905161372e-05,
      "loss": 0.6526,
      "step": 4513
    },
    {
      "epoch": 0.7831367106176267,
      "grad_norm": 0.8740278482437134,
      "learning_rate": 3.4603974845020754e-05,
      "loss": 0.5209,
      "step": 4514
    },
    {
      "epoch": 0.7833102012491325,
      "grad_norm": 0.7870283722877502,
      "learning_rate": 3.459931894361949e-05,
      "loss": 0.5172,
      "step": 4515
    },
    {
      "epoch": 0.7834836918806385,
      "grad_norm": 0.7225337624549866,
      "learning_rate": 3.4594661347950255e-05,
      "loss": 0.6119,
      "step": 4516
    },
    {
      "epoch": 0.7836571825121443,
      "grad_norm": 0.9904224276542664,
      "learning_rate": 3.459000205855356e-05,
      "loss": 0.571,
      "step": 4517
    },
    {
      "epoch": 0.7838306731436503,
      "grad_norm": 0.6905709505081177,
      "learning_rate": 3.458534107597013e-05,
      "loss": 0.6287,
      "step": 4518
    },
    {
      "epoch": 0.7840041637751561,
      "grad_norm": 0.8519802689552307,
      "learning_rate": 3.458067840074087e-05,
      "loss": 0.5784,
      "step": 4519
    },
    {
      "epoch": 0.7841776544066621,
      "grad_norm": 1.2368385791778564,
      "learning_rate": 3.457601403340689e-05,
      "loss": 0.5349,
      "step": 4520
    },
    {
      "epoch": 0.7843511450381679,
      "grad_norm": 0.8582872748374939,
      "learning_rate": 3.45713479745095e-05,
      "loss": 0.55,
      "step": 4521
    },
    {
      "epoch": 0.7845246356696738,
      "grad_norm": 0.8825411200523376,
      "learning_rate": 3.45666802245902e-05,
      "loss": 0.5625,
      "step": 4522
    },
    {
      "epoch": 0.7846981263011797,
      "grad_norm": 0.8189730048179626,
      "learning_rate": 3.456201078419068e-05,
      "loss": 0.4833,
      "step": 4523
    },
    {
      "epoch": 0.7848716169326856,
      "grad_norm": 0.8698748350143433,
      "learning_rate": 3.455733965385284e-05,
      "loss": 0.5504,
      "step": 4524
    },
    {
      "epoch": 0.7850451075641915,
      "grad_norm": 2.195985794067383,
      "learning_rate": 3.455266683411878e-05,
      "loss": 0.5345,
      "step": 4525
    },
    {
      "epoch": 0.7852185981956974,
      "grad_norm": 0.6514781713485718,
      "learning_rate": 3.454799232553077e-05,
      "loss": 0.5372,
      "step": 4526
    },
    {
      "epoch": 0.7853920888272033,
      "grad_norm": 1.0037038326263428,
      "learning_rate": 3.45433161286313e-05,
      "loss": 0.4568,
      "step": 4527
    },
    {
      "epoch": 0.7855655794587092,
      "grad_norm": 2.1790852546691895,
      "learning_rate": 3.4538638243963045e-05,
      "loss": 0.5914,
      "step": 4528
    },
    {
      "epoch": 0.7857390700902152,
      "grad_norm": 0.7083577513694763,
      "learning_rate": 3.453395867206888e-05,
      "loss": 0.5159,
      "step": 4529
    },
    {
      "epoch": 0.785912560721721,
      "grad_norm": 0.737430989742279,
      "learning_rate": 3.4529277413491885e-05,
      "loss": 0.6462,
      "step": 4530
    },
    {
      "epoch": 0.786086051353227,
      "grad_norm": 0.9406954646110535,
      "learning_rate": 3.452459446877531e-05,
      "loss": 0.5854,
      "step": 4531
    },
    {
      "epoch": 0.7862595419847328,
      "grad_norm": 0.6856603026390076,
      "learning_rate": 3.451990983846262e-05,
      "loss": 0.5818,
      "step": 4532
    },
    {
      "epoch": 0.7864330326162388,
      "grad_norm": 1.1390035152435303,
      "learning_rate": 3.4515223523097476e-05,
      "loss": 0.5536,
      "step": 4533
    },
    {
      "epoch": 0.7866065232477446,
      "grad_norm": 0.7555428147315979,
      "learning_rate": 3.451053552322373e-05,
      "loss": 0.52,
      "step": 4534
    },
    {
      "epoch": 0.7867800138792506,
      "grad_norm": 1.1655290126800537,
      "learning_rate": 3.4505845839385426e-05,
      "loss": 0.5337,
      "step": 4535
    },
    {
      "epoch": 0.7869535045107564,
      "grad_norm": 0.8107268214225769,
      "learning_rate": 3.4501154472126815e-05,
      "loss": 0.531,
      "step": 4536
    },
    {
      "epoch": 0.7871269951422624,
      "grad_norm": 0.6117629408836365,
      "learning_rate": 3.449646142199233e-05,
      "loss": 0.5676,
      "step": 4537
    },
    {
      "epoch": 0.7873004857737682,
      "grad_norm": 0.5959532856941223,
      "learning_rate": 3.4491766689526596e-05,
      "loss": 0.703,
      "step": 4538
    },
    {
      "epoch": 0.7874739764052742,
      "grad_norm": 1.2634477615356445,
      "learning_rate": 3.4487070275274454e-05,
      "loss": 0.5743,
      "step": 4539
    },
    {
      "epoch": 0.78764746703678,
      "grad_norm": 0.7086024880409241,
      "learning_rate": 3.448237217978093e-05,
      "loss": 0.6033,
      "step": 4540
    },
    {
      "epoch": 0.7878209576682859,
      "grad_norm": 0.9089869260787964,
      "learning_rate": 3.447767240359124e-05,
      "loss": 0.7173,
      "step": 4541
    },
    {
      "epoch": 0.7879944482997918,
      "grad_norm": 0.8030762672424316,
      "learning_rate": 3.4472970947250794e-05,
      "loss": 0.5846,
      "step": 4542
    },
    {
      "epoch": 0.7881679389312977,
      "grad_norm": 0.5570784211158752,
      "learning_rate": 3.4468267811305206e-05,
      "loss": 0.5676,
      "step": 4543
    },
    {
      "epoch": 0.7883414295628036,
      "grad_norm": 0.7393231391906738,
      "learning_rate": 3.446356299630028e-05,
      "loss": 0.5032,
      "step": 4544
    },
    {
      "epoch": 0.7885149201943095,
      "grad_norm": 0.7207363843917847,
      "learning_rate": 3.4458856502782016e-05,
      "loss": 0.5482,
      "step": 4545
    },
    {
      "epoch": 0.7886884108258154,
      "grad_norm": 0.9359233975410461,
      "learning_rate": 3.4454148331296606e-05,
      "loss": 0.5486,
      "step": 4546
    },
    {
      "epoch": 0.7888619014573213,
      "grad_norm": 0.8161929249763489,
      "learning_rate": 3.444943848239044e-05,
      "loss": 0.5615,
      "step": 4547
    },
    {
      "epoch": 0.7890353920888272,
      "grad_norm": 1.0180940628051758,
      "learning_rate": 3.44447269566101e-05,
      "loss": 0.5215,
      "step": 4548
    },
    {
      "epoch": 0.7892088827203331,
      "grad_norm": 0.7428413033485413,
      "learning_rate": 3.444001375450237e-05,
      "loss": 0.5164,
      "step": 4549
    },
    {
      "epoch": 0.789382373351839,
      "grad_norm": 1.1766921281814575,
      "learning_rate": 3.443529887661421e-05,
      "loss": 0.5972,
      "step": 4550
    },
    {
      "epoch": 0.7895558639833449,
      "grad_norm": 0.6415600180625916,
      "learning_rate": 3.4430582323492805e-05,
      "loss": 0.5316,
      "step": 4551
    },
    {
      "epoch": 0.7897293546148508,
      "grad_norm": 0.6624864339828491,
      "learning_rate": 3.44258640956855e-05,
      "loss": 0.5677,
      "step": 4552
    },
    {
      "epoch": 0.7899028452463567,
      "grad_norm": 0.7604926228523254,
      "learning_rate": 3.4421144193739856e-05,
      "loss": 0.6213,
      "step": 4553
    },
    {
      "epoch": 0.7900763358778626,
      "grad_norm": 0.6163475513458252,
      "learning_rate": 3.441642261820363e-05,
      "loss": 0.5591,
      "step": 4554
    },
    {
      "epoch": 0.7902498265093685,
      "grad_norm": 1.185647964477539,
      "learning_rate": 3.441169936962475e-05,
      "loss": 0.5846,
      "step": 4555
    },
    {
      "epoch": 0.7904233171408744,
      "grad_norm": 0.9624270796775818,
      "learning_rate": 3.440697444855137e-05,
      "loss": 0.5084,
      "step": 4556
    },
    {
      "epoch": 0.7905968077723803,
      "grad_norm": 0.5973623991012573,
      "learning_rate": 3.440224785553183e-05,
      "loss": 0.6094,
      "step": 4557
    },
    {
      "epoch": 0.7907702984038862,
      "grad_norm": 0.6004007458686829,
      "learning_rate": 3.439751959111463e-05,
      "loss": 0.6891,
      "step": 4558
    },
    {
      "epoch": 0.7909437890353921,
      "grad_norm": 0.9249184131622314,
      "learning_rate": 3.439278965584851e-05,
      "loss": 0.5569,
      "step": 4559
    },
    {
      "epoch": 0.7911172796668979,
      "grad_norm": 0.8773783445358276,
      "learning_rate": 3.438805805028238e-05,
      "loss": 0.5096,
      "step": 4560
    },
    {
      "epoch": 0.7912907702984039,
      "grad_norm": 0.7292712330818176,
      "learning_rate": 3.438332477496534e-05,
      "loss": 0.6202,
      "step": 4561
    },
    {
      "epoch": 0.7914642609299097,
      "grad_norm": 1.348038911819458,
      "learning_rate": 3.4378589830446714e-05,
      "loss": 0.6636,
      "step": 4562
    },
    {
      "epoch": 0.7916377515614157,
      "grad_norm": 0.7116172909736633,
      "learning_rate": 3.437385321727597e-05,
      "loss": 0.6177,
      "step": 4563
    },
    {
      "epoch": 0.7918112421929215,
      "grad_norm": 0.6307856440544128,
      "learning_rate": 3.436911493600282e-05,
      "loss": 0.698,
      "step": 4564
    },
    {
      "epoch": 0.7919847328244275,
      "grad_norm": 0.7494140267372131,
      "learning_rate": 3.436437498717713e-05,
      "loss": 0.6954,
      "step": 4565
    },
    {
      "epoch": 0.7921582234559333,
      "grad_norm": 0.830976128578186,
      "learning_rate": 3.4359633371349e-05,
      "loss": 0.5254,
      "step": 4566
    },
    {
      "epoch": 0.7923317140874393,
      "grad_norm": 0.6158875226974487,
      "learning_rate": 3.435489008906867e-05,
      "loss": 0.6903,
      "step": 4567
    },
    {
      "epoch": 0.7925052047189451,
      "grad_norm": 0.6615442037582397,
      "learning_rate": 3.435014514088662e-05,
      "loss": 0.694,
      "step": 4568
    },
    {
      "epoch": 0.7926786953504511,
      "grad_norm": 1.397637963294983,
      "learning_rate": 3.434539852735352e-05,
      "loss": 0.5242,
      "step": 4569
    },
    {
      "epoch": 0.792852185981957,
      "grad_norm": 1.1062660217285156,
      "learning_rate": 3.434065024902019e-05,
      "loss": 0.5393,
      "step": 4570
    },
    {
      "epoch": 0.7930256766134629,
      "grad_norm": 0.7684928178787231,
      "learning_rate": 3.4335900306437694e-05,
      "loss": 0.6217,
      "step": 4571
    },
    {
      "epoch": 0.7931991672449688,
      "grad_norm": 0.6392742395401001,
      "learning_rate": 3.4331148700157263e-05,
      "loss": 0.5267,
      "step": 4572
    },
    {
      "epoch": 0.7933726578764747,
      "grad_norm": 0.6729644536972046,
      "learning_rate": 3.4326395430730325e-05,
      "loss": 0.7158,
      "step": 4573
    },
    {
      "epoch": 0.7935461485079806,
      "grad_norm": 1.0804266929626465,
      "learning_rate": 3.43216404987085e-05,
      "loss": 0.5745,
      "step": 4574
    },
    {
      "epoch": 0.7937196391394865,
      "grad_norm": 0.7597807049751282,
      "learning_rate": 3.431688390464361e-05,
      "loss": 0.5743,
      "step": 4575
    },
    {
      "epoch": 0.7938931297709924,
      "grad_norm": 0.9156981110572815,
      "learning_rate": 3.4312125649087664e-05,
      "loss": 0.5352,
      "step": 4576
    },
    {
      "epoch": 0.7940666204024983,
      "grad_norm": 0.7099980115890503,
      "learning_rate": 3.4307365732592854e-05,
      "loss": 0.6041,
      "step": 4577
    },
    {
      "epoch": 0.7942401110340042,
      "grad_norm": 0.8593411445617676,
      "learning_rate": 3.430260415571158e-05,
      "loss": 0.4979,
      "step": 4578
    },
    {
      "epoch": 0.7944136016655101,
      "grad_norm": 0.7119089365005493,
      "learning_rate": 3.429784091899644e-05,
      "loss": 0.5009,
      "step": 4579
    },
    {
      "epoch": 0.794587092297016,
      "grad_norm": 0.6532433032989502,
      "learning_rate": 3.429307602300019e-05,
      "loss": 0.6216,
      "step": 4580
    },
    {
      "epoch": 0.7947605829285218,
      "grad_norm": 0.6321079134941101,
      "learning_rate": 3.428830946827581e-05,
      "loss": 0.6637,
      "step": 4581
    },
    {
      "epoch": 0.7949340735600278,
      "grad_norm": 0.6303501129150391,
      "learning_rate": 3.4283541255376466e-05,
      "loss": 0.6283,
      "step": 4582
    },
    {
      "epoch": 0.7951075641915336,
      "grad_norm": 0.6134404540061951,
      "learning_rate": 3.427877138485552e-05,
      "loss": 0.626,
      "step": 4583
    },
    {
      "epoch": 0.7952810548230396,
      "grad_norm": 0.7335180044174194,
      "learning_rate": 3.427399985726652e-05,
      "loss": 0.4523,
      "step": 4584
    },
    {
      "epoch": 0.7954545454545454,
      "grad_norm": 0.836480975151062,
      "learning_rate": 3.4269226673163204e-05,
      "loss": 0.5232,
      "step": 4585
    },
    {
      "epoch": 0.7956280360860514,
      "grad_norm": 0.7442110180854797,
      "learning_rate": 3.42644518330995e-05,
      "loss": 0.5751,
      "step": 4586
    },
    {
      "epoch": 0.7958015267175572,
      "grad_norm": 0.6837741136550903,
      "learning_rate": 3.425967533762954e-05,
      "loss": 0.6285,
      "step": 4587
    },
    {
      "epoch": 0.7959750173490632,
      "grad_norm": 0.8320590257644653,
      "learning_rate": 3.425489718730765e-05,
      "loss": 0.6012,
      "step": 4588
    },
    {
      "epoch": 0.796148507980569,
      "grad_norm": 0.7570087313652039,
      "learning_rate": 3.425011738268832e-05,
      "loss": 0.5479,
      "step": 4589
    },
    {
      "epoch": 0.796321998612075,
      "grad_norm": 0.69992595911026,
      "learning_rate": 3.4245335924326274e-05,
      "loss": 0.5217,
      "step": 4590
    },
    {
      "epoch": 0.7964954892435808,
      "grad_norm": 0.8826947808265686,
      "learning_rate": 3.424055281277638e-05,
      "loss": 0.5259,
      "step": 4591
    },
    {
      "epoch": 0.7966689798750868,
      "grad_norm": 0.7894955277442932,
      "learning_rate": 3.423576804859375e-05,
      "loss": 0.507,
      "step": 4592
    },
    {
      "epoch": 0.7968424705065926,
      "grad_norm": 1.5117219686508179,
      "learning_rate": 3.423098163233365e-05,
      "loss": 0.5177,
      "step": 4593
    },
    {
      "epoch": 0.7970159611380986,
      "grad_norm": 0.9371479153633118,
      "learning_rate": 3.422619356455154e-05,
      "loss": 0.6943,
      "step": 4594
    },
    {
      "epoch": 0.7971894517696044,
      "grad_norm": 0.8013246059417725,
      "learning_rate": 3.42214038458031e-05,
      "loss": 0.5144,
      "step": 4595
    },
    {
      "epoch": 0.7973629424011104,
      "grad_norm": 0.6921529173851013,
      "learning_rate": 3.421661247664417e-05,
      "loss": 0.5118,
      "step": 4596
    },
    {
      "epoch": 0.7975364330326162,
      "grad_norm": 0.7774759531021118,
      "learning_rate": 3.4211819457630795e-05,
      "loss": 0.578,
      "step": 4597
    },
    {
      "epoch": 0.7977099236641222,
      "grad_norm": 0.7389329671859741,
      "learning_rate": 3.420702478931921e-05,
      "loss": 0.5549,
      "step": 4598
    },
    {
      "epoch": 0.797883414295628,
      "grad_norm": 1.7281907796859741,
      "learning_rate": 3.420222847226585e-05,
      "loss": 0.7001,
      "step": 4599
    },
    {
      "epoch": 0.7980569049271339,
      "grad_norm": 0.6511208415031433,
      "learning_rate": 3.419743050702732e-05,
      "loss": 0.5558,
      "step": 4600
    },
    {
      "epoch": 0.7982303955586398,
      "grad_norm": 0.9398014545440674,
      "learning_rate": 3.4192630894160435e-05,
      "loss": 0.4885,
      "step": 4601
    },
    {
      "epoch": 0.7984038861901457,
      "grad_norm": 0.6670880913734436,
      "learning_rate": 3.4187829634222205e-05,
      "loss": 0.5802,
      "step": 4602
    },
    {
      "epoch": 0.7985773768216516,
      "grad_norm": 1.0299681425094604,
      "learning_rate": 3.4183026727769806e-05,
      "loss": 0.4952,
      "step": 4603
    },
    {
      "epoch": 0.7987508674531575,
      "grad_norm": 0.7633277773857117,
      "learning_rate": 3.417822217536064e-05,
      "loss": 0.5039,
      "step": 4604
    },
    {
      "epoch": 0.7989243580846634,
      "grad_norm": 0.5957623720169067,
      "learning_rate": 3.417341597755226e-05,
      "loss": 0.689,
      "step": 4605
    },
    {
      "epoch": 0.7990978487161693,
      "grad_norm": 0.8645452260971069,
      "learning_rate": 3.4168608134902443e-05,
      "loss": 0.6707,
      "step": 4606
    },
    {
      "epoch": 0.7992713393476752,
      "grad_norm": 0.6606653928756714,
      "learning_rate": 3.416379864796914e-05,
      "loss": 0.6157,
      "step": 4607
    },
    {
      "epoch": 0.7994448299791811,
      "grad_norm": 0.6899295449256897,
      "learning_rate": 3.4158987517310506e-05,
      "loss": 0.6846,
      "step": 4608
    },
    {
      "epoch": 0.799618320610687,
      "grad_norm": 0.7298061847686768,
      "learning_rate": 3.4154174743484865e-05,
      "loss": 0.5833,
      "step": 4609
    },
    {
      "epoch": 0.7997918112421929,
      "grad_norm": 0.6306266784667969,
      "learning_rate": 3.414936032705075e-05,
      "loss": 0.6777,
      "step": 4610
    },
    {
      "epoch": 0.7999653018736989,
      "grad_norm": 0.6362050175666809,
      "learning_rate": 3.414454426856689e-05,
      "loss": 0.5823,
      "step": 4611
    },
    {
      "epoch": 0.8001387925052047,
      "grad_norm": 0.563105583190918,
      "learning_rate": 3.413972656859218e-05,
      "loss": 0.5461,
      "step": 4612
    },
    {
      "epoch": 0.8003122831367107,
      "grad_norm": 0.7751722931861877,
      "learning_rate": 3.413490722768573e-05,
      "loss": 0.5826,
      "step": 4613
    },
    {
      "epoch": 0.8004857737682165,
      "grad_norm": 0.6509333252906799,
      "learning_rate": 3.413008624640683e-05,
      "loss": 0.5813,
      "step": 4614
    },
    {
      "epoch": 0.8006592643997225,
      "grad_norm": 0.697648823261261,
      "learning_rate": 3.412526362531495e-05,
      "loss": 0.5458,
      "step": 4615
    },
    {
      "epoch": 0.8008327550312283,
      "grad_norm": 0.7336834073066711,
      "learning_rate": 3.4120439364969766e-05,
      "loss": 0.5037,
      "step": 4616
    },
    {
      "epoch": 0.8010062456627343,
      "grad_norm": 0.643248438835144,
      "learning_rate": 3.4115613465931145e-05,
      "loss": 0.6239,
      "step": 4617
    },
    {
      "epoch": 0.8011797362942401,
      "grad_norm": 0.860309898853302,
      "learning_rate": 3.411078592875912e-05,
      "loss": 0.5322,
      "step": 4618
    },
    {
      "epoch": 0.8013532269257461,
      "grad_norm": 0.6955852508544922,
      "learning_rate": 3.4105956754013966e-05,
      "loss": 0.6305,
      "step": 4619
    },
    {
      "epoch": 0.8015267175572519,
      "grad_norm": 0.5716445446014404,
      "learning_rate": 3.410112594225607e-05,
      "loss": 0.5771,
      "step": 4620
    },
    {
      "epoch": 0.8017002081887578,
      "grad_norm": 0.517090380191803,
      "learning_rate": 3.4096293494046103e-05,
      "loss": 0.6381,
      "step": 4621
    },
    {
      "epoch": 0.8018736988202637,
      "grad_norm": 0.6566348671913147,
      "learning_rate": 3.4091459409944836e-05,
      "loss": 0.6127,
      "step": 4622
    },
    {
      "epoch": 0.8020471894517696,
      "grad_norm": 0.6508132815361023,
      "learning_rate": 3.408662369051329e-05,
      "loss": 0.5328,
      "step": 4623
    },
    {
      "epoch": 0.8022206800832755,
      "grad_norm": 0.7261322736740112,
      "learning_rate": 3.408178633631265e-05,
      "loss": 0.5547,
      "step": 4624
    },
    {
      "epoch": 0.8023941707147814,
      "grad_norm": 0.6083239316940308,
      "learning_rate": 3.407694734790429e-05,
      "loss": 0.5652,
      "step": 4625
    },
    {
      "epoch": 0.8025676613462873,
      "grad_norm": 0.6434155702590942,
      "learning_rate": 3.407210672584979e-05,
      "loss": 0.5436,
      "step": 4626
    },
    {
      "epoch": 0.8027411519777932,
      "grad_norm": 0.9602734446525574,
      "learning_rate": 3.406726447071091e-05,
      "loss": 0.5203,
      "step": 4627
    },
    {
      "epoch": 0.8029146426092991,
      "grad_norm": 0.6316145062446594,
      "learning_rate": 3.40624205830496e-05,
      "loss": 0.5953,
      "step": 4628
    },
    {
      "epoch": 0.803088133240805,
      "grad_norm": 0.5899636745452881,
      "learning_rate": 3.405757506342799e-05,
      "loss": 0.6345,
      "step": 4629
    },
    {
      "epoch": 0.8032616238723109,
      "grad_norm": 0.681353747844696,
      "learning_rate": 3.4052727912408414e-05,
      "loss": 0.6948,
      "step": 4630
    },
    {
      "epoch": 0.8034351145038168,
      "grad_norm": 0.6250098943710327,
      "learning_rate": 3.404787913055339e-05,
      "loss": 0.598,
      "step": 4631
    },
    {
      "epoch": 0.8036086051353227,
      "grad_norm": 0.7942088842391968,
      "learning_rate": 3.404302871842563e-05,
      "loss": 0.6128,
      "step": 4632
    },
    {
      "epoch": 0.8037820957668286,
      "grad_norm": 0.765593945980072,
      "learning_rate": 3.4038176676588014e-05,
      "loss": 0.5671,
      "step": 4633
    },
    {
      "epoch": 0.8039555863983345,
      "grad_norm": 0.6088782548904419,
      "learning_rate": 3.403332300560364e-05,
      "loss": 0.6742,
      "step": 4634
    },
    {
      "epoch": 0.8041290770298404,
      "grad_norm": 1.0885931253433228,
      "learning_rate": 3.402846770603578e-05,
      "loss": 0.6034,
      "step": 4635
    },
    {
      "epoch": 0.8043025676613463,
      "grad_norm": 0.6098711490631104,
      "learning_rate": 3.4023610778447895e-05,
      "loss": 0.5867,
      "step": 4636
    },
    {
      "epoch": 0.8044760582928522,
      "grad_norm": 0.7530845403671265,
      "learning_rate": 3.4018752223403634e-05,
      "loss": 0.5793,
      "step": 4637
    },
    {
      "epoch": 0.8046495489243581,
      "grad_norm": 0.7427805662155151,
      "learning_rate": 3.401389204146685e-05,
      "loss": 0.6084,
      "step": 4638
    },
    {
      "epoch": 0.804823039555864,
      "grad_norm": 0.734031617641449,
      "learning_rate": 3.400903023320156e-05,
      "loss": 0.5615,
      "step": 4639
    },
    {
      "epoch": 0.8049965301873698,
      "grad_norm": 0.6565466523170471,
      "learning_rate": 3.4004166799172004e-05,
      "loss": 0.5858,
      "step": 4640
    },
    {
      "epoch": 0.8051700208188758,
      "grad_norm": 0.8570353984832764,
      "learning_rate": 3.399930173994255e-05,
      "loss": 0.7297,
      "step": 4641
    },
    {
      "epoch": 0.8053435114503816,
      "grad_norm": 0.7440707683563232,
      "learning_rate": 3.3994435056077827e-05,
      "loss": 0.5283,
      "step": 4642
    },
    {
      "epoch": 0.8055170020818876,
      "grad_norm": 0.7319448590278625,
      "learning_rate": 3.398956674814261e-05,
      "loss": 0.6826,
      "step": 4643
    },
    {
      "epoch": 0.8056904927133934,
      "grad_norm": 0.7692334651947021,
      "learning_rate": 3.3984696816701865e-05,
      "loss": 0.5645,
      "step": 4644
    },
    {
      "epoch": 0.8058639833448994,
      "grad_norm": 0.7705518007278442,
      "learning_rate": 3.397982526232077e-05,
      "loss": 0.5481,
      "step": 4645
    },
    {
      "epoch": 0.8060374739764052,
      "grad_norm": 0.7228432297706604,
      "learning_rate": 3.397495208556465e-05,
      "loss": 0.5607,
      "step": 4646
    },
    {
      "epoch": 0.8062109646079112,
      "grad_norm": 0.7670116424560547,
      "learning_rate": 3.397007728699907e-05,
      "loss": 0.6168,
      "step": 4647
    },
    {
      "epoch": 0.806384455239417,
      "grad_norm": 0.7331372499465942,
      "learning_rate": 3.3965200867189734e-05,
      "loss": 0.4906,
      "step": 4648
    },
    {
      "epoch": 0.806557945870923,
      "grad_norm": 0.5422827005386353,
      "learning_rate": 3.3960322826702565e-05,
      "loss": 0.5981,
      "step": 4649
    },
    {
      "epoch": 0.8067314365024288,
      "grad_norm": 0.6020506620407104,
      "learning_rate": 3.395544316610367e-05,
      "loss": 0.5762,
      "step": 4650
    },
    {
      "epoch": 0.8069049271339348,
      "grad_norm": 0.8605988025665283,
      "learning_rate": 3.395056188595933e-05,
      "loss": 0.5676,
      "step": 4651
    },
    {
      "epoch": 0.8070784177654406,
      "grad_norm": 0.6065489649772644,
      "learning_rate": 3.394567898683602e-05,
      "loss": 0.5508,
      "step": 4652
    },
    {
      "epoch": 0.8072519083969466,
      "grad_norm": 0.6497067213058472,
      "learning_rate": 3.394079446930043e-05,
      "loss": 0.5767,
      "step": 4653
    },
    {
      "epoch": 0.8074253990284525,
      "grad_norm": 0.7564646601676941,
      "learning_rate": 3.3935908333919385e-05,
      "loss": 0.6957,
      "step": 4654
    },
    {
      "epoch": 0.8075988896599584,
      "grad_norm": 0.6681103706359863,
      "learning_rate": 3.393102058125995e-05,
      "loss": 0.5319,
      "step": 4655
    },
    {
      "epoch": 0.8077723802914643,
      "grad_norm": 0.7304568886756897,
      "learning_rate": 3.3926131211889336e-05,
      "loss": 0.5475,
      "step": 4656
    },
    {
      "epoch": 0.8079458709229702,
      "grad_norm": 0.6326201558113098,
      "learning_rate": 3.392124022637497e-05,
      "loss": 0.6721,
      "step": 4657
    },
    {
      "epoch": 0.8081193615544761,
      "grad_norm": 0.622427225112915,
      "learning_rate": 3.391634762528445e-05,
      "loss": 0.5912,
      "step": 4658
    },
    {
      "epoch": 0.8082928521859819,
      "grad_norm": 0.850075900554657,
      "learning_rate": 3.391145340918557e-05,
      "loss": 0.5068,
      "step": 4659
    },
    {
      "epoch": 0.8084663428174879,
      "grad_norm": 0.8585726022720337,
      "learning_rate": 3.3906557578646317e-05,
      "loss": 0.5958,
      "step": 4660
    },
    {
      "epoch": 0.8086398334489937,
      "grad_norm": 0.7498711347579956,
      "learning_rate": 3.390166013423485e-05,
      "loss": 0.4838,
      "step": 4661
    },
    {
      "epoch": 0.8088133240804997,
      "grad_norm": 0.7035397887229919,
      "learning_rate": 3.389676107651953e-05,
      "loss": 0.6271,
      "step": 4662
    },
    {
      "epoch": 0.8089868147120055,
      "grad_norm": 0.642154335975647,
      "learning_rate": 3.38918604060689e-05,
      "loss": 0.6163,
      "step": 4663
    },
    {
      "epoch": 0.8091603053435115,
      "grad_norm": 0.7695229053497314,
      "learning_rate": 3.388695812345168e-05,
      "loss": 0.5443,
      "step": 4664
    },
    {
      "epoch": 0.8093337959750173,
      "grad_norm": 0.7467740178108215,
      "learning_rate": 3.388205422923678e-05,
      "loss": 0.6442,
      "step": 4665
    },
    {
      "epoch": 0.8095072866065233,
      "grad_norm": 0.5992782115936279,
      "learning_rate": 3.3877148723993306e-05,
      "loss": 0.5465,
      "step": 4666
    },
    {
      "epoch": 0.8096807772380291,
      "grad_norm": 0.7369151711463928,
      "learning_rate": 3.387224160829057e-05,
      "loss": 0.5757,
      "step": 4667
    },
    {
      "epoch": 0.8098542678695351,
      "grad_norm": 0.7979943156242371,
      "learning_rate": 3.3867332882698016e-05,
      "loss": 0.671,
      "step": 4668
    },
    {
      "epoch": 0.8100277585010409,
      "grad_norm": 0.5905467867851257,
      "learning_rate": 3.386242254778533e-05,
      "loss": 0.6077,
      "step": 4669
    },
    {
      "epoch": 0.8102012491325469,
      "grad_norm": 0.781458854675293,
      "learning_rate": 3.385751060412235e-05,
      "loss": 0.6287,
      "step": 4670
    },
    {
      "epoch": 0.8103747397640527,
      "grad_norm": 0.7113485336303711,
      "learning_rate": 3.3852597052279124e-05,
      "loss": 0.5155,
      "step": 4671
    },
    {
      "epoch": 0.8105482303955587,
      "grad_norm": 0.7800762057304382,
      "learning_rate": 3.3847681892825865e-05,
      "loss": 0.536,
      "step": 4672
    },
    {
      "epoch": 0.8107217210270645,
      "grad_norm": 0.6829761266708374,
      "learning_rate": 3.3842765126332984e-05,
      "loss": 0.6334,
      "step": 4673
    },
    {
      "epoch": 0.8108952116585705,
      "grad_norm": 0.6885767579078674,
      "learning_rate": 3.383784675337108e-05,
      "loss": 0.4915,
      "step": 4674
    },
    {
      "epoch": 0.8110687022900763,
      "grad_norm": 0.5690920352935791,
      "learning_rate": 3.383292677451094e-05,
      "loss": 0.5786,
      "step": 4675
    },
    {
      "epoch": 0.8112421929215823,
      "grad_norm": 0.6843216419219971,
      "learning_rate": 3.382800519032353e-05,
      "loss": 0.5209,
      "step": 4676
    },
    {
      "epoch": 0.8114156835530881,
      "grad_norm": 0.588300883769989,
      "learning_rate": 3.382308200138e-05,
      "loss": 0.6359,
      "step": 4677
    },
    {
      "epoch": 0.8115891741845941,
      "grad_norm": 0.682434618473053,
      "learning_rate": 3.38181572082517e-05,
      "loss": 0.6154,
      "step": 4678
    },
    {
      "epoch": 0.8117626648160999,
      "grad_norm": 1.069860577583313,
      "learning_rate": 3.381323081151015e-05,
      "loss": 0.5594,
      "step": 4679
    },
    {
      "epoch": 0.8119361554476058,
      "grad_norm": 0.8422788381576538,
      "learning_rate": 3.3808302811727074e-05,
      "loss": 0.516,
      "step": 4680
    },
    {
      "epoch": 0.8121096460791117,
      "grad_norm": 1.0733751058578491,
      "learning_rate": 3.380337320947437e-05,
      "loss": 0.498,
      "step": 4681
    },
    {
      "epoch": 0.8122831367106176,
      "grad_norm": 0.6904345750808716,
      "learning_rate": 3.379844200532411e-05,
      "loss": 0.6714,
      "step": 4682
    },
    {
      "epoch": 0.8124566273421235,
      "grad_norm": 0.8521090745925903,
      "learning_rate": 3.379350919984858e-05,
      "loss": 0.541,
      "step": 4683
    },
    {
      "epoch": 0.8126301179736294,
      "grad_norm": 0.7869723439216614,
      "learning_rate": 3.378857479362024e-05,
      "loss": 0.5409,
      "step": 4684
    },
    {
      "epoch": 0.8128036086051353,
      "grad_norm": 0.8046953678131104,
      "learning_rate": 3.378363878721173e-05,
      "loss": 0.5199,
      "step": 4685
    },
    {
      "epoch": 0.8129770992366412,
      "grad_norm": 0.5668148994445801,
      "learning_rate": 3.377870118119587e-05,
      "loss": 0.6606,
      "step": 4686
    },
    {
      "epoch": 0.8131505898681471,
      "grad_norm": 0.6504331231117249,
      "learning_rate": 3.377376197614568e-05,
      "loss": 0.631,
      "step": 4687
    },
    {
      "epoch": 0.813324080499653,
      "grad_norm": 0.6131774187088013,
      "learning_rate": 3.376882117263437e-05,
      "loss": 0.6415,
      "step": 4688
    },
    {
      "epoch": 0.813497571131159,
      "grad_norm": 0.905603289604187,
      "learning_rate": 3.376387877123533e-05,
      "loss": 0.5316,
      "step": 4689
    },
    {
      "epoch": 0.8136710617626648,
      "grad_norm": 0.8098279237747192,
      "learning_rate": 3.37589347725221e-05,
      "loss": 0.5935,
      "step": 4690
    },
    {
      "epoch": 0.8138445523941708,
      "grad_norm": 0.5495930314064026,
      "learning_rate": 3.375398917706847e-05,
      "loss": 0.7113,
      "step": 4691
    },
    {
      "epoch": 0.8140180430256766,
      "grad_norm": 0.8750534653663635,
      "learning_rate": 3.374904198544836e-05,
      "loss": 0.6096,
      "step": 4692
    },
    {
      "epoch": 0.8141915336571826,
      "grad_norm": 0.7372435331344604,
      "learning_rate": 3.374409319823592e-05,
      "loss": 0.6508,
      "step": 4693
    },
    {
      "epoch": 0.8143650242886884,
      "grad_norm": 0.7002584934234619,
      "learning_rate": 3.373914281600544e-05,
      "loss": 0.5582,
      "step": 4694
    },
    {
      "epoch": 0.8145385149201944,
      "grad_norm": 0.6547242403030396,
      "learning_rate": 3.3734190839331425e-05,
      "loss": 0.5031,
      "step": 4695
    },
    {
      "epoch": 0.8147120055517002,
      "grad_norm": 0.6443631649017334,
      "learning_rate": 3.372923726878856e-05,
      "loss": 0.6936,
      "step": 4696
    },
    {
      "epoch": 0.8148854961832062,
      "grad_norm": 0.5747844576835632,
      "learning_rate": 3.372428210495172e-05,
      "loss": 0.6733,
      "step": 4697
    },
    {
      "epoch": 0.815058986814712,
      "grad_norm": 0.6448113918304443,
      "learning_rate": 3.371932534839594e-05,
      "loss": 0.66,
      "step": 4698
    },
    {
      "epoch": 0.8152324774462179,
      "grad_norm": 0.8336846828460693,
      "learning_rate": 3.371436699969648e-05,
      "loss": 0.4602,
      "step": 4699
    },
    {
      "epoch": 0.8154059680777238,
      "grad_norm": 0.7222806215286255,
      "learning_rate": 3.370940705942874e-05,
      "loss": 0.5013,
      "step": 4700
    },
    {
      "epoch": 0.8155794587092297,
      "grad_norm": 1.2259315252304077,
      "learning_rate": 3.3704445528168335e-05,
      "loss": 0.6086,
      "step": 4701
    },
    {
      "epoch": 0.8157529493407356,
      "grad_norm": 0.8463292121887207,
      "learning_rate": 3.369948240649106e-05,
      "loss": 0.4943,
      "step": 4702
    },
    {
      "epoch": 0.8159264399722415,
      "grad_norm": 0.6418718695640564,
      "learning_rate": 3.369451769497289e-05,
      "loss": 0.5498,
      "step": 4703
    },
    {
      "epoch": 0.8160999306037474,
      "grad_norm": 0.634361207485199,
      "learning_rate": 3.368955139418998e-05,
      "loss": 0.5438,
      "step": 4704
    },
    {
      "epoch": 0.8162734212352533,
      "grad_norm": 0.6433486938476562,
      "learning_rate": 3.368458350471868e-05,
      "loss": 0.6433,
      "step": 4705
    },
    {
      "epoch": 0.8164469118667592,
      "grad_norm": 0.7039868235588074,
      "learning_rate": 3.367961402713553e-05,
      "loss": 0.6018,
      "step": 4706
    },
    {
      "epoch": 0.8166204024982651,
      "grad_norm": 2.6011016368865967,
      "learning_rate": 3.3674642962017215e-05,
      "loss": 0.5925,
      "step": 4707
    },
    {
      "epoch": 0.816793893129771,
      "grad_norm": 0.8203527331352234,
      "learning_rate": 3.3669670309940663e-05,
      "loss": 0.4731,
      "step": 4708
    },
    {
      "epoch": 0.8169673837612769,
      "grad_norm": 0.862680196762085,
      "learning_rate": 3.366469607148293e-05,
      "loss": 0.6145,
      "step": 4709
    },
    {
      "epoch": 0.8171408743927828,
      "grad_norm": 0.8152139782905579,
      "learning_rate": 3.365972024722131e-05,
      "loss": 0.6744,
      "step": 4710
    },
    {
      "epoch": 0.8173143650242887,
      "grad_norm": 0.7853912711143494,
      "learning_rate": 3.365474283773323e-05,
      "loss": 0.5402,
      "step": 4711
    },
    {
      "epoch": 0.8174878556557946,
      "grad_norm": 0.8514853715896606,
      "learning_rate": 3.3649763843596334e-05,
      "loss": 0.5544,
      "step": 4712
    },
    {
      "epoch": 0.8176613462873005,
      "grad_norm": 0.7401768565177917,
      "learning_rate": 3.364478326538844e-05,
      "loss": 0.5277,
      "step": 4713
    },
    {
      "epoch": 0.8178348369188064,
      "grad_norm": 0.9554262161254883,
      "learning_rate": 3.363980110368755e-05,
      "loss": 0.5295,
      "step": 4714
    },
    {
      "epoch": 0.8180083275503123,
      "grad_norm": 0.6825421452522278,
      "learning_rate": 3.363481735907185e-05,
      "loss": 0.6483,
      "step": 4715
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.6086795926094055,
      "learning_rate": 3.36298320321197e-05,
      "loss": 0.5706,
      "step": 4716
    },
    {
      "epoch": 0.8183553088133241,
      "grad_norm": 0.9270070791244507,
      "learning_rate": 3.3624845123409665e-05,
      "loss": 0.4869,
      "step": 4717
    },
    {
      "epoch": 0.8185287994448299,
      "grad_norm": 0.7197990417480469,
      "learning_rate": 3.361985663352048e-05,
      "loss": 0.5343,
      "step": 4718
    },
    {
      "epoch": 0.8187022900763359,
      "grad_norm": 0.7212938070297241,
      "learning_rate": 3.361486656303106e-05,
      "loss": 0.5411,
      "step": 4719
    },
    {
      "epoch": 0.8188757807078417,
      "grad_norm": 0.613197386264801,
      "learning_rate": 3.360987491252051e-05,
      "loss": 0.6571,
      "step": 4720
    },
    {
      "epoch": 0.8190492713393477,
      "grad_norm": 0.7823191285133362,
      "learning_rate": 3.3604881682568126e-05,
      "loss": 0.4731,
      "step": 4721
    },
    {
      "epoch": 0.8192227619708535,
      "grad_norm": 0.8566967844963074,
      "learning_rate": 3.359988687375336e-05,
      "loss": 0.4845,
      "step": 4722
    },
    {
      "epoch": 0.8193962526023595,
      "grad_norm": 0.6343022584915161,
      "learning_rate": 3.359489048665587e-05,
      "loss": 0.5663,
      "step": 4723
    },
    {
      "epoch": 0.8195697432338653,
      "grad_norm": 1.0479716062545776,
      "learning_rate": 3.3589892521855515e-05,
      "loss": 0.526,
      "step": 4724
    },
    {
      "epoch": 0.8197432338653713,
      "grad_norm": 1.333949327468872,
      "learning_rate": 3.3584892979932284e-05,
      "loss": 0.5077,
      "step": 4725
    },
    {
      "epoch": 0.8199167244968771,
      "grad_norm": 0.8642805814743042,
      "learning_rate": 3.35798918614664e-05,
      "loss": 0.5461,
      "step": 4726
    },
    {
      "epoch": 0.8200902151283831,
      "grad_norm": 1.1042691469192505,
      "learning_rate": 3.357488916703824e-05,
      "loss": 0.475,
      "step": 4727
    },
    {
      "epoch": 0.8202637057598889,
      "grad_norm": 0.8307474851608276,
      "learning_rate": 3.356988489722837e-05,
      "loss": 0.5586,
      "step": 4728
    },
    {
      "epoch": 0.8204371963913949,
      "grad_norm": 0.8251770734786987,
      "learning_rate": 3.3564879052617555e-05,
      "loss": 0.6156,
      "step": 4729
    },
    {
      "epoch": 0.8206106870229007,
      "grad_norm": 0.7211297750473022,
      "learning_rate": 3.355987163378671e-05,
      "loss": 0.5853,
      "step": 4730
    },
    {
      "epoch": 0.8207841776544067,
      "grad_norm": 0.8894070386886597,
      "learning_rate": 3.3554862641316965e-05,
      "loss": 0.5624,
      "step": 4731
    },
    {
      "epoch": 0.8209576682859125,
      "grad_norm": 0.6238900423049927,
      "learning_rate": 3.354985207578961e-05,
      "loss": 0.6147,
      "step": 4732
    },
    {
      "epoch": 0.8211311589174185,
      "grad_norm": 0.5919173359870911,
      "learning_rate": 3.354483993778614e-05,
      "loss": 0.6302,
      "step": 4733
    },
    {
      "epoch": 0.8213046495489243,
      "grad_norm": 0.5965079665184021,
      "learning_rate": 3.3539826227888216e-05,
      "loss": 0.7605,
      "step": 4734
    },
    {
      "epoch": 0.8214781401804303,
      "grad_norm": 0.6897684931755066,
      "learning_rate": 3.3534810946677676e-05,
      "loss": 0.6838,
      "step": 4735
    },
    {
      "epoch": 0.8216516308119362,
      "grad_norm": 0.7165369987487793,
      "learning_rate": 3.352979409473656e-05,
      "loss": 0.6315,
      "step": 4736
    },
    {
      "epoch": 0.8218251214434421,
      "grad_norm": 1.2224963903427124,
      "learning_rate": 3.3524775672647064e-05,
      "loss": 0.5433,
      "step": 4737
    },
    {
      "epoch": 0.821998612074948,
      "grad_norm": 0.7387986779212952,
      "learning_rate": 3.351975568099159e-05,
      "loss": 0.5634,
      "step": 4738
    },
    {
      "epoch": 0.8221721027064538,
      "grad_norm": 0.7433927059173584,
      "learning_rate": 3.3514734120352735e-05,
      "loss": 0.552,
      "step": 4739
    },
    {
      "epoch": 0.8223455933379598,
      "grad_norm": 0.7800529599189758,
      "learning_rate": 3.350971099131322e-05,
      "loss": 0.4694,
      "step": 4740
    },
    {
      "epoch": 0.8225190839694656,
      "grad_norm": 0.8435859084129333,
      "learning_rate": 3.350468629445601e-05,
      "loss": 0.5874,
      "step": 4741
    },
    {
      "epoch": 0.8226925746009716,
      "grad_norm": 0.8645241856575012,
      "learning_rate": 3.349966003036421e-05,
      "loss": 0.6349,
      "step": 4742
    },
    {
      "epoch": 0.8228660652324774,
      "grad_norm": 0.9837126135826111,
      "learning_rate": 3.3494632199621146e-05,
      "loss": 0.5375,
      "step": 4743
    },
    {
      "epoch": 0.8230395558639834,
      "grad_norm": 0.8240866661071777,
      "learning_rate": 3.3489602802810276e-05,
      "loss": 0.4889,
      "step": 4744
    },
    {
      "epoch": 0.8232130464954892,
      "grad_norm": 0.8092725276947021,
      "learning_rate": 3.3484571840515295e-05,
      "loss": 0.4625,
      "step": 4745
    },
    {
      "epoch": 0.8233865371269952,
      "grad_norm": 0.857721209526062,
      "learning_rate": 3.347953931332004e-05,
      "loss": 0.5581,
      "step": 4746
    },
    {
      "epoch": 0.823560027758501,
      "grad_norm": 4.3697943687438965,
      "learning_rate": 3.347450522180854e-05,
      "loss": 0.5386,
      "step": 4747
    },
    {
      "epoch": 0.823733518390007,
      "grad_norm": 0.7056412696838379,
      "learning_rate": 3.3469469566565e-05,
      "loss": 0.5598,
      "step": 4748
    },
    {
      "epoch": 0.8239070090215128,
      "grad_norm": 0.6662498116493225,
      "learning_rate": 3.3464432348173827e-05,
      "loss": 0.5824,
      "step": 4749
    },
    {
      "epoch": 0.8240804996530188,
      "grad_norm": 0.8417497873306274,
      "learning_rate": 3.345939356721959e-05,
      "loss": 0.5107,
      "step": 4750
    },
    {
      "epoch": 0.8242539902845246,
      "grad_norm": 0.5518035888671875,
      "learning_rate": 3.345435322428705e-05,
      "loss": 0.5029,
      "step": 4751
    },
    {
      "epoch": 0.8244274809160306,
      "grad_norm": 0.5892915725708008,
      "learning_rate": 3.3449311319961134e-05,
      "loss": 0.6656,
      "step": 4752
    },
    {
      "epoch": 0.8246009715475364,
      "grad_norm": 0.5732632279396057,
      "learning_rate": 3.344426785482697e-05,
      "loss": 0.6272,
      "step": 4753
    },
    {
      "epoch": 0.8247744621790424,
      "grad_norm": 0.6002928018569946,
      "learning_rate": 3.343922282946985e-05,
      "loss": 0.5825,
      "step": 4754
    },
    {
      "epoch": 0.8249479528105482,
      "grad_norm": 0.7098351120948792,
      "learning_rate": 3.343417624447527e-05,
      "loss": 0.5216,
      "step": 4755
    },
    {
      "epoch": 0.8251214434420542,
      "grad_norm": 0.5916649103164673,
      "learning_rate": 3.342912810042888e-05,
      "loss": 0.7095,
      "step": 4756
    },
    {
      "epoch": 0.82529493407356,
      "grad_norm": 0.5721748471260071,
      "learning_rate": 3.342407839791653e-05,
      "loss": 0.6305,
      "step": 4757
    },
    {
      "epoch": 0.8254684247050659,
      "grad_norm": 0.7761891484260559,
      "learning_rate": 3.3419027137524236e-05,
      "loss": 0.6144,
      "step": 4758
    },
    {
      "epoch": 0.8256419153365718,
      "grad_norm": 0.7536807060241699,
      "learning_rate": 3.34139743198382e-05,
      "loss": 0.5024,
      "step": 4759
    },
    {
      "epoch": 0.8258154059680777,
      "grad_norm": 0.8090083003044128,
      "learning_rate": 3.340891994544483e-05,
      "loss": 0.5054,
      "step": 4760
    },
    {
      "epoch": 0.8259888965995836,
      "grad_norm": 1.0500702857971191,
      "learning_rate": 3.3403864014930665e-05,
      "loss": 0.6921,
      "step": 4761
    },
    {
      "epoch": 0.8261623872310895,
      "grad_norm": 0.6257863640785217,
      "learning_rate": 3.339880652888246e-05,
      "loss": 0.7422,
      "step": 4762
    },
    {
      "epoch": 0.8263358778625954,
      "grad_norm": 0.6666653752326965,
      "learning_rate": 3.339374748788715e-05,
      "loss": 0.573,
      "step": 4763
    },
    {
      "epoch": 0.8265093684941013,
      "grad_norm": 0.8072382211685181,
      "learning_rate": 3.338868689253183e-05,
      "loss": 0.4584,
      "step": 4764
    },
    {
      "epoch": 0.8266828591256072,
      "grad_norm": 0.6549413800239563,
      "learning_rate": 3.338362474340381e-05,
      "loss": 0.5607,
      "step": 4765
    },
    {
      "epoch": 0.8268563497571131,
      "grad_norm": 0.6903956532478333,
      "learning_rate": 3.337856104109053e-05,
      "loss": 0.6548,
      "step": 4766
    },
    {
      "epoch": 0.827029840388619,
      "grad_norm": 0.9219315052032471,
      "learning_rate": 3.337349578617965e-05,
      "loss": 0.6171,
      "step": 4767
    },
    {
      "epoch": 0.8272033310201249,
      "grad_norm": 1.2927672863006592,
      "learning_rate": 3.3368428979259006e-05,
      "loss": 0.687,
      "step": 4768
    },
    {
      "epoch": 0.8273768216516308,
      "grad_norm": 1.2503191232681274,
      "learning_rate": 3.336336062091661e-05,
      "loss": 0.6322,
      "step": 4769
    },
    {
      "epoch": 0.8275503122831367,
      "grad_norm": 0.7192893028259277,
      "learning_rate": 3.335829071174063e-05,
      "loss": 0.5916,
      "step": 4770
    },
    {
      "epoch": 0.8277238029146426,
      "grad_norm": 0.8918823003768921,
      "learning_rate": 3.335321925231946e-05,
      "loss": 0.5858,
      "step": 4771
    },
    {
      "epoch": 0.8278972935461485,
      "grad_norm": 1.0696288347244263,
      "learning_rate": 3.334814624324163e-05,
      "loss": 0.5559,
      "step": 4772
    },
    {
      "epoch": 0.8280707841776545,
      "grad_norm": 2.7750024795532227,
      "learning_rate": 3.334307168509587e-05,
      "loss": 0.5511,
      "step": 4773
    },
    {
      "epoch": 0.8282442748091603,
      "grad_norm": 0.6778345704078674,
      "learning_rate": 3.333799557847109e-05,
      "loss": 0.656,
      "step": 4774
    },
    {
      "epoch": 0.8284177654406663,
      "grad_norm": 0.6657668948173523,
      "learning_rate": 3.3332917923956394e-05,
      "loss": 0.5795,
      "step": 4775
    },
    {
      "epoch": 0.8285912560721721,
      "grad_norm": 0.9768241047859192,
      "learning_rate": 3.332783872214103e-05,
      "loss": 0.5366,
      "step": 4776
    },
    {
      "epoch": 0.8287647467036781,
      "grad_norm": 0.6983128786087036,
      "learning_rate": 3.332275797361446e-05,
      "loss": 0.5173,
      "step": 4777
    },
    {
      "epoch": 0.8289382373351839,
      "grad_norm": 0.9408507943153381,
      "learning_rate": 3.331767567896629e-05,
      "loss": 0.6117,
      "step": 4778
    },
    {
      "epoch": 0.8291117279666897,
      "grad_norm": 0.7272319197654724,
      "learning_rate": 3.331259183878635e-05,
      "loss": 0.5349,
      "step": 4779
    },
    {
      "epoch": 0.8292852185981957,
      "grad_norm": 0.9621340036392212,
      "learning_rate": 3.330750645366461e-05,
      "loss": 0.7329,
      "step": 4780
    },
    {
      "epoch": 0.8294587092297016,
      "grad_norm": 1.641162633895874,
      "learning_rate": 3.330241952419123e-05,
      "loss": 0.552,
      "step": 4781
    },
    {
      "epoch": 0.8296321998612075,
      "grad_norm": 1.0812734365463257,
      "learning_rate": 3.3297331050956576e-05,
      "loss": 0.6604,
      "step": 4782
    },
    {
      "epoch": 0.8298056904927134,
      "grad_norm": 0.7657483816146851,
      "learning_rate": 3.329224103455116e-05,
      "loss": 0.5968,
      "step": 4783
    },
    {
      "epoch": 0.8299791811242193,
      "grad_norm": 0.7395285367965698,
      "learning_rate": 3.328714947556568e-05,
      "loss": 0.5719,
      "step": 4784
    },
    {
      "epoch": 0.8301526717557252,
      "grad_norm": 0.6676485538482666,
      "learning_rate": 3.328205637459102e-05,
      "loss": 0.6592,
      "step": 4785
    },
    {
      "epoch": 0.8303261623872311,
      "grad_norm": 0.6954163312911987,
      "learning_rate": 3.327696173221824e-05,
      "loss": 0.6749,
      "step": 4786
    },
    {
      "epoch": 0.830499653018737,
      "grad_norm": 0.6690672636032104,
      "learning_rate": 3.327186554903859e-05,
      "loss": 0.5436,
      "step": 4787
    },
    {
      "epoch": 0.8306731436502429,
      "grad_norm": 0.5956690311431885,
      "learning_rate": 3.326676782564347e-05,
      "loss": 0.6643,
      "step": 4788
    },
    {
      "epoch": 0.8308466342817488,
      "grad_norm": 0.6999792456626892,
      "learning_rate": 3.3261668562624484e-05,
      "loss": 0.6108,
      "step": 4789
    },
    {
      "epoch": 0.8310201249132547,
      "grad_norm": 0.7433859705924988,
      "learning_rate": 3.325656776057341e-05,
      "loss": 0.5802,
      "step": 4790
    },
    {
      "epoch": 0.8311936155447606,
      "grad_norm": 1.4846487045288086,
      "learning_rate": 3.32514654200822e-05,
      "loss": 0.5504,
      "step": 4791
    },
    {
      "epoch": 0.8313671061762665,
      "grad_norm": 0.6759796738624573,
      "learning_rate": 3.324636154174299e-05,
      "loss": 0.6409,
      "step": 4792
    },
    {
      "epoch": 0.8315405968077724,
      "grad_norm": 0.6552556753158569,
      "learning_rate": 3.3241256126148084e-05,
      "loss": 0.6464,
      "step": 4793
    },
    {
      "epoch": 0.8317140874392783,
      "grad_norm": 0.8450214862823486,
      "learning_rate": 3.3236149173889975e-05,
      "loss": 0.5894,
      "step": 4794
    },
    {
      "epoch": 0.8318875780707842,
      "grad_norm": 0.7100221514701843,
      "learning_rate": 3.323104068556133e-05,
      "loss": 0.6774,
      "step": 4795
    },
    {
      "epoch": 0.8320610687022901,
      "grad_norm": 0.6351094841957092,
      "learning_rate": 3.3225930661755005e-05,
      "loss": 0.5955,
      "step": 4796
    },
    {
      "epoch": 0.832234559333796,
      "grad_norm": 1.1856613159179688,
      "learning_rate": 3.322081910306401e-05,
      "loss": 0.585,
      "step": 4797
    },
    {
      "epoch": 0.8324080499653018,
      "grad_norm": 0.9470002055168152,
      "learning_rate": 3.321570601008155e-05,
      "loss": 0.5363,
      "step": 4798
    },
    {
      "epoch": 0.8325815405968078,
      "grad_norm": 0.7687327861785889,
      "learning_rate": 3.321059138340101e-05,
      "loss": 0.5703,
      "step": 4799
    },
    {
      "epoch": 0.8327550312283136,
      "grad_norm": 0.8524781465530396,
      "learning_rate": 3.320547522361595e-05,
      "loss": 0.4939,
      "step": 4800
    },
    {
      "epoch": 0.8329285218598196,
      "grad_norm": 1.0657740831375122,
      "learning_rate": 3.32003575313201e-05,
      "loss": 0.5284,
      "step": 4801
    },
    {
      "epoch": 0.8331020124913254,
      "grad_norm": 0.6151628494262695,
      "learning_rate": 3.3195238307107375e-05,
      "loss": 0.5389,
      "step": 4802
    },
    {
      "epoch": 0.8332755031228314,
      "grad_norm": 0.9582716226577759,
      "learning_rate": 3.3190117551571876e-05,
      "loss": 0.623,
      "step": 4803
    },
    {
      "epoch": 0.8334489937543372,
      "grad_norm": 1.1062990427017212,
      "learning_rate": 3.318499526530786e-05,
      "loss": 0.5428,
      "step": 4804
    },
    {
      "epoch": 0.8336224843858432,
      "grad_norm": 0.8115580081939697,
      "learning_rate": 3.317987144890978e-05,
      "loss": 0.6187,
      "step": 4805
    },
    {
      "epoch": 0.833795975017349,
      "grad_norm": 0.6153755784034729,
      "learning_rate": 3.317474610297226e-05,
      "loss": 0.5103,
      "step": 4806
    },
    {
      "epoch": 0.833969465648855,
      "grad_norm": 0.586530864238739,
      "learning_rate": 3.31696192280901e-05,
      "loss": 0.6226,
      "step": 4807
    },
    {
      "epoch": 0.8341429562803608,
      "grad_norm": 0.7791578769683838,
      "learning_rate": 3.316449082485829e-05,
      "loss": 0.5255,
      "step": 4808
    },
    {
      "epoch": 0.8343164469118668,
      "grad_norm": 0.6278956532478333,
      "learning_rate": 3.315936089387198e-05,
      "loss": 0.5488,
      "step": 4809
    },
    {
      "epoch": 0.8344899375433726,
      "grad_norm": 0.779247522354126,
      "learning_rate": 3.31542294357265e-05,
      "loss": 0.5624,
      "step": 4810
    },
    {
      "epoch": 0.8346634281748786,
      "grad_norm": 0.8130978345870972,
      "learning_rate": 3.314909645101737e-05,
      "loss": 0.5956,
      "step": 4811
    },
    {
      "epoch": 0.8348369188063844,
      "grad_norm": 0.5932303667068481,
      "learning_rate": 3.3143961940340274e-05,
      "loss": 0.4766,
      "step": 4812
    },
    {
      "epoch": 0.8350104094378904,
      "grad_norm": 0.8169277906417847,
      "learning_rate": 3.313882590429108e-05,
      "loss": 0.5641,
      "step": 4813
    },
    {
      "epoch": 0.8351839000693962,
      "grad_norm": 0.6386431455612183,
      "learning_rate": 3.313368834346583e-05,
      "loss": 0.5211,
      "step": 4814
    },
    {
      "epoch": 0.8353573907009022,
      "grad_norm": 0.9650004506111145,
      "learning_rate": 3.3128549258460734e-05,
      "loss": 0.6526,
      "step": 4815
    },
    {
      "epoch": 0.835530881332408,
      "grad_norm": 0.6431168913841248,
      "learning_rate": 3.312340864987221e-05,
      "loss": 0.6299,
      "step": 4816
    },
    {
      "epoch": 0.8357043719639139,
      "grad_norm": 0.5430099368095398,
      "learning_rate": 3.311826651829682e-05,
      "loss": 0.6063,
      "step": 4817
    },
    {
      "epoch": 0.8358778625954199,
      "grad_norm": 0.6411396265029907,
      "learning_rate": 3.311312286433131e-05,
      "loss": 0.6183,
      "step": 4818
    },
    {
      "epoch": 0.8360513532269257,
      "grad_norm": 0.5702820420265198,
      "learning_rate": 3.310797768857262e-05,
      "loss": 0.5219,
      "step": 4819
    },
    {
      "epoch": 0.8362248438584317,
      "grad_norm": 0.6720927953720093,
      "learning_rate": 3.310283099161783e-05,
      "loss": 0.6099,
      "step": 4820
    },
    {
      "epoch": 0.8363983344899375,
      "grad_norm": 0.7756640315055847,
      "learning_rate": 3.3097682774064255e-05,
      "loss": 0.5182,
      "step": 4821
    },
    {
      "epoch": 0.8365718251214435,
      "grad_norm": 0.7168863415718079,
      "learning_rate": 3.309253303650932e-05,
      "loss": 0.5314,
      "step": 4822
    },
    {
      "epoch": 0.8367453157529493,
      "grad_norm": 0.7782425880432129,
      "learning_rate": 3.308738177955067e-05,
      "loss": 0.6135,
      "step": 4823
    },
    {
      "epoch": 0.8369188063844553,
      "grad_norm": 0.6969422101974487,
      "learning_rate": 3.3082229003786114e-05,
      "loss": 0.5763,
      "step": 4824
    },
    {
      "epoch": 0.8370922970159611,
      "grad_norm": 0.8078770637512207,
      "learning_rate": 3.307707470981364e-05,
      "loss": 0.675,
      "step": 4825
    },
    {
      "epoch": 0.8372657876474671,
      "grad_norm": 0.9803861975669861,
      "learning_rate": 3.3071918898231413e-05,
      "loss": 0.5399,
      "step": 4826
    },
    {
      "epoch": 0.8374392782789729,
      "grad_norm": 0.6594573855400085,
      "learning_rate": 3.306676156963776e-05,
      "loss": 0.6039,
      "step": 4827
    },
    {
      "epoch": 0.8376127689104789,
      "grad_norm": 0.7573655843734741,
      "learning_rate": 3.3061602724631205e-05,
      "loss": 0.5188,
      "step": 4828
    },
    {
      "epoch": 0.8377862595419847,
      "grad_norm": 0.7867535352706909,
      "learning_rate": 3.3056442363810435e-05,
      "loss": 0.573,
      "step": 4829
    },
    {
      "epoch": 0.8379597501734907,
      "grad_norm": 0.9551301002502441,
      "learning_rate": 3.3051280487774316e-05,
      "loss": 0.4702,
      "step": 4830
    },
    {
      "epoch": 0.8381332408049965,
      "grad_norm": 0.8941601514816284,
      "learning_rate": 3.3046117097121884e-05,
      "loss": 0.517,
      "step": 4831
    },
    {
      "epoch": 0.8383067314365025,
      "grad_norm": 0.9159838557243347,
      "learning_rate": 3.304095219245236e-05,
      "loss": 0.507,
      "step": 4832
    },
    {
      "epoch": 0.8384802220680083,
      "grad_norm": 0.8108102083206177,
      "learning_rate": 3.3035785774365136e-05,
      "loss": 0.5586,
      "step": 4833
    },
    {
      "epoch": 0.8386537126995143,
      "grad_norm": 1.6117689609527588,
      "learning_rate": 3.303061784345979e-05,
      "loss": 0.5632,
      "step": 4834
    },
    {
      "epoch": 0.8388272033310201,
      "grad_norm": 0.6373217105865479,
      "learning_rate": 3.3025448400336064e-05,
      "loss": 0.5707,
      "step": 4835
    },
    {
      "epoch": 0.8390006939625261,
      "grad_norm": 0.6130152344703674,
      "learning_rate": 3.302027744559387e-05,
      "loss": 0.6497,
      "step": 4836
    },
    {
      "epoch": 0.8391741845940319,
      "grad_norm": 0.7247080206871033,
      "learning_rate": 3.30151049798333e-05,
      "loss": 0.4744,
      "step": 4837
    },
    {
      "epoch": 0.8393476752255378,
      "grad_norm": 0.9034555554389954,
      "learning_rate": 3.300993100365463e-05,
      "loss": 0.4786,
      "step": 4838
    },
    {
      "epoch": 0.8395211658570437,
      "grad_norm": 0.7661714553833008,
      "learning_rate": 3.3004755517658314e-05,
      "loss": 0.5989,
      "step": 4839
    },
    {
      "epoch": 0.8396946564885496,
      "grad_norm": 0.7545140385627747,
      "learning_rate": 3.299957852244496e-05,
      "loss": 0.6145,
      "step": 4840
    },
    {
      "epoch": 0.8398681471200555,
      "grad_norm": 0.705784261226654,
      "learning_rate": 3.299440001861538e-05,
      "loss": 0.5802,
      "step": 4841
    },
    {
      "epoch": 0.8400416377515614,
      "grad_norm": 0.6998658776283264,
      "learning_rate": 3.298922000677053e-05,
      "loss": 0.5647,
      "step": 4842
    },
    {
      "epoch": 0.8402151283830673,
      "grad_norm": 0.852400004863739,
      "learning_rate": 3.298403848751157e-05,
      "loss": 0.5278,
      "step": 4843
    },
    {
      "epoch": 0.8403886190145732,
      "grad_norm": 0.774298369884491,
      "learning_rate": 3.2978855461439806e-05,
      "loss": 0.5186,
      "step": 4844
    },
    {
      "epoch": 0.8405621096460791,
      "grad_norm": 1.1349924802780151,
      "learning_rate": 3.297367092915675e-05,
      "loss": 0.5896,
      "step": 4845
    },
    {
      "epoch": 0.840735600277585,
      "grad_norm": 1.0255142450332642,
      "learning_rate": 3.296848489126406e-05,
      "loss": 0.4928,
      "step": 4846
    },
    {
      "epoch": 0.8409090909090909,
      "grad_norm": 0.716189980506897,
      "learning_rate": 3.296329734836359e-05,
      "loss": 0.5927,
      "step": 4847
    },
    {
      "epoch": 0.8410825815405968,
      "grad_norm": 0.7085785269737244,
      "learning_rate": 3.295810830105736e-05,
      "loss": 0.4711,
      "step": 4848
    },
    {
      "epoch": 0.8412560721721027,
      "grad_norm": 0.7188493013381958,
      "learning_rate": 3.2952917749947556e-05,
      "loss": 0.6251,
      "step": 4849
    },
    {
      "epoch": 0.8414295628036086,
      "grad_norm": 0.7988995909690857,
      "learning_rate": 3.294772569563656e-05,
      "loss": 0.5269,
      "step": 4850
    },
    {
      "epoch": 0.8416030534351145,
      "grad_norm": 0.8568524122238159,
      "learning_rate": 3.2942532138726906e-05,
      "loss": 0.4526,
      "step": 4851
    },
    {
      "epoch": 0.8417765440666204,
      "grad_norm": 0.9650561809539795,
      "learning_rate": 3.293733707982132e-05,
      "loss": 0.6852,
      "step": 4852
    },
    {
      "epoch": 0.8419500346981263,
      "grad_norm": 0.7690786719322205,
      "learning_rate": 3.2932140519522676e-05,
      "loss": 0.7271,
      "step": 4853
    },
    {
      "epoch": 0.8421235253296322,
      "grad_norm": 2.4386205673217773,
      "learning_rate": 3.292694245843407e-05,
      "loss": 0.5551,
      "step": 4854
    },
    {
      "epoch": 0.8422970159611382,
      "grad_norm": 0.6074408292770386,
      "learning_rate": 3.2921742897158726e-05,
      "loss": 0.616,
      "step": 4855
    },
    {
      "epoch": 0.842470506592644,
      "grad_norm": 0.6939220428466797,
      "learning_rate": 3.2916541836300065e-05,
      "loss": 0.5975,
      "step": 4856
    },
    {
      "epoch": 0.8426439972241498,
      "grad_norm": 0.5517061352729797,
      "learning_rate": 3.2911339276461665e-05,
      "loss": 0.5438,
      "step": 4857
    },
    {
      "epoch": 0.8428174878556558,
      "grad_norm": 0.7376974821090698,
      "learning_rate": 3.290613521824731e-05,
      "loss": 0.5941,
      "step": 4858
    },
    {
      "epoch": 0.8429909784871616,
      "grad_norm": 0.7541755437850952,
      "learning_rate": 3.290092966226092e-05,
      "loss": 0.6024,
      "step": 4859
    },
    {
      "epoch": 0.8431644691186676,
      "grad_norm": 0.6830940246582031,
      "learning_rate": 3.28957226091066e-05,
      "loss": 0.6503,
      "step": 4860
    },
    {
      "epoch": 0.8433379597501734,
      "grad_norm": 0.683634877204895,
      "learning_rate": 3.289051405938865e-05,
      "loss": 0.7583,
      "step": 4861
    },
    {
      "epoch": 0.8435114503816794,
      "grad_norm": 0.5917184948921204,
      "learning_rate": 3.2885304013711525e-05,
      "loss": 0.6101,
      "step": 4862
    },
    {
      "epoch": 0.8436849410131853,
      "grad_norm": 0.7042409777641296,
      "learning_rate": 3.2880092472679854e-05,
      "loss": 0.5956,
      "step": 4863
    },
    {
      "epoch": 0.8438584316446912,
      "grad_norm": 0.6945961713790894,
      "learning_rate": 3.2874879436898444e-05,
      "loss": 0.6509,
      "step": 4864
    },
    {
      "epoch": 0.844031922276197,
      "grad_norm": 0.8403766751289368,
      "learning_rate": 3.286966490697227e-05,
      "loss": 0.5544,
      "step": 4865
    },
    {
      "epoch": 0.844205412907703,
      "grad_norm": 0.6023331880569458,
      "learning_rate": 3.286444888350649e-05,
      "loss": 0.7223,
      "step": 4866
    },
    {
      "epoch": 0.8443789035392089,
      "grad_norm": 0.7682532668113708,
      "learning_rate": 3.285923136710643e-05,
      "loss": 0.5638,
      "step": 4867
    },
    {
      "epoch": 0.8445523941707148,
      "grad_norm": 0.8390847444534302,
      "learning_rate": 3.285401235837758e-05,
      "loss": 0.5304,
      "step": 4868
    },
    {
      "epoch": 0.8447258848022207,
      "grad_norm": 0.837834894657135,
      "learning_rate": 3.284879185792562e-05,
      "loss": 0.5387,
      "step": 4869
    },
    {
      "epoch": 0.8448993754337266,
      "grad_norm": 0.8863024115562439,
      "learning_rate": 3.28435698663564e-05,
      "loss": 0.5472,
      "step": 4870
    },
    {
      "epoch": 0.8450728660652325,
      "grad_norm": 0.7301396727561951,
      "learning_rate": 3.2838346384275924e-05,
      "loss": 0.6384,
      "step": 4871
    },
    {
      "epoch": 0.8452463566967384,
      "grad_norm": 0.9071313142776489,
      "learning_rate": 3.283312141229039e-05,
      "loss": 0.64,
      "step": 4872
    },
    {
      "epoch": 0.8454198473282443,
      "grad_norm": 0.9178149104118347,
      "learning_rate": 3.282789495100616e-05,
      "loss": 0.4503,
      "step": 4873
    },
    {
      "epoch": 0.8455933379597502,
      "grad_norm": 0.6926846504211426,
      "learning_rate": 3.282266700102978e-05,
      "loss": 0.6289,
      "step": 4874
    },
    {
      "epoch": 0.8457668285912561,
      "grad_norm": 0.7270438075065613,
      "learning_rate": 3.281743756296795e-05,
      "loss": 0.5001,
      "step": 4875
    },
    {
      "epoch": 0.8459403192227619,
      "grad_norm": 0.7516143321990967,
      "learning_rate": 3.281220663742756e-05,
      "loss": 0.4733,
      "step": 4876
    },
    {
      "epoch": 0.8461138098542679,
      "grad_norm": 0.7802963256835938,
      "learning_rate": 3.280697422501565e-05,
      "loss": 0.475,
      "step": 4877
    },
    {
      "epoch": 0.8462873004857737,
      "grad_norm": 0.7416410446166992,
      "learning_rate": 3.280174032633947e-05,
      "loss": 0.5905,
      "step": 4878
    },
    {
      "epoch": 0.8464607911172797,
      "grad_norm": 1.9238371849060059,
      "learning_rate": 3.27965049420064e-05,
      "loss": 0.5709,
      "step": 4879
    },
    {
      "epoch": 0.8466342817487855,
      "grad_norm": 0.6889614462852478,
      "learning_rate": 3.279126807262403e-05,
      "loss": 0.5857,
      "step": 4880
    },
    {
      "epoch": 0.8468077723802915,
      "grad_norm": 0.8448830842971802,
      "learning_rate": 3.278602971880009e-05,
      "loss": 0.455,
      "step": 4881
    },
    {
      "epoch": 0.8469812630117973,
      "grad_norm": 1.147859811782837,
      "learning_rate": 3.27807898811425e-05,
      "loss": 0.4927,
      "step": 4882
    },
    {
      "epoch": 0.8471547536433033,
      "grad_norm": 0.738511860370636,
      "learning_rate": 3.2775548560259355e-05,
      "loss": 0.5715,
      "step": 4883
    },
    {
      "epoch": 0.8473282442748091,
      "grad_norm": 1.2310166358947754,
      "learning_rate": 3.277030575675891e-05,
      "loss": 0.5648,
      "step": 4884
    },
    {
      "epoch": 0.8475017349063151,
      "grad_norm": 0.9854730367660522,
      "learning_rate": 3.27650614712496e-05,
      "loss": 0.6177,
      "step": 4885
    },
    {
      "epoch": 0.8476752255378209,
      "grad_norm": 0.8511006832122803,
      "learning_rate": 3.2759815704340034e-05,
      "loss": 0.649,
      "step": 4886
    },
    {
      "epoch": 0.8478487161693269,
      "grad_norm": 0.7844124436378479,
      "learning_rate": 3.275456845663899e-05,
      "loss": 0.5099,
      "step": 4887
    },
    {
      "epoch": 0.8480222068008327,
      "grad_norm": 0.7616360783576965,
      "learning_rate": 3.2749319728755415e-05,
      "loss": 0.5283,
      "step": 4888
    },
    {
      "epoch": 0.8481956974323387,
      "grad_norm": 1.0567359924316406,
      "learning_rate": 3.2744069521298424e-05,
      "loss": 0.5382,
      "step": 4889
    },
    {
      "epoch": 0.8483691880638445,
      "grad_norm": 0.6821542978286743,
      "learning_rate": 3.273881783487732e-05,
      "loss": 0.6072,
      "step": 4890
    },
    {
      "epoch": 0.8485426786953505,
      "grad_norm": 0.6456993222236633,
      "learning_rate": 3.273356467010156e-05,
      "loss": 0.5413,
      "step": 4891
    },
    {
      "epoch": 0.8487161693268563,
      "grad_norm": 0.6628501415252686,
      "learning_rate": 3.2728310027580786e-05,
      "loss": 0.5604,
      "step": 4892
    },
    {
      "epoch": 0.8488896599583623,
      "grad_norm": 0.921438992023468,
      "learning_rate": 3.27230539079248e-05,
      "loss": 0.5522,
      "step": 4893
    },
    {
      "epoch": 0.8490631505898681,
      "grad_norm": 0.8634639978408813,
      "learning_rate": 3.271779631174358e-05,
      "loss": 0.4839,
      "step": 4894
    },
    {
      "epoch": 0.8492366412213741,
      "grad_norm": 0.6726203560829163,
      "learning_rate": 3.271253723964728e-05,
      "loss": 0.5292,
      "step": 4895
    },
    {
      "epoch": 0.84941013185288,
      "grad_norm": 1.0867524147033691,
      "learning_rate": 3.270727669224622e-05,
      "loss": 0.6208,
      "step": 4896
    },
    {
      "epoch": 0.8495836224843858,
      "grad_norm": 0.9534117579460144,
      "learning_rate": 3.2702014670150904e-05,
      "loss": 0.5372,
      "step": 4897
    },
    {
      "epoch": 0.8497571131158917,
      "grad_norm": 0.7047215700149536,
      "learning_rate": 3.269675117397196e-05,
      "loss": 0.4513,
      "step": 4898
    },
    {
      "epoch": 0.8499306037473976,
      "grad_norm": 0.722929835319519,
      "learning_rate": 3.269148620432027e-05,
      "loss": 0.5305,
      "step": 4899
    },
    {
      "epoch": 0.8501040943789036,
      "grad_norm": 0.7983660697937012,
      "learning_rate": 3.268621976180681e-05,
      "loss": 0.5753,
      "step": 4900
    },
    {
      "epoch": 0.8502775850104094,
      "grad_norm": 0.5249261856079102,
      "learning_rate": 3.2680951847042766e-05,
      "loss": 0.6467,
      "step": 4901
    },
    {
      "epoch": 0.8504510756419154,
      "grad_norm": 0.8097899556159973,
      "learning_rate": 3.267568246063948e-05,
      "loss": 0.5132,
      "step": 4902
    },
    {
      "epoch": 0.8506245662734212,
      "grad_norm": 0.8358068466186523,
      "learning_rate": 3.2670411603208484e-05,
      "loss": 0.4952,
      "step": 4903
    },
    {
      "epoch": 0.8507980569049272,
      "grad_norm": 1.2445707321166992,
      "learning_rate": 3.2665139275361446e-05,
      "loss": 0.6091,
      "step": 4904
    },
    {
      "epoch": 0.850971547536433,
      "grad_norm": 0.6666598916053772,
      "learning_rate": 3.2659865477710244e-05,
      "loss": 0.6417,
      "step": 4905
    },
    {
      "epoch": 0.851145038167939,
      "grad_norm": 0.7324227690696716,
      "learning_rate": 3.26545902108669e-05,
      "loss": 0.4935,
      "step": 4906
    },
    {
      "epoch": 0.8513185287994448,
      "grad_norm": 0.7859541177749634,
      "learning_rate": 3.264931347544361e-05,
      "loss": 0.5149,
      "step": 4907
    },
    {
      "epoch": 0.8514920194309508,
      "grad_norm": 0.5554445385932922,
      "learning_rate": 3.2644035272052756e-05,
      "loss": 0.6406,
      "step": 4908
    },
    {
      "epoch": 0.8516655100624566,
      "grad_norm": 0.7216631174087524,
      "learning_rate": 3.263875560130689e-05,
      "loss": 0.5939,
      "step": 4909
    },
    {
      "epoch": 0.8518390006939626,
      "grad_norm": 0.9014060497283936,
      "learning_rate": 3.263347446381869e-05,
      "loss": 0.5608,
      "step": 4910
    },
    {
      "epoch": 0.8520124913254684,
      "grad_norm": 0.6206343173980713,
      "learning_rate": 3.262819186020106e-05,
      "loss": 0.5126,
      "step": 4911
    },
    {
      "epoch": 0.8521859819569744,
      "grad_norm": 1.0631262063980103,
      "learning_rate": 3.2622907791067056e-05,
      "loss": 0.5747,
      "step": 4912
    },
    {
      "epoch": 0.8523594725884802,
      "grad_norm": 0.7039932012557983,
      "learning_rate": 3.261762225702989e-05,
      "loss": 0.6125,
      "step": 4913
    },
    {
      "epoch": 0.8525329632199862,
      "grad_norm": 3.6588799953460693,
      "learning_rate": 3.261233525870296e-05,
      "loss": 0.4834,
      "step": 4914
    },
    {
      "epoch": 0.852706453851492,
      "grad_norm": 0.9658022522926331,
      "learning_rate": 3.2607046796699824e-05,
      "loss": 0.6519,
      "step": 4915
    },
    {
      "epoch": 0.8528799444829979,
      "grad_norm": 0.6569541096687317,
      "learning_rate": 3.260175687163423e-05,
      "loss": 0.552,
      "step": 4916
    },
    {
      "epoch": 0.8530534351145038,
      "grad_norm": 0.8620007634162903,
      "learning_rate": 3.259646548412005e-05,
      "loss": 0.5516,
      "step": 4917
    },
    {
      "epoch": 0.8532269257460097,
      "grad_norm": 0.7383683919906616,
      "learning_rate": 3.259117263477138e-05,
      "loss": 0.5127,
      "step": 4918
    },
    {
      "epoch": 0.8534004163775156,
      "grad_norm": 0.8829156160354614,
      "learning_rate": 3.258587832420246e-05,
      "loss": 0.7289,
      "step": 4919
    },
    {
      "epoch": 0.8535739070090215,
      "grad_norm": 0.6492002606391907,
      "learning_rate": 3.2580582553027684e-05,
      "loss": 0.717,
      "step": 4920
    },
    {
      "epoch": 0.8537473976405274,
      "grad_norm": 0.7520597577095032,
      "learning_rate": 3.2575285321861656e-05,
      "loss": 0.5946,
      "step": 4921
    },
    {
      "epoch": 0.8539208882720333,
      "grad_norm": 0.6571438312530518,
      "learning_rate": 3.2569986631319104e-05,
      "loss": 0.6296,
      "step": 4922
    },
    {
      "epoch": 0.8540943789035392,
      "grad_norm": 0.831369161605835,
      "learning_rate": 3.256468648201496e-05,
      "loss": 0.5238,
      "step": 4923
    },
    {
      "epoch": 0.8542678695350451,
      "grad_norm": 0.7360967397689819,
      "learning_rate": 3.2559384874564305e-05,
      "loss": 0.556,
      "step": 4924
    },
    {
      "epoch": 0.854441360166551,
      "grad_norm": 0.8607320785522461,
      "learning_rate": 3.25540818095824e-05,
      "loss": 0.4917,
      "step": 4925
    },
    {
      "epoch": 0.8546148507980569,
      "grad_norm": 1.5490577220916748,
      "learning_rate": 3.254877728768468e-05,
      "loss": 0.509,
      "step": 4926
    },
    {
      "epoch": 0.8547883414295628,
      "grad_norm": 0.9422374963760376,
      "learning_rate": 3.254347130948673e-05,
      "loss": 0.6012,
      "step": 4927
    },
    {
      "epoch": 0.8549618320610687,
      "grad_norm": 1.096207857131958,
      "learning_rate": 3.2538163875604316e-05,
      "loss": 0.5375,
      "step": 4928
    },
    {
      "epoch": 0.8551353226925746,
      "grad_norm": 0.7442988157272339,
      "learning_rate": 3.2532854986653375e-05,
      "loss": 0.5046,
      "step": 4929
    },
    {
      "epoch": 0.8553088133240805,
      "grad_norm": 0.9817276000976562,
      "learning_rate": 3.252754464325001e-05,
      "loss": 0.491,
      "step": 4930
    },
    {
      "epoch": 0.8554823039555864,
      "grad_norm": 1.6339360475540161,
      "learning_rate": 3.2522232846010496e-05,
      "loss": 0.4955,
      "step": 4931
    },
    {
      "epoch": 0.8556557945870923,
      "grad_norm": 0.9660579562187195,
      "learning_rate": 3.251691959555126e-05,
      "loss": 0.4913,
      "step": 4932
    },
    {
      "epoch": 0.8558292852185982,
      "grad_norm": 0.8408375382423401,
      "learning_rate": 3.251160489248893e-05,
      "loss": 0.5836,
      "step": 4933
    },
    {
      "epoch": 0.8560027758501041,
      "grad_norm": 0.7440459132194519,
      "learning_rate": 3.2506288737440265e-05,
      "loss": 0.6317,
      "step": 4934
    },
    {
      "epoch": 0.8561762664816099,
      "grad_norm": 0.6120437383651733,
      "learning_rate": 3.250097113102222e-05,
      "loss": 0.6022,
      "step": 4935
    },
    {
      "epoch": 0.8563497571131159,
      "grad_norm": 0.9216663241386414,
      "learning_rate": 3.249565207385192e-05,
      "loss": 0.5349,
      "step": 4936
    },
    {
      "epoch": 0.8565232477446217,
      "grad_norm": 0.7753810882568359,
      "learning_rate": 3.249033156654663e-05,
      "loss": 0.5497,
      "step": 4937
    },
    {
      "epoch": 0.8566967383761277,
      "grad_norm": 1.2120603322982788,
      "learning_rate": 3.24850096097238e-05,
      "loss": 0.6997,
      "step": 4938
    },
    {
      "epoch": 0.8568702290076335,
      "grad_norm": 0.9648410677909851,
      "learning_rate": 3.247968620400106e-05,
      "loss": 0.527,
      "step": 4939
    },
    {
      "epoch": 0.8570437196391395,
      "grad_norm": 0.8988485336303711,
      "learning_rate": 3.2474361349996205e-05,
      "loss": 0.5538,
      "step": 4940
    },
    {
      "epoch": 0.8572172102706453,
      "grad_norm": 2.031445264816284,
      "learning_rate": 3.2469035048327166e-05,
      "loss": 0.6179,
      "step": 4941
    },
    {
      "epoch": 0.8573907009021513,
      "grad_norm": 0.8860834240913391,
      "learning_rate": 3.2463707299612086e-05,
      "loss": 0.5106,
      "step": 4942
    },
    {
      "epoch": 0.8575641915336571,
      "grad_norm": 0.7643118500709534,
      "learning_rate": 3.245837810446925e-05,
      "loss": 0.5621,
      "step": 4943
    },
    {
      "epoch": 0.8577376821651631,
      "grad_norm": 0.6657795906066895,
      "learning_rate": 3.245304746351712e-05,
      "loss": 0.5447,
      "step": 4944
    },
    {
      "epoch": 0.857911172796669,
      "grad_norm": 1.121260643005371,
      "learning_rate": 3.244771537737432e-05,
      "loss": 0.5657,
      "step": 4945
    },
    {
      "epoch": 0.8580846634281749,
      "grad_norm": 0.910545825958252,
      "learning_rate": 3.2442381846659644e-05,
      "loss": 0.6078,
      "step": 4946
    },
    {
      "epoch": 0.8582581540596808,
      "grad_norm": 0.6867709159851074,
      "learning_rate": 3.243704687199206e-05,
      "loss": 0.6078,
      "step": 4947
    },
    {
      "epoch": 0.8584316446911867,
      "grad_norm": 0.9394713044166565,
      "learning_rate": 3.243171045399069e-05,
      "loss": 0.5994,
      "step": 4948
    },
    {
      "epoch": 0.8586051353226926,
      "grad_norm": 0.8011114597320557,
      "learning_rate": 3.2426372593274834e-05,
      "loss": 0.6193,
      "step": 4949
    },
    {
      "epoch": 0.8587786259541985,
      "grad_norm": 1.425828456878662,
      "learning_rate": 3.2421033290463966e-05,
      "loss": 0.4648,
      "step": 4950
    },
    {
      "epoch": 0.8589521165857044,
      "grad_norm": 0.8111879229545593,
      "learning_rate": 3.2415692546177714e-05,
      "loss": 0.5406,
      "step": 4951
    },
    {
      "epoch": 0.8591256072172103,
      "grad_norm": 0.621471107006073,
      "learning_rate": 3.241035036103587e-05,
      "loss": 0.6207,
      "step": 4952
    },
    {
      "epoch": 0.8592990978487162,
      "grad_norm": 0.7024137377738953,
      "learning_rate": 3.24050067356584e-05,
      "loss": 0.6758,
      "step": 4953
    },
    {
      "epoch": 0.8594725884802221,
      "grad_norm": 0.7092933058738708,
      "learning_rate": 3.239966167066545e-05,
      "loss": 0.5184,
      "step": 4954
    },
    {
      "epoch": 0.859646079111728,
      "grad_norm": 0.6382533311843872,
      "learning_rate": 3.239431516667732e-05,
      "loss": 0.4784,
      "step": 4955
    },
    {
      "epoch": 0.8598195697432338,
      "grad_norm": 0.899437427520752,
      "learning_rate": 3.2388967224314464e-05,
      "loss": 0.6016,
      "step": 4956
    },
    {
      "epoch": 0.8599930603747398,
      "grad_norm": 0.6273519396781921,
      "learning_rate": 3.238361784419753e-05,
      "loss": 0.6202,
      "step": 4957
    },
    {
      "epoch": 0.8601665510062456,
      "grad_norm": 0.7345899939537048,
      "learning_rate": 3.2378267026947314e-05,
      "loss": 0.5728,
      "step": 4958
    },
    {
      "epoch": 0.8603400416377516,
      "grad_norm": 1.0373185873031616,
      "learning_rate": 3.237291477318478e-05,
      "loss": 0.577,
      "step": 4959
    },
    {
      "epoch": 0.8605135322692574,
      "grad_norm": 0.5629845261573792,
      "learning_rate": 3.2367561083531074e-05,
      "loss": 0.6731,
      "step": 4960
    },
    {
      "epoch": 0.8606870229007634,
      "grad_norm": 0.7160812020301819,
      "learning_rate": 3.2362205958607495e-05,
      "loss": 0.6404,
      "step": 4961
    },
    {
      "epoch": 0.8608605135322692,
      "grad_norm": 0.814015805721283,
      "learning_rate": 3.235684939903551e-05,
      "loss": 0.51,
      "step": 4962
    },
    {
      "epoch": 0.8610340041637752,
      "grad_norm": 0.7279803156852722,
      "learning_rate": 3.235149140543675e-05,
      "loss": 0.5006,
      "step": 4963
    },
    {
      "epoch": 0.861207494795281,
      "grad_norm": 0.5915574431419373,
      "learning_rate": 3.234613197843302e-05,
      "loss": 0.6266,
      "step": 4964
    },
    {
      "epoch": 0.861380985426787,
      "grad_norm": 0.649194598197937,
      "learning_rate": 3.234077111864629e-05,
      "loss": 0.5829,
      "step": 4965
    },
    {
      "epoch": 0.8615544760582928,
      "grad_norm": 0.8546715974807739,
      "learning_rate": 3.233540882669869e-05,
      "loss": 0.4796,
      "step": 4966
    },
    {
      "epoch": 0.8617279666897988,
      "grad_norm": 0.623928964138031,
      "learning_rate": 3.233004510321253e-05,
      "loss": 0.6228,
      "step": 4967
    },
    {
      "epoch": 0.8619014573213046,
      "grad_norm": 0.5498993396759033,
      "learning_rate": 3.232467994881026e-05,
      "loss": 0.5118,
      "step": 4968
    },
    {
      "epoch": 0.8620749479528106,
      "grad_norm": 0.5465213656425476,
      "learning_rate": 3.2319313364114524e-05,
      "loss": 0.6534,
      "step": 4969
    },
    {
      "epoch": 0.8622484385843164,
      "grad_norm": 0.6918518543243408,
      "learning_rate": 3.2313945349748116e-05,
      "loss": 0.5933,
      "step": 4970
    },
    {
      "epoch": 0.8624219292158224,
      "grad_norm": 0.7475275993347168,
      "learning_rate": 3.2308575906334004e-05,
      "loss": 0.6959,
      "step": 4971
    },
    {
      "epoch": 0.8625954198473282,
      "grad_norm": 0.9270092248916626,
      "learning_rate": 3.230320503449531e-05,
      "loss": 0.6343,
      "step": 4972
    },
    {
      "epoch": 0.8627689104788342,
      "grad_norm": 0.9913419485092163,
      "learning_rate": 3.229783273485534e-05,
      "loss": 0.5769,
      "step": 4973
    },
    {
      "epoch": 0.86294240111034,
      "grad_norm": 0.8261295557022095,
      "learning_rate": 3.2292459008037554e-05,
      "loss": 0.564,
      "step": 4974
    },
    {
      "epoch": 0.8631158917418459,
      "grad_norm": 0.8295417428016663,
      "learning_rate": 3.2287083854665566e-05,
      "loss": 0.5098,
      "step": 4975
    },
    {
      "epoch": 0.8632893823733518,
      "grad_norm": 0.6348004341125488,
      "learning_rate": 3.228170727536319e-05,
      "loss": 0.6146,
      "step": 4976
    },
    {
      "epoch": 0.8634628730048577,
      "grad_norm": 0.6621866822242737,
      "learning_rate": 3.227632927075437e-05,
      "loss": 0.5645,
      "step": 4977
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 0.821857213973999,
      "learning_rate": 3.227094984146323e-05,
      "loss": 0.5233,
      "step": 4978
    },
    {
      "epoch": 0.8638098542678695,
      "grad_norm": 0.7312668561935425,
      "learning_rate": 3.2265568988114075e-05,
      "loss": 0.6039,
      "step": 4979
    },
    {
      "epoch": 0.8639833448993754,
      "grad_norm": 0.576884388923645,
      "learning_rate": 3.226018671133134e-05,
      "loss": 0.6044,
      "step": 4980
    },
    {
      "epoch": 0.8641568355308813,
      "grad_norm": 1.4261698722839355,
      "learning_rate": 3.225480301173964e-05,
      "loss": 0.5365,
      "step": 4981
    },
    {
      "epoch": 0.8643303261623873,
      "grad_norm": 0.7893202304840088,
      "learning_rate": 3.224941788996378e-05,
      "loss": 0.4792,
      "step": 4982
    },
    {
      "epoch": 0.8645038167938931,
      "grad_norm": 0.7231020927429199,
      "learning_rate": 3.22440313466287e-05,
      "loss": 0.4773,
      "step": 4983
    },
    {
      "epoch": 0.864677307425399,
      "grad_norm": 0.8429858088493347,
      "learning_rate": 3.223864338235951e-05,
      "loss": 0.4828,
      "step": 4984
    },
    {
      "epoch": 0.8648507980569049,
      "grad_norm": 0.813918948173523,
      "learning_rate": 3.22332539977815e-05,
      "loss": 0.5181,
      "step": 4985
    },
    {
      "epoch": 0.8650242886884109,
      "grad_norm": 0.6264095306396484,
      "learning_rate": 3.2227863193520115e-05,
      "loss": 0.5846,
      "step": 4986
    },
    {
      "epoch": 0.8651977793199167,
      "grad_norm": 0.7643534541130066,
      "learning_rate": 3.222247097020095e-05,
      "loss": 0.5066,
      "step": 4987
    },
    {
      "epoch": 0.8653712699514227,
      "grad_norm": 0.8795880675315857,
      "learning_rate": 3.221707732844979e-05,
      "loss": 0.5547,
      "step": 4988
    },
    {
      "epoch": 0.8655447605829285,
      "grad_norm": 0.7697674632072449,
      "learning_rate": 3.221168226889257e-05,
      "loss": 0.692,
      "step": 4989
    },
    {
      "epoch": 0.8657182512144345,
      "grad_norm": 0.575239896774292,
      "learning_rate": 3.220628579215539e-05,
      "loss": 0.5803,
      "step": 4990
    },
    {
      "epoch": 0.8658917418459403,
      "grad_norm": 0.7778034806251526,
      "learning_rate": 3.220088789886452e-05,
      "loss": 0.6017,
      "step": 4991
    },
    {
      "epoch": 0.8660652324774463,
      "grad_norm": 0.7716045379638672,
      "learning_rate": 3.21954885896464e-05,
      "loss": 0.5767,
      "step": 4992
    },
    {
      "epoch": 0.8662387231089521,
      "grad_norm": 0.7348926663398743,
      "learning_rate": 3.219008786512762e-05,
      "loss": 0.5782,
      "step": 4993
    },
    {
      "epoch": 0.8664122137404581,
      "grad_norm": 0.9667244553565979,
      "learning_rate": 3.2184685725934926e-05,
      "loss": 0.5251,
      "step": 4994
    },
    {
      "epoch": 0.8665857043719639,
      "grad_norm": 0.9431669116020203,
      "learning_rate": 3.217928217269527e-05,
      "loss": 0.5309,
      "step": 4995
    },
    {
      "epoch": 0.8667591950034698,
      "grad_norm": 0.7136666774749756,
      "learning_rate": 3.2173877206035714e-05,
      "loss": 0.5741,
      "step": 4996
    },
    {
      "epoch": 0.8669326856349757,
      "grad_norm": 0.7457221150398254,
      "learning_rate": 3.2168470826583525e-05,
      "loss": 0.611,
      "step": 4997
    },
    {
      "epoch": 0.8671061762664816,
      "grad_norm": 0.7214698195457458,
      "learning_rate": 3.2163063034966126e-05,
      "loss": 0.5966,
      "step": 4998
    },
    {
      "epoch": 0.8672796668979875,
      "grad_norm": 0.6944583654403687,
      "learning_rate": 3.215765383181108e-05,
      "loss": 0.5519,
      "step": 4999
    },
    {
      "epoch": 0.8674531575294934,
      "grad_norm": 0.8013060092926025,
      "learning_rate": 3.215224321774614e-05,
      "loss": 0.5673,
      "step": 5000
    },
    {
      "epoch": 0.8676266481609993,
      "grad_norm": 1.4525957107543945,
      "learning_rate": 3.2146831193399225e-05,
      "loss": 0.5699,
      "step": 5001
    },
    {
      "epoch": 0.8678001387925052,
      "grad_norm": 0.9097924828529358,
      "learning_rate": 3.214141775939839e-05,
      "loss": 0.5128,
      "step": 5002
    },
    {
      "epoch": 0.8679736294240111,
      "grad_norm": 0.6187672019004822,
      "learning_rate": 3.213600291637187e-05,
      "loss": 0.542,
      "step": 5003
    },
    {
      "epoch": 0.868147120055517,
      "grad_norm": 0.7136866450309753,
      "learning_rate": 3.213058666494807e-05,
      "loss": 0.567,
      "step": 5004
    },
    {
      "epoch": 0.8683206106870229,
      "grad_norm": 0.7285737991333008,
      "learning_rate": 3.2125169005755566e-05,
      "loss": 0.5281,
      "step": 5005
    },
    {
      "epoch": 0.8684941013185288,
      "grad_norm": 0.6819010376930237,
      "learning_rate": 3.2119749939423063e-05,
      "loss": 0.5353,
      "step": 5006
    },
    {
      "epoch": 0.8686675919500347,
      "grad_norm": 0.632310688495636,
      "learning_rate": 3.211432946657946e-05,
      "loss": 0.7002,
      "step": 5007
    },
    {
      "epoch": 0.8688410825815406,
      "grad_norm": 0.5707098841667175,
      "learning_rate": 3.2108907587853794e-05,
      "loss": 0.6343,
      "step": 5008
    },
    {
      "epoch": 0.8690145732130465,
      "grad_norm": 0.6051770448684692,
      "learning_rate": 3.210348430387531e-05,
      "loss": 0.6757,
      "step": 5009
    },
    {
      "epoch": 0.8691880638445524,
      "grad_norm": 1.1853944063186646,
      "learning_rate": 3.2098059615273354e-05,
      "loss": 0.6776,
      "step": 5010
    },
    {
      "epoch": 0.8693615544760583,
      "grad_norm": 0.6617575287818909,
      "learning_rate": 3.209263352267749e-05,
      "loss": 0.6189,
      "step": 5011
    },
    {
      "epoch": 0.8695350451075642,
      "grad_norm": 0.8673063516616821,
      "learning_rate": 3.2087206026717415e-05,
      "loss": 0.6863,
      "step": 5012
    },
    {
      "epoch": 0.8697085357390701,
      "grad_norm": 0.602722704410553,
      "learning_rate": 3.2081777128023005e-05,
      "loss": 0.5613,
      "step": 5013
    },
    {
      "epoch": 0.869882026370576,
      "grad_norm": 0.6934711933135986,
      "learning_rate": 3.207634682722427e-05,
      "loss": 0.5569,
      "step": 5014
    },
    {
      "epoch": 0.8700555170020818,
      "grad_norm": 0.7566132545471191,
      "learning_rate": 3.2070915124951406e-05,
      "loss": 0.5197,
      "step": 5015
    },
    {
      "epoch": 0.8702290076335878,
      "grad_norm": 0.8444889783859253,
      "learning_rate": 3.206548202183479e-05,
      "loss": 0.4972,
      "step": 5016
    },
    {
      "epoch": 0.8704024982650936,
      "grad_norm": 0.7059224843978882,
      "learning_rate": 3.206004751850493e-05,
      "loss": 0.5042,
      "step": 5017
    },
    {
      "epoch": 0.8705759888965996,
      "grad_norm": 0.7606081366539001,
      "learning_rate": 3.20546116155925e-05,
      "loss": 0.5923,
      "step": 5018
    },
    {
      "epoch": 0.8707494795281054,
      "grad_norm": 0.6471020579338074,
      "learning_rate": 3.204917431372833e-05,
      "loss": 0.4932,
      "step": 5019
    },
    {
      "epoch": 0.8709229701596114,
      "grad_norm": 0.716760516166687,
      "learning_rate": 3.204373561354345e-05,
      "loss": 0.5991,
      "step": 5020
    },
    {
      "epoch": 0.8710964607911172,
      "grad_norm": 0.8604891300201416,
      "learning_rate": 3.2038295515669024e-05,
      "loss": 0.6353,
      "step": 5021
    },
    {
      "epoch": 0.8712699514226232,
      "grad_norm": 0.7452986836433411,
      "learning_rate": 3.203285402073637e-05,
      "loss": 0.5154,
      "step": 5022
    },
    {
      "epoch": 0.871443442054129,
      "grad_norm": 0.981040894985199,
      "learning_rate": 3.202741112937699e-05,
      "loss": 0.5051,
      "step": 5023
    },
    {
      "epoch": 0.871616932685635,
      "grad_norm": 1.1439461708068848,
      "learning_rate": 3.202196684222253e-05,
      "loss": 0.5151,
      "step": 5024
    },
    {
      "epoch": 0.8717904233171409,
      "grad_norm": 0.6975328326225281,
      "learning_rate": 3.201652115990481e-05,
      "loss": 0.6628,
      "step": 5025
    },
    {
      "epoch": 0.8719639139486468,
      "grad_norm": 0.7577537298202515,
      "learning_rate": 3.2011074083055814e-05,
      "loss": 0.5559,
      "step": 5026
    },
    {
      "epoch": 0.8721374045801527,
      "grad_norm": 0.6863942742347717,
      "learning_rate": 3.200562561230767e-05,
      "loss": 0.5581,
      "step": 5027
    },
    {
      "epoch": 0.8723108952116586,
      "grad_norm": 2.8689303398132324,
      "learning_rate": 3.200017574829268e-05,
      "loss": 0.6234,
      "step": 5028
    },
    {
      "epoch": 0.8724843858431645,
      "grad_norm": 0.6341760158538818,
      "learning_rate": 3.199472449164332e-05,
      "loss": 0.5452,
      "step": 5029
    },
    {
      "epoch": 0.8726578764746704,
      "grad_norm": 0.5848334431648254,
      "learning_rate": 3.19892718429922e-05,
      "loss": 0.6141,
      "step": 5030
    },
    {
      "epoch": 0.8728313671061763,
      "grad_norm": 0.8144198060035706,
      "learning_rate": 3.198381780297211e-05,
      "loss": 0.6423,
      "step": 5031
    },
    {
      "epoch": 0.8730048577376822,
      "grad_norm": 0.6849162578582764,
      "learning_rate": 3.1978362372216006e-05,
      "loss": 0.6716,
      "step": 5032
    },
    {
      "epoch": 0.8731783483691881,
      "grad_norm": 0.6544033885002136,
      "learning_rate": 3.1972905551356995e-05,
      "loss": 0.6765,
      "step": 5033
    },
    {
      "epoch": 0.8733518390006939,
      "grad_norm": 0.5681815147399902,
      "learning_rate": 3.196744734102833e-05,
      "loss": 0.6099,
      "step": 5034
    },
    {
      "epoch": 0.8735253296321999,
      "grad_norm": 0.7293042540550232,
      "learning_rate": 3.196198774186347e-05,
      "loss": 0.5864,
      "step": 5035
    },
    {
      "epoch": 0.8736988202637057,
      "grad_norm": 0.7678490281105042,
      "learning_rate": 3.195652675449599e-05,
      "loss": 0.4551,
      "step": 5036
    },
    {
      "epoch": 0.8738723108952117,
      "grad_norm": 0.6535952091217041,
      "learning_rate": 3.195106437955964e-05,
      "loss": 0.6561,
      "step": 5037
    },
    {
      "epoch": 0.8740458015267175,
      "grad_norm": 0.6233806610107422,
      "learning_rate": 3.194560061768835e-05,
      "loss": 0.5675,
      "step": 5038
    },
    {
      "epoch": 0.8742192921582235,
      "grad_norm": 0.580977201461792,
      "learning_rate": 3.194013546951619e-05,
      "loss": 0.5573,
      "step": 5039
    },
    {
      "epoch": 0.8743927827897293,
      "grad_norm": 0.5699653029441833,
      "learning_rate": 3.193466893567739e-05,
      "loss": 0.6335,
      "step": 5040
    },
    {
      "epoch": 0.8745662734212353,
      "grad_norm": 0.6607775688171387,
      "learning_rate": 3.1929201016806355e-05,
      "loss": 0.5901,
      "step": 5041
    },
    {
      "epoch": 0.8747397640527411,
      "grad_norm": 0.6944895386695862,
      "learning_rate": 3.192373171353765e-05,
      "loss": 0.6528,
      "step": 5042
    },
    {
      "epoch": 0.8749132546842471,
      "grad_norm": 0.6482436060905457,
      "learning_rate": 3.1918261026505974e-05,
      "loss": 0.6279,
      "step": 5043
    },
    {
      "epoch": 0.8750867453157529,
      "grad_norm": 1.018430471420288,
      "learning_rate": 3.1912788956346226e-05,
      "loss": 0.5323,
      "step": 5044
    },
    {
      "epoch": 0.8752602359472589,
      "grad_norm": 0.5560790300369263,
      "learning_rate": 3.1907315503693434e-05,
      "loss": 0.5754,
      "step": 5045
    },
    {
      "epoch": 0.8754337265787647,
      "grad_norm": 1.4722284078598022,
      "learning_rate": 3.190184066918281e-05,
      "loss": 0.748,
      "step": 5046
    },
    {
      "epoch": 0.8756072172102707,
      "grad_norm": 0.856460452079773,
      "learning_rate": 3.18963644534497e-05,
      "loss": 0.5693,
      "step": 5047
    },
    {
      "epoch": 0.8757807078417765,
      "grad_norm": 0.6080543398857117,
      "learning_rate": 3.189088685712964e-05,
      "loss": 0.6155,
      "step": 5048
    },
    {
      "epoch": 0.8759541984732825,
      "grad_norm": 0.8170772194862366,
      "learning_rate": 3.18854078808583e-05,
      "loss": 0.5282,
      "step": 5049
    },
    {
      "epoch": 0.8761276891047883,
      "grad_norm": 0.8213574886322021,
      "learning_rate": 3.187992752527153e-05,
      "loss": 0.5027,
      "step": 5050
    },
    {
      "epoch": 0.8763011797362943,
      "grad_norm": 0.8203826546669006,
      "learning_rate": 3.187444579100533e-05,
      "loss": 0.5299,
      "step": 5051
    },
    {
      "epoch": 0.8764746703678001,
      "grad_norm": 0.7871171236038208,
      "learning_rate": 3.186896267869585e-05,
      "loss": 0.4895,
      "step": 5052
    },
    {
      "epoch": 0.8766481609993061,
      "grad_norm": 0.7084567546844482,
      "learning_rate": 3.1863478188979424e-05,
      "loss": 0.4972,
      "step": 5053
    },
    {
      "epoch": 0.8768216516308119,
      "grad_norm": 1.004349946975708,
      "learning_rate": 3.1857992322492525e-05,
      "loss": 0.567,
      "step": 5054
    },
    {
      "epoch": 0.8769951422623178,
      "grad_norm": 0.5509437322616577,
      "learning_rate": 3.185250507987181e-05,
      "loss": 0.5276,
      "step": 5055
    },
    {
      "epoch": 0.8771686328938237,
      "grad_norm": 0.6013137698173523,
      "learning_rate": 3.184701646175407e-05,
      "loss": 0.5765,
      "step": 5056
    },
    {
      "epoch": 0.8773421235253296,
      "grad_norm": 0.759196937084198,
      "learning_rate": 3.184152646877626e-05,
      "loss": 0.6095,
      "step": 5057
    },
    {
      "epoch": 0.8775156141568355,
      "grad_norm": 0.6360790133476257,
      "learning_rate": 3.183603510157551e-05,
      "loss": 0.6708,
      "step": 5058
    },
    {
      "epoch": 0.8776891047883414,
      "grad_norm": 0.7660230994224548,
      "learning_rate": 3.183054236078909e-05,
      "loss": 0.4877,
      "step": 5059
    },
    {
      "epoch": 0.8778625954198473,
      "grad_norm": 0.7693997025489807,
      "learning_rate": 3.1825048247054444e-05,
      "loss": 0.4686,
      "step": 5060
    },
    {
      "epoch": 0.8780360860513532,
      "grad_norm": 0.8580908179283142,
      "learning_rate": 3.181955276100917e-05,
      "loss": 0.4996,
      "step": 5061
    },
    {
      "epoch": 0.8782095766828591,
      "grad_norm": 0.6530462503433228,
      "learning_rate": 3.181405590329102e-05,
      "loss": 0.5463,
      "step": 5062
    },
    {
      "epoch": 0.878383067314365,
      "grad_norm": 0.7426706552505493,
      "learning_rate": 3.180855767453793e-05,
      "loss": 0.63,
      "step": 5063
    },
    {
      "epoch": 0.878556557945871,
      "grad_norm": 0.7013078927993774,
      "learning_rate": 3.180305807538796e-05,
      "loss": 0.5508,
      "step": 5064
    },
    {
      "epoch": 0.8787300485773768,
      "grad_norm": 0.564922571182251,
      "learning_rate": 3.179755710647934e-05,
      "loss": 0.5793,
      "step": 5065
    },
    {
      "epoch": 0.8789035392088828,
      "grad_norm": 0.8199964761734009,
      "learning_rate": 3.1792054768450466e-05,
      "loss": 0.5695,
      "step": 5066
    },
    {
      "epoch": 0.8790770298403886,
      "grad_norm": 0.8828762173652649,
      "learning_rate": 3.1786551061939905e-05,
      "loss": 0.54,
      "step": 5067
    },
    {
      "epoch": 0.8792505204718946,
      "grad_norm": 0.6814677119255066,
      "learning_rate": 3.178104598758636e-05,
      "loss": 0.7408,
      "step": 5068
    },
    {
      "epoch": 0.8794240111034004,
      "grad_norm": 0.5756251215934753,
      "learning_rate": 3.17755395460287e-05,
      "loss": 0.5538,
      "step": 5069
    },
    {
      "epoch": 0.8795975017349064,
      "grad_norm": 0.7447443008422852,
      "learning_rate": 3.1770031737905946e-05,
      "loss": 0.6128,
      "step": 5070
    },
    {
      "epoch": 0.8797709923664122,
      "grad_norm": 0.7762458324432373,
      "learning_rate": 3.17645225638573e-05,
      "loss": 0.5146,
      "step": 5071
    },
    {
      "epoch": 0.8799444829979182,
      "grad_norm": 0.6583542823791504,
      "learning_rate": 3.1759012024522103e-05,
      "loss": 0.4659,
      "step": 5072
    },
    {
      "epoch": 0.880117973629424,
      "grad_norm": 1.224825143814087,
      "learning_rate": 3.1753500120539856e-05,
      "loss": 0.575,
      "step": 5073
    },
    {
      "epoch": 0.8802914642609299,
      "grad_norm": 1.3071458339691162,
      "learning_rate": 3.1747986852550225e-05,
      "loss": 0.4731,
      "step": 5074
    },
    {
      "epoch": 0.8804649548924358,
      "grad_norm": 0.7071096301078796,
      "learning_rate": 3.174247222119303e-05,
      "loss": 0.4604,
      "step": 5075
    },
    {
      "epoch": 0.8806384455239417,
      "grad_norm": 0.9300988912582397,
      "learning_rate": 3.173695622710826e-05,
      "loss": 0.54,
      "step": 5076
    },
    {
      "epoch": 0.8808119361554476,
      "grad_norm": 0.9893963932991028,
      "learning_rate": 3.173143887093603e-05,
      "loss": 0.6067,
      "step": 5077
    },
    {
      "epoch": 0.8809854267869535,
      "grad_norm": 0.5781874656677246,
      "learning_rate": 3.172592015331666e-05,
      "loss": 0.5546,
      "step": 5078
    },
    {
      "epoch": 0.8811589174184594,
      "grad_norm": 1.6471365690231323,
      "learning_rate": 3.172040007489058e-05,
      "loss": 0.6376,
      "step": 5079
    },
    {
      "epoch": 0.8813324080499653,
      "grad_norm": 0.8680630326271057,
      "learning_rate": 3.171487863629843e-05,
      "loss": 0.4967,
      "step": 5080
    },
    {
      "epoch": 0.8815058986814712,
      "grad_norm": 0.7255361676216125,
      "learning_rate": 3.1709355838180953e-05,
      "loss": 0.6582,
      "step": 5081
    },
    {
      "epoch": 0.8816793893129771,
      "grad_norm": 0.594342827796936,
      "learning_rate": 3.1703831681179096e-05,
      "loss": 0.7122,
      "step": 5082
    },
    {
      "epoch": 0.881852879944483,
      "grad_norm": 0.7831090688705444,
      "learning_rate": 3.169830616593392e-05,
      "loss": 0.5392,
      "step": 5083
    },
    {
      "epoch": 0.8820263705759889,
      "grad_norm": 0.9640688896179199,
      "learning_rate": 3.169277929308669e-05,
      "loss": 0.5665,
      "step": 5084
    },
    {
      "epoch": 0.8821998612074948,
      "grad_norm": 0.7106096744537354,
      "learning_rate": 3.1687251063278806e-05,
      "loss": 0.5957,
      "step": 5085
    },
    {
      "epoch": 0.8823733518390007,
      "grad_norm": 0.7489754557609558,
      "learning_rate": 3.168172147715181e-05,
      "loss": 0.5967,
      "step": 5086
    },
    {
      "epoch": 0.8825468424705066,
      "grad_norm": 0.893450915813446,
      "learning_rate": 3.167619053534742e-05,
      "loss": 0.5494,
      "step": 5087
    },
    {
      "epoch": 0.8827203331020125,
      "grad_norm": 0.6532497406005859,
      "learning_rate": 3.1670658238507524e-05,
      "loss": 0.546,
      "step": 5088
    },
    {
      "epoch": 0.8828938237335184,
      "grad_norm": 0.684929370880127,
      "learning_rate": 3.1665124587274134e-05,
      "loss": 0.5731,
      "step": 5089
    },
    {
      "epoch": 0.8830673143650243,
      "grad_norm": 0.8249834179878235,
      "learning_rate": 3.1659589582289446e-05,
      "loss": 0.5719,
      "step": 5090
    },
    {
      "epoch": 0.8832408049965302,
      "grad_norm": 0.6517502069473267,
      "learning_rate": 3.16540532241958e-05,
      "loss": 0.6447,
      "step": 5091
    },
    {
      "epoch": 0.8834142956280361,
      "grad_norm": 0.6734473705291748,
      "learning_rate": 3.16485155136357e-05,
      "loss": 0.5737,
      "step": 5092
    },
    {
      "epoch": 0.8835877862595419,
      "grad_norm": 0.6809418201446533,
      "learning_rate": 3.164297645125179e-05,
      "loss": 0.5717,
      "step": 5093
    },
    {
      "epoch": 0.8837612768910479,
      "grad_norm": 0.8024628162384033,
      "learning_rate": 3.16374360376869e-05,
      "loss": 0.5487,
      "step": 5094
    },
    {
      "epoch": 0.8839347675225537,
      "grad_norm": 1.124971866607666,
      "learning_rate": 3.1631894273584007e-05,
      "loss": 0.5197,
      "step": 5095
    },
    {
      "epoch": 0.8841082581540597,
      "grad_norm": 0.6369608044624329,
      "learning_rate": 3.1626351159586224e-05,
      "loss": 0.5383,
      "step": 5096
    },
    {
      "epoch": 0.8842817487855655,
      "grad_norm": 1.3009787797927856,
      "learning_rate": 3.1620806696336845e-05,
      "loss": 0.5071,
      "step": 5097
    },
    {
      "epoch": 0.8844552394170715,
      "grad_norm": 0.6499316692352295,
      "learning_rate": 3.1615260884479304e-05,
      "loss": 0.5649,
      "step": 5098
    },
    {
      "epoch": 0.8846287300485773,
      "grad_norm": 0.7309339046478271,
      "learning_rate": 3.16097137246572e-05,
      "loss": 0.6074,
      "step": 5099
    },
    {
      "epoch": 0.8848022206800833,
      "grad_norm": 0.7429511547088623,
      "learning_rate": 3.1604165217514296e-05,
      "loss": 0.5323,
      "step": 5100
    },
    {
      "epoch": 0.8849757113115891,
      "grad_norm": 1.0571507215499878,
      "learning_rate": 3.15986153636945e-05,
      "loss": 0.6587,
      "step": 5101
    },
    {
      "epoch": 0.8851492019430951,
      "grad_norm": 0.7839773893356323,
      "learning_rate": 3.159306416384187e-05,
      "loss": 0.6249,
      "step": 5102
    },
    {
      "epoch": 0.885322692574601,
      "grad_norm": 0.7556633353233337,
      "learning_rate": 3.158751161860063e-05,
      "loss": 0.527,
      "step": 5103
    },
    {
      "epoch": 0.8854961832061069,
      "grad_norm": 0.9725896120071411,
      "learning_rate": 3.158195772861517e-05,
      "loss": 0.5485,
      "step": 5104
    },
    {
      "epoch": 0.8856696738376127,
      "grad_norm": 0.8777176141738892,
      "learning_rate": 3.1576402494530025e-05,
      "loss": 0.5334,
      "step": 5105
    },
    {
      "epoch": 0.8858431644691187,
      "grad_norm": 0.7924165725708008,
      "learning_rate": 3.1570845916989875e-05,
      "loss": 0.5017,
      "step": 5106
    },
    {
      "epoch": 0.8860166551006246,
      "grad_norm": 0.8182539343833923,
      "learning_rate": 3.156528799663957e-05,
      "loss": 0.5979,
      "step": 5107
    },
    {
      "epoch": 0.8861901457321305,
      "grad_norm": 0.946678102016449,
      "learning_rate": 3.1559728734124125e-05,
      "loss": 0.5708,
      "step": 5108
    },
    {
      "epoch": 0.8863636363636364,
      "grad_norm": 1.404183268547058,
      "learning_rate": 3.155416813008869e-05,
      "loss": 0.5416,
      "step": 5109
    },
    {
      "epoch": 0.8865371269951423,
      "grad_norm": 0.7550862431526184,
      "learning_rate": 3.154860618517858e-05,
      "loss": 0.5841,
      "step": 5110
    },
    {
      "epoch": 0.8867106176266482,
      "grad_norm": 0.7476509213447571,
      "learning_rate": 3.154304290003926e-05,
      "loss": 0.5048,
      "step": 5111
    },
    {
      "epoch": 0.8868841082581541,
      "grad_norm": 0.7353691458702087,
      "learning_rate": 3.1537478275316364e-05,
      "loss": 0.5107,
      "step": 5112
    },
    {
      "epoch": 0.88705759888966,
      "grad_norm": 0.8673321008682251,
      "learning_rate": 3.153191231165568e-05,
      "loss": 0.5299,
      "step": 5113
    },
    {
      "epoch": 0.8872310895211658,
      "grad_norm": 0.8236722350120544,
      "learning_rate": 3.152634500970312e-05,
      "loss": 0.5636,
      "step": 5114
    },
    {
      "epoch": 0.8874045801526718,
      "grad_norm": 1.140663504600525,
      "learning_rate": 3.15207763701048e-05,
      "loss": 0.5178,
      "step": 5115
    },
    {
      "epoch": 0.8875780707841776,
      "grad_norm": 0.6625383496284485,
      "learning_rate": 3.151520639350695e-05,
      "loss": 0.5139,
      "step": 5116
    },
    {
      "epoch": 0.8877515614156836,
      "grad_norm": 0.7425408959388733,
      "learning_rate": 3.150963508055599e-05,
      "loss": 0.5006,
      "step": 5117
    },
    {
      "epoch": 0.8879250520471894,
      "grad_norm": 0.6644777059555054,
      "learning_rate": 3.150406243189846e-05,
      "loss": 0.5924,
      "step": 5118
    },
    {
      "epoch": 0.8880985426786954,
      "grad_norm": 0.8685725331306458,
      "learning_rate": 3.1498488448181074e-05,
      "loss": 0.6484,
      "step": 5119
    },
    {
      "epoch": 0.8882720333102012,
      "grad_norm": 0.6809495687484741,
      "learning_rate": 3.1492913130050715e-05,
      "loss": 0.6633,
      "step": 5120
    },
    {
      "epoch": 0.8884455239417072,
      "grad_norm": 0.7087438106536865,
      "learning_rate": 3.14873364781544e-05,
      "loss": 0.6554,
      "step": 5121
    },
    {
      "epoch": 0.888619014573213,
      "grad_norm": 0.7695276737213135,
      "learning_rate": 3.1481758493139295e-05,
      "loss": 0.6027,
      "step": 5122
    },
    {
      "epoch": 0.888792505204719,
      "grad_norm": 0.7297627329826355,
      "learning_rate": 3.147617917565273e-05,
      "loss": 0.6149,
      "step": 5123
    },
    {
      "epoch": 0.8889659958362248,
      "grad_norm": 0.9996668696403503,
      "learning_rate": 3.147059852634221e-05,
      "loss": 0.5967,
      "step": 5124
    },
    {
      "epoch": 0.8891394864677308,
      "grad_norm": 0.8751716613769531,
      "learning_rate": 3.146501654585537e-05,
      "loss": 0.5002,
      "step": 5125
    },
    {
      "epoch": 0.8893129770992366,
      "grad_norm": 0.7023032307624817,
      "learning_rate": 3.145943323483999e-05,
      "loss": 0.5453,
      "step": 5126
    },
    {
      "epoch": 0.8894864677307426,
      "grad_norm": 0.9588670134544373,
      "learning_rate": 3.145384859394403e-05,
      "loss": 0.5043,
      "step": 5127
    },
    {
      "epoch": 0.8896599583622484,
      "grad_norm": 0.957583487033844,
      "learning_rate": 3.144826262381559e-05,
      "loss": 0.4735,
      "step": 5128
    },
    {
      "epoch": 0.8898334489937544,
      "grad_norm": 1.4724335670471191,
      "learning_rate": 3.144267532510295e-05,
      "loss": 0.5228,
      "step": 5129
    },
    {
      "epoch": 0.8900069396252602,
      "grad_norm": 0.8515888452529907,
      "learning_rate": 3.143708669845449e-05,
      "loss": 0.5482,
      "step": 5130
    },
    {
      "epoch": 0.8901804302567662,
      "grad_norm": 0.7285219430923462,
      "learning_rate": 3.1431496744518794e-05,
      "loss": 0.6484,
      "step": 5131
    },
    {
      "epoch": 0.890353920888272,
      "grad_norm": 1.1804567575454712,
      "learning_rate": 3.142590546394459e-05,
      "loss": 0.5201,
      "step": 5132
    },
    {
      "epoch": 0.8905274115197779,
      "grad_norm": 0.6425312161445618,
      "learning_rate": 3.142031285738073e-05,
      "loss": 0.5853,
      "step": 5133
    },
    {
      "epoch": 0.8907009021512838,
      "grad_norm": 0.804465651512146,
      "learning_rate": 3.141471892547627e-05,
      "loss": 0.5792,
      "step": 5134
    },
    {
      "epoch": 0.8908743927827897,
      "grad_norm": 0.670696496963501,
      "learning_rate": 3.140912366888037e-05,
      "loss": 0.5276,
      "step": 5135
    },
    {
      "epoch": 0.8910478834142956,
      "grad_norm": 0.8929148316383362,
      "learning_rate": 3.140352708824237e-05,
      "loss": 0.4873,
      "step": 5136
    },
    {
      "epoch": 0.8912213740458015,
      "grad_norm": 0.965196967124939,
      "learning_rate": 3.139792918421177e-05,
      "loss": 0.4818,
      "step": 5137
    },
    {
      "epoch": 0.8913948646773074,
      "grad_norm": 1.160454273223877,
      "learning_rate": 3.1392329957438203e-05,
      "loss": 0.4968,
      "step": 5138
    },
    {
      "epoch": 0.8915683553088133,
      "grad_norm": 0.9319005608558655,
      "learning_rate": 3.1386729408571467e-05,
      "loss": 0.5962,
      "step": 5139
    },
    {
      "epoch": 0.8917418459403192,
      "grad_norm": 0.6847009658813477,
      "learning_rate": 3.1381127538261524e-05,
      "loss": 0.6313,
      "step": 5140
    },
    {
      "epoch": 0.8919153365718251,
      "grad_norm": 0.7343131303787231,
      "learning_rate": 3.137552434715846e-05,
      "loss": 0.5214,
      "step": 5141
    },
    {
      "epoch": 0.892088827203331,
      "grad_norm": 1.1229071617126465,
      "learning_rate": 3.136991983591255e-05,
      "loss": 0.5236,
      "step": 5142
    },
    {
      "epoch": 0.8922623178348369,
      "grad_norm": 0.7773802280426025,
      "learning_rate": 3.1364314005174184e-05,
      "loss": 0.6536,
      "step": 5143
    },
    {
      "epoch": 0.8924358084663429,
      "grad_norm": 0.7838901281356812,
      "learning_rate": 3.1358706855593935e-05,
      "loss": 0.5215,
      "step": 5144
    },
    {
      "epoch": 0.8926092990978487,
      "grad_norm": 0.7665859460830688,
      "learning_rate": 3.135309838782253e-05,
      "loss": 0.6309,
      "step": 5145
    },
    {
      "epoch": 0.8927827897293547,
      "grad_norm": 0.8027396202087402,
      "learning_rate": 3.1347488602510824e-05,
      "loss": 0.614,
      "step": 5146
    },
    {
      "epoch": 0.8929562803608605,
      "grad_norm": 0.8135119676589966,
      "learning_rate": 3.134187750030984e-05,
      "loss": 0.6162,
      "step": 5147
    },
    {
      "epoch": 0.8931297709923665,
      "grad_norm": 0.7477341890335083,
      "learning_rate": 3.133626508187076e-05,
      "loss": 0.5168,
      "step": 5148
    },
    {
      "epoch": 0.8933032616238723,
      "grad_norm": 1.039930820465088,
      "learning_rate": 3.133065134784491e-05,
      "loss": 0.4841,
      "step": 5149
    },
    {
      "epoch": 0.8934767522553783,
      "grad_norm": 0.8588441610336304,
      "learning_rate": 3.132503629888376e-05,
      "loss": 0.6981,
      "step": 5150
    },
    {
      "epoch": 0.8936502428868841,
      "grad_norm": 0.6859707236289978,
      "learning_rate": 3.131941993563896e-05,
      "loss": 0.6285,
      "step": 5151
    },
    {
      "epoch": 0.8938237335183901,
      "grad_norm": 1.0702961683273315,
      "learning_rate": 3.131380225876228e-05,
      "loss": 0.4751,
      "step": 5152
    },
    {
      "epoch": 0.8939972241498959,
      "grad_norm": 0.7685006260871887,
      "learning_rate": 3.130818326890567e-05,
      "loss": 0.5216,
      "step": 5153
    },
    {
      "epoch": 0.8941707147814018,
      "grad_norm": 1.5758066177368164,
      "learning_rate": 3.130256296672121e-05,
      "loss": 0.6899,
      "step": 5154
    },
    {
      "epoch": 0.8943442054129077,
      "grad_norm": 1.3756191730499268,
      "learning_rate": 3.1296941352861146e-05,
      "loss": 0.4948,
      "step": 5155
    },
    {
      "epoch": 0.8945176960444136,
      "grad_norm": 0.7226235270500183,
      "learning_rate": 3.1291318427977876e-05,
      "loss": 0.5996,
      "step": 5156
    },
    {
      "epoch": 0.8946911866759195,
      "grad_norm": 0.7024737596511841,
      "learning_rate": 3.128569419272395e-05,
      "loss": 0.5797,
      "step": 5157
    },
    {
      "epoch": 0.8948646773074254,
      "grad_norm": 0.9813096523284912,
      "learning_rate": 3.1280068647752066e-05,
      "loss": 0.5109,
      "step": 5158
    },
    {
      "epoch": 0.8950381679389313,
      "grad_norm": 0.7015733122825623,
      "learning_rate": 3.127444179371506e-05,
      "loss": 0.4841,
      "step": 5159
    },
    {
      "epoch": 0.8952116585704372,
      "grad_norm": 0.7428815960884094,
      "learning_rate": 3.126881363126595e-05,
      "loss": 0.6926,
      "step": 5160
    },
    {
      "epoch": 0.8953851492019431,
      "grad_norm": 0.5810912847518921,
      "learning_rate": 3.126318416105789e-05,
      "loss": 0.6013,
      "step": 5161
    },
    {
      "epoch": 0.895558639833449,
      "grad_norm": 0.7125701904296875,
      "learning_rate": 3.1257553383744186e-05,
      "loss": 0.5754,
      "step": 5162
    },
    {
      "epoch": 0.8957321304649549,
      "grad_norm": 0.6725031733512878,
      "learning_rate": 3.125192129997829e-05,
      "loss": 0.5127,
      "step": 5163
    },
    {
      "epoch": 0.8959056210964608,
      "grad_norm": 0.626672089099884,
      "learning_rate": 3.1246287910413824e-05,
      "loss": 0.6064,
      "step": 5164
    },
    {
      "epoch": 0.8960791117279667,
      "grad_norm": 0.8559987545013428,
      "learning_rate": 3.124065321570453e-05,
      "loss": 0.569,
      "step": 5165
    },
    {
      "epoch": 0.8962526023594726,
      "grad_norm": 0.7856817841529846,
      "learning_rate": 3.123501721650434e-05,
      "loss": 0.5181,
      "step": 5166
    },
    {
      "epoch": 0.8964260929909785,
      "grad_norm": 1.0046130418777466,
      "learning_rate": 3.1229379913467305e-05,
      "loss": 0.5107,
      "step": 5167
    },
    {
      "epoch": 0.8965995836224844,
      "grad_norm": 0.7973633408546448,
      "learning_rate": 3.122374130724765e-05,
      "loss": 0.5603,
      "step": 5168
    },
    {
      "epoch": 0.8967730742539903,
      "grad_norm": 1.8391568660736084,
      "learning_rate": 3.1218101398499736e-05,
      "loss": 0.522,
      "step": 5169
    },
    {
      "epoch": 0.8969465648854962,
      "grad_norm": 1.0152584314346313,
      "learning_rate": 3.1212460187878085e-05,
      "loss": 0.5153,
      "step": 5170
    },
    {
      "epoch": 0.8971200555170021,
      "grad_norm": 0.8015640377998352,
      "learning_rate": 3.1206817676037365e-05,
      "loss": 0.5974,
      "step": 5171
    },
    {
      "epoch": 0.897293546148508,
      "grad_norm": 0.6515125036239624,
      "learning_rate": 3.1201173863632396e-05,
      "loss": 0.6613,
      "step": 5172
    },
    {
      "epoch": 0.8974670367800138,
      "grad_norm": 0.8756774067878723,
      "learning_rate": 3.119552875131814e-05,
      "loss": 0.5544,
      "step": 5173
    },
    {
      "epoch": 0.8976405274115198,
      "grad_norm": 1.1011831760406494,
      "learning_rate": 3.1189882339749735e-05,
      "loss": 0.5172,
      "step": 5174
    },
    {
      "epoch": 0.8978140180430256,
      "grad_norm": 0.6310820579528809,
      "learning_rate": 3.1184234629582444e-05,
      "loss": 0.5692,
      "step": 5175
    },
    {
      "epoch": 0.8979875086745316,
      "grad_norm": 0.8873246312141418,
      "learning_rate": 3.117858562147169e-05,
      "loss": 0.5406,
      "step": 5176
    },
    {
      "epoch": 0.8981609993060374,
      "grad_norm": 0.6612831950187683,
      "learning_rate": 3.117293531607305e-05,
      "loss": 0.5819,
      "step": 5177
    },
    {
      "epoch": 0.8983344899375434,
      "grad_norm": 0.8151834011077881,
      "learning_rate": 3.116728371404225e-05,
      "loss": 0.5066,
      "step": 5178
    },
    {
      "epoch": 0.8985079805690492,
      "grad_norm": 0.9618954062461853,
      "learning_rate": 3.116163081603516e-05,
      "loss": 0.5043,
      "step": 5179
    },
    {
      "epoch": 0.8986814712005552,
      "grad_norm": 0.8373923301696777,
      "learning_rate": 3.115597662270781e-05,
      "loss": 0.5439,
      "step": 5180
    },
    {
      "epoch": 0.898854961832061,
      "grad_norm": 0.8901704549789429,
      "learning_rate": 3.115032113471637e-05,
      "loss": 0.6293,
      "step": 5181
    },
    {
      "epoch": 0.899028452463567,
      "grad_norm": 1.0294442176818848,
      "learning_rate": 3.114466435271717e-05,
      "loss": 0.6533,
      "step": 5182
    },
    {
      "epoch": 0.8992019430950728,
      "grad_norm": 0.6956663727760315,
      "learning_rate": 3.113900627736669e-05,
      "loss": 0.6963,
      "step": 5183
    },
    {
      "epoch": 0.8993754337265788,
      "grad_norm": 0.9930188655853271,
      "learning_rate": 3.113334690932155e-05,
      "loss": 0.5216,
      "step": 5184
    },
    {
      "epoch": 0.8995489243580846,
      "grad_norm": 1.244781494140625,
      "learning_rate": 3.112768624923853e-05,
      "loss": 0.5474,
      "step": 5185
    },
    {
      "epoch": 0.8997224149895906,
      "grad_norm": 0.908898651599884,
      "learning_rate": 3.1122024297774545e-05,
      "loss": 0.4548,
      "step": 5186
    },
    {
      "epoch": 0.8998959056210964,
      "grad_norm": 0.6766604781150818,
      "learning_rate": 3.111636105558669e-05,
      "loss": 0.6635,
      "step": 5187
    },
    {
      "epoch": 0.9000693962526024,
      "grad_norm": 0.8646974563598633,
      "learning_rate": 3.111069652333219e-05,
      "loss": 0.6058,
      "step": 5188
    },
    {
      "epoch": 0.9002428868841083,
      "grad_norm": 0.6690236926078796,
      "learning_rate": 3.1105030701668395e-05,
      "loss": 0.5758,
      "step": 5189
    },
    {
      "epoch": 0.9004163775156142,
      "grad_norm": 1.2098963260650635,
      "learning_rate": 3.1099363591252844e-05,
      "loss": 0.4812,
      "step": 5190
    },
    {
      "epoch": 0.90058986814712,
      "grad_norm": 0.7784439325332642,
      "learning_rate": 3.109369519274323e-05,
      "loss": 0.4951,
      "step": 5191
    },
    {
      "epoch": 0.9007633587786259,
      "grad_norm": 0.55345219373703,
      "learning_rate": 3.1088025506797356e-05,
      "loss": 0.6191,
      "step": 5192
    },
    {
      "epoch": 0.9009368494101319,
      "grad_norm": 0.8035297989845276,
      "learning_rate": 3.1082354534073206e-05,
      "loss": 0.5957,
      "step": 5193
    },
    {
      "epoch": 0.9011103400416377,
      "grad_norm": 0.7231811881065369,
      "learning_rate": 3.107668227522889e-05,
      "loss": 0.5792,
      "step": 5194
    },
    {
      "epoch": 0.9012838306731437,
      "grad_norm": 1.1240482330322266,
      "learning_rate": 3.107100873092269e-05,
      "loss": 0.5188,
      "step": 5195
    },
    {
      "epoch": 0.9014573213046495,
      "grad_norm": 0.7079700231552124,
      "learning_rate": 3.106533390181304e-05,
      "loss": 0.6602,
      "step": 5196
    },
    {
      "epoch": 0.9016308119361555,
      "grad_norm": 0.772443950176239,
      "learning_rate": 3.105965778855848e-05,
      "loss": 0.5796,
      "step": 5197
    },
    {
      "epoch": 0.9018043025676613,
      "grad_norm": 0.6135613322257996,
      "learning_rate": 3.105398039181775e-05,
      "loss": 0.6349,
      "step": 5198
    },
    {
      "epoch": 0.9019777931991673,
      "grad_norm": 0.7034465670585632,
      "learning_rate": 3.1048301712249726e-05,
      "loss": 0.7047,
      "step": 5199
    },
    {
      "epoch": 0.9021512838306731,
      "grad_norm": 0.9171649813652039,
      "learning_rate": 3.1042621750513405e-05,
      "loss": 0.4851,
      "step": 5200
    },
    {
      "epoch": 0.9023247744621791,
      "grad_norm": 0.836614727973938,
      "learning_rate": 3.103694050726797e-05,
      "loss": 0.462,
      "step": 5201
    },
    {
      "epoch": 0.9024982650936849,
      "grad_norm": 0.726462721824646,
      "learning_rate": 3.103125798317272e-05,
      "loss": 0.4623,
      "step": 5202
    },
    {
      "epoch": 0.9026717557251909,
      "grad_norm": 0.9646469950675964,
      "learning_rate": 3.102557417888713e-05,
      "loss": 0.5901,
      "step": 5203
    },
    {
      "epoch": 0.9028452463566967,
      "grad_norm": 0.6644760966300964,
      "learning_rate": 3.101988909507081e-05,
      "loss": 0.5316,
      "step": 5204
    },
    {
      "epoch": 0.9030187369882027,
      "grad_norm": 0.7346537113189697,
      "learning_rate": 3.1014202732383525e-05,
      "loss": 0.6246,
      "step": 5205
    },
    {
      "epoch": 0.9031922276197085,
      "grad_norm": 0.6510661840438843,
      "learning_rate": 3.100851509148517e-05,
      "loss": 0.5361,
      "step": 5206
    },
    {
      "epoch": 0.9033657182512145,
      "grad_norm": 1.276292324066162,
      "learning_rate": 3.100282617303581e-05,
      "loss": 0.6156,
      "step": 5207
    },
    {
      "epoch": 0.9035392088827203,
      "grad_norm": 0.8766646385192871,
      "learning_rate": 3.099713597769566e-05,
      "loss": 0.51,
      "step": 5208
    },
    {
      "epoch": 0.9037126995142263,
      "grad_norm": 0.7327955365180969,
      "learning_rate": 3.0991444506125066e-05,
      "loss": 0.5787,
      "step": 5209
    },
    {
      "epoch": 0.9038861901457321,
      "grad_norm": 0.9742059111595154,
      "learning_rate": 3.098575175898452e-05,
      "loss": 0.5641,
      "step": 5210
    },
    {
      "epoch": 0.9040596807772381,
      "grad_norm": 0.7599835991859436,
      "learning_rate": 3.098005773693469e-05,
      "loss": 0.5162,
      "step": 5211
    },
    {
      "epoch": 0.9042331714087439,
      "grad_norm": 0.8517898321151733,
      "learning_rate": 3.097436244063636e-05,
      "loss": 0.6124,
      "step": 5212
    },
    {
      "epoch": 0.9044066620402498,
      "grad_norm": 0.7128757834434509,
      "learning_rate": 3.0968665870750484e-05,
      "loss": 0.4829,
      "step": 5213
    },
    {
      "epoch": 0.9045801526717557,
      "grad_norm": 0.6965377330780029,
      "learning_rate": 3.0962968027938156e-05,
      "loss": 0.5793,
      "step": 5214
    },
    {
      "epoch": 0.9047536433032616,
      "grad_norm": 0.9639315009117126,
      "learning_rate": 3.09572689128606e-05,
      "loss": 0.5276,
      "step": 5215
    },
    {
      "epoch": 0.9049271339347675,
      "grad_norm": 0.8930005431175232,
      "learning_rate": 3.0951568526179235e-05,
      "loss": 0.5297,
      "step": 5216
    },
    {
      "epoch": 0.9051006245662734,
      "grad_norm": 0.8374587893486023,
      "learning_rate": 3.094586686855558e-05,
      "loss": 0.64,
      "step": 5217
    },
    {
      "epoch": 0.9052741151977793,
      "grad_norm": 0.5651605725288391,
      "learning_rate": 3.094016394065131e-05,
      "loss": 0.5555,
      "step": 5218
    },
    {
      "epoch": 0.9054476058292852,
      "grad_norm": 0.7753821611404419,
      "learning_rate": 3.093445974312828e-05,
      "loss": 0.5393,
      "step": 5219
    },
    {
      "epoch": 0.9056210964607911,
      "grad_norm": 1.0455563068389893,
      "learning_rate": 3.0928754276648443e-05,
      "loss": 0.5767,
      "step": 5220
    },
    {
      "epoch": 0.905794587092297,
      "grad_norm": 0.8705766201019287,
      "learning_rate": 3.092304754187394e-05,
      "loss": 0.6367,
      "step": 5221
    },
    {
      "epoch": 0.905968077723803,
      "grad_norm": 0.8444671630859375,
      "learning_rate": 3.091733953946705e-05,
      "loss": 0.4584,
      "step": 5222
    },
    {
      "epoch": 0.9061415683553088,
      "grad_norm": 0.722236156463623,
      "learning_rate": 3.0911630270090177e-05,
      "loss": 0.5236,
      "step": 5223
    },
    {
      "epoch": 0.9063150589868147,
      "grad_norm": 0.7913243770599365,
      "learning_rate": 3.09059197344059e-05,
      "loss": 0.541,
      "step": 5224
    },
    {
      "epoch": 0.9064885496183206,
      "grad_norm": 0.8392624258995056,
      "learning_rate": 3.090020793307693e-05,
      "loss": 0.6556,
      "step": 5225
    },
    {
      "epoch": 0.9066620402498266,
      "grad_norm": 0.7073938846588135,
      "learning_rate": 3.089449486676613e-05,
      "loss": 0.5879,
      "step": 5226
    },
    {
      "epoch": 0.9068355308813324,
      "grad_norm": 0.5899890661239624,
      "learning_rate": 3.08887805361365e-05,
      "loss": 0.6183,
      "step": 5227
    },
    {
      "epoch": 0.9070090215128384,
      "grad_norm": 0.8052083849906921,
      "learning_rate": 3.08830649418512e-05,
      "loss": 0.5619,
      "step": 5228
    },
    {
      "epoch": 0.9071825121443442,
      "grad_norm": 0.8192055225372314,
      "learning_rate": 3.087734808457354e-05,
      "loss": 0.5341,
      "step": 5229
    },
    {
      "epoch": 0.9073560027758502,
      "grad_norm": 0.5747373700141907,
      "learning_rate": 3.087162996496696e-05,
      "loss": 0.6653,
      "step": 5230
    },
    {
      "epoch": 0.907529493407356,
      "grad_norm": 0.9766119122505188,
      "learning_rate": 3.086591058369505e-05,
      "loss": 0.5201,
      "step": 5231
    },
    {
      "epoch": 0.9077029840388618,
      "grad_norm": 0.8069945573806763,
      "learning_rate": 3.086018994142156e-05,
      "loss": 0.5443,
      "step": 5232
    },
    {
      "epoch": 0.9078764746703678,
      "grad_norm": 0.829116940498352,
      "learning_rate": 3.0854468038810365e-05,
      "loss": 0.4589,
      "step": 5233
    },
    {
      "epoch": 0.9080499653018737,
      "grad_norm": 0.6612994074821472,
      "learning_rate": 3.084874487652551e-05,
      "loss": 0.5837,
      "step": 5234
    },
    {
      "epoch": 0.9082234559333796,
      "grad_norm": 0.7349043488502502,
      "learning_rate": 3.0843020455231173e-05,
      "loss": 0.5742,
      "step": 5235
    },
    {
      "epoch": 0.9083969465648855,
      "grad_norm": 0.6258236765861511,
      "learning_rate": 3.0837294775591675e-05,
      "loss": 0.5515,
      "step": 5236
    },
    {
      "epoch": 0.9085704371963914,
      "grad_norm": 0.6086591482162476,
      "learning_rate": 3.08315678382715e-05,
      "loss": 0.5005,
      "step": 5237
    },
    {
      "epoch": 0.9087439278278973,
      "grad_norm": 0.6610521078109741,
      "learning_rate": 3.082583964393524e-05,
      "loss": 0.5731,
      "step": 5238
    },
    {
      "epoch": 0.9089174184594032,
      "grad_norm": 0.6950920820236206,
      "learning_rate": 3.082011019324768e-05,
      "loss": 0.6082,
      "step": 5239
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.6199288964271545,
      "learning_rate": 3.081437948687373e-05,
      "loss": 0.5906,
      "step": 5240
    },
    {
      "epoch": 0.909264399722415,
      "grad_norm": 0.6856377124786377,
      "learning_rate": 3.0808647525478434e-05,
      "loss": 0.502,
      "step": 5241
    },
    {
      "epoch": 0.9094378903539209,
      "grad_norm": 0.829742431640625,
      "learning_rate": 3.0802914309727004e-05,
      "loss": 0.6536,
      "step": 5242
    },
    {
      "epoch": 0.9096113809854268,
      "grad_norm": 0.7659375667572021,
      "learning_rate": 3.079717984028478e-05,
      "loss": 0.5575,
      "step": 5243
    },
    {
      "epoch": 0.9097848716169327,
      "grad_norm": 0.6697655916213989,
      "learning_rate": 3.0791444117817247e-05,
      "loss": 0.6537,
      "step": 5244
    },
    {
      "epoch": 0.9099583622484386,
      "grad_norm": 0.6022835373878479,
      "learning_rate": 3.078570714299005e-05,
      "loss": 0.5839,
      "step": 5245
    },
    {
      "epoch": 0.9101318528799445,
      "grad_norm": 0.6736202239990234,
      "learning_rate": 3.0779968916468974e-05,
      "loss": 0.6902,
      "step": 5246
    },
    {
      "epoch": 0.9103053435114504,
      "grad_norm": 0.8430968523025513,
      "learning_rate": 3.0774229438919944e-05,
      "loss": 0.5851,
      "step": 5247
    },
    {
      "epoch": 0.9104788341429563,
      "grad_norm": 1.3374613523483276,
      "learning_rate": 3.076848871100904e-05,
      "loss": 0.5179,
      "step": 5248
    },
    {
      "epoch": 0.9106523247744622,
      "grad_norm": 0.73630291223526,
      "learning_rate": 3.0762746733402456e-05,
      "loss": 0.561,
      "step": 5249
    },
    {
      "epoch": 0.9108258154059681,
      "grad_norm": 0.6396110653877258,
      "learning_rate": 3.075700350676659e-05,
      "loss": 0.5751,
      "step": 5250
    },
    {
      "epoch": 0.9109993060374739,
      "grad_norm": 0.7337170839309692,
      "learning_rate": 3.075125903176792e-05,
      "loss": 0.6136,
      "step": 5251
    },
    {
      "epoch": 0.9111727966689799,
      "grad_norm": 1.0147831439971924,
      "learning_rate": 3.074551330907312e-05,
      "loss": 0.5428,
      "step": 5252
    },
    {
      "epoch": 0.9113462873004857,
      "grad_norm": 0.7059173583984375,
      "learning_rate": 3.073976633934898e-05,
      "loss": 0.6399,
      "step": 5253
    },
    {
      "epoch": 0.9115197779319917,
      "grad_norm": 0.7515703439712524,
      "learning_rate": 3.073401812326244e-05,
      "loss": 0.6021,
      "step": 5254
    },
    {
      "epoch": 0.9116932685634975,
      "grad_norm": 0.9002457857131958,
      "learning_rate": 3.072826866148058e-05,
      "loss": 0.5867,
      "step": 5255
    },
    {
      "epoch": 0.9118667591950035,
      "grad_norm": 0.7314658164978027,
      "learning_rate": 3.072251795467065e-05,
      "loss": 0.516,
      "step": 5256
    },
    {
      "epoch": 0.9120402498265093,
      "grad_norm": 0.9362332224845886,
      "learning_rate": 3.071676600350002e-05,
      "loss": 0.52,
      "step": 5257
    },
    {
      "epoch": 0.9122137404580153,
      "grad_norm": 0.8564382195472717,
      "learning_rate": 3.071101280863621e-05,
      "loss": 0.5596,
      "step": 5258
    },
    {
      "epoch": 0.9123872310895211,
      "grad_norm": 0.974568784236908,
      "learning_rate": 3.0705258370746874e-05,
      "loss": 0.5802,
      "step": 5259
    },
    {
      "epoch": 0.9125607217210271,
      "grad_norm": 0.6499694585800171,
      "learning_rate": 3.069950269049983e-05,
      "loss": 0.5847,
      "step": 5260
    },
    {
      "epoch": 0.9127342123525329,
      "grad_norm": 0.7674784660339355,
      "learning_rate": 3.069374576856304e-05,
      "loss": 0.4797,
      "step": 5261
    },
    {
      "epoch": 0.9129077029840389,
      "grad_norm": 0.6690618991851807,
      "learning_rate": 3.068798760560458e-05,
      "loss": 0.5447,
      "step": 5262
    },
    {
      "epoch": 0.9130811936155447,
      "grad_norm": 0.6676790118217468,
      "learning_rate": 3.068222820229272e-05,
      "loss": 0.5741,
      "step": 5263
    },
    {
      "epoch": 0.9132546842470507,
      "grad_norm": 0.7896708250045776,
      "learning_rate": 3.067646755929582e-05,
      "loss": 0.488,
      "step": 5264
    },
    {
      "epoch": 0.9134281748785565,
      "grad_norm": 0.7204405665397644,
      "learning_rate": 3.067070567728242e-05,
      "loss": 0.563,
      "step": 5265
    },
    {
      "epoch": 0.9136016655100625,
      "grad_norm": 0.7507115006446838,
      "learning_rate": 3.066494255692119e-05,
      "loss": 0.5887,
      "step": 5266
    },
    {
      "epoch": 0.9137751561415683,
      "grad_norm": 0.6853576302528381,
      "learning_rate": 3.065917819888095e-05,
      "loss": 0.6367,
      "step": 5267
    },
    {
      "epoch": 0.9139486467730743,
      "grad_norm": 0.6107186675071716,
      "learning_rate": 3.0653412603830665e-05,
      "loss": 0.6069,
      "step": 5268
    },
    {
      "epoch": 0.9141221374045801,
      "grad_norm": 1.0904207229614258,
      "learning_rate": 3.064764577243943e-05,
      "loss": 0.5908,
      "step": 5269
    },
    {
      "epoch": 0.9142956280360861,
      "grad_norm": 1.1321725845336914,
      "learning_rate": 3.06418777053765e-05,
      "loss": 0.5096,
      "step": 5270
    },
    {
      "epoch": 0.914469118667592,
      "grad_norm": 0.7792344093322754,
      "learning_rate": 3.063610840331125e-05,
      "loss": 0.5742,
      "step": 5271
    },
    {
      "epoch": 0.9146426092990978,
      "grad_norm": 0.6540992259979248,
      "learning_rate": 3.0630337866913236e-05,
      "loss": 0.665,
      "step": 5272
    },
    {
      "epoch": 0.9148160999306038,
      "grad_norm": 0.7271228432655334,
      "learning_rate": 3.0624566096852124e-05,
      "loss": 0.4648,
      "step": 5273
    },
    {
      "epoch": 0.9149895905621096,
      "grad_norm": 0.6793175339698792,
      "learning_rate": 3.061879309379774e-05,
      "loss": 0.5343,
      "step": 5274
    },
    {
      "epoch": 0.9151630811936156,
      "grad_norm": 0.8328024744987488,
      "learning_rate": 3.061301885842004e-05,
      "loss": 0.5597,
      "step": 5275
    },
    {
      "epoch": 0.9153365718251214,
      "grad_norm": 1.870966911315918,
      "learning_rate": 3.060724339138913e-05,
      "loss": 0.5872,
      "step": 5276
    },
    {
      "epoch": 0.9155100624566274,
      "grad_norm": 0.8721336722373962,
      "learning_rate": 3.060146669337528e-05,
      "loss": 0.601,
      "step": 5277
    },
    {
      "epoch": 0.9156835530881332,
      "grad_norm": 0.8151071667671204,
      "learning_rate": 3.0595688765048855e-05,
      "loss": 0.5375,
      "step": 5278
    },
    {
      "epoch": 0.9158570437196392,
      "grad_norm": 0.8064154982566833,
      "learning_rate": 3.05899096070804e-05,
      "loss": 0.6492,
      "step": 5279
    },
    {
      "epoch": 0.916030534351145,
      "grad_norm": 0.7032491564750671,
      "learning_rate": 3.058412922014061e-05,
      "loss": 0.6439,
      "step": 5280
    },
    {
      "epoch": 0.916204024982651,
      "grad_norm": 0.7188166379928589,
      "learning_rate": 3.057834760490027e-05,
      "loss": 0.59,
      "step": 5281
    },
    {
      "epoch": 0.9163775156141568,
      "grad_norm": 0.8670480847358704,
      "learning_rate": 3.057256476203038e-05,
      "loss": 0.5725,
      "step": 5282
    },
    {
      "epoch": 0.9165510062456628,
      "grad_norm": 0.822883129119873,
      "learning_rate": 3.056678069220203e-05,
      "loss": 0.686,
      "step": 5283
    },
    {
      "epoch": 0.9167244968771686,
      "grad_norm": 0.8245817422866821,
      "learning_rate": 3.056099539608646e-05,
      "loss": 0.6072,
      "step": 5284
    },
    {
      "epoch": 0.9168979875086746,
      "grad_norm": 0.7782118916511536,
      "learning_rate": 3.055520887435507e-05,
      "loss": 0.5344,
      "step": 5285
    },
    {
      "epoch": 0.9170714781401804,
      "grad_norm": 0.7215374112129211,
      "learning_rate": 3.0549421127679395e-05,
      "loss": 0.5449,
      "step": 5286
    },
    {
      "epoch": 0.9172449687716864,
      "grad_norm": 0.8895929455757141,
      "learning_rate": 3.0543632156731105e-05,
      "loss": 0.5015,
      "step": 5287
    },
    {
      "epoch": 0.9174184594031922,
      "grad_norm": 0.8760278820991516,
      "learning_rate": 3.053784196218201e-05,
      "loss": 0.5188,
      "step": 5288
    },
    {
      "epoch": 0.9175919500346982,
      "grad_norm": 0.7011651992797852,
      "learning_rate": 3.053205054470408e-05,
      "loss": 0.6781,
      "step": 5289
    },
    {
      "epoch": 0.917765440666204,
      "grad_norm": 0.9101936221122742,
      "learning_rate": 3.052625790496942e-05,
      "loss": 0.5417,
      "step": 5290
    },
    {
      "epoch": 0.9179389312977099,
      "grad_norm": 0.796743631362915,
      "learning_rate": 3.052046404365025e-05,
      "loss": 0.5387,
      "step": 5291
    },
    {
      "epoch": 0.9181124219292158,
      "grad_norm": 0.8621236085891724,
      "learning_rate": 3.0514668961418984e-05,
      "loss": 0.6255,
      "step": 5292
    },
    {
      "epoch": 0.9182859125607217,
      "grad_norm": 0.7842460870742798,
      "learning_rate": 3.0508872658948125e-05,
      "loss": 0.5413,
      "step": 5293
    },
    {
      "epoch": 0.9184594031922276,
      "grad_norm": 0.7441717386245728,
      "learning_rate": 3.050307513691035e-05,
      "loss": 0.5526,
      "step": 5294
    },
    {
      "epoch": 0.9186328938237335,
      "grad_norm": 0.6309664249420166,
      "learning_rate": 3.0497276395978468e-05,
      "loss": 0.6365,
      "step": 5295
    },
    {
      "epoch": 0.9188063844552394,
      "grad_norm": 0.7494913935661316,
      "learning_rate": 3.0491476436825427e-05,
      "loss": 0.5692,
      "step": 5296
    },
    {
      "epoch": 0.9189798750867453,
      "grad_norm": 0.650999128818512,
      "learning_rate": 3.048567526012432e-05,
      "loss": 0.6404,
      "step": 5297
    },
    {
      "epoch": 0.9191533657182512,
      "grad_norm": 0.604469358921051,
      "learning_rate": 3.047987286654838e-05,
      "loss": 0.6122,
      "step": 5298
    },
    {
      "epoch": 0.9193268563497571,
      "grad_norm": 0.5945268273353577,
      "learning_rate": 3.0474069256770983e-05,
      "loss": 0.6539,
      "step": 5299
    },
    {
      "epoch": 0.919500346981263,
      "grad_norm": 0.5838721394538879,
      "learning_rate": 3.0468264431465643e-05,
      "loss": 0.7271,
      "step": 5300
    },
    {
      "epoch": 0.9196738376127689,
      "grad_norm": 0.8042100667953491,
      "learning_rate": 3.0462458391306023e-05,
      "loss": 0.5717,
      "step": 5301
    },
    {
      "epoch": 0.9198473282442748,
      "grad_norm": 1.0784622430801392,
      "learning_rate": 3.045665113696591e-05,
      "loss": 0.4805,
      "step": 5302
    },
    {
      "epoch": 0.9200208188757807,
      "grad_norm": 0.7842101454734802,
      "learning_rate": 3.0450842669119255e-05,
      "loss": 0.5531,
      "step": 5303
    },
    {
      "epoch": 0.9201943095072866,
      "grad_norm": 0.760572612285614,
      "learning_rate": 3.0445032988440126e-05,
      "loss": 0.5067,
      "step": 5304
    },
    {
      "epoch": 0.9203678001387925,
      "grad_norm": 0.7464740872383118,
      "learning_rate": 3.0439222095602744e-05,
      "loss": 0.5059,
      "step": 5305
    },
    {
      "epoch": 0.9205412907702984,
      "grad_norm": 1.0330287218093872,
      "learning_rate": 3.0433409991281483e-05,
      "loss": 0.4354,
      "step": 5306
    },
    {
      "epoch": 0.9207147814018043,
      "grad_norm": 1.0916588306427002,
      "learning_rate": 3.042759667615083e-05,
      "loss": 0.5347,
      "step": 5307
    },
    {
      "epoch": 0.9208882720333103,
      "grad_norm": 0.7047819495201111,
      "learning_rate": 3.042178215088543e-05,
      "loss": 0.6464,
      "step": 5308
    },
    {
      "epoch": 0.9210617626648161,
      "grad_norm": 1.1781091690063477,
      "learning_rate": 3.041596641616007e-05,
      "loss": 0.4905,
      "step": 5309
    },
    {
      "epoch": 0.9212352532963219,
      "grad_norm": 0.8421317338943481,
      "learning_rate": 3.041014947264967e-05,
      "loss": 0.6125,
      "step": 5310
    },
    {
      "epoch": 0.9214087439278279,
      "grad_norm": 0.7191337943077087,
      "learning_rate": 3.0404331321029293e-05,
      "loss": 0.6671,
      "step": 5311
    },
    {
      "epoch": 0.9215822345593337,
      "grad_norm": 0.6631479859352112,
      "learning_rate": 3.0398511961974143e-05,
      "loss": 0.5836,
      "step": 5312
    },
    {
      "epoch": 0.9217557251908397,
      "grad_norm": 0.5192256569862366,
      "learning_rate": 3.0392691396159562e-05,
      "loss": 0.6151,
      "step": 5313
    },
    {
      "epoch": 0.9219292158223455,
      "grad_norm": 0.9425125122070312,
      "learning_rate": 3.0386869624261036e-05,
      "loss": 0.495,
      "step": 5314
    },
    {
      "epoch": 0.9221027064538515,
      "grad_norm": 0.7365497946739197,
      "learning_rate": 3.0381046646954185e-05,
      "loss": 0.5093,
      "step": 5315
    },
    {
      "epoch": 0.9222761970853574,
      "grad_norm": 0.7596287131309509,
      "learning_rate": 3.0375222464914782e-05,
      "loss": 0.4493,
      "step": 5316
    },
    {
      "epoch": 0.9224496877168633,
      "grad_norm": 0.7143353223800659,
      "learning_rate": 3.036939707881871e-05,
      "loss": 0.5822,
      "step": 5317
    },
    {
      "epoch": 0.9226231783483692,
      "grad_norm": 0.9538487195968628,
      "learning_rate": 3.0363570489342033e-05,
      "loss": 0.5338,
      "step": 5318
    },
    {
      "epoch": 0.9227966689798751,
      "grad_norm": 0.79997718334198,
      "learning_rate": 3.0357742697160924e-05,
      "loss": 0.6605,
      "step": 5319
    },
    {
      "epoch": 0.922970159611381,
      "grad_norm": 0.7166290283203125,
      "learning_rate": 3.03519137029517e-05,
      "loss": 0.5365,
      "step": 5320
    },
    {
      "epoch": 0.9231436502428869,
      "grad_norm": 0.6760879158973694,
      "learning_rate": 3.034608350739084e-05,
      "loss": 0.5872,
      "step": 5321
    },
    {
      "epoch": 0.9233171408743928,
      "grad_norm": 0.6162000894546509,
      "learning_rate": 3.034025211115492e-05,
      "loss": 0.6427,
      "step": 5322
    },
    {
      "epoch": 0.9234906315058987,
      "grad_norm": 0.7952109575271606,
      "learning_rate": 3.03344195149207e-05,
      "loss": 0.567,
      "step": 5323
    },
    {
      "epoch": 0.9236641221374046,
      "grad_norm": 0.6535466313362122,
      "learning_rate": 3.0328585719365057e-05,
      "loss": 0.5482,
      "step": 5324
    },
    {
      "epoch": 0.9238376127689105,
      "grad_norm": 0.8834538459777832,
      "learning_rate": 3.0322750725165e-05,
      "loss": 0.5892,
      "step": 5325
    },
    {
      "epoch": 0.9240111034004164,
      "grad_norm": 0.8313627243041992,
      "learning_rate": 3.0316914532997694e-05,
      "loss": 0.4695,
      "step": 5326
    },
    {
      "epoch": 0.9241845940319223,
      "grad_norm": 0.7623519897460938,
      "learning_rate": 3.031107714354044e-05,
      "loss": 0.5778,
      "step": 5327
    },
    {
      "epoch": 0.9243580846634282,
      "grad_norm": 0.53902268409729,
      "learning_rate": 3.030523855747066e-05,
      "loss": 0.6056,
      "step": 5328
    },
    {
      "epoch": 0.9245315752949341,
      "grad_norm": 0.9953489303588867,
      "learning_rate": 3.0299398775465945e-05,
      "loss": 0.5792,
      "step": 5329
    },
    {
      "epoch": 0.92470506592644,
      "grad_norm": 0.88249671459198,
      "learning_rate": 3.0293557798203998e-05,
      "loss": 0.5087,
      "step": 5330
    },
    {
      "epoch": 0.9248785565579458,
      "grad_norm": 0.7026335597038269,
      "learning_rate": 3.0287715626362676e-05,
      "loss": 0.5618,
      "step": 5331
    },
    {
      "epoch": 0.9250520471894518,
      "grad_norm": 0.6941031813621521,
      "learning_rate": 3.0281872260619965e-05,
      "loss": 0.5142,
      "step": 5332
    },
    {
      "epoch": 0.9252255378209576,
      "grad_norm": 1.039933204650879,
      "learning_rate": 3.0276027701654e-05,
      "loss": 0.4713,
      "step": 5333
    },
    {
      "epoch": 0.9253990284524636,
      "grad_norm": 0.6781054735183716,
      "learning_rate": 3.0270181950143045e-05,
      "loss": 0.6986,
      "step": 5334
    },
    {
      "epoch": 0.9255725190839694,
      "grad_norm": 0.7049393653869629,
      "learning_rate": 3.0264335006765506e-05,
      "loss": 0.5482,
      "step": 5335
    },
    {
      "epoch": 0.9257460097154754,
      "grad_norm": 0.8201839923858643,
      "learning_rate": 3.025848687219993e-05,
      "loss": 0.6687,
      "step": 5336
    },
    {
      "epoch": 0.9259195003469812,
      "grad_norm": 0.9204058647155762,
      "learning_rate": 3.0252637547125e-05,
      "loss": 0.5258,
      "step": 5337
    },
    {
      "epoch": 0.9260929909784872,
      "grad_norm": 0.86858731508255,
      "learning_rate": 3.0246787032219535e-05,
      "loss": 0.4951,
      "step": 5338
    },
    {
      "epoch": 0.926266481609993,
      "grad_norm": 0.9537758231163025,
      "learning_rate": 3.0240935328162498e-05,
      "loss": 0.5529,
      "step": 5339
    },
    {
      "epoch": 0.926439972241499,
      "grad_norm": 0.8528037071228027,
      "learning_rate": 3.0235082435632984e-05,
      "loss": 0.65,
      "step": 5340
    },
    {
      "epoch": 0.9266134628730048,
      "grad_norm": 1.1834839582443237,
      "learning_rate": 3.0229228355310218e-05,
      "loss": 0.5521,
      "step": 5341
    },
    {
      "epoch": 0.9267869535045108,
      "grad_norm": 0.7489802837371826,
      "learning_rate": 3.022337308787359e-05,
      "loss": 0.5699,
      "step": 5342
    },
    {
      "epoch": 0.9269604441360166,
      "grad_norm": 0.9030627608299255,
      "learning_rate": 3.0217516634002596e-05,
      "loss": 0.5197,
      "step": 5343
    },
    {
      "epoch": 0.9271339347675226,
      "grad_norm": 0.6515082716941833,
      "learning_rate": 3.02116589943769e-05,
      "loss": 0.6346,
      "step": 5344
    },
    {
      "epoch": 0.9273074253990284,
      "grad_norm": 0.791456937789917,
      "learning_rate": 3.020580016967627e-05,
      "loss": 0.6565,
      "step": 5345
    },
    {
      "epoch": 0.9274809160305344,
      "grad_norm": 0.8554321527481079,
      "learning_rate": 3.019994016058064e-05,
      "loss": 0.5292,
      "step": 5346
    },
    {
      "epoch": 0.9276544066620402,
      "grad_norm": 0.7764866948127747,
      "learning_rate": 3.019407896777007e-05,
      "loss": 0.5233,
      "step": 5347
    },
    {
      "epoch": 0.9278278972935462,
      "grad_norm": 0.7593770623207092,
      "learning_rate": 3.018821659192476e-05,
      "loss": 0.5081,
      "step": 5348
    },
    {
      "epoch": 0.928001387925052,
      "grad_norm": 0.6704859733581543,
      "learning_rate": 3.018235303372504e-05,
      "loss": 0.5856,
      "step": 5349
    },
    {
      "epoch": 0.9281748785565579,
      "grad_norm": 0.6950547099113464,
      "learning_rate": 3.0176488293851388e-05,
      "loss": 0.511,
      "step": 5350
    },
    {
      "epoch": 0.9283483691880638,
      "grad_norm": 0.5945006012916565,
      "learning_rate": 3.017062237298441e-05,
      "loss": 0.6611,
      "step": 5351
    },
    {
      "epoch": 0.9285218598195697,
      "grad_norm": 0.925314724445343,
      "learning_rate": 3.0164755271804856e-05,
      "loss": 0.6042,
      "step": 5352
    },
    {
      "epoch": 0.9286953504510757,
      "grad_norm": 0.9462495446205139,
      "learning_rate": 3.0158886990993612e-05,
      "loss": 0.5857,
      "step": 5353
    },
    {
      "epoch": 0.9288688410825815,
      "grad_norm": 0.7837413549423218,
      "learning_rate": 3.015301753123169e-05,
      "loss": 0.6163,
      "step": 5354
    },
    {
      "epoch": 0.9290423317140875,
      "grad_norm": 0.8495484590530396,
      "learning_rate": 3.0147146893200248e-05,
      "loss": 0.6177,
      "step": 5355
    },
    {
      "epoch": 0.9292158223455933,
      "grad_norm": 0.8009089827537537,
      "learning_rate": 3.0141275077580592e-05,
      "loss": 0.5399,
      "step": 5356
    },
    {
      "epoch": 0.9293893129770993,
      "grad_norm": 1.9470219612121582,
      "learning_rate": 3.0135402085054148e-05,
      "loss": 0.5857,
      "step": 5357
    },
    {
      "epoch": 0.9295628036086051,
      "grad_norm": 0.6537600159645081,
      "learning_rate": 3.0129527916302482e-05,
      "loss": 0.7056,
      "step": 5358
    },
    {
      "epoch": 0.9297362942401111,
      "grad_norm": 0.8429149985313416,
      "learning_rate": 3.0123652572007295e-05,
      "loss": 0.6906,
      "step": 5359
    },
    {
      "epoch": 0.9299097848716169,
      "grad_norm": 0.8222671747207642,
      "learning_rate": 3.0117776052850427e-05,
      "loss": 0.6055,
      "step": 5360
    },
    {
      "epoch": 0.9300832755031229,
      "grad_norm": 0.5313491225242615,
      "learning_rate": 3.0111898359513865e-05,
      "loss": 0.6307,
      "step": 5361
    },
    {
      "epoch": 0.9302567661346287,
      "grad_norm": 0.6201162934303284,
      "learning_rate": 3.0106019492679714e-05,
      "loss": 0.6304,
      "step": 5362
    },
    {
      "epoch": 0.9304302567661347,
      "grad_norm": 0.9020854234695435,
      "learning_rate": 3.0100139453030222e-05,
      "loss": 0.4982,
      "step": 5363
    },
    {
      "epoch": 0.9306037473976405,
      "grad_norm": 0.7810084223747253,
      "learning_rate": 3.009425824124778e-05,
      "loss": 0.6063,
      "step": 5364
    },
    {
      "epoch": 0.9307772380291465,
      "grad_norm": 0.7074047923088074,
      "learning_rate": 3.0088375858014905e-05,
      "loss": 0.6418,
      "step": 5365
    },
    {
      "epoch": 0.9309507286606523,
      "grad_norm": 0.7246010899543762,
      "learning_rate": 3.008249230401426e-05,
      "loss": 0.6698,
      "step": 5366
    },
    {
      "epoch": 0.9311242192921583,
      "grad_norm": 0.7391355037689209,
      "learning_rate": 3.007660757992863e-05,
      "loss": 0.6702,
      "step": 5367
    },
    {
      "epoch": 0.9312977099236641,
      "grad_norm": 0.7674587368965149,
      "learning_rate": 3.0070721686440953e-05,
      "loss": 0.5809,
      "step": 5368
    },
    {
      "epoch": 0.9314712005551701,
      "grad_norm": 0.7705550193786621,
      "learning_rate": 3.0064834624234283e-05,
      "loss": 0.6299,
      "step": 5369
    },
    {
      "epoch": 0.9316446911866759,
      "grad_norm": 1.0173293352127075,
      "learning_rate": 3.0058946393991833e-05,
      "loss": 0.4689,
      "step": 5370
    },
    {
      "epoch": 0.9318181818181818,
      "grad_norm": 0.5605577230453491,
      "learning_rate": 3.005305699639693e-05,
      "loss": 0.6779,
      "step": 5371
    },
    {
      "epoch": 0.9319916724496877,
      "grad_norm": 0.6631309986114502,
      "learning_rate": 3.004716643213305e-05,
      "loss": 0.6848,
      "step": 5372
    },
    {
      "epoch": 0.9321651630811936,
      "grad_norm": 0.7009269595146179,
      "learning_rate": 3.0041274701883794e-05,
      "loss": 0.4752,
      "step": 5373
    },
    {
      "epoch": 0.9323386537126995,
      "grad_norm": 0.7363549470901489,
      "learning_rate": 3.003538180633292e-05,
      "loss": 0.5668,
      "step": 5374
    },
    {
      "epoch": 0.9325121443442054,
      "grad_norm": 2.7398874759674072,
      "learning_rate": 3.002948774616429e-05,
      "loss": 0.6873,
      "step": 5375
    },
    {
      "epoch": 0.9326856349757113,
      "grad_norm": 0.7422850131988525,
      "learning_rate": 3.0023592522061916e-05,
      "loss": 0.6235,
      "step": 5376
    },
    {
      "epoch": 0.9328591256072172,
      "grad_norm": 0.642112672328949,
      "learning_rate": 3.0017696134709946e-05,
      "loss": 0.6691,
      "step": 5377
    },
    {
      "epoch": 0.9330326162387231,
      "grad_norm": 0.731460690498352,
      "learning_rate": 3.0011798584792672e-05,
      "loss": 0.5582,
      "step": 5378
    },
    {
      "epoch": 0.933206106870229,
      "grad_norm": 1.0289613008499146,
      "learning_rate": 3.000589987299451e-05,
      "loss": 0.5909,
      "step": 5379
    },
    {
      "epoch": 0.9333795975017349,
      "grad_norm": 0.6110485792160034,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.6539,
      "step": 5380
    },
    {
      "epoch": 0.9335530881332408,
      "grad_norm": 0.7026828527450562,
      "learning_rate": 2.9994098966493842e-05,
      "loss": 0.5867,
      "step": 5381
    },
    {
      "epoch": 0.9337265787647467,
      "grad_norm": 0.675056517124176,
      "learning_rate": 2.9988196773160857e-05,
      "loss": 0.5767,
      "step": 5382
    },
    {
      "epoch": 0.9339000693962526,
      "grad_norm": 0.8703925013542175,
      "learning_rate": 2.9982293420685998e-05,
      "loss": 0.6873,
      "step": 5383
    },
    {
      "epoch": 0.9340735600277585,
      "grad_norm": 0.7216558456420898,
      "learning_rate": 2.9976388909754348e-05,
      "loss": 0.5837,
      "step": 5384
    },
    {
      "epoch": 0.9342470506592644,
      "grad_norm": 1.0611991882324219,
      "learning_rate": 2.997048324105115e-05,
      "loss": 0.5745,
      "step": 5385
    },
    {
      "epoch": 0.9344205412907703,
      "grad_norm": 0.6006755828857422,
      "learning_rate": 2.996457641526174e-05,
      "loss": 0.4956,
      "step": 5386
    },
    {
      "epoch": 0.9345940319222762,
      "grad_norm": 0.713334858417511,
      "learning_rate": 2.995866843307164e-05,
      "loss": 0.6038,
      "step": 5387
    },
    {
      "epoch": 0.9347675225537821,
      "grad_norm": 0.6840294599533081,
      "learning_rate": 2.995275929516646e-05,
      "loss": 0.5118,
      "step": 5388
    },
    {
      "epoch": 0.934941013185288,
      "grad_norm": 0.853123128414154,
      "learning_rate": 2.9946849002231962e-05,
      "loss": 0.4916,
      "step": 5389
    },
    {
      "epoch": 0.9351145038167938,
      "grad_norm": 0.750608503818512,
      "learning_rate": 2.9940937554954053e-05,
      "loss": 0.4921,
      "step": 5390
    },
    {
      "epoch": 0.9352879944482998,
      "grad_norm": 0.769507110118866,
      "learning_rate": 2.993502495401875e-05,
      "loss": 0.6527,
      "step": 5391
    },
    {
      "epoch": 0.9354614850798056,
      "grad_norm": 0.6452317237854004,
      "learning_rate": 2.9929111200112233e-05,
      "loss": 0.5686,
      "step": 5392
    },
    {
      "epoch": 0.9356349757113116,
      "grad_norm": 0.6807578802108765,
      "learning_rate": 2.9923196293920786e-05,
      "loss": 0.6285,
      "step": 5393
    },
    {
      "epoch": 0.9358084663428174,
      "grad_norm": 0.6604334712028503,
      "learning_rate": 2.991728023613085e-05,
      "loss": 0.6748,
      "step": 5394
    },
    {
      "epoch": 0.9359819569743234,
      "grad_norm": 0.6205240488052368,
      "learning_rate": 2.9911363027428982e-05,
      "loss": 0.662,
      "step": 5395
    },
    {
      "epoch": 0.9361554476058292,
      "grad_norm": 0.8346847891807556,
      "learning_rate": 2.990544466850189e-05,
      "loss": 0.6627,
      "step": 5396
    },
    {
      "epoch": 0.9363289382373352,
      "grad_norm": 0.9589856863021851,
      "learning_rate": 2.9899525160036398e-05,
      "loss": 0.675,
      "step": 5397
    },
    {
      "epoch": 0.936502428868841,
      "grad_norm": 0.6498943567276001,
      "learning_rate": 2.9893604502719474e-05,
      "loss": 0.6348,
      "step": 5398
    },
    {
      "epoch": 0.936675919500347,
      "grad_norm": 1.012377142906189,
      "learning_rate": 2.9887682697238226e-05,
      "loss": 0.5652,
      "step": 5399
    },
    {
      "epoch": 0.9368494101318529,
      "grad_norm": 0.6812121868133545,
      "learning_rate": 2.9881759744279875e-05,
      "loss": 0.6731,
      "step": 5400
    },
    {
      "epoch": 0.9370229007633588,
      "grad_norm": 0.7409417629241943,
      "learning_rate": 2.9875835644531793e-05,
      "loss": 0.5255,
      "step": 5401
    },
    {
      "epoch": 0.9371963913948647,
      "grad_norm": 0.7330713272094727,
      "learning_rate": 2.986991039868148e-05,
      "loss": 0.5776,
      "step": 5402
    },
    {
      "epoch": 0.9373698820263706,
      "grad_norm": 0.6653503775596619,
      "learning_rate": 2.986398400741656e-05,
      "loss": 0.5612,
      "step": 5403
    },
    {
      "epoch": 0.9375433726578765,
      "grad_norm": 0.6653831601142883,
      "learning_rate": 2.9858056471424804e-05,
      "loss": 0.6423,
      "step": 5404
    },
    {
      "epoch": 0.9377168632893824,
      "grad_norm": 0.7596984505653381,
      "learning_rate": 2.9852127791394106e-05,
      "loss": 0.5795,
      "step": 5405
    },
    {
      "epoch": 0.9378903539208883,
      "grad_norm": 0.9415115714073181,
      "learning_rate": 2.98461979680125e-05,
      "loss": 0.5104,
      "step": 5406
    },
    {
      "epoch": 0.9380638445523942,
      "grad_norm": 0.8615516424179077,
      "learning_rate": 2.984026700196814e-05,
      "loss": 0.4844,
      "step": 5407
    },
    {
      "epoch": 0.9382373351839001,
      "grad_norm": 0.7577638626098633,
      "learning_rate": 2.983433489394934e-05,
      "loss": 0.4819,
      "step": 5408
    },
    {
      "epoch": 0.9384108258154059,
      "grad_norm": 1.619036316871643,
      "learning_rate": 2.982840164464451e-05,
      "loss": 0.5795,
      "step": 5409
    },
    {
      "epoch": 0.9385843164469119,
      "grad_norm": 0.6858264207839966,
      "learning_rate": 2.9822467254742212e-05,
      "loss": 0.4938,
      "step": 5410
    },
    {
      "epoch": 0.9387578070784177,
      "grad_norm": 0.8433195352554321,
      "learning_rate": 2.9816531724931152e-05,
      "loss": 0.4526,
      "step": 5411
    },
    {
      "epoch": 0.9389312977099237,
      "grad_norm": 0.7492510676383972,
      "learning_rate": 2.9810595055900148e-05,
      "loss": 0.5171,
      "step": 5412
    },
    {
      "epoch": 0.9391047883414295,
      "grad_norm": 1.151039958000183,
      "learning_rate": 2.9804657248338146e-05,
      "loss": 0.4989,
      "step": 5413
    },
    {
      "epoch": 0.9392782789729355,
      "grad_norm": 0.6459309458732605,
      "learning_rate": 2.9798718302934255e-05,
      "loss": 0.5631,
      "step": 5414
    },
    {
      "epoch": 0.9394517696044413,
      "grad_norm": 0.7519208788871765,
      "learning_rate": 2.9792778220377675e-05,
      "loss": 0.5808,
      "step": 5415
    },
    {
      "epoch": 0.9396252602359473,
      "grad_norm": 0.8628186583518982,
      "learning_rate": 2.9786837001357782e-05,
      "loss": 0.6256,
      "step": 5416
    },
    {
      "epoch": 0.9397987508674531,
      "grad_norm": 0.5924263000488281,
      "learning_rate": 2.978089464656405e-05,
      "loss": 0.6926,
      "step": 5417
    },
    {
      "epoch": 0.9399722414989591,
      "grad_norm": 0.7776699662208557,
      "learning_rate": 2.9774951156686094e-05,
      "loss": 0.5999,
      "step": 5418
    },
    {
      "epoch": 0.9401457321304649,
      "grad_norm": 0.7869418263435364,
      "learning_rate": 2.9769006532413667e-05,
      "loss": 0.5306,
      "step": 5419
    },
    {
      "epoch": 0.9403192227619709,
      "grad_norm": 0.9118232727050781,
      "learning_rate": 2.976306077443665e-05,
      "loss": 0.5358,
      "step": 5420
    },
    {
      "epoch": 0.9404927133934767,
      "grad_norm": 0.5323087573051453,
      "learning_rate": 2.9757113883445056e-05,
      "loss": 0.4699,
      "step": 5421
    },
    {
      "epoch": 0.9406662040249827,
      "grad_norm": 1.0102770328521729,
      "learning_rate": 2.9751165860129024e-05,
      "loss": 0.4867,
      "step": 5422
    },
    {
      "epoch": 0.9408396946564885,
      "grad_norm": 0.6074740886688232,
      "learning_rate": 2.974521670517883e-05,
      "loss": 0.6996,
      "step": 5423
    },
    {
      "epoch": 0.9410131852879945,
      "grad_norm": 0.9579439163208008,
      "learning_rate": 2.973926641928489e-05,
      "loss": 0.5344,
      "step": 5424
    },
    {
      "epoch": 0.9411866759195003,
      "grad_norm": 0.7391860485076904,
      "learning_rate": 2.9733315003137725e-05,
      "loss": 0.4725,
      "step": 5425
    },
    {
      "epoch": 0.9413601665510063,
      "grad_norm": 1.3167799711227417,
      "learning_rate": 2.9727362457428012e-05,
      "loss": 0.4675,
      "step": 5426
    },
    {
      "epoch": 0.9415336571825121,
      "grad_norm": 1.0091986656188965,
      "learning_rate": 2.9721408782846554e-05,
      "loss": 0.5354,
      "step": 5427
    },
    {
      "epoch": 0.9417071478140181,
      "grad_norm": 0.6604625582695007,
      "learning_rate": 2.971545398008428e-05,
      "loss": 0.5947,
      "step": 5428
    },
    {
      "epoch": 0.9418806384455239,
      "grad_norm": 0.723139226436615,
      "learning_rate": 2.970949804983225e-05,
      "loss": 0.6659,
      "step": 5429
    },
    {
      "epoch": 0.9420541290770298,
      "grad_norm": 0.6684121489524841,
      "learning_rate": 2.970354099278166e-05,
      "loss": 0.5742,
      "step": 5430
    },
    {
      "epoch": 0.9422276197085357,
      "grad_norm": 0.7406533360481262,
      "learning_rate": 2.9697582809623828e-05,
      "loss": 0.4503,
      "step": 5431
    },
    {
      "epoch": 0.9424011103400416,
      "grad_norm": 1.013691782951355,
      "learning_rate": 2.9691623501050212e-05,
      "loss": 0.6056,
      "step": 5432
    },
    {
      "epoch": 0.9425746009715475,
      "grad_norm": 0.9413995146751404,
      "learning_rate": 2.968566306775239e-05,
      "loss": 0.6575,
      "step": 5433
    },
    {
      "epoch": 0.9427480916030534,
      "grad_norm": 0.6486615538597107,
      "learning_rate": 2.967970151042209e-05,
      "loss": 0.5552,
      "step": 5434
    },
    {
      "epoch": 0.9429215822345594,
      "grad_norm": 0.7300331592559814,
      "learning_rate": 2.9673738829751148e-05,
      "loss": 0.6271,
      "step": 5435
    },
    {
      "epoch": 0.9430950728660652,
      "grad_norm": 0.7458223700523376,
      "learning_rate": 2.9667775026431544e-05,
      "loss": 0.4535,
      "step": 5436
    },
    {
      "epoch": 0.9432685634975712,
      "grad_norm": 0.8116676807403564,
      "learning_rate": 2.9661810101155387e-05,
      "loss": 0.5339,
      "step": 5437
    },
    {
      "epoch": 0.943442054129077,
      "grad_norm": 0.9047504663467407,
      "learning_rate": 2.96558440546149e-05,
      "loss": 0.5934,
      "step": 5438
    },
    {
      "epoch": 0.943615544760583,
      "grad_norm": 0.8103256225585938,
      "learning_rate": 2.9649876887502467e-05,
      "loss": 0.5372,
      "step": 5439
    },
    {
      "epoch": 0.9437890353920888,
      "grad_norm": 0.9196052551269531,
      "learning_rate": 2.9643908600510572e-05,
      "loss": 0.5477,
      "step": 5440
    },
    {
      "epoch": 0.9439625260235948,
      "grad_norm": 0.8215181827545166,
      "learning_rate": 2.9637939194331848e-05,
      "loss": 0.5109,
      "step": 5441
    },
    {
      "epoch": 0.9441360166551006,
      "grad_norm": 0.7650948166847229,
      "learning_rate": 2.9631968669659047e-05,
      "loss": 0.5592,
      "step": 5442
    },
    {
      "epoch": 0.9443095072866066,
      "grad_norm": 0.8600346446037292,
      "learning_rate": 2.9625997027185064e-05,
      "loss": 0.5566,
      "step": 5443
    },
    {
      "epoch": 0.9444829979181124,
      "grad_norm": 0.6058077216148376,
      "learning_rate": 2.9620024267602906e-05,
      "loss": 0.6108,
      "step": 5444
    },
    {
      "epoch": 0.9446564885496184,
      "grad_norm": 0.6449906229972839,
      "learning_rate": 2.9614050391605725e-05,
      "loss": 0.6129,
      "step": 5445
    },
    {
      "epoch": 0.9448299791811242,
      "grad_norm": 0.6964638829231262,
      "learning_rate": 2.9608075399886792e-05,
      "loss": 0.5753,
      "step": 5446
    },
    {
      "epoch": 0.9450034698126302,
      "grad_norm": 0.840380072593689,
      "learning_rate": 2.9602099293139512e-05,
      "loss": 0.5975,
      "step": 5447
    },
    {
      "epoch": 0.945176960444136,
      "grad_norm": 0.8918290734291077,
      "learning_rate": 2.9596122072057424e-05,
      "loss": 0.5198,
      "step": 5448
    },
    {
      "epoch": 0.9453504510756419,
      "grad_norm": 0.6563118100166321,
      "learning_rate": 2.9590143737334182e-05,
      "loss": 0.5398,
      "step": 5449
    },
    {
      "epoch": 0.9455239417071478,
      "grad_norm": 0.7787469625473022,
      "learning_rate": 2.958416428966359e-05,
      "loss": 0.6144,
      "step": 5450
    },
    {
      "epoch": 0.9456974323386537,
      "grad_norm": 0.6563782691955566,
      "learning_rate": 2.9578183729739566e-05,
      "loss": 0.5747,
      "step": 5451
    },
    {
      "epoch": 0.9458709229701596,
      "grad_norm": 1.1232762336730957,
      "learning_rate": 2.957220205825615e-05,
      "loss": 0.5459,
      "step": 5452
    },
    {
      "epoch": 0.9460444136016655,
      "grad_norm": 1.3598390817642212,
      "learning_rate": 2.956621927590754e-05,
      "loss": 0.5029,
      "step": 5453
    },
    {
      "epoch": 0.9462179042331714,
      "grad_norm": 0.5797144174575806,
      "learning_rate": 2.956023538338803e-05,
      "loss": 0.5321,
      "step": 5454
    },
    {
      "epoch": 0.9463913948646773,
      "grad_norm": 0.8988631367683411,
      "learning_rate": 2.9554250381392067e-05,
      "loss": 0.5005,
      "step": 5455
    },
    {
      "epoch": 0.9465648854961832,
      "grad_norm": 0.6104587316513062,
      "learning_rate": 2.9548264270614217e-05,
      "loss": 0.5681,
      "step": 5456
    },
    {
      "epoch": 0.9467383761276891,
      "grad_norm": 0.6378024816513062,
      "learning_rate": 2.954227705174917e-05,
      "loss": 0.5446,
      "step": 5457
    },
    {
      "epoch": 0.946911866759195,
      "grad_norm": 0.6203073859214783,
      "learning_rate": 2.953628872549175e-05,
      "loss": 0.5626,
      "step": 5458
    },
    {
      "epoch": 0.9470853573907009,
      "grad_norm": 1.1577155590057373,
      "learning_rate": 2.9530299292536918e-05,
      "loss": 0.6069,
      "step": 5459
    },
    {
      "epoch": 0.9472588480222068,
      "grad_norm": 0.7365251779556274,
      "learning_rate": 2.952430875357974e-05,
      "loss": 0.6189,
      "step": 5460
    },
    {
      "epoch": 0.9474323386537127,
      "grad_norm": 0.8239716291427612,
      "learning_rate": 2.951831710931544e-05,
      "loss": 0.5234,
      "step": 5461
    },
    {
      "epoch": 0.9476058292852186,
      "grad_norm": 0.8461446166038513,
      "learning_rate": 2.9512324360439347e-05,
      "loss": 0.4756,
      "step": 5462
    },
    {
      "epoch": 0.9477793199167245,
      "grad_norm": 0.7904476523399353,
      "learning_rate": 2.950633050764692e-05,
      "loss": 0.635,
      "step": 5463
    },
    {
      "epoch": 0.9479528105482304,
      "grad_norm": 0.792469322681427,
      "learning_rate": 2.9500335551633773e-05,
      "loss": 0.5447,
      "step": 5464
    },
    {
      "epoch": 0.9481263011797363,
      "grad_norm": 0.7433012127876282,
      "learning_rate": 2.9494339493095605e-05,
      "loss": 0.4449,
      "step": 5465
    },
    {
      "epoch": 0.9482997918112422,
      "grad_norm": 0.8306467533111572,
      "learning_rate": 2.9488342332728276e-05,
      "loss": 0.4879,
      "step": 5466
    },
    {
      "epoch": 0.9484732824427481,
      "grad_norm": 0.7154919505119324,
      "learning_rate": 2.9482344071227767e-05,
      "loss": 0.4698,
      "step": 5467
    },
    {
      "epoch": 0.9486467730742539,
      "grad_norm": 0.6896546483039856,
      "learning_rate": 2.9476344709290175e-05,
      "loss": 0.6769,
      "step": 5468
    },
    {
      "epoch": 0.9488202637057599,
      "grad_norm": 0.977916955947876,
      "learning_rate": 2.9470344247611744e-05,
      "loss": 0.5424,
      "step": 5469
    },
    {
      "epoch": 0.9489937543372657,
      "grad_norm": 0.7709428071975708,
      "learning_rate": 2.9464342686888826e-05,
      "loss": 0.6493,
      "step": 5470
    },
    {
      "epoch": 0.9491672449687717,
      "grad_norm": 0.7862691879272461,
      "learning_rate": 2.9458340027817912e-05,
      "loss": 0.6556,
      "step": 5471
    },
    {
      "epoch": 0.9493407356002775,
      "grad_norm": 0.9022845029830933,
      "learning_rate": 2.9452336271095613e-05,
      "loss": 0.6194,
      "step": 5472
    },
    {
      "epoch": 0.9495142262317835,
      "grad_norm": 0.6442118883132935,
      "learning_rate": 2.9446331417418678e-05,
      "loss": 0.5963,
      "step": 5473
    },
    {
      "epoch": 0.9496877168632893,
      "grad_norm": 0.9894945621490479,
      "learning_rate": 2.9440325467483974e-05,
      "loss": 0.5223,
      "step": 5474
    },
    {
      "epoch": 0.9498612074947953,
      "grad_norm": 0.7461522817611694,
      "learning_rate": 2.94343184219885e-05,
      "loss": 0.6187,
      "step": 5475
    },
    {
      "epoch": 0.9500346981263011,
      "grad_norm": 0.615246057510376,
      "learning_rate": 2.942831028162938e-05,
      "loss": 0.6678,
      "step": 5476
    },
    {
      "epoch": 0.9502081887578071,
      "grad_norm": 0.7718769311904907,
      "learning_rate": 2.942230104710387e-05,
      "loss": 0.5106,
      "step": 5477
    },
    {
      "epoch": 0.950381679389313,
      "grad_norm": 0.45673322677612305,
      "learning_rate": 2.9416290719109333e-05,
      "loss": 0.6527,
      "step": 5478
    },
    {
      "epoch": 0.9505551700208189,
      "grad_norm": 0.7970420122146606,
      "learning_rate": 2.9410279298343294e-05,
      "loss": 0.6426,
      "step": 5479
    },
    {
      "epoch": 0.9507286606523248,
      "grad_norm": 0.7532060146331787,
      "learning_rate": 2.9404266785503376e-05,
      "loss": 0.4949,
      "step": 5480
    },
    {
      "epoch": 0.9509021512838307,
      "grad_norm": 0.6640486717224121,
      "learning_rate": 2.939825318128734e-05,
      "loss": 0.6591,
      "step": 5481
    },
    {
      "epoch": 0.9510756419153366,
      "grad_norm": 0.5723349452018738,
      "learning_rate": 2.9392238486393068e-05,
      "loss": 0.7288,
      "step": 5482
    },
    {
      "epoch": 0.9512491325468425,
      "grad_norm": 0.6749547123908997,
      "learning_rate": 2.9386222701518582e-05,
      "loss": 0.5728,
      "step": 5483
    },
    {
      "epoch": 0.9514226231783484,
      "grad_norm": 1.8638474941253662,
      "learning_rate": 2.9380205827362007e-05,
      "loss": 0.519,
      "step": 5484
    },
    {
      "epoch": 0.9515961138098543,
      "grad_norm": 0.7419898509979248,
      "learning_rate": 2.937418786462162e-05,
      "loss": 0.6128,
      "step": 5485
    },
    {
      "epoch": 0.9517696044413602,
      "grad_norm": 1.0060269832611084,
      "learning_rate": 2.9368168813995806e-05,
      "loss": 0.5758,
      "step": 5486
    },
    {
      "epoch": 0.9519430950728661,
      "grad_norm": 0.7519690990447998,
      "learning_rate": 2.9362148676183087e-05,
      "loss": 0.5039,
      "step": 5487
    },
    {
      "epoch": 0.952116585704372,
      "grad_norm": 0.8661027550697327,
      "learning_rate": 2.9356127451882105e-05,
      "loss": 0.6177,
      "step": 5488
    },
    {
      "epoch": 0.9522900763358778,
      "grad_norm": 0.7929277420043945,
      "learning_rate": 2.9350105141791627e-05,
      "loss": 0.5514,
      "step": 5489
    },
    {
      "epoch": 0.9524635669673838,
      "grad_norm": 1.2426139116287231,
      "learning_rate": 2.934408174661055e-05,
      "loss": 0.6614,
      "step": 5490
    },
    {
      "epoch": 0.9526370575988896,
      "grad_norm": 1.2336061000823975,
      "learning_rate": 2.9338057267037906e-05,
      "loss": 0.5786,
      "step": 5491
    },
    {
      "epoch": 0.9528105482303956,
      "grad_norm": 0.7504399418830872,
      "learning_rate": 2.933203170377283e-05,
      "loss": 0.5699,
      "step": 5492
    },
    {
      "epoch": 0.9529840388619014,
      "grad_norm": 0.7243566513061523,
      "learning_rate": 2.9326005057514605e-05,
      "loss": 0.5646,
      "step": 5493
    },
    {
      "epoch": 0.9531575294934074,
      "grad_norm": 0.7563011050224304,
      "learning_rate": 2.931997732896262e-05,
      "loss": 0.5378,
      "step": 5494
    },
    {
      "epoch": 0.9533310201249132,
      "grad_norm": 0.727756917476654,
      "learning_rate": 2.9313948518816417e-05,
      "loss": 0.5839,
      "step": 5495
    },
    {
      "epoch": 0.9535045107564192,
      "grad_norm": 0.7686708569526672,
      "learning_rate": 2.9307918627775627e-05,
      "loss": 0.6272,
      "step": 5496
    },
    {
      "epoch": 0.953678001387925,
      "grad_norm": 0.8600760698318481,
      "learning_rate": 2.9301887656540034e-05,
      "loss": 0.4978,
      "step": 5497
    },
    {
      "epoch": 0.953851492019431,
      "grad_norm": 1.7401413917541504,
      "learning_rate": 2.9295855605809543e-05,
      "loss": 0.521,
      "step": 5498
    },
    {
      "epoch": 0.9540249826509368,
      "grad_norm": 1.0921040773391724,
      "learning_rate": 2.9289822476284172e-05,
      "loss": 0.6127,
      "step": 5499
    },
    {
      "epoch": 0.9541984732824428,
      "grad_norm": 0.7612084746360779,
      "learning_rate": 2.9283788268664085e-05,
      "loss": 0.6558,
      "step": 5500
    },
    {
      "epoch": 0.9543719639139486,
      "grad_norm": 0.7520211338996887,
      "learning_rate": 2.9277752983649548e-05,
      "loss": 0.5397,
      "step": 5501
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.8717694878578186,
      "learning_rate": 2.9271716621940965e-05,
      "loss": 0.6041,
      "step": 5502
    },
    {
      "epoch": 0.9547189451769604,
      "grad_norm": 0.8843144774436951,
      "learning_rate": 2.9265679184238867e-05,
      "loss": 0.6799,
      "step": 5503
    },
    {
      "epoch": 0.9548924358084664,
      "grad_norm": 0.555732250213623,
      "learning_rate": 2.9259640671243903e-05,
      "loss": 0.658,
      "step": 5504
    },
    {
      "epoch": 0.9550659264399722,
      "grad_norm": 0.6608411073684692,
      "learning_rate": 2.9253601083656846e-05,
      "loss": 0.5549,
      "step": 5505
    },
    {
      "epoch": 0.9552394170714782,
      "grad_norm": 0.6673864722251892,
      "learning_rate": 2.9247560422178604e-05,
      "loss": 0.6238,
      "step": 5506
    },
    {
      "epoch": 0.955412907702984,
      "grad_norm": 0.8510689735412598,
      "learning_rate": 2.92415186875102e-05,
      "loss": 0.6591,
      "step": 5507
    },
    {
      "epoch": 0.9555863983344899,
      "grad_norm": 0.6071183681488037,
      "learning_rate": 2.923547588035278e-05,
      "loss": 0.6448,
      "step": 5508
    },
    {
      "epoch": 0.9557598889659958,
      "grad_norm": 0.6564860939979553,
      "learning_rate": 2.922943200140763e-05,
      "loss": 0.609,
      "step": 5509
    },
    {
      "epoch": 0.9559333795975017,
      "grad_norm": 0.9711827635765076,
      "learning_rate": 2.9223387051376133e-05,
      "loss": 0.6007,
      "step": 5510
    },
    {
      "epoch": 0.9561068702290076,
      "grad_norm": 0.6167823076248169,
      "learning_rate": 2.9217341030959824e-05,
      "loss": 0.6223,
      "step": 5511
    },
    {
      "epoch": 0.9562803608605135,
      "grad_norm": 0.6490203738212585,
      "learning_rate": 2.921129394086035e-05,
      "loss": 0.5486,
      "step": 5512
    },
    {
      "epoch": 0.9564538514920194,
      "grad_norm": 0.8874254822731018,
      "learning_rate": 2.920524578177948e-05,
      "loss": 0.5619,
      "step": 5513
    },
    {
      "epoch": 0.9566273421235253,
      "grad_norm": 0.8154889941215515,
      "learning_rate": 2.919919655441911e-05,
      "loss": 0.6151,
      "step": 5514
    },
    {
      "epoch": 0.9568008327550312,
      "grad_norm": 0.6881371736526489,
      "learning_rate": 2.9193146259481265e-05,
      "loss": 0.6265,
      "step": 5515
    },
    {
      "epoch": 0.9569743233865371,
      "grad_norm": 0.6467137336730957,
      "learning_rate": 2.9187094897668076e-05,
      "loss": 0.7117,
      "step": 5516
    },
    {
      "epoch": 0.957147814018043,
      "grad_norm": 0.7240439057350159,
      "learning_rate": 2.9181042469681825e-05,
      "loss": 0.7471,
      "step": 5517
    },
    {
      "epoch": 0.9573213046495489,
      "grad_norm": 0.6998037099838257,
      "learning_rate": 2.9174988976224897e-05,
      "loss": 0.5426,
      "step": 5518
    },
    {
      "epoch": 0.9574947952810549,
      "grad_norm": 0.7013698816299438,
      "learning_rate": 2.9168934417999803e-05,
      "loss": 0.6407,
      "step": 5519
    },
    {
      "epoch": 0.9576682859125607,
      "grad_norm": 0.5440288782119751,
      "learning_rate": 2.916287879570919e-05,
      "loss": 0.5219,
      "step": 5520
    },
    {
      "epoch": 0.9578417765440667,
      "grad_norm": 0.8371912837028503,
      "learning_rate": 2.9156822110055816e-05,
      "loss": 0.5396,
      "step": 5521
    },
    {
      "epoch": 0.9580152671755725,
      "grad_norm": 0.7916197180747986,
      "learning_rate": 2.9150764361742564e-05,
      "loss": 0.4921,
      "step": 5522
    },
    {
      "epoch": 0.9581887578070785,
      "grad_norm": 0.7627412676811218,
      "learning_rate": 2.9144705551472445e-05,
      "loss": 0.5939,
      "step": 5523
    },
    {
      "epoch": 0.9583622484385843,
      "grad_norm": 0.6743812561035156,
      "learning_rate": 2.9138645679948596e-05,
      "loss": 0.5223,
      "step": 5524
    },
    {
      "epoch": 0.9585357390700903,
      "grad_norm": 0.8483150601387024,
      "learning_rate": 2.9132584747874265e-05,
      "loss": 0.5143,
      "step": 5525
    },
    {
      "epoch": 0.9587092297015961,
      "grad_norm": 0.8951861262321472,
      "learning_rate": 2.912652275595283e-05,
      "loss": 0.502,
      "step": 5526
    },
    {
      "epoch": 0.9588827203331021,
      "grad_norm": 0.8937050104141235,
      "learning_rate": 2.91204597048878e-05,
      "loss": 0.5006,
      "step": 5527
    },
    {
      "epoch": 0.9590562109646079,
      "grad_norm": 0.7550032138824463,
      "learning_rate": 2.9114395595382786e-05,
      "loss": 0.5442,
      "step": 5528
    },
    {
      "epoch": 0.9592297015961138,
      "grad_norm": 0.800572395324707,
      "learning_rate": 2.9108330428141544e-05,
      "loss": 0.5387,
      "step": 5529
    },
    {
      "epoch": 0.9594031922276197,
      "grad_norm": 0.8911494612693787,
      "learning_rate": 2.9102264203867948e-05,
      "loss": 0.6061,
      "step": 5530
    },
    {
      "epoch": 0.9595766828591256,
      "grad_norm": 0.7251107096672058,
      "learning_rate": 2.9096196923265984e-05,
      "loss": 0.5394,
      "step": 5531
    },
    {
      "epoch": 0.9597501734906315,
      "grad_norm": 0.8301741480827332,
      "learning_rate": 2.9090128587039763e-05,
      "loss": 0.5825,
      "step": 5532
    },
    {
      "epoch": 0.9599236641221374,
      "grad_norm": 0.7158095836639404,
      "learning_rate": 2.9084059195893536e-05,
      "loss": 0.5177,
      "step": 5533
    },
    {
      "epoch": 0.9600971547536433,
      "grad_norm": 1.2218570709228516,
      "learning_rate": 2.9077988750531645e-05,
      "loss": 0.4814,
      "step": 5534
    },
    {
      "epoch": 0.9602706453851492,
      "grad_norm": 0.881252110004425,
      "learning_rate": 2.9071917251658583e-05,
      "loss": 0.6406,
      "step": 5535
    },
    {
      "epoch": 0.9604441360166551,
      "grad_norm": 0.8924074172973633,
      "learning_rate": 2.9065844699978957e-05,
      "loss": 0.5591,
      "step": 5536
    },
    {
      "epoch": 0.960617626648161,
      "grad_norm": 0.8034957051277161,
      "learning_rate": 2.9059771096197488e-05,
      "loss": 0.4939,
      "step": 5537
    },
    {
      "epoch": 0.9607911172796669,
      "grad_norm": 0.9034562706947327,
      "learning_rate": 2.9053696441019022e-05,
      "loss": 0.5133,
      "step": 5538
    },
    {
      "epoch": 0.9609646079111728,
      "grad_norm": 0.6369368433952332,
      "learning_rate": 2.9047620735148542e-05,
      "loss": 0.5203,
      "step": 5539
    },
    {
      "epoch": 0.9611380985426787,
      "grad_norm": 0.6362287402153015,
      "learning_rate": 2.9041543979291125e-05,
      "loss": 0.558,
      "step": 5540
    },
    {
      "epoch": 0.9613115891741846,
      "grad_norm": 0.7860467433929443,
      "learning_rate": 2.9035466174152004e-05,
      "loss": 0.6414,
      "step": 5541
    },
    {
      "epoch": 0.9614850798056905,
      "grad_norm": 0.8414846062660217,
      "learning_rate": 2.90293873204365e-05,
      "loss": 0.5997,
      "step": 5542
    },
    {
      "epoch": 0.9616585704371964,
      "grad_norm": 0.5543221831321716,
      "learning_rate": 2.9023307418850074e-05,
      "loss": 0.6477,
      "step": 5543
    },
    {
      "epoch": 0.9618320610687023,
      "grad_norm": 0.8663203120231628,
      "learning_rate": 2.9017226470098307e-05,
      "loss": 0.6692,
      "step": 5544
    },
    {
      "epoch": 0.9620055517002082,
      "grad_norm": 0.7039609551429749,
      "learning_rate": 2.9011144474886905e-05,
      "loss": 0.5739,
      "step": 5545
    },
    {
      "epoch": 0.9621790423317141,
      "grad_norm": 0.7383800148963928,
      "learning_rate": 2.9005061433921685e-05,
      "loss": 0.5753,
      "step": 5546
    },
    {
      "epoch": 0.96235253296322,
      "grad_norm": 1.215936303138733,
      "learning_rate": 2.8998977347908593e-05,
      "loss": 0.6697,
      "step": 5547
    },
    {
      "epoch": 0.9625260235947258,
      "grad_norm": 0.8985759615898132,
      "learning_rate": 2.8992892217553693e-05,
      "loss": 0.5801,
      "step": 5548
    },
    {
      "epoch": 0.9626995142262318,
      "grad_norm": 0.7822864055633545,
      "learning_rate": 2.8986806043563174e-05,
      "loss": 0.5975,
      "step": 5549
    },
    {
      "epoch": 0.9628730048577376,
      "grad_norm": 0.5498910546302795,
      "learning_rate": 2.898071882664334e-05,
      "loss": 0.5688,
      "step": 5550
    },
    {
      "epoch": 0.9630464954892436,
      "grad_norm": 0.6806402802467346,
      "learning_rate": 2.8974630567500627e-05,
      "loss": 0.5627,
      "step": 5551
    },
    {
      "epoch": 0.9632199861207494,
      "grad_norm": 0.7067112326622009,
      "learning_rate": 2.896854126684157e-05,
      "loss": 0.6111,
      "step": 5552
    },
    {
      "epoch": 0.9633934767522554,
      "grad_norm": 0.7096337080001831,
      "learning_rate": 2.8962450925372855e-05,
      "loss": 0.5426,
      "step": 5553
    },
    {
      "epoch": 0.9635669673837612,
      "grad_norm": 0.6538836359977722,
      "learning_rate": 2.895635954380127e-05,
      "loss": 0.6331,
      "step": 5554
    },
    {
      "epoch": 0.9637404580152672,
      "grad_norm": 0.6250468492507935,
      "learning_rate": 2.8950267122833712e-05,
      "loss": 0.5671,
      "step": 5555
    },
    {
      "epoch": 0.963913948646773,
      "grad_norm": 0.6868553161621094,
      "learning_rate": 2.894417366317724e-05,
      "loss": 0.5541,
      "step": 5556
    },
    {
      "epoch": 0.964087439278279,
      "grad_norm": 1.0072047710418701,
      "learning_rate": 2.893807916553898e-05,
      "loss": 0.5559,
      "step": 5557
    },
    {
      "epoch": 0.9642609299097848,
      "grad_norm": 0.5556061267852783,
      "learning_rate": 2.8931983630626218e-05,
      "loss": 0.6761,
      "step": 5558
    },
    {
      "epoch": 0.9644344205412908,
      "grad_norm": 0.7940955758094788,
      "learning_rate": 2.8925887059146357e-05,
      "loss": 0.647,
      "step": 5559
    },
    {
      "epoch": 0.9646079111727967,
      "grad_norm": 0.6780730485916138,
      "learning_rate": 2.8919789451806893e-05,
      "loss": 0.5903,
      "step": 5560
    },
    {
      "epoch": 0.9647814018043026,
      "grad_norm": 0.7893752455711365,
      "learning_rate": 2.891369080931548e-05,
      "loss": 0.625,
      "step": 5561
    },
    {
      "epoch": 0.9649548924358085,
      "grad_norm": 0.701627254486084,
      "learning_rate": 2.890759113237985e-05,
      "loss": 0.5433,
      "step": 5562
    },
    {
      "epoch": 0.9651283830673144,
      "grad_norm": 0.5994738340377808,
      "learning_rate": 2.89014904217079e-05,
      "loss": 0.5035,
      "step": 5563
    },
    {
      "epoch": 0.9653018736988203,
      "grad_norm": 1.1059223413467407,
      "learning_rate": 2.8895388678007602e-05,
      "loss": 0.4935,
      "step": 5564
    },
    {
      "epoch": 0.9654753643303262,
      "grad_norm": 0.6889150142669678,
      "learning_rate": 2.8889285901987097e-05,
      "loss": 0.502,
      "step": 5565
    },
    {
      "epoch": 0.9656488549618321,
      "grad_norm": 0.5731109380722046,
      "learning_rate": 2.8883182094354594e-05,
      "loss": 0.5271,
      "step": 5566
    },
    {
      "epoch": 0.9658223455933379,
      "grad_norm": 0.678790271282196,
      "learning_rate": 2.8877077255818464e-05,
      "loss": 0.6252,
      "step": 5567
    },
    {
      "epoch": 0.9659958362248439,
      "grad_norm": 0.6979020237922668,
      "learning_rate": 2.887097138708717e-05,
      "loss": 0.5166,
      "step": 5568
    },
    {
      "epoch": 0.9661693268563497,
      "grad_norm": 0.8345228433609009,
      "learning_rate": 2.8864864488869314e-05,
      "loss": 0.5016,
      "step": 5569
    },
    {
      "epoch": 0.9663428174878557,
      "grad_norm": 0.6399042010307312,
      "learning_rate": 2.8858756561873605e-05,
      "loss": 0.5341,
      "step": 5570
    },
    {
      "epoch": 0.9665163081193615,
      "grad_norm": 0.7701651453971863,
      "learning_rate": 2.885264760680887e-05,
      "loss": 0.6294,
      "step": 5571
    },
    {
      "epoch": 0.9666897987508675,
      "grad_norm": 0.8324815630912781,
      "learning_rate": 2.884653762438407e-05,
      "loss": 0.5715,
      "step": 5572
    },
    {
      "epoch": 0.9668632893823733,
      "grad_norm": 1.1969701051712036,
      "learning_rate": 2.8840426615308267e-05,
      "loss": 0.6501,
      "step": 5573
    },
    {
      "epoch": 0.9670367800138793,
      "grad_norm": 0.6024868488311768,
      "learning_rate": 2.8834314580290655e-05,
      "loss": 0.5708,
      "step": 5574
    },
    {
      "epoch": 0.9672102706453851,
      "grad_norm": 0.8180156946182251,
      "learning_rate": 2.8828201520040546e-05,
      "loss": 0.5875,
      "step": 5575
    },
    {
      "epoch": 0.9673837612768911,
      "grad_norm": 0.7437762022018433,
      "learning_rate": 2.882208743526736e-05,
      "loss": 0.6156,
      "step": 5576
    },
    {
      "epoch": 0.9675572519083969,
      "grad_norm": 0.721993625164032,
      "learning_rate": 2.8815972326680648e-05,
      "loss": 0.4769,
      "step": 5577
    },
    {
      "epoch": 0.9677307425399029,
      "grad_norm": 0.56771320104599,
      "learning_rate": 2.8809856194990067e-05,
      "loss": 0.6461,
      "step": 5578
    },
    {
      "epoch": 0.9679042331714087,
      "grad_norm": 0.8299348950386047,
      "learning_rate": 2.880373904090542e-05,
      "loss": 0.4725,
      "step": 5579
    },
    {
      "epoch": 0.9680777238029147,
      "grad_norm": 0.789548933506012,
      "learning_rate": 2.8797620865136594e-05,
      "loss": 0.4939,
      "step": 5580
    },
    {
      "epoch": 0.9682512144344205,
      "grad_norm": 0.7346118688583374,
      "learning_rate": 2.8791501668393614e-05,
      "loss": 0.5712,
      "step": 5581
    },
    {
      "epoch": 0.9684247050659265,
      "grad_norm": 0.6299523711204529,
      "learning_rate": 2.8785381451386628e-05,
      "loss": 0.6372,
      "step": 5582
    },
    {
      "epoch": 0.9685981956974323,
      "grad_norm": 0.5474387407302856,
      "learning_rate": 2.8779260214825883e-05,
      "loss": 0.5952,
      "step": 5583
    },
    {
      "epoch": 0.9687716863289383,
      "grad_norm": 0.9798077940940857,
      "learning_rate": 2.877313795942176e-05,
      "loss": 0.5024,
      "step": 5584
    },
    {
      "epoch": 0.9689451769604441,
      "grad_norm": 0.5727953314781189,
      "learning_rate": 2.8767014685884755e-05,
      "loss": 0.6917,
      "step": 5585
    },
    {
      "epoch": 0.9691186675919501,
      "grad_norm": 0.7917117476463318,
      "learning_rate": 2.8760890394925477e-05,
      "loss": 0.5623,
      "step": 5586
    },
    {
      "epoch": 0.9692921582234559,
      "grad_norm": 0.8574097156524658,
      "learning_rate": 2.8754765087254665e-05,
      "loss": 0.4679,
      "step": 5587
    },
    {
      "epoch": 0.9694656488549618,
      "grad_norm": 3.6565754413604736,
      "learning_rate": 2.8748638763583158e-05,
      "loss": 0.6477,
      "step": 5588
    },
    {
      "epoch": 0.9696391394864677,
      "grad_norm": 0.7693340182304382,
      "learning_rate": 2.874251142462193e-05,
      "loss": 0.5785,
      "step": 5589
    },
    {
      "epoch": 0.9698126301179736,
      "grad_norm": 0.8621879816055298,
      "learning_rate": 2.8736383071082065e-05,
      "loss": 0.492,
      "step": 5590
    },
    {
      "epoch": 0.9699861207494795,
      "grad_norm": 0.7890018820762634,
      "learning_rate": 2.8730253703674765e-05,
      "loss": 0.5876,
      "step": 5591
    },
    {
      "epoch": 0.9701596113809854,
      "grad_norm": 0.5805262923240662,
      "learning_rate": 2.872412332311135e-05,
      "loss": 0.7037,
      "step": 5592
    },
    {
      "epoch": 0.9703331020124913,
      "grad_norm": 0.784298837184906,
      "learning_rate": 2.8717991930103258e-05,
      "loss": 0.5427,
      "step": 5593
    },
    {
      "epoch": 0.9705065926439972,
      "grad_norm": 0.8179328441619873,
      "learning_rate": 2.8711859525362045e-05,
      "loss": 0.469,
      "step": 5594
    },
    {
      "epoch": 0.9706800832755031,
      "grad_norm": 0.7602160573005676,
      "learning_rate": 2.8705726109599382e-05,
      "loss": 0.4801,
      "step": 5595
    },
    {
      "epoch": 0.970853573907009,
      "grad_norm": 1.113956332206726,
      "learning_rate": 2.8699591683527058e-05,
      "loss": 0.5945,
      "step": 5596
    },
    {
      "epoch": 0.971027064538515,
      "grad_norm": 1.1886241436004639,
      "learning_rate": 2.869345624785698e-05,
      "loss": 0.4926,
      "step": 5597
    },
    {
      "epoch": 0.9712005551700208,
      "grad_norm": 0.6510448455810547,
      "learning_rate": 2.8687319803301183e-05,
      "loss": 0.6802,
      "step": 5598
    },
    {
      "epoch": 0.9713740458015268,
      "grad_norm": 0.7982162237167358,
      "learning_rate": 2.8681182350571795e-05,
      "loss": 0.5955,
      "step": 5599
    },
    {
      "epoch": 0.9715475364330326,
      "grad_norm": 0.6104753017425537,
      "learning_rate": 2.867504389038108e-05,
      "loss": 0.5809,
      "step": 5600
    },
    {
      "epoch": 0.9717210270645386,
      "grad_norm": 0.6703321933746338,
      "learning_rate": 2.8668904423441413e-05,
      "loss": 0.5819,
      "step": 5601
    },
    {
      "epoch": 0.9718945176960444,
      "grad_norm": 0.7472213506698608,
      "learning_rate": 2.8662763950465284e-05,
      "loss": 0.4871,
      "step": 5602
    },
    {
      "epoch": 0.9720680083275504,
      "grad_norm": 0.7371354699134827,
      "learning_rate": 2.865662247216531e-05,
      "loss": 0.4873,
      "step": 5603
    },
    {
      "epoch": 0.9722414989590562,
      "grad_norm": 0.8960732817649841,
      "learning_rate": 2.8650479989254206e-05,
      "loss": 0.5057,
      "step": 5604
    },
    {
      "epoch": 0.9724149895905622,
      "grad_norm": 0.9224213361740112,
      "learning_rate": 2.864433650244482e-05,
      "loss": 0.4885,
      "step": 5605
    },
    {
      "epoch": 0.972588480222068,
      "grad_norm": 0.6924038529396057,
      "learning_rate": 2.8638192012450115e-05,
      "loss": 0.6155,
      "step": 5606
    },
    {
      "epoch": 0.9727619708535739,
      "grad_norm": 1.2647117376327515,
      "learning_rate": 2.8632046519983157e-05,
      "loss": 0.5897,
      "step": 5607
    },
    {
      "epoch": 0.9729354614850798,
      "grad_norm": 0.5417555570602417,
      "learning_rate": 2.862590002575714e-05,
      "loss": 0.5942,
      "step": 5608
    },
    {
      "epoch": 0.9731089521165857,
      "grad_norm": 1.6354655027389526,
      "learning_rate": 2.861975253048538e-05,
      "loss": 0.4634,
      "step": 5609
    },
    {
      "epoch": 0.9732824427480916,
      "grad_norm": 0.838060736656189,
      "learning_rate": 2.861360403488129e-05,
      "loss": 0.618,
      "step": 5610
    },
    {
      "epoch": 0.9734559333795975,
      "grad_norm": 0.8327847719192505,
      "learning_rate": 2.8607454539658417e-05,
      "loss": 0.4602,
      "step": 5611
    },
    {
      "epoch": 0.9736294240111034,
      "grad_norm": 0.6787186861038208,
      "learning_rate": 2.8601304045530414e-05,
      "loss": 0.5684,
      "step": 5612
    },
    {
      "epoch": 0.9738029146426093,
      "grad_norm": 0.8590366840362549,
      "learning_rate": 2.859515255321105e-05,
      "loss": 0.562,
      "step": 5613
    },
    {
      "epoch": 0.9739764052741152,
      "grad_norm": 0.8135871291160583,
      "learning_rate": 2.858900006341422e-05,
      "loss": 0.4685,
      "step": 5614
    },
    {
      "epoch": 0.9741498959056211,
      "grad_norm": 1.042456865310669,
      "learning_rate": 2.858284657685393e-05,
      "loss": 0.4919,
      "step": 5615
    },
    {
      "epoch": 0.974323386537127,
      "grad_norm": 0.6399260759353638,
      "learning_rate": 2.8576692094244286e-05,
      "loss": 0.5925,
      "step": 5616
    },
    {
      "epoch": 0.9744968771686329,
      "grad_norm": 0.6510165333747864,
      "learning_rate": 2.8570536616299538e-05,
      "loss": 0.5336,
      "step": 5617
    },
    {
      "epoch": 0.9746703678001388,
      "grad_norm": 0.6991555094718933,
      "learning_rate": 2.856438014373402e-05,
      "loss": 0.4651,
      "step": 5618
    },
    {
      "epoch": 0.9748438584316447,
      "grad_norm": 0.7268310189247131,
      "learning_rate": 2.855822267726222e-05,
      "loss": 0.5339,
      "step": 5619
    },
    {
      "epoch": 0.9750173490631506,
      "grad_norm": 0.6740028858184814,
      "learning_rate": 2.8552064217598694e-05,
      "loss": 0.524,
      "step": 5620
    },
    {
      "epoch": 0.9751908396946565,
      "grad_norm": 0.7472015023231506,
      "learning_rate": 2.8545904765458153e-05,
      "loss": 0.5776,
      "step": 5621
    },
    {
      "epoch": 0.9753643303261624,
      "grad_norm": 0.7645635604858398,
      "learning_rate": 2.853974432155541e-05,
      "loss": 0.5522,
      "step": 5622
    },
    {
      "epoch": 0.9755378209576683,
      "grad_norm": 0.8446869254112244,
      "learning_rate": 2.8533582886605383e-05,
      "loss": 0.6638,
      "step": 5623
    },
    {
      "epoch": 0.9757113115891742,
      "grad_norm": 0.760636568069458,
      "learning_rate": 2.852742046132312e-05,
      "loss": 0.5383,
      "step": 5624
    },
    {
      "epoch": 0.9758848022206801,
      "grad_norm": 0.7456691265106201,
      "learning_rate": 2.8521257046423782e-05,
      "loss": 0.6187,
      "step": 5625
    },
    {
      "epoch": 0.9760582928521859,
      "grad_norm": 0.6363016963005066,
      "learning_rate": 2.8515092642622625e-05,
      "loss": 0.6128,
      "step": 5626
    },
    {
      "epoch": 0.9762317834836919,
      "grad_norm": 0.6683642864227295,
      "learning_rate": 2.850892725063505e-05,
      "loss": 0.532,
      "step": 5627
    },
    {
      "epoch": 0.9764052741151977,
      "grad_norm": 1.0434778928756714,
      "learning_rate": 2.8502760871176546e-05,
      "loss": 0.6349,
      "step": 5628
    },
    {
      "epoch": 0.9765787647467037,
      "grad_norm": 0.871713399887085,
      "learning_rate": 2.849659350496275e-05,
      "loss": 0.5149,
      "step": 5629
    },
    {
      "epoch": 0.9767522553782095,
      "grad_norm": 0.6335962414741516,
      "learning_rate": 2.8490425152709367e-05,
      "loss": 0.6923,
      "step": 5630
    },
    {
      "epoch": 0.9769257460097155,
      "grad_norm": 0.8314769268035889,
      "learning_rate": 2.848425581513225e-05,
      "loss": 0.5887,
      "step": 5631
    },
    {
      "epoch": 0.9770992366412213,
      "grad_norm": 0.5725652575492859,
      "learning_rate": 2.847808549294736e-05,
      "loss": 0.6077,
      "step": 5632
    },
    {
      "epoch": 0.9772727272727273,
      "grad_norm": 1.0846760272979736,
      "learning_rate": 2.847191418687078e-05,
      "loss": 0.6786,
      "step": 5633
    },
    {
      "epoch": 0.9774462179042331,
      "grad_norm": 0.6418253779411316,
      "learning_rate": 2.8465741897618673e-05,
      "loss": 0.6667,
      "step": 5634
    },
    {
      "epoch": 0.9776197085357391,
      "grad_norm": 0.5402447581291199,
      "learning_rate": 2.845956862590736e-05,
      "loss": 0.5862,
      "step": 5635
    },
    {
      "epoch": 0.9777931991672449,
      "grad_norm": 0.6820518970489502,
      "learning_rate": 2.8453394372453253e-05,
      "loss": 0.589,
      "step": 5636
    },
    {
      "epoch": 0.9779666897987509,
      "grad_norm": 0.9091362357139587,
      "learning_rate": 2.844721913797287e-05,
      "loss": 0.6671,
      "step": 5637
    },
    {
      "epoch": 0.9781401804302567,
      "grad_norm": 0.7631445527076721,
      "learning_rate": 2.8441042923182872e-05,
      "loss": 0.5135,
      "step": 5638
    },
    {
      "epoch": 0.9783136710617627,
      "grad_norm": 1.4382708072662354,
      "learning_rate": 2.84348657288e-05,
      "loss": 0.5564,
      "step": 5639
    },
    {
      "epoch": 0.9784871616932685,
      "grad_norm": 0.7542216181755066,
      "learning_rate": 2.842868755554114e-05,
      "loss": 0.4821,
      "step": 5640
    },
    {
      "epoch": 0.9786606523247745,
      "grad_norm": 0.823925256729126,
      "learning_rate": 2.8422508404123264e-05,
      "loss": 0.46,
      "step": 5641
    },
    {
      "epoch": 0.9788341429562804,
      "grad_norm": 0.7728753089904785,
      "learning_rate": 2.8416328275263472e-05,
      "loss": 0.5385,
      "step": 5642
    },
    {
      "epoch": 0.9790076335877863,
      "grad_norm": 0.7813199758529663,
      "learning_rate": 2.841014716967898e-05,
      "loss": 0.6272,
      "step": 5643
    },
    {
      "epoch": 0.9791811242192922,
      "grad_norm": 0.7330827713012695,
      "learning_rate": 2.8403965088087105e-05,
      "loss": 0.6547,
      "step": 5644
    },
    {
      "epoch": 0.9793546148507981,
      "grad_norm": 0.7045038342475891,
      "learning_rate": 2.8397782031205295e-05,
      "loss": 0.4763,
      "step": 5645
    },
    {
      "epoch": 0.979528105482304,
      "grad_norm": 1.028846263885498,
      "learning_rate": 2.839159799975109e-05,
      "loss": 0.6293,
      "step": 5646
    },
    {
      "epoch": 0.9797015961138098,
      "grad_norm": 0.7624039649963379,
      "learning_rate": 2.838541299444216e-05,
      "loss": 0.5862,
      "step": 5647
    },
    {
      "epoch": 0.9798750867453158,
      "grad_norm": 0.6337740421295166,
      "learning_rate": 2.8379227015996283e-05,
      "loss": 0.5353,
      "step": 5648
    },
    {
      "epoch": 0.9800485773768216,
      "grad_norm": 0.6328892111778259,
      "learning_rate": 2.837304006513135e-05,
      "loss": 0.6538,
      "step": 5649
    },
    {
      "epoch": 0.9802220680083276,
      "grad_norm": 1.041323184967041,
      "learning_rate": 2.8366852142565352e-05,
      "loss": 0.575,
      "step": 5650
    },
    {
      "epoch": 0.9803955586398334,
      "grad_norm": 0.763387143611908,
      "learning_rate": 2.8360663249016417e-05,
      "loss": 0.5221,
      "step": 5651
    },
    {
      "epoch": 0.9805690492713394,
      "grad_norm": 0.7254363298416138,
      "learning_rate": 2.8354473385202772e-05,
      "loss": 0.584,
      "step": 5652
    },
    {
      "epoch": 0.9807425399028452,
      "grad_norm": 0.6395940780639648,
      "learning_rate": 2.8348282551842756e-05,
      "loss": 0.6965,
      "step": 5653
    },
    {
      "epoch": 0.9809160305343512,
      "grad_norm": 1.0305383205413818,
      "learning_rate": 2.834209074965482e-05,
      "loss": 0.5254,
      "step": 5654
    },
    {
      "epoch": 0.981089521165857,
      "grad_norm": 1.0242565870285034,
      "learning_rate": 2.833589797935753e-05,
      "loss": 0.4504,
      "step": 5655
    },
    {
      "epoch": 0.981263011797363,
      "grad_norm": 0.6797073483467102,
      "learning_rate": 2.8329704241669574e-05,
      "loss": 0.6068,
      "step": 5656
    },
    {
      "epoch": 0.9814365024288688,
      "grad_norm": 0.5994985699653625,
      "learning_rate": 2.8323509537309725e-05,
      "loss": 0.5944,
      "step": 5657
    },
    {
      "epoch": 0.9816099930603748,
      "grad_norm": 0.935546338558197,
      "learning_rate": 2.8317313866996897e-05,
      "loss": 0.4546,
      "step": 5658
    },
    {
      "epoch": 0.9817834836918806,
      "grad_norm": 0.7284477353096008,
      "learning_rate": 2.831111723145011e-05,
      "loss": 0.5553,
      "step": 5659
    },
    {
      "epoch": 0.9819569743233866,
      "grad_norm": 0.919700026512146,
      "learning_rate": 2.830491963138848e-05,
      "loss": 0.491,
      "step": 5660
    },
    {
      "epoch": 0.9821304649548924,
      "grad_norm": 0.7011570334434509,
      "learning_rate": 2.8298721067531248e-05,
      "loss": 0.5303,
      "step": 5661
    },
    {
      "epoch": 0.9823039555863984,
      "grad_norm": 1.7850123643875122,
      "learning_rate": 2.8292521540597767e-05,
      "loss": 0.5337,
      "step": 5662
    },
    {
      "epoch": 0.9824774462179042,
      "grad_norm": 0.6571459770202637,
      "learning_rate": 2.8286321051307494e-05,
      "loss": 0.4933,
      "step": 5663
    },
    {
      "epoch": 0.9826509368494102,
      "grad_norm": 0.6553524136543274,
      "learning_rate": 2.828011960038002e-05,
      "loss": 0.5347,
      "step": 5664
    },
    {
      "epoch": 0.982824427480916,
      "grad_norm": 0.7700760960578918,
      "learning_rate": 2.827391718853501e-05,
      "loss": 0.4912,
      "step": 5665
    },
    {
      "epoch": 0.9829979181124219,
      "grad_norm": 1.0490473508834839,
      "learning_rate": 2.826771381649227e-05,
      "loss": 0.5784,
      "step": 5666
    },
    {
      "epoch": 0.9831714087439278,
      "grad_norm": 0.749843418598175,
      "learning_rate": 2.826150948497171e-05,
      "loss": 0.4961,
      "step": 5667
    },
    {
      "epoch": 0.9833448993754337,
      "grad_norm": 0.8119151592254639,
      "learning_rate": 2.8255304194693343e-05,
      "loss": 0.5551,
      "step": 5668
    },
    {
      "epoch": 0.9835183900069396,
      "grad_norm": 0.6026073098182678,
      "learning_rate": 2.8249097946377307e-05,
      "loss": 0.5271,
      "step": 5669
    },
    {
      "epoch": 0.9836918806384455,
      "grad_norm": 0.6521915793418884,
      "learning_rate": 2.8242890740743844e-05,
      "loss": 0.5321,
      "step": 5670
    },
    {
      "epoch": 0.9838653712699514,
      "grad_norm": 0.8184699416160583,
      "learning_rate": 2.8236682578513302e-05,
      "loss": 0.6404,
      "step": 5671
    },
    {
      "epoch": 0.9840388619014573,
      "grad_norm": 0.5834811925888062,
      "learning_rate": 2.8230473460406154e-05,
      "loss": 0.6033,
      "step": 5672
    },
    {
      "epoch": 0.9842123525329632,
      "grad_norm": 1.45811128616333,
      "learning_rate": 2.8224263387142963e-05,
      "loss": 0.5653,
      "step": 5673
    },
    {
      "epoch": 0.9843858431644691,
      "grad_norm": 0.9198524355888367,
      "learning_rate": 2.8218052359444434e-05,
      "loss": 0.556,
      "step": 5674
    },
    {
      "epoch": 0.984559333795975,
      "grad_norm": 0.6337687969207764,
      "learning_rate": 2.821184037803135e-05,
      "loss": 0.5148,
      "step": 5675
    },
    {
      "epoch": 0.9847328244274809,
      "grad_norm": 1.02034330368042,
      "learning_rate": 2.8205627443624616e-05,
      "loss": 0.4585,
      "step": 5676
    },
    {
      "epoch": 0.9849063150589868,
      "grad_norm": 0.5336766839027405,
      "learning_rate": 2.8199413556945256e-05,
      "loss": 0.6122,
      "step": 5677
    },
    {
      "epoch": 0.9850798056904927,
      "grad_norm": 0.7545201182365417,
      "learning_rate": 2.8193198718714402e-05,
      "loss": 0.52,
      "step": 5678
    },
    {
      "epoch": 0.9852532963219987,
      "grad_norm": 0.6611053943634033,
      "learning_rate": 2.8186982929653287e-05,
      "loss": 0.5245,
      "step": 5679
    },
    {
      "epoch": 0.9854267869535045,
      "grad_norm": 0.827336311340332,
      "learning_rate": 2.8180766190483263e-05,
      "loss": 0.535,
      "step": 5680
    },
    {
      "epoch": 0.9856002775850105,
      "grad_norm": 0.975527286529541,
      "learning_rate": 2.817454850192579e-05,
      "loss": 0.5465,
      "step": 5681
    },
    {
      "epoch": 0.9857737682165163,
      "grad_norm": 0.9336392879486084,
      "learning_rate": 2.8168329864702443e-05,
      "loss": 0.5498,
      "step": 5682
    },
    {
      "epoch": 0.9859472588480223,
      "grad_norm": 0.7438031435012817,
      "learning_rate": 2.8162110279534893e-05,
      "loss": 0.6061,
      "step": 5683
    },
    {
      "epoch": 0.9861207494795281,
      "grad_norm": 0.5993937253952026,
      "learning_rate": 2.8155889747144933e-05,
      "loss": 0.6034,
      "step": 5684
    },
    {
      "epoch": 0.9862942401110341,
      "grad_norm": 0.5549498200416565,
      "learning_rate": 2.8149668268254465e-05,
      "loss": 0.6147,
      "step": 5685
    },
    {
      "epoch": 0.9864677307425399,
      "grad_norm": 1.1180832386016846,
      "learning_rate": 2.8143445843585498e-05,
      "loss": 0.5997,
      "step": 5686
    },
    {
      "epoch": 0.9866412213740458,
      "grad_norm": 0.8614696860313416,
      "learning_rate": 2.8137222473860154e-05,
      "loss": 0.6512,
      "step": 5687
    },
    {
      "epoch": 0.9868147120055517,
      "grad_norm": 0.5003150105476379,
      "learning_rate": 2.8130998159800663e-05,
      "loss": 0.7042,
      "step": 5688
    },
    {
      "epoch": 0.9869882026370576,
      "grad_norm": 0.699743390083313,
      "learning_rate": 2.8124772902129353e-05,
      "loss": 0.6189,
      "step": 5689
    },
    {
      "epoch": 0.9871616932685635,
      "grad_norm": 0.646152913570404,
      "learning_rate": 2.8118546701568687e-05,
      "loss": 0.5361,
      "step": 5690
    },
    {
      "epoch": 0.9873351839000694,
      "grad_norm": 0.6917030811309814,
      "learning_rate": 2.8112319558841216e-05,
      "loss": 0.6029,
      "step": 5691
    },
    {
      "epoch": 0.9875086745315753,
      "grad_norm": 0.9531429409980774,
      "learning_rate": 2.81060914746696e-05,
      "loss": 0.4884,
      "step": 5692
    },
    {
      "epoch": 0.9876821651630812,
      "grad_norm": 0.6426189541816711,
      "learning_rate": 2.8099862449776637e-05,
      "loss": 0.5273,
      "step": 5693
    },
    {
      "epoch": 0.9878556557945871,
      "grad_norm": 0.6073868274688721,
      "learning_rate": 2.8093632484885182e-05,
      "loss": 0.576,
      "step": 5694
    },
    {
      "epoch": 0.988029146426093,
      "grad_norm": 0.6698435544967651,
      "learning_rate": 2.8087401580718258e-05,
      "loss": 0.4551,
      "step": 5695
    },
    {
      "epoch": 0.9882026370575989,
      "grad_norm": 0.7084513306617737,
      "learning_rate": 2.8081169737998956e-05,
      "loss": 0.5062,
      "step": 5696
    },
    {
      "epoch": 0.9883761276891048,
      "grad_norm": 0.8095421195030212,
      "learning_rate": 2.8074936957450485e-05,
      "loss": 0.5396,
      "step": 5697
    },
    {
      "epoch": 0.9885496183206107,
      "grad_norm": 0.8599531650543213,
      "learning_rate": 2.8068703239796175e-05,
      "loss": 0.5001,
      "step": 5698
    },
    {
      "epoch": 0.9887231089521166,
      "grad_norm": 0.5939335823059082,
      "learning_rate": 2.806246858575945e-05,
      "loss": 0.5999,
      "step": 5699
    },
    {
      "epoch": 0.9888965995836225,
      "grad_norm": 0.7356471419334412,
      "learning_rate": 2.805623299606385e-05,
      "loss": 0.5591,
      "step": 5700
    },
    {
      "epoch": 0.9890700902151284,
      "grad_norm": 0.6846799850463867,
      "learning_rate": 2.8049996471433022e-05,
      "loss": 0.5365,
      "step": 5701
    },
    {
      "epoch": 0.9892435808466343,
      "grad_norm": 1.5366616249084473,
      "learning_rate": 2.8043759012590723e-05,
      "loss": 0.603,
      "step": 5702
    },
    {
      "epoch": 0.9894170714781402,
      "grad_norm": 0.6540420055389404,
      "learning_rate": 2.8037520620260826e-05,
      "loss": 0.6267,
      "step": 5703
    },
    {
      "epoch": 0.9895905621096461,
      "grad_norm": 0.7610870599746704,
      "learning_rate": 2.803128129516729e-05,
      "loss": 0.4932,
      "step": 5704
    },
    {
      "epoch": 0.989764052741152,
      "grad_norm": 0.8002392053604126,
      "learning_rate": 2.8025041038034197e-05,
      "loss": 0.5995,
      "step": 5705
    },
    {
      "epoch": 0.9899375433726578,
      "grad_norm": 0.940723180770874,
      "learning_rate": 2.801879984958575e-05,
      "loss": 0.4882,
      "step": 5706
    },
    {
      "epoch": 0.9901110340041638,
      "grad_norm": 0.7501986026763916,
      "learning_rate": 2.8012557730546224e-05,
      "loss": 0.5312,
      "step": 5707
    },
    {
      "epoch": 0.9902845246356696,
      "grad_norm": 0.745395839214325,
      "learning_rate": 2.800631468164005e-05,
      "loss": 0.5092,
      "step": 5708
    },
    {
      "epoch": 0.9904580152671756,
      "grad_norm": 0.9255587458610535,
      "learning_rate": 2.800007070359172e-05,
      "loss": 0.5626,
      "step": 5709
    },
    {
      "epoch": 0.9906315058986814,
      "grad_norm": 0.6927536725997925,
      "learning_rate": 2.7993825797125866e-05,
      "loss": 0.5692,
      "step": 5710
    },
    {
      "epoch": 0.9908049965301874,
      "grad_norm": 0.8264254927635193,
      "learning_rate": 2.798757996296721e-05,
      "loss": 0.5314,
      "step": 5711
    },
    {
      "epoch": 0.9909784871616932,
      "grad_norm": 0.7039380073547363,
      "learning_rate": 2.7981333201840595e-05,
      "loss": 0.547,
      "step": 5712
    },
    {
      "epoch": 0.9911519777931992,
      "grad_norm": 0.7396357655525208,
      "learning_rate": 2.7975085514470958e-05,
      "loss": 0.5565,
      "step": 5713
    },
    {
      "epoch": 0.991325468424705,
      "grad_norm": 0.8510541319847107,
      "learning_rate": 2.7968836901583364e-05,
      "loss": 0.5774,
      "step": 5714
    },
    {
      "epoch": 0.991498959056211,
      "grad_norm": 0.5891192555427551,
      "learning_rate": 2.7962587363902952e-05,
      "loss": 0.6868,
      "step": 5715
    },
    {
      "epoch": 0.9916724496877168,
      "grad_norm": 0.6883437633514404,
      "learning_rate": 2.7956336902155003e-05,
      "loss": 0.6849,
      "step": 5716
    },
    {
      "epoch": 0.9918459403192228,
      "grad_norm": 0.8984211683273315,
      "learning_rate": 2.7950085517064884e-05,
      "loss": 0.5027,
      "step": 5717
    },
    {
      "epoch": 0.9920194309507286,
      "grad_norm": 0.796939492225647,
      "learning_rate": 2.7943833209358076e-05,
      "loss": 0.644,
      "step": 5718
    },
    {
      "epoch": 0.9921929215822346,
      "grad_norm": 0.7189422845840454,
      "learning_rate": 2.793757997976017e-05,
      "loss": 0.5895,
      "step": 5719
    },
    {
      "epoch": 0.9923664122137404,
      "grad_norm": 0.780312180519104,
      "learning_rate": 2.793132582899686e-05,
      "loss": 0.5587,
      "step": 5720
    },
    {
      "epoch": 0.9925399028452464,
      "grad_norm": 0.8351185321807861,
      "learning_rate": 2.7925070757793943e-05,
      "loss": 0.5417,
      "step": 5721
    },
    {
      "epoch": 0.9927133934767522,
      "grad_norm": 4.6577019691467285,
      "learning_rate": 2.791881476687733e-05,
      "loss": 0.5024,
      "step": 5722
    },
    {
      "epoch": 0.9928868841082582,
      "grad_norm": 0.6500504016876221,
      "learning_rate": 2.7912557856973035e-05,
      "loss": 0.5898,
      "step": 5723
    },
    {
      "epoch": 0.993060374739764,
      "grad_norm": 0.8028048276901245,
      "learning_rate": 2.790630002880718e-05,
      "loss": 0.67,
      "step": 5724
    },
    {
      "epoch": 0.9932338653712699,
      "grad_norm": 0.9770735502243042,
      "learning_rate": 2.7900041283106e-05,
      "loss": 0.5446,
      "step": 5725
    },
    {
      "epoch": 0.9934073560027759,
      "grad_norm": 0.9949390888214111,
      "learning_rate": 2.7893781620595818e-05,
      "loss": 0.5074,
      "step": 5726
    },
    {
      "epoch": 0.9935808466342817,
      "grad_norm": 0.6422823667526245,
      "learning_rate": 2.7887521042003084e-05,
      "loss": 0.588,
      "step": 5727
    },
    {
      "epoch": 0.9937543372657877,
      "grad_norm": 0.697289764881134,
      "learning_rate": 2.788125954805434e-05,
      "loss": 0.5432,
      "step": 5728
    },
    {
      "epoch": 0.9939278278972935,
      "grad_norm": 0.6521013975143433,
      "learning_rate": 2.787499713947624e-05,
      "loss": 0.559,
      "step": 5729
    },
    {
      "epoch": 0.9941013185287995,
      "grad_norm": 0.7985821962356567,
      "learning_rate": 2.7868733816995553e-05,
      "loss": 0.6085,
      "step": 5730
    },
    {
      "epoch": 0.9942748091603053,
      "grad_norm": 0.817812979221344,
      "learning_rate": 2.786246958133913e-05,
      "loss": 0.5981,
      "step": 5731
    },
    {
      "epoch": 0.9944482997918113,
      "grad_norm": 0.7825731635093689,
      "learning_rate": 2.7856204433233954e-05,
      "loss": 0.4961,
      "step": 5732
    },
    {
      "epoch": 0.9946217904233171,
      "grad_norm": 0.6842442154884338,
      "learning_rate": 2.7849938373407095e-05,
      "loss": 0.5099,
      "step": 5733
    },
    {
      "epoch": 0.9947952810548231,
      "grad_norm": 1.0025458335876465,
      "learning_rate": 2.7843671402585747e-05,
      "loss": 0.5641,
      "step": 5734
    },
    {
      "epoch": 0.9949687716863289,
      "grad_norm": 0.9568683505058289,
      "learning_rate": 2.783740352149719e-05,
      "loss": 0.5083,
      "step": 5735
    },
    {
      "epoch": 0.9951422623178349,
      "grad_norm": 0.6607463955879211,
      "learning_rate": 2.783113473086882e-05,
      "loss": 0.499,
      "step": 5736
    },
    {
      "epoch": 0.9953157529493407,
      "grad_norm": 0.6710206866264343,
      "learning_rate": 2.7824865031428144e-05,
      "loss": 0.5931,
      "step": 5737
    },
    {
      "epoch": 0.9954892435808467,
      "grad_norm": 0.637427806854248,
      "learning_rate": 2.781859442390276e-05,
      "loss": 0.4478,
      "step": 5738
    },
    {
      "epoch": 0.9956627342123525,
      "grad_norm": 1.1940019130706787,
      "learning_rate": 2.7812322909020385e-05,
      "loss": 0.5146,
      "step": 5739
    },
    {
      "epoch": 0.9958362248438585,
      "grad_norm": 0.8191688060760498,
      "learning_rate": 2.780605048750883e-05,
      "loss": 0.5422,
      "step": 5740
    },
    {
      "epoch": 0.9960097154753643,
      "grad_norm": 0.7413262724876404,
      "learning_rate": 2.7799777160096025e-05,
      "loss": 0.5463,
      "step": 5741
    },
    {
      "epoch": 0.9961832061068703,
      "grad_norm": 0.672318160533905,
      "learning_rate": 2.7793502927509988e-05,
      "loss": 0.6199,
      "step": 5742
    },
    {
      "epoch": 0.9963566967383761,
      "grad_norm": 0.687721848487854,
      "learning_rate": 2.7787227790478856e-05,
      "loss": 0.5553,
      "step": 5743
    },
    {
      "epoch": 0.9965301873698821,
      "grad_norm": 0.7222127318382263,
      "learning_rate": 2.7780951749730864e-05,
      "loss": 0.589,
      "step": 5744
    },
    {
      "epoch": 0.9967036780013879,
      "grad_norm": 0.6709451079368591,
      "learning_rate": 2.7774674805994356e-05,
      "loss": 0.5746,
      "step": 5745
    },
    {
      "epoch": 0.9968771686328938,
      "grad_norm": 0.7386717200279236,
      "learning_rate": 2.7768396959997783e-05,
      "loss": 0.5798,
      "step": 5746
    },
    {
      "epoch": 0.9970506592643997,
      "grad_norm": 0.6941803097724915,
      "learning_rate": 2.7762118212469686e-05,
      "loss": 0.5363,
      "step": 5747
    },
    {
      "epoch": 0.9972241498959056,
      "grad_norm": 0.6951861381530762,
      "learning_rate": 2.7755838564138722e-05,
      "loss": 0.6049,
      "step": 5748
    },
    {
      "epoch": 0.9973976405274115,
      "grad_norm": 0.812441349029541,
      "learning_rate": 2.774955801573366e-05,
      "loss": 0.5265,
      "step": 5749
    },
    {
      "epoch": 0.9975711311589174,
      "grad_norm": 0.7692736983299255,
      "learning_rate": 2.7743276567983354e-05,
      "loss": 0.5095,
      "step": 5750
    },
    {
      "epoch": 0.9977446217904233,
      "grad_norm": 0.8522500395774841,
      "learning_rate": 2.7736994221616788e-05,
      "loss": 0.4911,
      "step": 5751
    },
    {
      "epoch": 0.9979181124219292,
      "grad_norm": 0.6917999386787415,
      "learning_rate": 2.7730710977363023e-05,
      "loss": 0.4574,
      "step": 5752
    },
    {
      "epoch": 0.9980916030534351,
      "grad_norm": 0.7634033560752869,
      "learning_rate": 2.7724426835951242e-05,
      "loss": 0.5029,
      "step": 5753
    },
    {
      "epoch": 0.998265093684941,
      "grad_norm": 0.9081072807312012,
      "learning_rate": 2.771814179811073e-05,
      "loss": 0.5377,
      "step": 5754
    },
    {
      "epoch": 0.9984385843164469,
      "grad_norm": 1.0557156801223755,
      "learning_rate": 2.7711855864570858e-05,
      "loss": 0.543,
      "step": 5755
    },
    {
      "epoch": 0.9986120749479528,
      "grad_norm": 1.009839415550232,
      "learning_rate": 2.7705569036061137e-05,
      "loss": 0.535,
      "step": 5756
    },
    {
      "epoch": 0.9987855655794587,
      "grad_norm": 0.6140355467796326,
      "learning_rate": 2.7699281313311144e-05,
      "loss": 0.4337,
      "step": 5757
    },
    {
      "epoch": 0.9989590562109646,
      "grad_norm": 0.7017039060592651,
      "learning_rate": 2.7692992697050587e-05,
      "loss": 0.5054,
      "step": 5758
    },
    {
      "epoch": 0.9991325468424705,
      "grad_norm": 0.6694011688232422,
      "learning_rate": 2.768670318800926e-05,
      "loss": 0.674,
      "step": 5759
    },
    {
      "epoch": 0.9993060374739764,
      "grad_norm": 0.9655817151069641,
      "learning_rate": 2.7680412786917074e-05,
      "loss": 0.5253,
      "step": 5760
    },
    {
      "epoch": 0.9994795281054824,
      "grad_norm": 0.8425588607788086,
      "learning_rate": 2.7674121494504032e-05,
      "loss": 0.5544,
      "step": 5761
    },
    {
      "epoch": 0.9996530187369882,
      "grad_norm": 0.7655010223388672,
      "learning_rate": 2.7667829311500255e-05,
      "loss": 0.6002,
      "step": 5762
    },
    {
      "epoch": 0.9998265093684942,
      "grad_norm": 0.8534713983535767,
      "learning_rate": 2.766153623863594e-05,
      "loss": 0.5909,
      "step": 5763
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6403183937072754,
      "learning_rate": 2.765524227664143e-05,
      "loss": 0.4885,
      "step": 5764
    },
    {
      "epoch": 1.0001734906315058,
      "grad_norm": 0.7058083415031433,
      "learning_rate": 2.7648947426247122e-05,
      "loss": 0.6553,
      "step": 5765
    },
    {
      "epoch": 1.0003469812630117,
      "grad_norm": 0.8651044368743896,
      "learning_rate": 2.7642651688183558e-05,
      "loss": 0.5359,
      "step": 5766
    },
    {
      "epoch": 1.0005204718945178,
      "grad_norm": 0.7307851910591125,
      "learning_rate": 2.763635506318137e-05,
      "loss": 0.4159,
      "step": 5767
    },
    {
      "epoch": 1.0006939625260236,
      "grad_norm": 0.5823991298675537,
      "learning_rate": 2.763005755197126e-05,
      "loss": 0.4789,
      "step": 5768
    },
    {
      "epoch": 1.0008674531575295,
      "grad_norm": 0.9836352467536926,
      "learning_rate": 2.7623759155284093e-05,
      "loss": 0.463,
      "step": 5769
    },
    {
      "epoch": 1.0010409437890353,
      "grad_norm": 1.1837478876113892,
      "learning_rate": 2.7617459873850792e-05,
      "loss": 0.6111,
      "step": 5770
    },
    {
      "epoch": 1.0012144344205414,
      "grad_norm": 0.6720978617668152,
      "learning_rate": 2.7611159708402387e-05,
      "loss": 0.5699,
      "step": 5771
    },
    {
      "epoch": 1.0013879250520472,
      "grad_norm": 0.6235537528991699,
      "learning_rate": 2.760485865967004e-05,
      "loss": 0.6168,
      "step": 5772
    },
    {
      "epoch": 1.001561415683553,
      "grad_norm": 0.8067814111709595,
      "learning_rate": 2.759855672838498e-05,
      "loss": 0.3896,
      "step": 5773
    },
    {
      "epoch": 1.001734906315059,
      "grad_norm": 1.0587890148162842,
      "learning_rate": 2.7592253915278556e-05,
      "loss": 0.5634,
      "step": 5774
    },
    {
      "epoch": 1.001908396946565,
      "grad_norm": 0.7729614973068237,
      "learning_rate": 2.7585950221082223e-05,
      "loss": 0.5593,
      "step": 5775
    },
    {
      "epoch": 1.0020818875780708,
      "grad_norm": 0.7301870584487915,
      "learning_rate": 2.7579645646527522e-05,
      "loss": 0.4178,
      "step": 5776
    },
    {
      "epoch": 1.0022553782095767,
      "grad_norm": 0.9195185303688049,
      "learning_rate": 2.7573340192346117e-05,
      "loss": 0.5452,
      "step": 5777
    },
    {
      "epoch": 1.0024288688410825,
      "grad_norm": 1.3216475248336792,
      "learning_rate": 2.7567033859269754e-05,
      "loss": 0.4032,
      "step": 5778
    },
    {
      "epoch": 1.0026023594725886,
      "grad_norm": 0.8470260500907898,
      "learning_rate": 2.7560726648030294e-05,
      "loss": 0.448,
      "step": 5779
    },
    {
      "epoch": 1.0027758501040944,
      "grad_norm": 0.721225917339325,
      "learning_rate": 2.75544185593597e-05,
      "loss": 0.4741,
      "step": 5780
    },
    {
      "epoch": 1.0029493407356003,
      "grad_norm": 0.7026661038398743,
      "learning_rate": 2.7548109593990022e-05,
      "loss": 0.5828,
      "step": 5781
    },
    {
      "epoch": 1.0031228313671061,
      "grad_norm": 1.6201528310775757,
      "learning_rate": 2.754179975265344e-05,
      "loss": 0.6151,
      "step": 5782
    },
    {
      "epoch": 1.0032963219986122,
      "grad_norm": 0.8122080564498901,
      "learning_rate": 2.7535489036082198e-05,
      "loss": 0.4321,
      "step": 5783
    },
    {
      "epoch": 1.003469812630118,
      "grad_norm": 1.0058845281600952,
      "learning_rate": 2.752917744500868e-05,
      "loss": 0.5387,
      "step": 5784
    },
    {
      "epoch": 1.0036433032616239,
      "grad_norm": 0.8528434634208679,
      "learning_rate": 2.7522864980165346e-05,
      "loss": 0.6672,
      "step": 5785
    },
    {
      "epoch": 1.0038167938931297,
      "grad_norm": 1.0794541835784912,
      "learning_rate": 2.7516551642284765e-05,
      "loss": 0.5281,
      "step": 5786
    },
    {
      "epoch": 1.0039902845246356,
      "grad_norm": 1.606688380241394,
      "learning_rate": 2.7510237432099605e-05,
      "loss": 0.5308,
      "step": 5787
    },
    {
      "epoch": 1.0041637751561416,
      "grad_norm": 0.6467757225036621,
      "learning_rate": 2.7503922350342645e-05,
      "loss": 0.5367,
      "step": 5788
    },
    {
      "epoch": 1.0043372657876475,
      "grad_norm": 0.7525832056999207,
      "learning_rate": 2.7497606397746745e-05,
      "loss": 0.5387,
      "step": 5789
    },
    {
      "epoch": 1.0045107564191533,
      "grad_norm": 0.8825986981391907,
      "learning_rate": 2.7491289575044893e-05,
      "loss": 0.514,
      "step": 5790
    },
    {
      "epoch": 1.0046842470506592,
      "grad_norm": 0.8700129389762878,
      "learning_rate": 2.7484971882970156e-05,
      "loss": 0.521,
      "step": 5791
    },
    {
      "epoch": 1.0048577376821652,
      "grad_norm": 0.7932398319244385,
      "learning_rate": 2.7478653322255707e-05,
      "loss": 0.54,
      "step": 5792
    },
    {
      "epoch": 1.005031228313671,
      "grad_norm": 0.6790376305580139,
      "learning_rate": 2.7472333893634824e-05,
      "loss": 0.4625,
      "step": 5793
    },
    {
      "epoch": 1.005204718945177,
      "grad_norm": 0.6960176229476929,
      "learning_rate": 2.746601359784089e-05,
      "loss": 0.5845,
      "step": 5794
    },
    {
      "epoch": 1.0053782095766828,
      "grad_norm": 0.7734279036521912,
      "learning_rate": 2.7459692435607376e-05,
      "loss": 0.5856,
      "step": 5795
    },
    {
      "epoch": 1.0055517002081888,
      "grad_norm": 0.7826294302940369,
      "learning_rate": 2.745337040766787e-05,
      "loss": 0.4456,
      "step": 5796
    },
    {
      "epoch": 1.0057251908396947,
      "grad_norm": 0.6511348485946655,
      "learning_rate": 2.7447047514756032e-05,
      "loss": 0.5326,
      "step": 5797
    },
    {
      "epoch": 1.0058986814712005,
      "grad_norm": 0.9289817214012146,
      "learning_rate": 2.744072375760566e-05,
      "loss": 0.406,
      "step": 5798
    },
    {
      "epoch": 1.0060721721027064,
      "grad_norm": 0.8878532648086548,
      "learning_rate": 2.7434399136950625e-05,
      "loss": 0.4139,
      "step": 5799
    },
    {
      "epoch": 1.0062456627342125,
      "grad_norm": 0.8880317807197571,
      "learning_rate": 2.74280736535249e-05,
      "loss": 0.4412,
      "step": 5800
    },
    {
      "epoch": 1.0064191533657183,
      "grad_norm": 0.7840941548347473,
      "learning_rate": 2.742174730806258e-05,
      "loss": 0.4554,
      "step": 5801
    },
    {
      "epoch": 1.0065926439972241,
      "grad_norm": 0.8049899339675903,
      "learning_rate": 2.7415420101297836e-05,
      "loss": 0.4913,
      "step": 5802
    },
    {
      "epoch": 1.00676613462873,
      "grad_norm": 0.8128389716148376,
      "learning_rate": 2.7409092033964943e-05,
      "loss": 0.6445,
      "step": 5803
    },
    {
      "epoch": 1.0069396252602358,
      "grad_norm": 0.9518760442733765,
      "learning_rate": 2.7402763106798295e-05,
      "loss": 0.5773,
      "step": 5804
    },
    {
      "epoch": 1.007113115891742,
      "grad_norm": 0.845658004283905,
      "learning_rate": 2.7396433320532356e-05,
      "loss": 0.5273,
      "step": 5805
    },
    {
      "epoch": 1.0072866065232478,
      "grad_norm": 0.7681678533554077,
      "learning_rate": 2.739010267590171e-05,
      "loss": 0.5714,
      "step": 5806
    },
    {
      "epoch": 1.0074600971547536,
      "grad_norm": 1.1148523092269897,
      "learning_rate": 2.7383771173641037e-05,
      "loss": 0.4296,
      "step": 5807
    },
    {
      "epoch": 1.0076335877862594,
      "grad_norm": 0.7010675072669983,
      "learning_rate": 2.7377438814485117e-05,
      "loss": 0.4215,
      "step": 5808
    },
    {
      "epoch": 1.0078070784177655,
      "grad_norm": 0.8457129001617432,
      "learning_rate": 2.7371105599168833e-05,
      "loss": 0.5698,
      "step": 5809
    },
    {
      "epoch": 1.0079805690492714,
      "grad_norm": 0.8173154592514038,
      "learning_rate": 2.7364771528427145e-05,
      "loss": 0.5417,
      "step": 5810
    },
    {
      "epoch": 1.0081540596807772,
      "grad_norm": 0.8678355813026428,
      "learning_rate": 2.735843660299515e-05,
      "loss": 0.5249,
      "step": 5811
    },
    {
      "epoch": 1.008327550312283,
      "grad_norm": 0.7418676614761353,
      "learning_rate": 2.7352100823608006e-05,
      "loss": 0.5223,
      "step": 5812
    },
    {
      "epoch": 1.0085010409437891,
      "grad_norm": 1.0119794607162476,
      "learning_rate": 2.7345764191000993e-05,
      "loss": 0.3997,
      "step": 5813
    },
    {
      "epoch": 1.008674531575295,
      "grad_norm": 0.6230090856552124,
      "learning_rate": 2.733942670590949e-05,
      "loss": 0.6511,
      "step": 5814
    },
    {
      "epoch": 1.0088480222068008,
      "grad_norm": 0.6754621863365173,
      "learning_rate": 2.7333088369068967e-05,
      "loss": 0.5382,
      "step": 5815
    },
    {
      "epoch": 1.0090215128383067,
      "grad_norm": 1.1556576490402222,
      "learning_rate": 2.7326749181214992e-05,
      "loss": 0.3966,
      "step": 5816
    },
    {
      "epoch": 1.0091950034698127,
      "grad_norm": 0.7548868656158447,
      "learning_rate": 2.732040914308324e-05,
      "loss": 0.4733,
      "step": 5817
    },
    {
      "epoch": 1.0093684941013186,
      "grad_norm": 0.690839409828186,
      "learning_rate": 2.7314068255409466e-05,
      "loss": 0.5928,
      "step": 5818
    },
    {
      "epoch": 1.0095419847328244,
      "grad_norm": 0.7861557006835938,
      "learning_rate": 2.7307726518929562e-05,
      "loss": 0.4268,
      "step": 5819
    },
    {
      "epoch": 1.0097154753643303,
      "grad_norm": 0.936643660068512,
      "learning_rate": 2.7301383934379475e-05,
      "loss": 0.4392,
      "step": 5820
    },
    {
      "epoch": 1.0098889659958363,
      "grad_norm": 1.1157642602920532,
      "learning_rate": 2.7295040502495274e-05,
      "loss": 0.5464,
      "step": 5821
    },
    {
      "epoch": 1.0100624566273422,
      "grad_norm": 0.893825113773346,
      "learning_rate": 2.7288696224013124e-05,
      "loss": 0.4768,
      "step": 5822
    },
    {
      "epoch": 1.010235947258848,
      "grad_norm": 0.7682893872261047,
      "learning_rate": 2.728235109966928e-05,
      "loss": 0.4808,
      "step": 5823
    },
    {
      "epoch": 1.0104094378903539,
      "grad_norm": 0.7515265941619873,
      "learning_rate": 2.727600513020011e-05,
      "loss": 0.4105,
      "step": 5824
    },
    {
      "epoch": 1.0105829285218597,
      "grad_norm": 0.788699209690094,
      "learning_rate": 2.7269658316342065e-05,
      "loss": 0.4908,
      "step": 5825
    },
    {
      "epoch": 1.0107564191533658,
      "grad_norm": 0.8162208199501038,
      "learning_rate": 2.7263310658831697e-05,
      "loss": 0.6309,
      "step": 5826
    },
    {
      "epoch": 1.0109299097848716,
      "grad_norm": 0.7742594480514526,
      "learning_rate": 2.725696215840567e-05,
      "loss": 0.517,
      "step": 5827
    },
    {
      "epoch": 1.0111034004163775,
      "grad_norm": 0.7700182795524597,
      "learning_rate": 2.725061281580073e-05,
      "loss": 0.5311,
      "step": 5828
    },
    {
      "epoch": 1.0112768910478833,
      "grad_norm": 0.7160996794700623,
      "learning_rate": 2.724426263175372e-05,
      "loss": 0.5254,
      "step": 5829
    },
    {
      "epoch": 1.0114503816793894,
      "grad_norm": 0.7404236793518066,
      "learning_rate": 2.7237911607001586e-05,
      "loss": 0.4312,
      "step": 5830
    },
    {
      "epoch": 1.0116238723108952,
      "grad_norm": 1.0733016729354858,
      "learning_rate": 2.7231559742281382e-05,
      "loss": 0.6321,
      "step": 5831
    },
    {
      "epoch": 1.011797362942401,
      "grad_norm": 0.6563270688056946,
      "learning_rate": 2.722520703833024e-05,
      "loss": 0.6345,
      "step": 5832
    },
    {
      "epoch": 1.011970853573907,
      "grad_norm": 0.7572863698005676,
      "learning_rate": 2.7218853495885406e-05,
      "loss": 0.5138,
      "step": 5833
    },
    {
      "epoch": 1.012144344205413,
      "grad_norm": 0.7853530645370483,
      "learning_rate": 2.7212499115684204e-05,
      "loss": 0.6611,
      "step": 5834
    },
    {
      "epoch": 1.0123178348369188,
      "grad_norm": 0.724513053894043,
      "learning_rate": 2.7206143898464084e-05,
      "loss": 0.6832,
      "step": 5835
    },
    {
      "epoch": 1.0124913254684247,
      "grad_norm": 0.826076328754425,
      "learning_rate": 2.719978784496257e-05,
      "loss": 0.3856,
      "step": 5836
    },
    {
      "epoch": 1.0126648160999305,
      "grad_norm": 1.0083286762237549,
      "learning_rate": 2.719343095591728e-05,
      "loss": 0.5342,
      "step": 5837
    },
    {
      "epoch": 1.0128383067314366,
      "grad_norm": 0.6915223002433777,
      "learning_rate": 2.718707323206595e-05,
      "loss": 0.4821,
      "step": 5838
    },
    {
      "epoch": 1.0130117973629424,
      "grad_norm": 0.7911536693572998,
      "learning_rate": 2.7180714674146388e-05,
      "loss": 0.449,
      "step": 5839
    },
    {
      "epoch": 1.0131852879944483,
      "grad_norm": 0.8074101209640503,
      "learning_rate": 2.717435528289653e-05,
      "loss": 0.5811,
      "step": 5840
    },
    {
      "epoch": 1.0133587786259541,
      "grad_norm": 0.7548245787620544,
      "learning_rate": 2.7167995059054386e-05,
      "loss": 0.589,
      "step": 5841
    },
    {
      "epoch": 1.0135322692574602,
      "grad_norm": 0.8580533266067505,
      "learning_rate": 2.7161634003358056e-05,
      "loss": 0.3796,
      "step": 5842
    },
    {
      "epoch": 1.013705759888966,
      "grad_norm": 1.4111472368240356,
      "learning_rate": 2.715527211654575e-05,
      "loss": 0.5515,
      "step": 5843
    },
    {
      "epoch": 1.013879250520472,
      "grad_norm": 1.0080534219741821,
      "learning_rate": 2.7148909399355785e-05,
      "loss": 0.4456,
      "step": 5844
    },
    {
      "epoch": 1.0140527411519777,
      "grad_norm": 0.6560617685317993,
      "learning_rate": 2.7142545852526555e-05,
      "loss": 0.6492,
      "step": 5845
    },
    {
      "epoch": 1.0142262317834836,
      "grad_norm": 0.8145388960838318,
      "learning_rate": 2.713618147679655e-05,
      "loss": 0.371,
      "step": 5846
    },
    {
      "epoch": 1.0143997224149897,
      "grad_norm": 0.7898300290107727,
      "learning_rate": 2.7129816272904372e-05,
      "loss": 0.5088,
      "step": 5847
    },
    {
      "epoch": 1.0145732130464955,
      "grad_norm": 0.8385614156723022,
      "learning_rate": 2.712345024158871e-05,
      "loss": 0.5608,
      "step": 5848
    },
    {
      "epoch": 1.0147467036780013,
      "grad_norm": 0.7383617758750916,
      "learning_rate": 2.711708338358835e-05,
      "loss": 0.5056,
      "step": 5849
    },
    {
      "epoch": 1.0149201943095072,
      "grad_norm": 0.8515306115150452,
      "learning_rate": 2.711071569964216e-05,
      "loss": 0.5203,
      "step": 5850
    },
    {
      "epoch": 1.0150936849410133,
      "grad_norm": 0.7870065569877625,
      "learning_rate": 2.7104347190489134e-05,
      "loss": 0.5743,
      "step": 5851
    },
    {
      "epoch": 1.015267175572519,
      "grad_norm": 0.8708816766738892,
      "learning_rate": 2.7097977856868336e-05,
      "loss": 0.4529,
      "step": 5852
    },
    {
      "epoch": 1.015440666204025,
      "grad_norm": 1.0223897695541382,
      "learning_rate": 2.7091607699518936e-05,
      "loss": 0.3867,
      "step": 5853
    },
    {
      "epoch": 1.0156141568355308,
      "grad_norm": 0.8871798515319824,
      "learning_rate": 2.70852367191802e-05,
      "loss": 0.528,
      "step": 5854
    },
    {
      "epoch": 1.0157876474670369,
      "grad_norm": 0.9174553751945496,
      "learning_rate": 2.707886491659149e-05,
      "loss": 0.495,
      "step": 5855
    },
    {
      "epoch": 1.0159611380985427,
      "grad_norm": 0.9138659238815308,
      "learning_rate": 2.707249229249225e-05,
      "loss": 0.408,
      "step": 5856
    },
    {
      "epoch": 1.0161346287300486,
      "grad_norm": 0.7654126882553101,
      "learning_rate": 2.7066118847622053e-05,
      "loss": 0.545,
      "step": 5857
    },
    {
      "epoch": 1.0163081193615544,
      "grad_norm": 0.9373764991760254,
      "learning_rate": 2.7059744582720515e-05,
      "loss": 0.4532,
      "step": 5858
    },
    {
      "epoch": 1.0164816099930605,
      "grad_norm": 0.8717721104621887,
      "learning_rate": 2.7053369498527404e-05,
      "loss": 0.4827,
      "step": 5859
    },
    {
      "epoch": 1.0166551006245663,
      "grad_norm": 0.6174218654632568,
      "learning_rate": 2.7046993595782532e-05,
      "loss": 0.4834,
      "step": 5860
    },
    {
      "epoch": 1.0168285912560722,
      "grad_norm": 0.5347632765769958,
      "learning_rate": 2.704061687522585e-05,
      "loss": 0.6573,
      "step": 5861
    },
    {
      "epoch": 1.017002081887578,
      "grad_norm": 0.7912558913230896,
      "learning_rate": 2.7034239337597378e-05,
      "loss": 0.4384,
      "step": 5862
    },
    {
      "epoch": 1.017175572519084,
      "grad_norm": 0.6055047512054443,
      "learning_rate": 2.7027860983637223e-05,
      "loss": 0.6099,
      "step": 5863
    },
    {
      "epoch": 1.01734906315059,
      "grad_norm": 0.7508504986763,
      "learning_rate": 2.7021481814085622e-05,
      "loss": 0.446,
      "step": 5864
    },
    {
      "epoch": 1.0175225537820958,
      "grad_norm": 0.8576595783233643,
      "learning_rate": 2.7015101829682867e-05,
      "loss": 0.5245,
      "step": 5865
    },
    {
      "epoch": 1.0176960444136016,
      "grad_norm": 0.8401842713356018,
      "learning_rate": 2.7008721031169378e-05,
      "loss": 0.5066,
      "step": 5866
    },
    {
      "epoch": 1.0178695350451075,
      "grad_norm": 1.3381179571151733,
      "learning_rate": 2.7002339419285646e-05,
      "loss": 0.4844,
      "step": 5867
    },
    {
      "epoch": 1.0180430256766135,
      "grad_norm": 1.0883938074111938,
      "learning_rate": 2.699595699477226e-05,
      "loss": 0.5787,
      "step": 5868
    },
    {
      "epoch": 1.0182165163081194,
      "grad_norm": 0.7692278623580933,
      "learning_rate": 2.6989573758369915e-05,
      "loss": 0.4121,
      "step": 5869
    },
    {
      "epoch": 1.0183900069396252,
      "grad_norm": 0.6606525182723999,
      "learning_rate": 2.6983189710819396e-05,
      "loss": 0.4698,
      "step": 5870
    },
    {
      "epoch": 1.018563497571131,
      "grad_norm": 0.838937520980835,
      "learning_rate": 2.6976804852861564e-05,
      "loss": 0.3987,
      "step": 5871
    },
    {
      "epoch": 1.0187369882026371,
      "grad_norm": 0.7664431929588318,
      "learning_rate": 2.697041918523741e-05,
      "loss": 0.4128,
      "step": 5872
    },
    {
      "epoch": 1.018910478834143,
      "grad_norm": 0.7541807889938354,
      "learning_rate": 2.696403270868798e-05,
      "loss": 0.5994,
      "step": 5873
    },
    {
      "epoch": 1.0190839694656488,
      "grad_norm": 2.5131638050079346,
      "learning_rate": 2.6957645423954438e-05,
      "loss": 0.5707,
      "step": 5874
    },
    {
      "epoch": 1.0192574600971547,
      "grad_norm": 0.7696883082389832,
      "learning_rate": 2.6951257331778045e-05,
      "loss": 0.6599,
      "step": 5875
    },
    {
      "epoch": 1.0194309507286607,
      "grad_norm": 0.8057942390441895,
      "learning_rate": 2.694486843290013e-05,
      "loss": 0.3791,
      "step": 5876
    },
    {
      "epoch": 1.0196044413601666,
      "grad_norm": 0.7075010538101196,
      "learning_rate": 2.6938478728062148e-05,
      "loss": 0.4614,
      "step": 5877
    },
    {
      "epoch": 1.0197779319916724,
      "grad_norm": 0.8975362777709961,
      "learning_rate": 2.6932088218005623e-05,
      "loss": 0.4589,
      "step": 5878
    },
    {
      "epoch": 1.0199514226231783,
      "grad_norm": 0.728349506855011,
      "learning_rate": 2.692569690347218e-05,
      "loss": 0.5753,
      "step": 5879
    },
    {
      "epoch": 1.0201249132546844,
      "grad_norm": 0.7772476077079773,
      "learning_rate": 2.6919304785203543e-05,
      "loss": 0.4753,
      "step": 5880
    },
    {
      "epoch": 1.0202984038861902,
      "grad_norm": 0.7649878263473511,
      "learning_rate": 2.6912911863941525e-05,
      "loss": 0.4459,
      "step": 5881
    },
    {
      "epoch": 1.020471894517696,
      "grad_norm": 0.8078417181968689,
      "learning_rate": 2.6906518140428027e-05,
      "loss": 0.5367,
      "step": 5882
    },
    {
      "epoch": 1.020645385149202,
      "grad_norm": 0.7875314950942993,
      "learning_rate": 2.6900123615405052e-05,
      "loss": 0.3754,
      "step": 5883
    },
    {
      "epoch": 1.0208188757807077,
      "grad_norm": 0.8608262538909912,
      "learning_rate": 2.6893728289614693e-05,
      "loss": 0.3492,
      "step": 5884
    },
    {
      "epoch": 1.0209923664122138,
      "grad_norm": 0.7267404794692993,
      "learning_rate": 2.6887332163799133e-05,
      "loss": 0.6483,
      "step": 5885
    },
    {
      "epoch": 1.0211658570437196,
      "grad_norm": 0.8437882661819458,
      "learning_rate": 2.688093523870065e-05,
      "loss": 0.5643,
      "step": 5886
    },
    {
      "epoch": 1.0213393476752255,
      "grad_norm": 0.8284854292869568,
      "learning_rate": 2.6874537515061612e-05,
      "loss": 0.5295,
      "step": 5887
    },
    {
      "epoch": 1.0215128383067313,
      "grad_norm": 0.9316041469573975,
      "learning_rate": 2.6868138993624486e-05,
      "loss": 0.563,
      "step": 5888
    },
    {
      "epoch": 1.0216863289382374,
      "grad_norm": 3.2630345821380615,
      "learning_rate": 2.6861739675131823e-05,
      "loss": 0.4152,
      "step": 5889
    },
    {
      "epoch": 1.0218598195697433,
      "grad_norm": 0.5239765644073486,
      "learning_rate": 2.6855339560326284e-05,
      "loss": 0.5642,
      "step": 5890
    },
    {
      "epoch": 1.022033310201249,
      "grad_norm": 1.1041979789733887,
      "learning_rate": 2.6848938649950597e-05,
      "loss": 0.4003,
      "step": 5891
    },
    {
      "epoch": 1.022206800832755,
      "grad_norm": 0.6967913508415222,
      "learning_rate": 2.6842536944747597e-05,
      "loss": 0.5873,
      "step": 5892
    },
    {
      "epoch": 1.022380291464261,
      "grad_norm": 0.9682864546775818,
      "learning_rate": 2.6836134445460212e-05,
      "loss": 0.5067,
      "step": 5893
    },
    {
      "epoch": 1.0225537820957669,
      "grad_norm": 0.7582650780677795,
      "learning_rate": 2.682973115283146e-05,
      "loss": 0.3793,
      "step": 5894
    },
    {
      "epoch": 1.0227272727272727,
      "grad_norm": 0.7745401263237,
      "learning_rate": 2.6823327067604452e-05,
      "loss": 0.449,
      "step": 5895
    },
    {
      "epoch": 1.0229007633587786,
      "grad_norm": 0.7303228378295898,
      "learning_rate": 2.6816922190522386e-05,
      "loss": 0.5547,
      "step": 5896
    },
    {
      "epoch": 1.0230742539902846,
      "grad_norm": 0.7065463066101074,
      "learning_rate": 2.6810516522328553e-05,
      "loss": 0.3846,
      "step": 5897
    },
    {
      "epoch": 1.0232477446217905,
      "grad_norm": 0.5702760219573975,
      "learning_rate": 2.6804110063766345e-05,
      "loss": 0.5317,
      "step": 5898
    },
    {
      "epoch": 1.0234212352532963,
      "grad_norm": 0.6798083186149597,
      "learning_rate": 2.6797702815579234e-05,
      "loss": 0.5187,
      "step": 5899
    },
    {
      "epoch": 1.0235947258848022,
      "grad_norm": 0.8473564982414246,
      "learning_rate": 2.679129477851079e-05,
      "loss": 0.5884,
      "step": 5900
    },
    {
      "epoch": 1.0237682165163082,
      "grad_norm": 0.783953845500946,
      "learning_rate": 2.6784885953304676e-05,
      "loss": 0.4545,
      "step": 5901
    },
    {
      "epoch": 1.023941707147814,
      "grad_norm": 0.6012303829193115,
      "learning_rate": 2.6778476340704636e-05,
      "loss": 0.6201,
      "step": 5902
    },
    {
      "epoch": 1.02411519777932,
      "grad_norm": 1.1695358753204346,
      "learning_rate": 2.6772065941454527e-05,
      "loss": 0.5081,
      "step": 5903
    },
    {
      "epoch": 1.0242886884108258,
      "grad_norm": 0.6857214570045471,
      "learning_rate": 2.6765654756298264e-05,
      "loss": 0.5099,
      "step": 5904
    },
    {
      "epoch": 1.0244621790423316,
      "grad_norm": 0.6966187357902527,
      "learning_rate": 2.675924278597989e-05,
      "loss": 0.5002,
      "step": 5905
    },
    {
      "epoch": 1.0246356696738377,
      "grad_norm": 0.613068699836731,
      "learning_rate": 2.675283003124351e-05,
      "loss": 0.5436,
      "step": 5906
    },
    {
      "epoch": 1.0248091603053435,
      "grad_norm": 0.8523713946342468,
      "learning_rate": 2.6746416492833343e-05,
      "loss": 0.5601,
      "step": 5907
    },
    {
      "epoch": 1.0249826509368494,
      "grad_norm": 0.6661941409111023,
      "learning_rate": 2.6740002171493676e-05,
      "loss": 0.48,
      "step": 5908
    },
    {
      "epoch": 1.0251561415683552,
      "grad_norm": 0.6891001462936401,
      "learning_rate": 2.67335870679689e-05,
      "loss": 0.4889,
      "step": 5909
    },
    {
      "epoch": 1.0253296321998613,
      "grad_norm": 0.8027567863464355,
      "learning_rate": 2.6727171183003502e-05,
      "loss": 0.475,
      "step": 5910
    },
    {
      "epoch": 1.0255031228313671,
      "grad_norm": 1.59710693359375,
      "learning_rate": 2.6720754517342053e-05,
      "loss": 0.3439,
      "step": 5911
    },
    {
      "epoch": 1.025676613462873,
      "grad_norm": 0.9039287567138672,
      "learning_rate": 2.6714337071729207e-05,
      "loss": 0.4475,
      "step": 5912
    },
    {
      "epoch": 1.0258501040943788,
      "grad_norm": 0.7993444204330444,
      "learning_rate": 2.6707918846909722e-05,
      "loss": 0.4545,
      "step": 5913
    },
    {
      "epoch": 1.026023594725885,
      "grad_norm": 0.6327459216117859,
      "learning_rate": 2.6701499843628443e-05,
      "loss": 0.6672,
      "step": 5914
    },
    {
      "epoch": 1.0261970853573907,
      "grad_norm": 0.7319054007530212,
      "learning_rate": 2.669508006263029e-05,
      "loss": 0.4263,
      "step": 5915
    },
    {
      "epoch": 1.0263705759888966,
      "grad_norm": 0.7239307761192322,
      "learning_rate": 2.66886595046603e-05,
      "loss": 0.5853,
      "step": 5916
    },
    {
      "epoch": 1.0265440666204024,
      "grad_norm": 0.8016424179077148,
      "learning_rate": 2.6682238170463575e-05,
      "loss": 0.4938,
      "step": 5917
    },
    {
      "epoch": 1.0267175572519085,
      "grad_norm": 0.8668724298477173,
      "learning_rate": 2.6675816060785327e-05,
      "loss": 0.5533,
      "step": 5918
    },
    {
      "epoch": 1.0268910478834143,
      "grad_norm": 0.5603392720222473,
      "learning_rate": 2.666939317637085e-05,
      "loss": 0.4508,
      "step": 5919
    },
    {
      "epoch": 1.0270645385149202,
      "grad_norm": 0.5357174277305603,
      "learning_rate": 2.666296951796552e-05,
      "loss": 0.6528,
      "step": 5920
    },
    {
      "epoch": 1.027238029146426,
      "grad_norm": 0.7446817755699158,
      "learning_rate": 2.665654508631481e-05,
      "loss": 0.5292,
      "step": 5921
    },
    {
      "epoch": 1.027411519777932,
      "grad_norm": 0.7444019913673401,
      "learning_rate": 2.6650119882164292e-05,
      "loss": 0.5618,
      "step": 5922
    },
    {
      "epoch": 1.027585010409438,
      "grad_norm": 1.1233359575271606,
      "learning_rate": 2.664369390625961e-05,
      "loss": 0.5344,
      "step": 5923
    },
    {
      "epoch": 1.0277585010409438,
      "grad_norm": 0.8197610974311829,
      "learning_rate": 2.663726715934651e-05,
      "loss": 0.5358,
      "step": 5924
    },
    {
      "epoch": 1.0279319916724496,
      "grad_norm": 0.7967637777328491,
      "learning_rate": 2.663083964217082e-05,
      "loss": 0.5045,
      "step": 5925
    },
    {
      "epoch": 1.0281054823039555,
      "grad_norm": 0.8068925738334656,
      "learning_rate": 2.6624411355478463e-05,
      "loss": 0.5197,
      "step": 5926
    },
    {
      "epoch": 1.0282789729354616,
      "grad_norm": 0.9581280946731567,
      "learning_rate": 2.6617982300015457e-05,
      "loss": 0.5475,
      "step": 5927
    },
    {
      "epoch": 1.0284524635669674,
      "grad_norm": 0.6548795104026794,
      "learning_rate": 2.661155247652788e-05,
      "loss": 0.4966,
      "step": 5928
    },
    {
      "epoch": 1.0286259541984732,
      "grad_norm": 0.8756265044212341,
      "learning_rate": 2.6605121885761948e-05,
      "loss": 0.4913,
      "step": 5929
    },
    {
      "epoch": 1.028799444829979,
      "grad_norm": 0.7330228090286255,
      "learning_rate": 2.6598690528463916e-05,
      "loss": 0.4917,
      "step": 5930
    },
    {
      "epoch": 1.0289729354614852,
      "grad_norm": 0.7564222812652588,
      "learning_rate": 2.659225840538016e-05,
      "loss": 0.5046,
      "step": 5931
    },
    {
      "epoch": 1.029146426092991,
      "grad_norm": 1.3399449586868286,
      "learning_rate": 2.6585825517257133e-05,
      "loss": 0.5187,
      "step": 5932
    },
    {
      "epoch": 1.0293199167244969,
      "grad_norm": 0.6582682728767395,
      "learning_rate": 2.657939186484139e-05,
      "loss": 0.5457,
      "step": 5933
    },
    {
      "epoch": 1.0294934073560027,
      "grad_norm": 1.13554048538208,
      "learning_rate": 2.6572957448879547e-05,
      "loss": 0.4081,
      "step": 5934
    },
    {
      "epoch": 1.0296668979875088,
      "grad_norm": 1.3798998594284058,
      "learning_rate": 2.6566522270118333e-05,
      "loss": 0.4091,
      "step": 5935
    },
    {
      "epoch": 1.0298403886190146,
      "grad_norm": 1.0581423044204712,
      "learning_rate": 2.656008632930456e-05,
      "loss": 0.4448,
      "step": 5936
    },
    {
      "epoch": 1.0300138792505205,
      "grad_norm": 0.7870919108390808,
      "learning_rate": 2.6553649627185122e-05,
      "loss": 0.6226,
      "step": 5937
    },
    {
      "epoch": 1.0301873698820263,
      "grad_norm": 0.8297151923179626,
      "learning_rate": 2.654721216450701e-05,
      "loss": 0.6178,
      "step": 5938
    },
    {
      "epoch": 1.0303608605135324,
      "grad_norm": 1.201669454574585,
      "learning_rate": 2.65407739420173e-05,
      "loss": 0.5902,
      "step": 5939
    },
    {
      "epoch": 1.0305343511450382,
      "grad_norm": 0.7206531167030334,
      "learning_rate": 2.653433496046315e-05,
      "loss": 0.5454,
      "step": 5940
    },
    {
      "epoch": 1.030707841776544,
      "grad_norm": 0.9494001269340515,
      "learning_rate": 2.652789522059181e-05,
      "loss": 0.397,
      "step": 5941
    },
    {
      "epoch": 1.03088133240805,
      "grad_norm": 0.6626607179641724,
      "learning_rate": 2.652145472315063e-05,
      "loss": 0.6467,
      "step": 5942
    },
    {
      "epoch": 1.0310548230395558,
      "grad_norm": 0.6038725972175598,
      "learning_rate": 2.6515013468887026e-05,
      "loss": 0.5157,
      "step": 5943
    },
    {
      "epoch": 1.0312283136710618,
      "grad_norm": 0.885932445526123,
      "learning_rate": 2.650857145854852e-05,
      "loss": 0.5641,
      "step": 5944
    },
    {
      "epoch": 1.0314018043025677,
      "grad_norm": 0.7481269240379333,
      "learning_rate": 2.650212869288271e-05,
      "loss": 0.5497,
      "step": 5945
    },
    {
      "epoch": 1.0315752949340735,
      "grad_norm": 1.073958396911621,
      "learning_rate": 2.6495685172637292e-05,
      "loss": 0.4203,
      "step": 5946
    },
    {
      "epoch": 1.0317487855655794,
      "grad_norm": 1.1243047714233398,
      "learning_rate": 2.6489240898560035e-05,
      "loss": 0.3448,
      "step": 5947
    },
    {
      "epoch": 1.0319222761970854,
      "grad_norm": 0.5776740312576294,
      "learning_rate": 2.6482795871398815e-05,
      "loss": 0.4563,
      "step": 5948
    },
    {
      "epoch": 1.0320957668285913,
      "grad_norm": 0.9867730736732483,
      "learning_rate": 2.6476350091901583e-05,
      "loss": 0.5099,
      "step": 5949
    },
    {
      "epoch": 1.0322692574600971,
      "grad_norm": 0.8928215503692627,
      "learning_rate": 2.646990356081637e-05,
      "loss": 0.4133,
      "step": 5950
    },
    {
      "epoch": 1.032442748091603,
      "grad_norm": 0.6463452577590942,
      "learning_rate": 2.646345627889131e-05,
      "loss": 0.511,
      "step": 5951
    },
    {
      "epoch": 1.032616238723109,
      "grad_norm": 0.768272340297699,
      "learning_rate": 2.645700824687462e-05,
      "loss": 0.5146,
      "step": 5952
    },
    {
      "epoch": 1.0327897293546149,
      "grad_norm": 1.2430983781814575,
      "learning_rate": 2.6450559465514598e-05,
      "loss": 0.5798,
      "step": 5953
    },
    {
      "epoch": 1.0329632199861207,
      "grad_norm": 1.2361890077590942,
      "learning_rate": 2.644410993555963e-05,
      "loss": 0.541,
      "step": 5954
    },
    {
      "epoch": 1.0331367106176266,
      "grad_norm": 0.8147668838500977,
      "learning_rate": 2.6437659657758198e-05,
      "loss": 0.4257,
      "step": 5955
    },
    {
      "epoch": 1.0333102012491326,
      "grad_norm": 0.721660315990448,
      "learning_rate": 2.643120863285886e-05,
      "loss": 0.5403,
      "step": 5956
    },
    {
      "epoch": 1.0334836918806385,
      "grad_norm": 0.8418735861778259,
      "learning_rate": 2.6424756861610274e-05,
      "loss": 0.4564,
      "step": 5957
    },
    {
      "epoch": 1.0336571825121443,
      "grad_norm": 0.9149510860443115,
      "learning_rate": 2.6418304344761165e-05,
      "loss": 0.4342,
      "step": 5958
    },
    {
      "epoch": 1.0338306731436502,
      "grad_norm": 0.6601338982582092,
      "learning_rate": 2.6411851083060355e-05,
      "loss": 0.4678,
      "step": 5959
    },
    {
      "epoch": 1.0340041637751562,
      "grad_norm": 0.6818693280220032,
      "learning_rate": 2.6405397077256752e-05,
      "loss": 0.4143,
      "step": 5960
    },
    {
      "epoch": 1.034177654406662,
      "grad_norm": 0.9768335223197937,
      "learning_rate": 2.639894232809936e-05,
      "loss": 0.5117,
      "step": 5961
    },
    {
      "epoch": 1.034351145038168,
      "grad_norm": 0.6836830377578735,
      "learning_rate": 2.6392486836337256e-05,
      "loss": 0.5513,
      "step": 5962
    },
    {
      "epoch": 1.0345246356696738,
      "grad_norm": 0.8871609568595886,
      "learning_rate": 2.63860306027196e-05,
      "loss": 0.3868,
      "step": 5963
    },
    {
      "epoch": 1.0346981263011796,
      "grad_norm": 0.7213243842124939,
      "learning_rate": 2.637957362799566e-05,
      "loss": 0.4501,
      "step": 5964
    },
    {
      "epoch": 1.0348716169326857,
      "grad_norm": 1.380795955657959,
      "learning_rate": 2.637311591291476e-05,
      "loss": 0.3999,
      "step": 5965
    },
    {
      "epoch": 1.0350451075641915,
      "grad_norm": 0.9298617839813232,
      "learning_rate": 2.636665745822633e-05,
      "loss": 0.4683,
      "step": 5966
    },
    {
      "epoch": 1.0352185981956974,
      "grad_norm": 0.6938121914863586,
      "learning_rate": 2.636019826467989e-05,
      "loss": 0.5135,
      "step": 5967
    },
    {
      "epoch": 1.0353920888272032,
      "grad_norm": 1.094827651977539,
      "learning_rate": 2.6353738333025022e-05,
      "loss": 0.5461,
      "step": 5968
    },
    {
      "epoch": 1.0355655794587093,
      "grad_norm": 1.1556035280227661,
      "learning_rate": 2.6347277664011426e-05,
      "loss": 0.4418,
      "step": 5969
    },
    {
      "epoch": 1.0357390700902152,
      "grad_norm": 0.8316487669944763,
      "learning_rate": 2.6340816258388858e-05,
      "loss": 0.5878,
      "step": 5970
    },
    {
      "epoch": 1.035912560721721,
      "grad_norm": 0.8427453637123108,
      "learning_rate": 2.6334354116907173e-05,
      "loss": 0.512,
      "step": 5971
    },
    {
      "epoch": 1.0360860513532268,
      "grad_norm": 1.3676400184631348,
      "learning_rate": 2.6327891240316313e-05,
      "loss": 0.5244,
      "step": 5972
    },
    {
      "epoch": 1.036259541984733,
      "grad_norm": 1.054203987121582,
      "learning_rate": 2.6321427629366295e-05,
      "loss": 0.5144,
      "step": 5973
    },
    {
      "epoch": 1.0364330326162388,
      "grad_norm": 1.2002469301223755,
      "learning_rate": 2.6314963284807246e-05,
      "loss": 0.5376,
      "step": 5974
    },
    {
      "epoch": 1.0366065232477446,
      "grad_norm": 0.6875637769699097,
      "learning_rate": 2.6308498207389344e-05,
      "loss": 0.4641,
      "step": 5975
    },
    {
      "epoch": 1.0367800138792505,
      "grad_norm": 0.7891087532043457,
      "learning_rate": 2.630203239786287e-05,
      "loss": 0.4939,
      "step": 5976
    },
    {
      "epoch": 1.0369535045107565,
      "grad_norm": 1.255618929862976,
      "learning_rate": 2.6295565856978202e-05,
      "loss": 0.4891,
      "step": 5977
    },
    {
      "epoch": 1.0371269951422624,
      "grad_norm": 0.8953392505645752,
      "learning_rate": 2.628909858548577e-05,
      "loss": 0.4438,
      "step": 5978
    },
    {
      "epoch": 1.0373004857737682,
      "grad_norm": 0.8011636137962341,
      "learning_rate": 2.6282630584136123e-05,
      "loss": 0.4788,
      "step": 5979
    },
    {
      "epoch": 1.037473976405274,
      "grad_norm": 1.016404628753662,
      "learning_rate": 2.6276161853679877e-05,
      "loss": 0.5754,
      "step": 5980
    },
    {
      "epoch": 1.0376474670367801,
      "grad_norm": 0.8635426759719849,
      "learning_rate": 2.626969239486773e-05,
      "loss": 0.5616,
      "step": 5981
    },
    {
      "epoch": 1.037820957668286,
      "grad_norm": 0.9815588593482971,
      "learning_rate": 2.626322220845048e-05,
      "loss": 0.499,
      "step": 5982
    },
    {
      "epoch": 1.0379944482997918,
      "grad_norm": 1.184328317642212,
      "learning_rate": 2.6256751295179e-05,
      "loss": 0.4153,
      "step": 5983
    },
    {
      "epoch": 1.0381679389312977,
      "grad_norm": 0.6641515493392944,
      "learning_rate": 2.6250279655804232e-05,
      "loss": 0.505,
      "step": 5984
    },
    {
      "epoch": 1.0383414295628035,
      "grad_norm": 2.3295421600341797,
      "learning_rate": 2.624380729107723e-05,
      "loss": 0.4332,
      "step": 5985
    },
    {
      "epoch": 1.0385149201943096,
      "grad_norm": 0.8141412138938904,
      "learning_rate": 2.6237334201749126e-05,
      "loss": 0.5208,
      "step": 5986
    },
    {
      "epoch": 1.0386884108258154,
      "grad_norm": 0.6626212000846863,
      "learning_rate": 2.6230860388571103e-05,
      "loss": 0.4482,
      "step": 5987
    },
    {
      "epoch": 1.0388619014573213,
      "grad_norm": 3.283660650253296,
      "learning_rate": 2.6224385852294484e-05,
      "loss": 0.488,
      "step": 5988
    },
    {
      "epoch": 1.0390353920888271,
      "grad_norm": 0.806022584438324,
      "learning_rate": 2.6217910593670626e-05,
      "loss": 0.5385,
      "step": 5989
    },
    {
      "epoch": 1.0392088827203332,
      "grad_norm": 0.8127049207687378,
      "learning_rate": 2.6211434613451006e-05,
      "loss": 0.5229,
      "step": 5990
    },
    {
      "epoch": 1.039382373351839,
      "grad_norm": 0.7175759673118591,
      "learning_rate": 2.6204957912387156e-05,
      "loss": 0.5304,
      "step": 5991
    },
    {
      "epoch": 1.0395558639833449,
      "grad_norm": 0.7946687936782837,
      "learning_rate": 2.6198480491230712e-05,
      "loss": 0.4648,
      "step": 5992
    },
    {
      "epoch": 1.0397293546148507,
      "grad_norm": 1.0387405157089233,
      "learning_rate": 2.6192002350733387e-05,
      "loss": 0.5369,
      "step": 5993
    },
    {
      "epoch": 1.0399028452463568,
      "grad_norm": 0.8700816631317139,
      "learning_rate": 2.618552349164697e-05,
      "loss": 0.4854,
      "step": 5994
    },
    {
      "epoch": 1.0400763358778626,
      "grad_norm": 0.8339025378227234,
      "learning_rate": 2.617904391472334e-05,
      "loss": 0.4008,
      "step": 5995
    },
    {
      "epoch": 1.0402498265093685,
      "grad_norm": 1.0166280269622803,
      "learning_rate": 2.6172563620714475e-05,
      "loss": 0.5493,
      "step": 5996
    },
    {
      "epoch": 1.0404233171408743,
      "grad_norm": 0.6536795496940613,
      "learning_rate": 2.61660826103724e-05,
      "loss": 0.5028,
      "step": 5997
    },
    {
      "epoch": 1.0405968077723804,
      "grad_norm": 0.661294162273407,
      "learning_rate": 2.6159600884449258e-05,
      "loss": 0.63,
      "step": 5998
    },
    {
      "epoch": 1.0407702984038862,
      "grad_norm": 0.7142971158027649,
      "learning_rate": 2.6153118443697255e-05,
      "loss": 0.4796,
      "step": 5999
    },
    {
      "epoch": 1.040943789035392,
      "grad_norm": 0.8321624994277954,
      "learning_rate": 2.6146635288868685e-05,
      "loss": 0.5194,
      "step": 6000
    },
    {
      "epoch": 1.041117279666898,
      "grad_norm": 0.7722504138946533,
      "learning_rate": 2.6140151420715932e-05,
      "loss": 0.4021,
      "step": 6001
    },
    {
      "epoch": 1.0412907702984038,
      "grad_norm": 1.652472972869873,
      "learning_rate": 2.6133666839991444e-05,
      "loss": 0.4716,
      "step": 6002
    },
    {
      "epoch": 1.0414642609299098,
      "grad_norm": 0.7369078397750854,
      "learning_rate": 2.6127181547447773e-05,
      "loss": 0.4482,
      "step": 6003
    },
    {
      "epoch": 1.0416377515614157,
      "grad_norm": 0.7246720790863037,
      "learning_rate": 2.612069554383755e-05,
      "loss": 0.4106,
      "step": 6004
    },
    {
      "epoch": 1.0418112421929215,
      "grad_norm": 1.4080967903137207,
      "learning_rate": 2.6114208829913473e-05,
      "loss": 0.5305,
      "step": 6005
    },
    {
      "epoch": 1.0419847328244274,
      "grad_norm": 1.1792443990707397,
      "learning_rate": 2.6107721406428338e-05,
      "loss": 0.5676,
      "step": 6006
    },
    {
      "epoch": 1.0421582234559335,
      "grad_norm": 1.539556860923767,
      "learning_rate": 2.6101233274135017e-05,
      "loss": 0.3627,
      "step": 6007
    },
    {
      "epoch": 1.0423317140874393,
      "grad_norm": 0.6607788801193237,
      "learning_rate": 2.6094744433786467e-05,
      "loss": 0.5356,
      "step": 6008
    },
    {
      "epoch": 1.0425052047189451,
      "grad_norm": 1.1622593402862549,
      "learning_rate": 2.608825488613572e-05,
      "loss": 0.4337,
      "step": 6009
    },
    {
      "epoch": 1.042678695350451,
      "grad_norm": 0.9463575482368469,
      "learning_rate": 2.6081764631935896e-05,
      "loss": 0.441,
      "step": 6010
    },
    {
      "epoch": 1.042852185981957,
      "grad_norm": 0.7776453495025635,
      "learning_rate": 2.6075273671940212e-05,
      "loss": 0.5146,
      "step": 6011
    },
    {
      "epoch": 1.043025676613463,
      "grad_norm": 0.8027953505516052,
      "learning_rate": 2.606878200690193e-05,
      "loss": 0.4731,
      "step": 6012
    },
    {
      "epoch": 1.0431991672449688,
      "grad_norm": 0.708922266960144,
      "learning_rate": 2.6062289637574428e-05,
      "loss": 0.3625,
      "step": 6013
    },
    {
      "epoch": 1.0433726578764746,
      "grad_norm": 0.7351964712142944,
      "learning_rate": 2.605579656471115e-05,
      "loss": 0.5157,
      "step": 6014
    },
    {
      "epoch": 1.0435461485079807,
      "grad_norm": 0.7151260375976562,
      "learning_rate": 2.6049302789065624e-05,
      "loss": 0.5406,
      "step": 6015
    },
    {
      "epoch": 1.0437196391394865,
      "grad_norm": 0.673349142074585,
      "learning_rate": 2.6042808311391456e-05,
      "loss": 0.5458,
      "step": 6016
    },
    {
      "epoch": 1.0438931297709924,
      "grad_norm": 1.6523752212524414,
      "learning_rate": 2.603631313244235e-05,
      "loss": 0.4021,
      "step": 6017
    },
    {
      "epoch": 1.0440666204024982,
      "grad_norm": 0.9373065233230591,
      "learning_rate": 2.6029817252972064e-05,
      "loss": 0.6046,
      "step": 6018
    },
    {
      "epoch": 1.0442401110340043,
      "grad_norm": 1.176289677619934,
      "learning_rate": 2.6023320673734462e-05,
      "loss": 0.4153,
      "step": 6019
    },
    {
      "epoch": 1.0444136016655101,
      "grad_norm": 0.937972366809845,
      "learning_rate": 2.6016823395483482e-05,
      "loss": 0.4359,
      "step": 6020
    },
    {
      "epoch": 1.044587092297016,
      "grad_norm": 0.8680937886238098,
      "learning_rate": 2.6010325418973127e-05,
      "loss": 0.4891,
      "step": 6021
    },
    {
      "epoch": 1.0447605829285218,
      "grad_norm": 0.8315525650978088,
      "learning_rate": 2.600382674495751e-05,
      "loss": 0.4292,
      "step": 6022
    },
    {
      "epoch": 1.0449340735600277,
      "grad_norm": 0.7118164896965027,
      "learning_rate": 2.5997327374190797e-05,
      "loss": 0.4984,
      "step": 6023
    },
    {
      "epoch": 1.0451075641915337,
      "grad_norm": 0.8397243618965149,
      "learning_rate": 2.5990827307427263e-05,
      "loss": 0.511,
      "step": 6024
    },
    {
      "epoch": 1.0452810548230396,
      "grad_norm": 0.7386336326599121,
      "learning_rate": 2.5984326545421238e-05,
      "loss": 0.5754,
      "step": 6025
    },
    {
      "epoch": 1.0454545454545454,
      "grad_norm": 0.6508328914642334,
      "learning_rate": 2.5977825088927135e-05,
      "loss": 0.6221,
      "step": 6026
    },
    {
      "epoch": 1.0456280360860513,
      "grad_norm": 0.9424730539321899,
      "learning_rate": 2.597132293869947e-05,
      "loss": 0.4105,
      "step": 6027
    },
    {
      "epoch": 1.0458015267175573,
      "grad_norm": 0.8315467238426208,
      "learning_rate": 2.5964820095492825e-05,
      "loss": 0.391,
      "step": 6028
    },
    {
      "epoch": 1.0459750173490632,
      "grad_norm": 1.1879899501800537,
      "learning_rate": 2.5958316560061853e-05,
      "loss": 0.5172,
      "step": 6029
    },
    {
      "epoch": 1.046148507980569,
      "grad_norm": 1.000832438468933,
      "learning_rate": 2.5951812333161298e-05,
      "loss": 0.5194,
      "step": 6030
    },
    {
      "epoch": 1.0463219986120749,
      "grad_norm": 1.8499864339828491,
      "learning_rate": 2.594530741554599e-05,
      "loss": 0.3949,
      "step": 6031
    },
    {
      "epoch": 1.046495489243581,
      "grad_norm": 2.039126396179199,
      "learning_rate": 2.593880180797083e-05,
      "loss": 0.5708,
      "step": 6032
    },
    {
      "epoch": 1.0466689798750868,
      "grad_norm": 1.0549595355987549,
      "learning_rate": 2.5932295511190803e-05,
      "loss": 0.4266,
      "step": 6033
    },
    {
      "epoch": 1.0468424705065926,
      "grad_norm": 1.0007039308547974,
      "learning_rate": 2.5925788525960964e-05,
      "loss": 0.519,
      "step": 6034
    },
    {
      "epoch": 1.0470159611380985,
      "grad_norm": 0.7571946382522583,
      "learning_rate": 2.5919280853036462e-05,
      "loss": 0.5032,
      "step": 6035
    },
    {
      "epoch": 1.0471894517696045,
      "grad_norm": 0.6391426920890808,
      "learning_rate": 2.5912772493172523e-05,
      "loss": 0.5856,
      "step": 6036
    },
    {
      "epoch": 1.0473629424011104,
      "grad_norm": 1.0349048376083374,
      "learning_rate": 2.590626344712444e-05,
      "loss": 0.4343,
      "step": 6037
    },
    {
      "epoch": 1.0475364330326162,
      "grad_norm": 0.7470468878746033,
      "learning_rate": 2.5899753715647614e-05,
      "loss": 0.4257,
      "step": 6038
    },
    {
      "epoch": 1.047709923664122,
      "grad_norm": 0.7566617727279663,
      "learning_rate": 2.5893243299497483e-05,
      "loss": 0.4045,
      "step": 6039
    },
    {
      "epoch": 1.0478834142956281,
      "grad_norm": 0.7339170575141907,
      "learning_rate": 2.5886732199429606e-05,
      "loss": 0.5164,
      "step": 6040
    },
    {
      "epoch": 1.048056904927134,
      "grad_norm": 1.0630372762680054,
      "learning_rate": 2.5880220416199598e-05,
      "loss": 0.4393,
      "step": 6041
    },
    {
      "epoch": 1.0482303955586398,
      "grad_norm": 0.6280874609947205,
      "learning_rate": 2.587370795056315e-05,
      "loss": 0.4855,
      "step": 6042
    },
    {
      "epoch": 1.0484038861901457,
      "grad_norm": 0.7635196447372437,
      "learning_rate": 2.5867194803276058e-05,
      "loss": 0.4342,
      "step": 6043
    },
    {
      "epoch": 1.0485773768216515,
      "grad_norm": 1.8262170553207397,
      "learning_rate": 2.5860680975094178e-05,
      "loss": 0.5427,
      "step": 6044
    },
    {
      "epoch": 1.0487508674531576,
      "grad_norm": 1.0261775255203247,
      "learning_rate": 2.585416646677343e-05,
      "loss": 0.462,
      "step": 6045
    },
    {
      "epoch": 1.0489243580846634,
      "grad_norm": 0.8317374587059021,
      "learning_rate": 2.5847651279069847e-05,
      "loss": 0.5562,
      "step": 6046
    },
    {
      "epoch": 1.0490978487161693,
      "grad_norm": 1.0046920776367188,
      "learning_rate": 2.584113541273952e-05,
      "loss": 0.4017,
      "step": 6047
    },
    {
      "epoch": 1.0492713393476751,
      "grad_norm": 1.0864875316619873,
      "learning_rate": 2.5834618868538623e-05,
      "loss": 0.4116,
      "step": 6048
    },
    {
      "epoch": 1.0494448299791812,
      "grad_norm": 0.9202406406402588,
      "learning_rate": 2.58281016472234e-05,
      "loss": 0.5406,
      "step": 6049
    },
    {
      "epoch": 1.049618320610687,
      "grad_norm": 0.7802914381027222,
      "learning_rate": 2.58215837495502e-05,
      "loss": 0.5032,
      "step": 6050
    },
    {
      "epoch": 1.049791811242193,
      "grad_norm": 0.687981128692627,
      "learning_rate": 2.5815065176275417e-05,
      "loss": 0.47,
      "step": 6051
    },
    {
      "epoch": 1.0499653018736987,
      "grad_norm": 0.8128355145454407,
      "learning_rate": 2.5808545928155547e-05,
      "loss": 0.4601,
      "step": 6052
    },
    {
      "epoch": 1.0501387925052048,
      "grad_norm": 0.8574787974357605,
      "learning_rate": 2.580202600594715e-05,
      "loss": 0.3673,
      "step": 6053
    },
    {
      "epoch": 1.0503122831367107,
      "grad_norm": 0.6471509337425232,
      "learning_rate": 2.5795505410406878e-05,
      "loss": 0.6451,
      "step": 6054
    },
    {
      "epoch": 1.0504857737682165,
      "grad_norm": 0.7630587816238403,
      "learning_rate": 2.5788984142291448e-05,
      "loss": 0.6399,
      "step": 6055
    },
    {
      "epoch": 1.0506592643997223,
      "grad_norm": 1.43451988697052,
      "learning_rate": 2.5782462202357664e-05,
      "loss": 0.4636,
      "step": 6056
    },
    {
      "epoch": 1.0508327550312284,
      "grad_norm": 0.7638126015663147,
      "learning_rate": 2.5775939591362403e-05,
      "loss": 0.5337,
      "step": 6057
    },
    {
      "epoch": 1.0510062456627343,
      "grad_norm": 0.7070881724357605,
      "learning_rate": 2.5769416310062622e-05,
      "loss": 0.3953,
      "step": 6058
    },
    {
      "epoch": 1.05117973629424,
      "grad_norm": 0.7772389054298401,
      "learning_rate": 2.576289235921536e-05,
      "loss": 0.6108,
      "step": 6059
    },
    {
      "epoch": 1.051353226925746,
      "grad_norm": 0.9139055609703064,
      "learning_rate": 2.5756367739577713e-05,
      "loss": 0.4846,
      "step": 6060
    },
    {
      "epoch": 1.0515267175572518,
      "grad_norm": 0.8612098097801208,
      "learning_rate": 2.574984245190689e-05,
      "loss": 0.5166,
      "step": 6061
    },
    {
      "epoch": 1.0517002081887579,
      "grad_norm": 0.5893256068229675,
      "learning_rate": 2.5743316496960154e-05,
      "loss": 0.6597,
      "step": 6062
    },
    {
      "epoch": 1.0518736988202637,
      "grad_norm": 1.2996785640716553,
      "learning_rate": 2.5736789875494844e-05,
      "loss": 0.4073,
      "step": 6063
    },
    {
      "epoch": 1.0520471894517696,
      "grad_norm": 1.8747106790542603,
      "learning_rate": 2.573026258826838e-05,
      "loss": 0.4684,
      "step": 6064
    },
    {
      "epoch": 1.0522206800832754,
      "grad_norm": 1.3142502307891846,
      "learning_rate": 2.5723734636038272e-05,
      "loss": 0.4611,
      "step": 6065
    },
    {
      "epoch": 1.0523941707147815,
      "grad_norm": 0.8775864839553833,
      "learning_rate": 2.571720601956208e-05,
      "loss": 0.465,
      "step": 6066
    },
    {
      "epoch": 1.0525676613462873,
      "grad_norm": 0.7700573205947876,
      "learning_rate": 2.571067673959748e-05,
      "loss": 0.4945,
      "step": 6067
    },
    {
      "epoch": 1.0527411519777932,
      "grad_norm": 0.9080457091331482,
      "learning_rate": 2.570414679690218e-05,
      "loss": 0.6381,
      "step": 6068
    },
    {
      "epoch": 1.052914642609299,
      "grad_norm": 0.7340862154960632,
      "learning_rate": 2.5697616192234005e-05,
      "loss": 0.4708,
      "step": 6069
    },
    {
      "epoch": 1.053088133240805,
      "grad_norm": 0.7591330409049988,
      "learning_rate": 2.5691084926350825e-05,
      "loss": 0.5574,
      "step": 6070
    },
    {
      "epoch": 1.053261623872311,
      "grad_norm": 0.7035334706306458,
      "learning_rate": 2.568455300001061e-05,
      "loss": 0.5569,
      "step": 6071
    },
    {
      "epoch": 1.0534351145038168,
      "grad_norm": 0.6817179918289185,
      "learning_rate": 2.56780204139714e-05,
      "loss": 0.391,
      "step": 6072
    },
    {
      "epoch": 1.0536086051353226,
      "grad_norm": 0.6882315278053284,
      "learning_rate": 2.56714871689913e-05,
      "loss": 0.402,
      "step": 6073
    },
    {
      "epoch": 1.0537820957668287,
      "grad_norm": 1.1013617515563965,
      "learning_rate": 2.5664953265828504e-05,
      "loss": 0.6382,
      "step": 6074
    },
    {
      "epoch": 1.0539555863983345,
      "grad_norm": 0.8133484125137329,
      "learning_rate": 2.5658418705241283e-05,
      "loss": 0.3861,
      "step": 6075
    },
    {
      "epoch": 1.0541290770298404,
      "grad_norm": 0.8575695157051086,
      "learning_rate": 2.565188348798798e-05,
      "loss": 0.6594,
      "step": 6076
    },
    {
      "epoch": 1.0543025676613462,
      "grad_norm": 0.928375780582428,
      "learning_rate": 2.5645347614827008e-05,
      "loss": 0.5104,
      "step": 6077
    },
    {
      "epoch": 1.0544760582928523,
      "grad_norm": 0.7670015692710876,
      "learning_rate": 2.5638811086516873e-05,
      "loss": 0.4182,
      "step": 6078
    },
    {
      "epoch": 1.0546495489243581,
      "grad_norm": 1.106449842453003,
      "learning_rate": 2.5632273903816133e-05,
      "loss": 0.5043,
      "step": 6079
    },
    {
      "epoch": 1.054823039555864,
      "grad_norm": 0.8924499154090881,
      "learning_rate": 2.562573606748345e-05,
      "loss": 0.4215,
      "step": 6080
    },
    {
      "epoch": 1.0549965301873698,
      "grad_norm": 0.6301459074020386,
      "learning_rate": 2.561919757827754e-05,
      "loss": 0.5708,
      "step": 6081
    },
    {
      "epoch": 1.0551700208188757,
      "grad_norm": 0.9665195345878601,
      "learning_rate": 2.5612658436957204e-05,
      "loss": 0.5227,
      "step": 6082
    },
    {
      "epoch": 1.0553435114503817,
      "grad_norm": 0.7565470337867737,
      "learning_rate": 2.5606118644281318e-05,
      "loss": 0.4502,
      "step": 6083
    },
    {
      "epoch": 1.0555170020818876,
      "grad_norm": 0.7726820707321167,
      "learning_rate": 2.5599578201008824e-05,
      "loss": 0.4883,
      "step": 6084
    },
    {
      "epoch": 1.0556904927133934,
      "grad_norm": 0.8903718590736389,
      "learning_rate": 2.559303710789876e-05,
      "loss": 0.3729,
      "step": 6085
    },
    {
      "epoch": 1.0558639833448993,
      "grad_norm": 1.2035996913909912,
      "learning_rate": 2.5586495365710225e-05,
      "loss": 0.538,
      "step": 6086
    },
    {
      "epoch": 1.0560374739764053,
      "grad_norm": 0.8257406949996948,
      "learning_rate": 2.557995297520239e-05,
      "loss": 0.4719,
      "step": 6087
    },
    {
      "epoch": 1.0562109646079112,
      "grad_norm": 0.6504576206207275,
      "learning_rate": 2.5573409937134508e-05,
      "loss": 0.4062,
      "step": 6088
    },
    {
      "epoch": 1.056384455239417,
      "grad_norm": 0.8960192203521729,
      "learning_rate": 2.5566866252265908e-05,
      "loss": 0.5537,
      "step": 6089
    },
    {
      "epoch": 1.0565579458709229,
      "grad_norm": 0.8303012251853943,
      "learning_rate": 2.5560321921355996e-05,
      "loss": 0.3818,
      "step": 6090
    },
    {
      "epoch": 1.056731436502429,
      "grad_norm": 0.6931776404380798,
      "learning_rate": 2.555377694516425e-05,
      "loss": 0.4065,
      "step": 6091
    },
    {
      "epoch": 1.0569049271339348,
      "grad_norm": 2.857333183288574,
      "learning_rate": 2.554723132445021e-05,
      "loss": 0.6019,
      "step": 6092
    },
    {
      "epoch": 1.0570784177654406,
      "grad_norm": 0.8212158679962158,
      "learning_rate": 2.5540685059973514e-05,
      "loss": 0.5443,
      "step": 6093
    },
    {
      "epoch": 1.0572519083969465,
      "grad_norm": 0.8178073763847351,
      "learning_rate": 2.5534138152493863e-05,
      "loss": 0.5602,
      "step": 6094
    },
    {
      "epoch": 1.0574253990284526,
      "grad_norm": 0.9276249408721924,
      "learning_rate": 2.5527590602771026e-05,
      "loss": 0.4221,
      "step": 6095
    },
    {
      "epoch": 1.0575988896599584,
      "grad_norm": 0.6286492347717285,
      "learning_rate": 2.5521042411564866e-05,
      "loss": 0.498,
      "step": 6096
    },
    {
      "epoch": 1.0577723802914643,
      "grad_norm": 0.7513982653617859,
      "learning_rate": 2.551449357963529e-05,
      "loss": 0.3982,
      "step": 6097
    },
    {
      "epoch": 1.05794587092297,
      "grad_norm": 0.6628233790397644,
      "learning_rate": 2.5507944107742314e-05,
      "loss": 0.5955,
      "step": 6098
    },
    {
      "epoch": 1.0581193615544762,
      "grad_norm": 1.5238407850265503,
      "learning_rate": 2.550139399664601e-05,
      "loss": 0.3729,
      "step": 6099
    },
    {
      "epoch": 1.058292852185982,
      "grad_norm": 0.517874538898468,
      "learning_rate": 2.549484324710652e-05,
      "loss": 0.5261,
      "step": 6100
    },
    {
      "epoch": 1.0584663428174879,
      "grad_norm": 0.6734583377838135,
      "learning_rate": 2.5488291859884067e-05,
      "loss": 0.402,
      "step": 6101
    },
    {
      "epoch": 1.0586398334489937,
      "grad_norm": 0.9803544878959656,
      "learning_rate": 2.548173983573895e-05,
      "loss": 0.4781,
      "step": 6102
    },
    {
      "epoch": 1.0588133240804996,
      "grad_norm": 0.7337921857833862,
      "learning_rate": 2.5475187175431532e-05,
      "loss": 0.4492,
      "step": 6103
    },
    {
      "epoch": 1.0589868147120056,
      "grad_norm": 0.8741667866706848,
      "learning_rate": 2.5468633879722272e-05,
      "loss": 0.4561,
      "step": 6104
    },
    {
      "epoch": 1.0591603053435115,
      "grad_norm": 0.7236998677253723,
      "learning_rate": 2.5462079949371665e-05,
      "loss": 0.6299,
      "step": 6105
    },
    {
      "epoch": 1.0593337959750173,
      "grad_norm": 0.7462474703788757,
      "learning_rate": 2.545552538514033e-05,
      "loss": 0.5455,
      "step": 6106
    },
    {
      "epoch": 1.0595072866065232,
      "grad_norm": 0.7738128304481506,
      "learning_rate": 2.5448970187788913e-05,
      "loss": 0.5048,
      "step": 6107
    },
    {
      "epoch": 1.0596807772380292,
      "grad_norm": 0.6994773149490356,
      "learning_rate": 2.5442414358078148e-05,
      "loss": 0.5266,
      "step": 6108
    },
    {
      "epoch": 1.059854267869535,
      "grad_norm": 0.85459965467453,
      "learning_rate": 2.5435857896768862e-05,
      "loss": 0.421,
      "step": 6109
    },
    {
      "epoch": 1.060027758501041,
      "grad_norm": 0.7872832417488098,
      "learning_rate": 2.5429300804621934e-05,
      "loss": 0.483,
      "step": 6110
    },
    {
      "epoch": 1.0602012491325468,
      "grad_norm": 0.6063489317893982,
      "learning_rate": 2.542274308239832e-05,
      "loss": 0.5454,
      "step": 6111
    },
    {
      "epoch": 1.0603747397640528,
      "grad_norm": 0.9173235893249512,
      "learning_rate": 2.541618473085905e-05,
      "loss": 0.5068,
      "step": 6112
    },
    {
      "epoch": 1.0605482303955587,
      "grad_norm": 0.8748695850372314,
      "learning_rate": 2.540962575076523e-05,
      "loss": 0.5232,
      "step": 6113
    },
    {
      "epoch": 1.0607217210270645,
      "grad_norm": 0.7439556121826172,
      "learning_rate": 2.5403066142878047e-05,
      "loss": 0.5273,
      "step": 6114
    },
    {
      "epoch": 1.0608952116585704,
      "grad_norm": 0.6065833568572998,
      "learning_rate": 2.5396505907958736e-05,
      "loss": 0.6476,
      "step": 6115
    },
    {
      "epoch": 1.0610687022900764,
      "grad_norm": 0.7149546146392822,
      "learning_rate": 2.538994504676862e-05,
      "loss": 0.519,
      "step": 6116
    },
    {
      "epoch": 1.0612421929215823,
      "grad_norm": 1.0197675228118896,
      "learning_rate": 2.5383383560069113e-05,
      "loss": 0.4003,
      "step": 6117
    },
    {
      "epoch": 1.0614156835530881,
      "grad_norm": 0.8125647902488708,
      "learning_rate": 2.537682144862166e-05,
      "loss": 0.6007,
      "step": 6118
    },
    {
      "epoch": 1.061589174184594,
      "grad_norm": 0.702014148235321,
      "learning_rate": 2.537025871318782e-05,
      "loss": 0.5608,
      "step": 6119
    },
    {
      "epoch": 1.0617626648160998,
      "grad_norm": 0.9729758501052856,
      "learning_rate": 2.53636953545292e-05,
      "loss": 0.3749,
      "step": 6120
    },
    {
      "epoch": 1.061936155447606,
      "grad_norm": 0.6251813173294067,
      "learning_rate": 2.5357131373407478e-05,
      "loss": 0.5125,
      "step": 6121
    },
    {
      "epoch": 1.0621096460791117,
      "grad_norm": 0.6916963458061218,
      "learning_rate": 2.5350566770584423e-05,
      "loss": 0.4899,
      "step": 6122
    },
    {
      "epoch": 1.0622831367106176,
      "grad_norm": 0.8575353622436523,
      "learning_rate": 2.534400154682185e-05,
      "loss": 0.5601,
      "step": 6123
    },
    {
      "epoch": 1.0624566273421234,
      "grad_norm": 0.8720142245292664,
      "learning_rate": 2.5337435702881683e-05,
      "loss": 0.425,
      "step": 6124
    },
    {
      "epoch": 1.0626301179736295,
      "grad_norm": 0.7958077192306519,
      "learning_rate": 2.5330869239525874e-05,
      "loss": 0.5974,
      "step": 6125
    },
    {
      "epoch": 1.0628036086051353,
      "grad_norm": 0.8834172487258911,
      "learning_rate": 2.5324302157516486e-05,
      "loss": 0.3673,
      "step": 6126
    },
    {
      "epoch": 1.0629770992366412,
      "grad_norm": 0.8853662610054016,
      "learning_rate": 2.531773445761562e-05,
      "loss": 0.3999,
      "step": 6127
    },
    {
      "epoch": 1.063150589868147,
      "grad_norm": 1.272106647491455,
      "learning_rate": 2.531116614058548e-05,
      "loss": 0.657,
      "step": 6128
    },
    {
      "epoch": 1.063324080499653,
      "grad_norm": 0.8048542141914368,
      "learning_rate": 2.5304597207188318e-05,
      "loss": 0.5995,
      "step": 6129
    },
    {
      "epoch": 1.063497571131159,
      "grad_norm": 0.7750091552734375,
      "learning_rate": 2.5298027658186472e-05,
      "loss": 0.4662,
      "step": 6130
    },
    {
      "epoch": 1.0636710617626648,
      "grad_norm": 1.1804490089416504,
      "learning_rate": 2.529145749434234e-05,
      "loss": 0.4143,
      "step": 6131
    },
    {
      "epoch": 1.0638445523941706,
      "grad_norm": 1.0123077630996704,
      "learning_rate": 2.52848867164184e-05,
      "loss": 0.5084,
      "step": 6132
    },
    {
      "epoch": 1.0640180430256767,
      "grad_norm": 0.7592995762825012,
      "learning_rate": 2.52783153251772e-05,
      "loss": 0.5007,
      "step": 6133
    },
    {
      "epoch": 1.0641915336571826,
      "grad_norm": 0.8198844790458679,
      "learning_rate": 2.5271743321381354e-05,
      "loss": 0.4343,
      "step": 6134
    },
    {
      "epoch": 1.0643650242886884,
      "grad_norm": 0.7294592261314392,
      "learning_rate": 2.5265170705793555e-05,
      "loss": 0.4335,
      "step": 6135
    },
    {
      "epoch": 1.0645385149201942,
      "grad_norm": 0.7923845648765564,
      "learning_rate": 2.525859747917656e-05,
      "loss": 0.5364,
      "step": 6136
    },
    {
      "epoch": 1.0647120055517,
      "grad_norm": 0.8322098851203918,
      "learning_rate": 2.52520236422932e-05,
      "loss": 0.6079,
      "step": 6137
    },
    {
      "epoch": 1.0648854961832062,
      "grad_norm": 0.8434830904006958,
      "learning_rate": 2.524544919590638e-05,
      "loss": 0.4244,
      "step": 6138
    },
    {
      "epoch": 1.065058986814712,
      "grad_norm": 0.5925361514091492,
      "learning_rate": 2.5238874140779057e-05,
      "loss": 0.6212,
      "step": 6139
    },
    {
      "epoch": 1.0652324774462179,
      "grad_norm": 0.8045483231544495,
      "learning_rate": 2.5232298477674297e-05,
      "loss": 0.554,
      "step": 6140
    },
    {
      "epoch": 1.065405968077724,
      "grad_norm": 0.843867301940918,
      "learning_rate": 2.5225722207355202e-05,
      "loss": 0.4384,
      "step": 6141
    },
    {
      "epoch": 1.0655794587092298,
      "grad_norm": 0.8407973051071167,
      "learning_rate": 2.5219145330584945e-05,
      "loss": 0.4971,
      "step": 6142
    },
    {
      "epoch": 1.0657529493407356,
      "grad_norm": 0.8980678915977478,
      "learning_rate": 2.5212567848126802e-05,
      "loss": 0.4048,
      "step": 6143
    },
    {
      "epoch": 1.0659264399722415,
      "grad_norm": 1.1637353897094727,
      "learning_rate": 2.5205989760744084e-05,
      "loss": 0.5775,
      "step": 6144
    },
    {
      "epoch": 1.0660999306037473,
      "grad_norm": 0.9472494721412659,
      "learning_rate": 2.519941106920018e-05,
      "loss": 0.5118,
      "step": 6145
    },
    {
      "epoch": 1.0662734212352534,
      "grad_norm": 0.69376540184021,
      "learning_rate": 2.5192831774258575e-05,
      "loss": 0.5542,
      "step": 6146
    },
    {
      "epoch": 1.0664469118667592,
      "grad_norm": 0.6257830262184143,
      "learning_rate": 2.5186251876682782e-05,
      "loss": 0.4934,
      "step": 6147
    },
    {
      "epoch": 1.066620402498265,
      "grad_norm": 0.7980381846427917,
      "learning_rate": 2.5179671377236422e-05,
      "loss": 0.5593,
      "step": 6148
    },
    {
      "epoch": 1.066793893129771,
      "grad_norm": 0.8958628177642822,
      "learning_rate": 2.5173090276683157e-05,
      "loss": 0.4069,
      "step": 6149
    },
    {
      "epoch": 1.066967383761277,
      "grad_norm": 0.6968737840652466,
      "learning_rate": 2.516650857578674e-05,
      "loss": 0.5197,
      "step": 6150
    },
    {
      "epoch": 1.0671408743927828,
      "grad_norm": 0.893977165222168,
      "learning_rate": 2.515992627531098e-05,
      "loss": 0.4059,
      "step": 6151
    },
    {
      "epoch": 1.0673143650242887,
      "grad_norm": 1.103964924812317,
      "learning_rate": 2.515334337601977e-05,
      "loss": 0.4072,
      "step": 6152
    },
    {
      "epoch": 1.0674878556557945,
      "grad_norm": 0.8834951519966125,
      "learning_rate": 2.514675987867705e-05,
      "loss": 0.5861,
      "step": 6153
    },
    {
      "epoch": 1.0676613462873006,
      "grad_norm": 0.5498542189598083,
      "learning_rate": 2.5140175784046858e-05,
      "loss": 0.6534,
      "step": 6154
    },
    {
      "epoch": 1.0678348369188064,
      "grad_norm": 0.9770853519439697,
      "learning_rate": 2.5133591092893265e-05,
      "loss": 0.5331,
      "step": 6155
    },
    {
      "epoch": 1.0680083275503123,
      "grad_norm": 0.7064255475997925,
      "learning_rate": 2.512700580598045e-05,
      "loss": 0.516,
      "step": 6156
    },
    {
      "epoch": 1.0681818181818181,
      "grad_norm": 0.8084623217582703,
      "learning_rate": 2.512041992407264e-05,
      "loss": 0.4167,
      "step": 6157
    },
    {
      "epoch": 1.0683553088133242,
      "grad_norm": 1.165953516960144,
      "learning_rate": 2.5113833447934126e-05,
      "loss": 0.3626,
      "step": 6158
    },
    {
      "epoch": 1.06852879944483,
      "grad_norm": 1.1164823770523071,
      "learning_rate": 2.5107246378329287e-05,
      "loss": 0.5253,
      "step": 6159
    },
    {
      "epoch": 1.0687022900763359,
      "grad_norm": 0.7830794453620911,
      "learning_rate": 2.510065871602255e-05,
      "loss": 0.4572,
      "step": 6160
    },
    {
      "epoch": 1.0688757807078417,
      "grad_norm": 1.0383309125900269,
      "learning_rate": 2.5094070461778424e-05,
      "loss": 0.4138,
      "step": 6161
    },
    {
      "epoch": 1.0690492713393476,
      "grad_norm": 0.8762059807777405,
      "learning_rate": 2.5087481616361493e-05,
      "loss": 0.4138,
      "step": 6162
    },
    {
      "epoch": 1.0692227619708536,
      "grad_norm": 0.7487605214118958,
      "learning_rate": 2.5080892180536386e-05,
      "loss": 0.4791,
      "step": 6163
    },
    {
      "epoch": 1.0693962526023595,
      "grad_norm": 0.8426170349121094,
      "learning_rate": 2.5074302155067823e-05,
      "loss": 0.5573,
      "step": 6164
    },
    {
      "epoch": 1.0695697432338653,
      "grad_norm": 0.9255799651145935,
      "learning_rate": 2.506771154072058e-05,
      "loss": 0.5547,
      "step": 6165
    },
    {
      "epoch": 1.0697432338653712,
      "grad_norm": 0.8196790218353271,
      "learning_rate": 2.5061120338259512e-05,
      "loss": 0.5496,
      "step": 6166
    },
    {
      "epoch": 1.0699167244968772,
      "grad_norm": 0.6261298060417175,
      "learning_rate": 2.505452854844953e-05,
      "loss": 0.6918,
      "step": 6167
    },
    {
      "epoch": 1.070090215128383,
      "grad_norm": 1.5703450441360474,
      "learning_rate": 2.5047936172055613e-05,
      "loss": 0.3839,
      "step": 6168
    },
    {
      "epoch": 1.070263705759889,
      "grad_norm": 0.6691255569458008,
      "learning_rate": 2.504134320984283e-05,
      "loss": 0.5,
      "step": 6169
    },
    {
      "epoch": 1.0704371963913948,
      "grad_norm": 1.0396575927734375,
      "learning_rate": 2.5034749662576293e-05,
      "loss": 0.3956,
      "step": 6170
    },
    {
      "epoch": 1.0706106870229009,
      "grad_norm": 0.9529989957809448,
      "learning_rate": 2.5028155531021186e-05,
      "loss": 0.4274,
      "step": 6171
    },
    {
      "epoch": 1.0707841776544067,
      "grad_norm": 1.0187219381332397,
      "learning_rate": 2.5021560815942777e-05,
      "loss": 0.3755,
      "step": 6172
    },
    {
      "epoch": 1.0709576682859125,
      "grad_norm": 0.7338615655899048,
      "learning_rate": 2.5014965518106372e-05,
      "loss": 0.5005,
      "step": 6173
    },
    {
      "epoch": 1.0711311589174184,
      "grad_norm": 1.0722684860229492,
      "learning_rate": 2.5008369638277382e-05,
      "loss": 0.4725,
      "step": 6174
    },
    {
      "epoch": 1.0713046495489245,
      "grad_norm": 0.5729461908340454,
      "learning_rate": 2.500177317722126e-05,
      "loss": 0.62,
      "step": 6175
    },
    {
      "epoch": 1.0714781401804303,
      "grad_norm": 1.2102477550506592,
      "learning_rate": 2.4995176135703533e-05,
      "loss": 0.5601,
      "step": 6176
    },
    {
      "epoch": 1.0716516308119362,
      "grad_norm": 0.7282514572143555,
      "learning_rate": 2.4988578514489797e-05,
      "loss": 0.408,
      "step": 6177
    },
    {
      "epoch": 1.071825121443442,
      "grad_norm": 0.8264929056167603,
      "learning_rate": 2.498198031434571e-05,
      "loss": 0.5389,
      "step": 6178
    },
    {
      "epoch": 1.0719986120749478,
      "grad_norm": 0.5981985926628113,
      "learning_rate": 2.4975381536037e-05,
      "loss": 0.5933,
      "step": 6179
    },
    {
      "epoch": 1.072172102706454,
      "grad_norm": 0.8617196083068848,
      "learning_rate": 2.496878218032947e-05,
      "loss": 0.4402,
      "step": 6180
    },
    {
      "epoch": 1.0723455933379598,
      "grad_norm": 1.0304713249206543,
      "learning_rate": 2.4962182247988974e-05,
      "loss": 0.5398,
      "step": 6181
    },
    {
      "epoch": 1.0725190839694656,
      "grad_norm": 0.7486497163772583,
      "learning_rate": 2.495558173978145e-05,
      "loss": 0.5283,
      "step": 6182
    },
    {
      "epoch": 1.0726925746009714,
      "grad_norm": 0.7354799509048462,
      "learning_rate": 2.494898065647289e-05,
      "loss": 0.5323,
      "step": 6183
    },
    {
      "epoch": 1.0728660652324775,
      "grad_norm": 0.7985234260559082,
      "learning_rate": 2.494237899882935e-05,
      "loss": 0.4429,
      "step": 6184
    },
    {
      "epoch": 1.0730395558639834,
      "grad_norm": 1.1000069379806519,
      "learning_rate": 2.4935776767616978e-05,
      "loss": 0.3912,
      "step": 6185
    },
    {
      "epoch": 1.0732130464954892,
      "grad_norm": 0.9294086694717407,
      "learning_rate": 2.4929173963601958e-05,
      "loss": 0.4817,
      "step": 6186
    },
    {
      "epoch": 1.073386537126995,
      "grad_norm": 0.6434595584869385,
      "learning_rate": 2.4922570587550552e-05,
      "loss": 0.5696,
      "step": 6187
    },
    {
      "epoch": 1.0735600277585011,
      "grad_norm": 0.7728206515312195,
      "learning_rate": 2.4915966640229098e-05,
      "loss": 0.5353,
      "step": 6188
    },
    {
      "epoch": 1.073733518390007,
      "grad_norm": 0.5889047980308533,
      "learning_rate": 2.4909362122403984e-05,
      "loss": 0.6283,
      "step": 6189
    },
    {
      "epoch": 1.0739070090215128,
      "grad_norm": 0.8491992354393005,
      "learning_rate": 2.4902757034841674e-05,
      "loss": 0.4575,
      "step": 6190
    },
    {
      "epoch": 1.0740804996530187,
      "grad_norm": 0.7175102233886719,
      "learning_rate": 2.4896151378308706e-05,
      "loss": 0.5492,
      "step": 6191
    },
    {
      "epoch": 1.0742539902845247,
      "grad_norm": 0.6997820138931274,
      "learning_rate": 2.4889545153571657e-05,
      "loss": 0.4293,
      "step": 6192
    },
    {
      "epoch": 1.0744274809160306,
      "grad_norm": 0.9219583868980408,
      "learning_rate": 2.48829383613972e-05,
      "loss": 0.6298,
      "step": 6193
    },
    {
      "epoch": 1.0746009715475364,
      "grad_norm": 0.6821068525314331,
      "learning_rate": 2.4876331002552055e-05,
      "loss": 0.5533,
      "step": 6194
    },
    {
      "epoch": 1.0747744621790423,
      "grad_norm": 0.7255200147628784,
      "learning_rate": 2.486972307780301e-05,
      "loss": 0.4609,
      "step": 6195
    },
    {
      "epoch": 1.0749479528105481,
      "grad_norm": 0.6879765391349792,
      "learning_rate": 2.4863114587916933e-05,
      "loss": 0.3905,
      "step": 6196
    },
    {
      "epoch": 1.0751214434420542,
      "grad_norm": 0.6694192886352539,
      "learning_rate": 2.485650553366074e-05,
      "loss": 0.6221,
      "step": 6197
    },
    {
      "epoch": 1.07529493407356,
      "grad_norm": 0.8467466235160828,
      "learning_rate": 2.484989591580142e-05,
      "loss": 0.5143,
      "step": 6198
    },
    {
      "epoch": 1.0754684247050659,
      "grad_norm": 0.7441421747207642,
      "learning_rate": 2.484328573510603e-05,
      "loss": 0.4595,
      "step": 6199
    },
    {
      "epoch": 1.075641915336572,
      "grad_norm": 1.1482595205307007,
      "learning_rate": 2.4836674992341684e-05,
      "loss": 0.5396,
      "step": 6200
    },
    {
      "epoch": 1.0758154059680778,
      "grad_norm": 0.7369665503501892,
      "learning_rate": 2.483006368827557e-05,
      "loss": 0.4896,
      "step": 6201
    },
    {
      "epoch": 1.0759888965995836,
      "grad_norm": 0.897925078868866,
      "learning_rate": 2.4823451823674943e-05,
      "loss": 0.4695,
      "step": 6202
    },
    {
      "epoch": 1.0761623872310895,
      "grad_norm": 0.7122481465339661,
      "learning_rate": 2.4816839399307102e-05,
      "loss": 0.5557,
      "step": 6203
    },
    {
      "epoch": 1.0763358778625953,
      "grad_norm": 1.171522617340088,
      "learning_rate": 2.481022641593944e-05,
      "loss": 0.3496,
      "step": 6204
    },
    {
      "epoch": 1.0765093684941014,
      "grad_norm": 0.8862089514732361,
      "learning_rate": 2.48036128743394e-05,
      "loss": 0.4128,
      "step": 6205
    },
    {
      "epoch": 1.0766828591256072,
      "grad_norm": 0.7411527037620544,
      "learning_rate": 2.4796998775274482e-05,
      "loss": 0.4863,
      "step": 6206
    },
    {
      "epoch": 1.076856349757113,
      "grad_norm": 0.9385290145874023,
      "learning_rate": 2.4790384119512275e-05,
      "loss": 0.4824,
      "step": 6207
    },
    {
      "epoch": 1.077029840388619,
      "grad_norm": 0.7270717024803162,
      "learning_rate": 2.4783768907820403e-05,
      "loss": 0.5505,
      "step": 6208
    },
    {
      "epoch": 1.077203331020125,
      "grad_norm": 2.0359649658203125,
      "learning_rate": 2.4777153140966583e-05,
      "loss": 0.4534,
      "step": 6209
    },
    {
      "epoch": 1.0773768216516308,
      "grad_norm": 0.7701866626739502,
      "learning_rate": 2.4770536819718562e-05,
      "loss": 0.3992,
      "step": 6210
    },
    {
      "epoch": 1.0775503122831367,
      "grad_norm": 0.6886849999427795,
      "learning_rate": 2.476391994484419e-05,
      "loss": 0.6028,
      "step": 6211
    },
    {
      "epoch": 1.0777238029146425,
      "grad_norm": 2.4191465377807617,
      "learning_rate": 2.475730251711136e-05,
      "loss": 0.4359,
      "step": 6212
    },
    {
      "epoch": 1.0778972935461486,
      "grad_norm": 0.8629088997840881,
      "learning_rate": 2.4750684537288024e-05,
      "loss": 0.582,
      "step": 6213
    },
    {
      "epoch": 1.0780707841776545,
      "grad_norm": 0.7598599791526794,
      "learning_rate": 2.4744066006142218e-05,
      "loss": 0.4397,
      "step": 6214
    },
    {
      "epoch": 1.0782442748091603,
      "grad_norm": 0.9446561932563782,
      "learning_rate": 2.4737446924442025e-05,
      "loss": 0.4485,
      "step": 6215
    },
    {
      "epoch": 1.0784177654406661,
      "grad_norm": 0.7072758674621582,
      "learning_rate": 2.4730827292955592e-05,
      "loss": 0.4698,
      "step": 6216
    },
    {
      "epoch": 1.0785912560721722,
      "grad_norm": 0.8763805627822876,
      "learning_rate": 2.472420711245114e-05,
      "loss": 0.4739,
      "step": 6217
    },
    {
      "epoch": 1.078764746703678,
      "grad_norm": 0.7658042311668396,
      "learning_rate": 2.4717586383696947e-05,
      "loss": 0.7263,
      "step": 6218
    },
    {
      "epoch": 1.078938237335184,
      "grad_norm": 0.678760826587677,
      "learning_rate": 2.4710965107461354e-05,
      "loss": 0.5195,
      "step": 6219
    },
    {
      "epoch": 1.0791117279666897,
      "grad_norm": 2.6144840717315674,
      "learning_rate": 2.470434328451278e-05,
      "loss": 0.5101,
      "step": 6220
    },
    {
      "epoch": 1.0792852185981956,
      "grad_norm": 0.6626024842262268,
      "learning_rate": 2.469772091561968e-05,
      "loss": 0.5975,
      "step": 6221
    },
    {
      "epoch": 1.0794587092297017,
      "grad_norm": 1.007093071937561,
      "learning_rate": 2.4691098001550588e-05,
      "loss": 0.5303,
      "step": 6222
    },
    {
      "epoch": 1.0796321998612075,
      "grad_norm": 0.8530629873275757,
      "learning_rate": 2.4684474543074116e-05,
      "loss": 0.5444,
      "step": 6223
    },
    {
      "epoch": 1.0798056904927134,
      "grad_norm": 0.6781681180000305,
      "learning_rate": 2.4677850540958906e-05,
      "loss": 0.53,
      "step": 6224
    },
    {
      "epoch": 1.0799791811242192,
      "grad_norm": 0.6077238321304321,
      "learning_rate": 2.46712259959737e-05,
      "loss": 0.5985,
      "step": 6225
    },
    {
      "epoch": 1.0801526717557253,
      "grad_norm": 2.4894297122955322,
      "learning_rate": 2.4664600908887272e-05,
      "loss": 0.6018,
      "step": 6226
    },
    {
      "epoch": 1.0803261623872311,
      "grad_norm": 0.8022773265838623,
      "learning_rate": 2.465797528046847e-05,
      "loss": 0.4695,
      "step": 6227
    },
    {
      "epoch": 1.080499653018737,
      "grad_norm": 0.6410969495773315,
      "learning_rate": 2.4651349111486212e-05,
      "loss": 0.4201,
      "step": 6228
    },
    {
      "epoch": 1.0806731436502428,
      "grad_norm": 0.9513944387435913,
      "learning_rate": 2.4644722402709467e-05,
      "loss": 0.6313,
      "step": 6229
    },
    {
      "epoch": 1.0808466342817489,
      "grad_norm": 0.890672504901886,
      "learning_rate": 2.4638095154907276e-05,
      "loss": 0.4587,
      "step": 6230
    },
    {
      "epoch": 1.0810201249132547,
      "grad_norm": 1.2217200994491577,
      "learning_rate": 2.463146736884874e-05,
      "loss": 0.5266,
      "step": 6231
    },
    {
      "epoch": 1.0811936155447606,
      "grad_norm": 0.7415736317634583,
      "learning_rate": 2.4624839045303014e-05,
      "loss": 0.5656,
      "step": 6232
    },
    {
      "epoch": 1.0813671061762664,
      "grad_norm": 1.1575895547866821,
      "learning_rate": 2.4618210185039333e-05,
      "loss": 0.4934,
      "step": 6233
    },
    {
      "epoch": 1.0815405968077725,
      "grad_norm": 1.0706515312194824,
      "learning_rate": 2.4611580788826973e-05,
      "loss": 0.4232,
      "step": 6234
    },
    {
      "epoch": 1.0817140874392783,
      "grad_norm": 2.1923775672912598,
      "learning_rate": 2.4604950857435297e-05,
      "loss": 0.5995,
      "step": 6235
    },
    {
      "epoch": 1.0818875780707842,
      "grad_norm": 0.788577139377594,
      "learning_rate": 2.4598320391633702e-05,
      "loss": 0.5975,
      "step": 6236
    },
    {
      "epoch": 1.08206106870229,
      "grad_norm": 1.1212619543075562,
      "learning_rate": 2.4591689392191667e-05,
      "loss": 0.4346,
      "step": 6237
    },
    {
      "epoch": 1.0822345593337959,
      "grad_norm": 0.6778693795204163,
      "learning_rate": 2.4585057859878732e-05,
      "loss": 0.4893,
      "step": 6238
    },
    {
      "epoch": 1.082408049965302,
      "grad_norm": 0.6982914805412292,
      "learning_rate": 2.4578425795464487e-05,
      "loss": 0.5394,
      "step": 6239
    },
    {
      "epoch": 1.0825815405968078,
      "grad_norm": 0.6827831864356995,
      "learning_rate": 2.4571793199718593e-05,
      "loss": 0.53,
      "step": 6240
    },
    {
      "epoch": 1.0827550312283136,
      "grad_norm": 0.6521167159080505,
      "learning_rate": 2.4565160073410774e-05,
      "loss": 0.5121,
      "step": 6241
    },
    {
      "epoch": 1.0829285218598195,
      "grad_norm": 1.1949217319488525,
      "learning_rate": 2.4558526417310805e-05,
      "loss": 0.4103,
      "step": 6242
    },
    {
      "epoch": 1.0831020124913255,
      "grad_norm": 0.7318608164787292,
      "learning_rate": 2.4551892232188535e-05,
      "loss": 0.4344,
      "step": 6243
    },
    {
      "epoch": 1.0832755031228314,
      "grad_norm": 0.9500438570976257,
      "learning_rate": 2.4545257518813866e-05,
      "loss": 0.4824,
      "step": 6244
    },
    {
      "epoch": 1.0834489937543372,
      "grad_norm": 0.7673509120941162,
      "learning_rate": 2.453862227795677e-05,
      "loss": 0.5067,
      "step": 6245
    },
    {
      "epoch": 1.083622484385843,
      "grad_norm": 0.9132089614868164,
      "learning_rate": 2.4531986510387268e-05,
      "loss": 0.5088,
      "step": 6246
    },
    {
      "epoch": 1.0837959750173491,
      "grad_norm": 1.1912236213684082,
      "learning_rate": 2.452535021687545e-05,
      "loss": 0.5366,
      "step": 6247
    },
    {
      "epoch": 1.083969465648855,
      "grad_norm": 0.8719744086265564,
      "learning_rate": 2.4518713398191464e-05,
      "loss": 0.5454,
      "step": 6248
    },
    {
      "epoch": 1.0841429562803608,
      "grad_norm": 0.7627516388893127,
      "learning_rate": 2.4512076055105527e-05,
      "loss": 0.5272,
      "step": 6249
    },
    {
      "epoch": 1.0843164469118667,
      "grad_norm": 0.7611238956451416,
      "learning_rate": 2.45054381883879e-05,
      "loss": 0.4019,
      "step": 6250
    },
    {
      "epoch": 1.0844899375433728,
      "grad_norm": 0.6928776502609253,
      "learning_rate": 2.4498799798808926e-05,
      "loss": 0.5763,
      "step": 6251
    },
    {
      "epoch": 1.0846634281748786,
      "grad_norm": 1.986975908279419,
      "learning_rate": 2.4492160887138998e-05,
      "loss": 0.4725,
      "step": 6252
    },
    {
      "epoch": 1.0848369188063844,
      "grad_norm": 0.8200432658195496,
      "learning_rate": 2.4485521454148558e-05,
      "loss": 0.4753,
      "step": 6253
    },
    {
      "epoch": 1.0850104094378903,
      "grad_norm": 0.7308650612831116,
      "learning_rate": 2.447888150060813e-05,
      "loss": 0.577,
      "step": 6254
    },
    {
      "epoch": 1.0851839000693964,
      "grad_norm": 0.8015735149383545,
      "learning_rate": 2.4472241027288276e-05,
      "loss": 0.5227,
      "step": 6255
    },
    {
      "epoch": 1.0853573907009022,
      "grad_norm": 1.1953891515731812,
      "learning_rate": 2.4465600034959654e-05,
      "loss": 0.4601,
      "step": 6256
    },
    {
      "epoch": 1.085530881332408,
      "grad_norm": 0.8995559215545654,
      "learning_rate": 2.4458958524392937e-05,
      "loss": 0.4825,
      "step": 6257
    },
    {
      "epoch": 1.085704371963914,
      "grad_norm": 1.450422763824463,
      "learning_rate": 2.4452316496358885e-05,
      "loss": 0.4043,
      "step": 6258
    },
    {
      "epoch": 1.08587786259542,
      "grad_norm": 0.616809070110321,
      "learning_rate": 2.444567395162832e-05,
      "loss": 0.5177,
      "step": 6259
    },
    {
      "epoch": 1.0860513532269258,
      "grad_norm": 0.7927566766738892,
      "learning_rate": 2.443903089097211e-05,
      "loss": 0.4994,
      "step": 6260
    },
    {
      "epoch": 1.0862248438584317,
      "grad_norm": 0.7003375887870789,
      "learning_rate": 2.4432387315161194e-05,
      "loss": 0.5319,
      "step": 6261
    },
    {
      "epoch": 1.0863983344899375,
      "grad_norm": 0.7656951546669006,
      "learning_rate": 2.4425743224966567e-05,
      "loss": 0.5613,
      "step": 6262
    },
    {
      "epoch": 1.0865718251214433,
      "grad_norm": 0.8551862835884094,
      "learning_rate": 2.4419098621159275e-05,
      "loss": 0.5496,
      "step": 6263
    },
    {
      "epoch": 1.0867453157529494,
      "grad_norm": 0.6719685792922974,
      "learning_rate": 2.4412453504510447e-05,
      "loss": 0.3932,
      "step": 6264
    },
    {
      "epoch": 1.0869188063844553,
      "grad_norm": 0.7628917098045349,
      "learning_rate": 2.4405807875791246e-05,
      "loss": 0.5336,
      "step": 6265
    },
    {
      "epoch": 1.087092297015961,
      "grad_norm": 0.5696899890899658,
      "learning_rate": 2.43991617357729e-05,
      "loss": 0.477,
      "step": 6266
    },
    {
      "epoch": 1.087265787647467,
      "grad_norm": 1.0026742219924927,
      "learning_rate": 2.4392515085226722e-05,
      "loss": 0.4136,
      "step": 6267
    },
    {
      "epoch": 1.087439278278973,
      "grad_norm": 1.2803168296813965,
      "learning_rate": 2.4385867924924037e-05,
      "loss": 0.4993,
      "step": 6268
    },
    {
      "epoch": 1.0876127689104789,
      "grad_norm": 1.0770628452301025,
      "learning_rate": 2.4379220255636278e-05,
      "loss": 0.5979,
      "step": 6269
    },
    {
      "epoch": 1.0877862595419847,
      "grad_norm": 0.5608989596366882,
      "learning_rate": 2.43725720781349e-05,
      "loss": 0.5822,
      "step": 6270
    },
    {
      "epoch": 1.0879597501734906,
      "grad_norm": 3.370060920715332,
      "learning_rate": 2.4365923393191443e-05,
      "loss": 0.4653,
      "step": 6271
    },
    {
      "epoch": 1.0881332408049966,
      "grad_norm": 0.8479523658752441,
      "learning_rate": 2.4359274201577478e-05,
      "loss": 0.4254,
      "step": 6272
    },
    {
      "epoch": 1.0883067314365025,
      "grad_norm": 0.6120976209640503,
      "learning_rate": 2.4352624504064672e-05,
      "loss": 0.5345,
      "step": 6273
    },
    {
      "epoch": 1.0884802220680083,
      "grad_norm": 1.6839667558670044,
      "learning_rate": 2.4345974301424717e-05,
      "loss": 0.4978,
      "step": 6274
    },
    {
      "epoch": 1.0886537126995142,
      "grad_norm": 0.5557832717895508,
      "learning_rate": 2.433932359442938e-05,
      "loss": 0.5186,
      "step": 6275
    },
    {
      "epoch": 1.0888272033310202,
      "grad_norm": 0.9949855208396912,
      "learning_rate": 2.433267238385048e-05,
      "loss": 0.4625,
      "step": 6276
    },
    {
      "epoch": 1.089000693962526,
      "grad_norm": 0.7671464681625366,
      "learning_rate": 2.4326020670459912e-05,
      "loss": 0.4021,
      "step": 6277
    },
    {
      "epoch": 1.089174184594032,
      "grad_norm": 0.6967581510543823,
      "learning_rate": 2.4319368455029598e-05,
      "loss": 0.4733,
      "step": 6278
    },
    {
      "epoch": 1.0893476752255378,
      "grad_norm": 0.8932548761367798,
      "learning_rate": 2.4312715738331542e-05,
      "loss": 0.4131,
      "step": 6279
    },
    {
      "epoch": 1.0895211658570436,
      "grad_norm": 0.8050430417060852,
      "learning_rate": 2.43060625211378e-05,
      "loss": 0.4819,
      "step": 6280
    },
    {
      "epoch": 1.0896946564885497,
      "grad_norm": 0.7218684554100037,
      "learning_rate": 2.4299408804220485e-05,
      "loss": 0.4854,
      "step": 6281
    },
    {
      "epoch": 1.0898681471200555,
      "grad_norm": 1.1423853635787964,
      "learning_rate": 2.4292754588351768e-05,
      "loss": 0.4004,
      "step": 6282
    },
    {
      "epoch": 1.0900416377515614,
      "grad_norm": 0.8457189202308655,
      "learning_rate": 2.4286099874303876e-05,
      "loss": 0.495,
      "step": 6283
    },
    {
      "epoch": 1.0902151283830672,
      "grad_norm": 1.2640057802200317,
      "learning_rate": 2.42794446628491e-05,
      "loss": 0.5071,
      "step": 6284
    },
    {
      "epoch": 1.0903886190145733,
      "grad_norm": 1.1674327850341797,
      "learning_rate": 2.4272788954759793e-05,
      "loss": 0.4744,
      "step": 6285
    },
    {
      "epoch": 1.0905621096460791,
      "grad_norm": 0.755543053150177,
      "learning_rate": 2.426613275080834e-05,
      "loss": 0.4208,
      "step": 6286
    },
    {
      "epoch": 1.090735600277585,
      "grad_norm": 0.7371096611022949,
      "learning_rate": 2.4259476051767213e-05,
      "loss": 0.3998,
      "step": 6287
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 1.3511090278625488,
      "learning_rate": 2.4252818858408923e-05,
      "loss": 0.3707,
      "step": 6288
    },
    {
      "epoch": 1.091082581540597,
      "grad_norm": 0.7861598134040833,
      "learning_rate": 2.4246161171506054e-05,
      "loss": 0.5133,
      "step": 6289
    },
    {
      "epoch": 1.0912560721721027,
      "grad_norm": 1.5742945671081543,
      "learning_rate": 2.4239502991831233e-05,
      "loss": 0.3892,
      "step": 6290
    },
    {
      "epoch": 1.0914295628036086,
      "grad_norm": 0.6407701969146729,
      "learning_rate": 2.4232844320157146e-05,
      "loss": 0.572,
      "step": 6291
    },
    {
      "epoch": 1.0916030534351144,
      "grad_norm": 0.5764026045799255,
      "learning_rate": 2.4226185157256546e-05,
      "loss": 0.6453,
      "step": 6292
    },
    {
      "epoch": 1.0917765440666205,
      "grad_norm": 1.245748519897461,
      "learning_rate": 2.4219525503902234e-05,
      "loss": 0.3743,
      "step": 6293
    },
    {
      "epoch": 1.0919500346981263,
      "grad_norm": 0.7878282070159912,
      "learning_rate": 2.421286536086707e-05,
      "loss": 0.4815,
      "step": 6294
    },
    {
      "epoch": 1.0921235253296322,
      "grad_norm": 0.9979248046875,
      "learning_rate": 2.4206204728923974e-05,
      "loss": 0.4969,
      "step": 6295
    },
    {
      "epoch": 1.092297015961138,
      "grad_norm": 0.8535788655281067,
      "learning_rate": 2.4199543608845916e-05,
      "loss": 0.4816,
      "step": 6296
    },
    {
      "epoch": 1.0924705065926439,
      "grad_norm": 1.433529257774353,
      "learning_rate": 2.419288200140593e-05,
      "loss": 0.386,
      "step": 6297
    },
    {
      "epoch": 1.09264399722415,
      "grad_norm": 0.6729404926300049,
      "learning_rate": 2.4186219907377097e-05,
      "loss": 0.4941,
      "step": 6298
    },
    {
      "epoch": 1.0928174878556558,
      "grad_norm": 0.5830033421516418,
      "learning_rate": 2.4179557327532574e-05,
      "loss": 0.4778,
      "step": 6299
    },
    {
      "epoch": 1.0929909784871616,
      "grad_norm": 0.6071484684944153,
      "learning_rate": 2.4172894262645544e-05,
      "loss": 0.4766,
      "step": 6300
    },
    {
      "epoch": 1.0931644691186675,
      "grad_norm": 0.9821061491966248,
      "learning_rate": 2.4166230713489277e-05,
      "loss": 0.4204,
      "step": 6301
    },
    {
      "epoch": 1.0933379597501736,
      "grad_norm": 0.8359472155570984,
      "learning_rate": 2.4159566680837086e-05,
      "loss": 0.5692,
      "step": 6302
    },
    {
      "epoch": 1.0935114503816794,
      "grad_norm": 0.5361720323562622,
      "learning_rate": 2.415290216546233e-05,
      "loss": 0.6711,
      "step": 6303
    },
    {
      "epoch": 1.0936849410131853,
      "grad_norm": 0.7677605152130127,
      "learning_rate": 2.414623716813844e-05,
      "loss": 0.3818,
      "step": 6304
    },
    {
      "epoch": 1.093858431644691,
      "grad_norm": 0.8098879456520081,
      "learning_rate": 2.4139571689638893e-05,
      "loss": 0.4827,
      "step": 6305
    },
    {
      "epoch": 1.0940319222761972,
      "grad_norm": 0.8161999583244324,
      "learning_rate": 2.413290573073723e-05,
      "loss": 0.3973,
      "step": 6306
    },
    {
      "epoch": 1.094205412907703,
      "grad_norm": 0.6727495789527893,
      "learning_rate": 2.412623929220704e-05,
      "loss": 0.5421,
      "step": 6307
    },
    {
      "epoch": 1.0943789035392089,
      "grad_norm": 0.7811001539230347,
      "learning_rate": 2.4119572374821968e-05,
      "loss": 0.405,
      "step": 6308
    },
    {
      "epoch": 1.0945523941707147,
      "grad_norm": 0.706080973148346,
      "learning_rate": 2.411290497935573e-05,
      "loss": 0.5844,
      "step": 6309
    },
    {
      "epoch": 1.0947258848022208,
      "grad_norm": 0.9299414753913879,
      "learning_rate": 2.4106237106582072e-05,
      "loss": 0.4591,
      "step": 6310
    },
    {
      "epoch": 1.0948993754337266,
      "grad_norm": 2.2729573249816895,
      "learning_rate": 2.4099568757274812e-05,
      "loss": 0.3805,
      "step": 6311
    },
    {
      "epoch": 1.0950728660652325,
      "grad_norm": 0.7673837542533875,
      "learning_rate": 2.4092899932207824e-05,
      "loss": 0.4224,
      "step": 6312
    },
    {
      "epoch": 1.0952463566967383,
      "grad_norm": 0.6698319315910339,
      "learning_rate": 2.408623063215503e-05,
      "loss": 0.4814,
      "step": 6313
    },
    {
      "epoch": 1.0954198473282444,
      "grad_norm": 1.160279631614685,
      "learning_rate": 2.4079560857890405e-05,
      "loss": 0.5238,
      "step": 6314
    },
    {
      "epoch": 1.0955933379597502,
      "grad_norm": 0.9155153036117554,
      "learning_rate": 2.4072890610187997e-05,
      "loss": 0.4193,
      "step": 6315
    },
    {
      "epoch": 1.095766828591256,
      "grad_norm": 0.9269014596939087,
      "learning_rate": 2.406621988982188e-05,
      "loss": 0.4788,
      "step": 6316
    },
    {
      "epoch": 1.095940319222762,
      "grad_norm": 0.8309975862503052,
      "learning_rate": 2.405954869756621e-05,
      "loss": 0.5477,
      "step": 6317
    },
    {
      "epoch": 1.096113809854268,
      "grad_norm": 0.7591610550880432,
      "learning_rate": 2.405287703419518e-05,
      "loss": 0.4873,
      "step": 6318
    },
    {
      "epoch": 1.0962873004857738,
      "grad_norm": 1.0823493003845215,
      "learning_rate": 2.4046204900483052e-05,
      "loss": 0.5029,
      "step": 6319
    },
    {
      "epoch": 1.0964607911172797,
      "grad_norm": 0.8195936679840088,
      "learning_rate": 2.4039532297204125e-05,
      "loss": 0.3933,
      "step": 6320
    },
    {
      "epoch": 1.0966342817487855,
      "grad_norm": 0.8526312112808228,
      "learning_rate": 2.403285922513277e-05,
      "loss": 0.3937,
      "step": 6321
    },
    {
      "epoch": 1.0968077723802914,
      "grad_norm": 2.5832276344299316,
      "learning_rate": 2.4026185685043405e-05,
      "loss": 0.5116,
      "step": 6322
    },
    {
      "epoch": 1.0969812630117974,
      "grad_norm": 0.7093821167945862,
      "learning_rate": 2.40195116777105e-05,
      "loss": 0.6206,
      "step": 6323
    },
    {
      "epoch": 1.0971547536433033,
      "grad_norm": 0.8102952241897583,
      "learning_rate": 2.4012837203908582e-05,
      "loss": 0.5405,
      "step": 6324
    },
    {
      "epoch": 1.0973282442748091,
      "grad_norm": 0.992218554019928,
      "learning_rate": 2.4006162264412227e-05,
      "loss": 0.5253,
      "step": 6325
    },
    {
      "epoch": 1.097501734906315,
      "grad_norm": 0.6748207211494446,
      "learning_rate": 2.3999486859996073e-05,
      "loss": 0.4137,
      "step": 6326
    },
    {
      "epoch": 1.097675225537821,
      "grad_norm": 0.8782357573509216,
      "learning_rate": 2.3992810991434815e-05,
      "loss": 0.5492,
      "step": 6327
    },
    {
      "epoch": 1.0978487161693269,
      "grad_norm": 0.6080440282821655,
      "learning_rate": 2.3986134659503187e-05,
      "loss": 0.4718,
      "step": 6328
    },
    {
      "epoch": 1.0980222068008327,
      "grad_norm": 0.7995212078094482,
      "learning_rate": 2.397945786497598e-05,
      "loss": 0.4375,
      "step": 6329
    },
    {
      "epoch": 1.0981956974323386,
      "grad_norm": 1.1713985204696655,
      "learning_rate": 2.3972780608628057e-05,
      "loss": 0.5001,
      "step": 6330
    },
    {
      "epoch": 1.0983691880638446,
      "grad_norm": 0.8009462952613831,
      "learning_rate": 2.3966102891234314e-05,
      "loss": 0.5101,
      "step": 6331
    },
    {
      "epoch": 1.0985426786953505,
      "grad_norm": 0.7381086349487305,
      "learning_rate": 2.3959424713569708e-05,
      "loss": 0.5708,
      "step": 6332
    },
    {
      "epoch": 1.0987161693268563,
      "grad_norm": 0.7604125738143921,
      "learning_rate": 2.395274607640925e-05,
      "loss": 0.5179,
      "step": 6333
    },
    {
      "epoch": 1.0988896599583622,
      "grad_norm": 0.8270940184593201,
      "learning_rate": 2.3946066980528e-05,
      "loss": 0.546,
      "step": 6334
    },
    {
      "epoch": 1.0990631505898683,
      "grad_norm": 0.7609905004501343,
      "learning_rate": 2.393938742670109e-05,
      "loss": 0.4622,
      "step": 6335
    },
    {
      "epoch": 1.099236641221374,
      "grad_norm": 0.6999890208244324,
      "learning_rate": 2.3932707415703673e-05,
      "loss": 0.557,
      "step": 6336
    },
    {
      "epoch": 1.09941013185288,
      "grad_norm": 0.7146739363670349,
      "learning_rate": 2.3926026948310975e-05,
      "loss": 0.631,
      "step": 6337
    },
    {
      "epoch": 1.0995836224843858,
      "grad_norm": 0.9941717386245728,
      "learning_rate": 2.391934602529828e-05,
      "loss": 0.5343,
      "step": 6338
    },
    {
      "epoch": 1.0997571131158916,
      "grad_norm": 1.3625564575195312,
      "learning_rate": 2.3912664647440903e-05,
      "loss": 0.5497,
      "step": 6339
    },
    {
      "epoch": 1.0999306037473977,
      "grad_norm": 0.8814844489097595,
      "learning_rate": 2.3905982815514243e-05,
      "loss": 0.5535,
      "step": 6340
    },
    {
      "epoch": 1.1001040943789036,
      "grad_norm": 1.0550538301467896,
      "learning_rate": 2.3899300530293728e-05,
      "loss": 0.4424,
      "step": 6341
    },
    {
      "epoch": 1.1002775850104094,
      "grad_norm": 0.6645036935806274,
      "learning_rate": 2.3892617792554833e-05,
      "loss": 0.4606,
      "step": 6342
    },
    {
      "epoch": 1.1004510756419152,
      "grad_norm": 0.6874741911888123,
      "learning_rate": 2.3885934603073117e-05,
      "loss": 0.6188,
      "step": 6343
    },
    {
      "epoch": 1.1006245662734213,
      "grad_norm": 0.8363273739814758,
      "learning_rate": 2.3879250962624152e-05,
      "loss": 0.4849,
      "step": 6344
    },
    {
      "epoch": 1.1007980569049272,
      "grad_norm": 0.7517639994621277,
      "learning_rate": 2.38725668719836e-05,
      "loss": 0.4653,
      "step": 6345
    },
    {
      "epoch": 1.100971547536433,
      "grad_norm": 0.671183168888092,
      "learning_rate": 2.386588233192715e-05,
      "loss": 0.4808,
      "step": 6346
    },
    {
      "epoch": 1.1011450381679388,
      "grad_norm": 0.7739551067352295,
      "learning_rate": 2.3859197343230546e-05,
      "loss": 0.5072,
      "step": 6347
    },
    {
      "epoch": 1.101318528799445,
      "grad_norm": 0.5653104782104492,
      "learning_rate": 2.38525119066696e-05,
      "loss": 0.5321,
      "step": 6348
    },
    {
      "epoch": 1.1014920194309508,
      "grad_norm": 0.7437562942504883,
      "learning_rate": 2.3845826023020156e-05,
      "loss": 0.5099,
      "step": 6349
    },
    {
      "epoch": 1.1016655100624566,
      "grad_norm": 1.2061591148376465,
      "learning_rate": 2.3839139693058116e-05,
      "loss": 0.5231,
      "step": 6350
    },
    {
      "epoch": 1.1018390006939625,
      "grad_norm": 0.8140386343002319,
      "learning_rate": 2.3832452917559446e-05,
      "loss": 0.516,
      "step": 6351
    },
    {
      "epoch": 1.1020124913254685,
      "grad_norm": 0.7298310995101929,
      "learning_rate": 2.382576569730015e-05,
      "loss": 0.4113,
      "step": 6352
    },
    {
      "epoch": 1.1021859819569744,
      "grad_norm": 1.2516953945159912,
      "learning_rate": 2.3819078033056284e-05,
      "loss": 0.5157,
      "step": 6353
    },
    {
      "epoch": 1.1023594725884802,
      "grad_norm": 0.9569164514541626,
      "learning_rate": 2.3812389925603963e-05,
      "loss": 0.4233,
      "step": 6354
    },
    {
      "epoch": 1.102532963219986,
      "grad_norm": 0.7385008335113525,
      "learning_rate": 2.380570137571935e-05,
      "loss": 0.5309,
      "step": 6355
    },
    {
      "epoch": 1.102706453851492,
      "grad_norm": 0.854784369468689,
      "learning_rate": 2.3799012384178654e-05,
      "loss": 0.4039,
      "step": 6356
    },
    {
      "epoch": 1.102879944482998,
      "grad_norm": 0.8272437453269958,
      "learning_rate": 2.3792322951758152e-05,
      "loss": 0.4584,
      "step": 6357
    },
    {
      "epoch": 1.1030534351145038,
      "grad_norm": 0.6209151148796082,
      "learning_rate": 2.3785633079234144e-05,
      "loss": 0.5188,
      "step": 6358
    },
    {
      "epoch": 1.1032269257460097,
      "grad_norm": 0.6605640649795532,
      "learning_rate": 2.3778942767383012e-05,
      "loss": 0.6174,
      "step": 6359
    },
    {
      "epoch": 1.1034004163775155,
      "grad_norm": 0.8699027299880981,
      "learning_rate": 2.377225201698117e-05,
      "loss": 0.4495,
      "step": 6360
    },
    {
      "epoch": 1.1035739070090216,
      "grad_norm": 1.1163221597671509,
      "learning_rate": 2.3765560828805075e-05,
      "loss": 0.5166,
      "step": 6361
    },
    {
      "epoch": 1.1037473976405274,
      "grad_norm": 0.8650309443473816,
      "learning_rate": 2.3758869203631266e-05,
      "loss": 0.418,
      "step": 6362
    },
    {
      "epoch": 1.1039208882720333,
      "grad_norm": 0.6874659061431885,
      "learning_rate": 2.3752177142236303e-05,
      "loss": 0.453,
      "step": 6363
    },
    {
      "epoch": 1.1040943789035391,
      "grad_norm": 0.7917651534080505,
      "learning_rate": 2.3745484645396816e-05,
      "loss": 0.3976,
      "step": 6364
    },
    {
      "epoch": 1.1042678695350452,
      "grad_norm": 0.704910397529602,
      "learning_rate": 2.3738791713889467e-05,
      "loss": 0.5693,
      "step": 6365
    },
    {
      "epoch": 1.104441360166551,
      "grad_norm": 1.8388170003890991,
      "learning_rate": 2.373209834849098e-05,
      "loss": 0.5923,
      "step": 6366
    },
    {
      "epoch": 1.1046148507980569,
      "grad_norm": 0.5804650783538818,
      "learning_rate": 2.3725404549978143e-05,
      "loss": 0.5958,
      "step": 6367
    },
    {
      "epoch": 1.1047883414295627,
      "grad_norm": 0.6623818278312683,
      "learning_rate": 2.3718710319127755e-05,
      "loss": 0.4172,
      "step": 6368
    },
    {
      "epoch": 1.1049618320610688,
      "grad_norm": 1.1011728048324585,
      "learning_rate": 2.3712015656716703e-05,
      "loss": 0.48,
      "step": 6369
    },
    {
      "epoch": 1.1051353226925746,
      "grad_norm": 1.700384259223938,
      "learning_rate": 2.370532056352191e-05,
      "loss": 0.37,
      "step": 6370
    },
    {
      "epoch": 1.1053088133240805,
      "grad_norm": 0.6900466680526733,
      "learning_rate": 2.3698625040320346e-05,
      "loss": 0.4216,
      "step": 6371
    },
    {
      "epoch": 1.1054823039555863,
      "grad_norm": 1.1002362966537476,
      "learning_rate": 2.3691929087889042e-05,
      "loss": 0.5177,
      "step": 6372
    },
    {
      "epoch": 1.1056557945870924,
      "grad_norm": 0.7985047698020935,
      "learning_rate": 2.3685232707005064e-05,
      "loss": 0.4059,
      "step": 6373
    },
    {
      "epoch": 1.1058292852185982,
      "grad_norm": 0.8414343595504761,
      "learning_rate": 2.3678535898445533e-05,
      "loss": 0.4829,
      "step": 6374
    },
    {
      "epoch": 1.106002775850104,
      "grad_norm": 0.9657633304595947,
      "learning_rate": 2.367183866298763e-05,
      "loss": 0.5266,
      "step": 6375
    },
    {
      "epoch": 1.10617626648161,
      "grad_norm": 0.7242559194564819,
      "learning_rate": 2.3665141001408562e-05,
      "loss": 0.5453,
      "step": 6376
    },
    {
      "epoch": 1.106349757113116,
      "grad_norm": 0.6628023982048035,
      "learning_rate": 2.365844291448562e-05,
      "loss": 0.5432,
      "step": 6377
    },
    {
      "epoch": 1.1065232477446219,
      "grad_norm": 0.8673304915428162,
      "learning_rate": 2.3651744402996114e-05,
      "loss": 0.4769,
      "step": 6378
    },
    {
      "epoch": 1.1066967383761277,
      "grad_norm": 0.6545118093490601,
      "learning_rate": 2.3645045467717405e-05,
      "loss": 0.5405,
      "step": 6379
    },
    {
      "epoch": 1.1068702290076335,
      "grad_norm": 0.8671982884407043,
      "learning_rate": 2.3638346109426932e-05,
      "loss": 0.4698,
      "step": 6380
    },
    {
      "epoch": 1.1070437196391394,
      "grad_norm": 0.8583928942680359,
      "learning_rate": 2.3631646328902153e-05,
      "loss": 0.3796,
      "step": 6381
    },
    {
      "epoch": 1.1072172102706455,
      "grad_norm": 1.0254278182983398,
      "learning_rate": 2.362494612692058e-05,
      "loss": 0.3856,
      "step": 6382
    },
    {
      "epoch": 1.1073907009021513,
      "grad_norm": 0.7909040451049805,
      "learning_rate": 2.361824550425979e-05,
      "loss": 0.3735,
      "step": 6383
    },
    {
      "epoch": 1.1075641915336571,
      "grad_norm": 0.7472656965255737,
      "learning_rate": 2.361154446169739e-05,
      "loss": 0.385,
      "step": 6384
    },
    {
      "epoch": 1.107737682165163,
      "grad_norm": 0.8990941643714905,
      "learning_rate": 2.360484300001105e-05,
      "loss": 0.5165,
      "step": 6385
    },
    {
      "epoch": 1.107911172796669,
      "grad_norm": 0.8338322639465332,
      "learning_rate": 2.3598141119978482e-05,
      "loss": 0.4694,
      "step": 6386
    },
    {
      "epoch": 1.108084663428175,
      "grad_norm": 0.9645411968231201,
      "learning_rate": 2.3591438822377434e-05,
      "loss": 0.4554,
      "step": 6387
    },
    {
      "epoch": 1.1082581540596808,
      "grad_norm": 1.1773382425308228,
      "learning_rate": 2.3584736107985737e-05,
      "loss": 0.5756,
      "step": 6388
    },
    {
      "epoch": 1.1084316446911866,
      "grad_norm": 0.9414104223251343,
      "learning_rate": 2.3578032977581234e-05,
      "loss": 0.4437,
      "step": 6389
    },
    {
      "epoch": 1.1086051353226927,
      "grad_norm": 0.7613778114318848,
      "learning_rate": 2.357132943194183e-05,
      "loss": 0.3824,
      "step": 6390
    },
    {
      "epoch": 1.1087786259541985,
      "grad_norm": 0.6773701906204224,
      "learning_rate": 2.356462547184549e-05,
      "loss": 0.5557,
      "step": 6391
    },
    {
      "epoch": 1.1089521165857044,
      "grad_norm": 0.7761004567146301,
      "learning_rate": 2.35579210980702e-05,
      "loss": 0.502,
      "step": 6392
    },
    {
      "epoch": 1.1091256072172102,
      "grad_norm": 0.8524529933929443,
      "learning_rate": 2.355121631139403e-05,
      "loss": 0.4213,
      "step": 6393
    },
    {
      "epoch": 1.1092990978487163,
      "grad_norm": 1.0654125213623047,
      "learning_rate": 2.3544511112595068e-05,
      "loss": 0.3893,
      "step": 6394
    },
    {
      "epoch": 1.1094725884802221,
      "grad_norm": 0.7196467518806458,
      "learning_rate": 2.353780550245146e-05,
      "loss": 0.4558,
      "step": 6395
    },
    {
      "epoch": 1.109646079111728,
      "grad_norm": 0.6411495208740234,
      "learning_rate": 2.3531099481741403e-05,
      "loss": 0.4595,
      "step": 6396
    },
    {
      "epoch": 1.1098195697432338,
      "grad_norm": 0.68552166223526,
      "learning_rate": 2.352439305124313e-05,
      "loss": 0.4657,
      "step": 6397
    },
    {
      "epoch": 1.1099930603747397,
      "grad_norm": 0.7618826031684875,
      "learning_rate": 2.351768621173495e-05,
      "loss": 0.4104,
      "step": 6398
    },
    {
      "epoch": 1.1101665510062457,
      "grad_norm": 0.6081966161727905,
      "learning_rate": 2.3510978963995176e-05,
      "loss": 0.6011,
      "step": 6399
    },
    {
      "epoch": 1.1103400416377516,
      "grad_norm": 0.8520535826683044,
      "learning_rate": 2.3504271308802204e-05,
      "loss": 0.438,
      "step": 6400
    },
    {
      "epoch": 1.1105135322692574,
      "grad_norm": 0.724348783493042,
      "learning_rate": 2.3497563246934464e-05,
      "loss": 0.6761,
      "step": 6401
    },
    {
      "epoch": 1.1106870229007633,
      "grad_norm": 0.5886250138282776,
      "learning_rate": 2.3490854779170436e-05,
      "loss": 0.6243,
      "step": 6402
    },
    {
      "epoch": 1.1108605135322693,
      "grad_norm": 0.6331730484962463,
      "learning_rate": 2.348414590628864e-05,
      "loss": 0.5045,
      "step": 6403
    },
    {
      "epoch": 1.1110340041637752,
      "grad_norm": 0.7192178964614868,
      "learning_rate": 2.347743662906765e-05,
      "loss": 0.3782,
      "step": 6404
    },
    {
      "epoch": 1.111207494795281,
      "grad_norm": 1.171298623085022,
      "learning_rate": 2.347072694828609e-05,
      "loss": 0.3865,
      "step": 6405
    },
    {
      "epoch": 1.1113809854267869,
      "grad_norm": 0.6927591562271118,
      "learning_rate": 2.3464016864722625e-05,
      "loss": 0.4514,
      "step": 6406
    },
    {
      "epoch": 1.111554476058293,
      "grad_norm": 1.1788617372512817,
      "learning_rate": 2.3457306379155965e-05,
      "loss": 0.5404,
      "step": 6407
    },
    {
      "epoch": 1.1117279666897988,
      "grad_norm": 2.0231029987335205,
      "learning_rate": 2.345059549236487e-05,
      "loss": 0.3983,
      "step": 6408
    },
    {
      "epoch": 1.1119014573213046,
      "grad_norm": 0.7832633852958679,
      "learning_rate": 2.3443884205128148e-05,
      "loss": 0.4203,
      "step": 6409
    },
    {
      "epoch": 1.1120749479528105,
      "grad_norm": 0.8718054890632629,
      "learning_rate": 2.343717251822465e-05,
      "loss": 0.4708,
      "step": 6410
    },
    {
      "epoch": 1.1122484385843165,
      "grad_norm": 0.8739398717880249,
      "learning_rate": 2.343046043243328e-05,
      "loss": 0.506,
      "step": 6411
    },
    {
      "epoch": 1.1124219292158224,
      "grad_norm": 0.8444788455963135,
      "learning_rate": 2.3423747948532976e-05,
      "loss": 0.5104,
      "step": 6412
    },
    {
      "epoch": 1.1125954198473282,
      "grad_norm": 0.9693549275398254,
      "learning_rate": 2.3417035067302733e-05,
      "loss": 0.473,
      "step": 6413
    },
    {
      "epoch": 1.112768910478834,
      "grad_norm": 0.640600323677063,
      "learning_rate": 2.341032178952159e-05,
      "loss": 0.4905,
      "step": 6414
    },
    {
      "epoch": 1.11294240111034,
      "grad_norm": 0.6085249781608582,
      "learning_rate": 2.340360811596863e-05,
      "loss": 0.5171,
      "step": 6415
    },
    {
      "epoch": 1.113115891741846,
      "grad_norm": 0.8247771263122559,
      "learning_rate": 2.339689404742298e-05,
      "loss": 0.4896,
      "step": 6416
    },
    {
      "epoch": 1.1132893823733518,
      "grad_norm": 0.7656151652336121,
      "learning_rate": 2.3390179584663815e-05,
      "loss": 0.5387,
      "step": 6417
    },
    {
      "epoch": 1.1134628730048577,
      "grad_norm": 0.899502158164978,
      "learning_rate": 2.338346472847037e-05,
      "loss": 0.4705,
      "step": 6418
    },
    {
      "epoch": 1.1136363636363635,
      "grad_norm": 0.7397540807723999,
      "learning_rate": 2.3376749479621886e-05,
      "loss": 0.4369,
      "step": 6419
    },
    {
      "epoch": 1.1138098542678696,
      "grad_norm": 0.9227269291877747,
      "learning_rate": 2.3370033838897702e-05,
      "loss": 0.3817,
      "step": 6420
    },
    {
      "epoch": 1.1139833448993754,
      "grad_norm": 0.7966392636299133,
      "learning_rate": 2.3363317807077157e-05,
      "loss": 0.4976,
      "step": 6421
    },
    {
      "epoch": 1.1141568355308813,
      "grad_norm": 0.6516497731208801,
      "learning_rate": 2.3356601384939665e-05,
      "loss": 0.5044,
      "step": 6422
    },
    {
      "epoch": 1.1143303261623871,
      "grad_norm": 0.8058405518531799,
      "learning_rate": 2.3349884573264673e-05,
      "loss": 0.4901,
      "step": 6423
    },
    {
      "epoch": 1.1145038167938932,
      "grad_norm": 0.6853688955307007,
      "learning_rate": 2.3343167372831665e-05,
      "loss": 0.428,
      "step": 6424
    },
    {
      "epoch": 1.114677307425399,
      "grad_norm": 0.7880139946937561,
      "learning_rate": 2.3336449784420197e-05,
      "loss": 0.5093,
      "step": 6425
    },
    {
      "epoch": 1.114850798056905,
      "grad_norm": 0.7561163306236267,
      "learning_rate": 2.3329731808809836e-05,
      "loss": 0.4244,
      "step": 6426
    },
    {
      "epoch": 1.1150242886884107,
      "grad_norm": 1.0141123533248901,
      "learning_rate": 2.3323013446780226e-05,
      "loss": 0.4185,
      "step": 6427
    },
    {
      "epoch": 1.1151977793199168,
      "grad_norm": 0.9526662826538086,
      "learning_rate": 2.331629469911103e-05,
      "loss": 0.4655,
      "step": 6428
    },
    {
      "epoch": 1.1153712699514227,
      "grad_norm": 0.9247124791145325,
      "learning_rate": 2.3309575566581968e-05,
      "loss": 0.4409,
      "step": 6429
    },
    {
      "epoch": 1.1155447605829285,
      "grad_norm": 0.7481706738471985,
      "learning_rate": 2.330285604997281e-05,
      "loss": 0.5492,
      "step": 6430
    },
    {
      "epoch": 1.1157182512144344,
      "grad_norm": 1.311652660369873,
      "learning_rate": 2.329613615006336e-05,
      "loss": 0.4746,
      "step": 6431
    },
    {
      "epoch": 1.1158917418459404,
      "grad_norm": 0.8068498969078064,
      "learning_rate": 2.328941586763346e-05,
      "loss": 0.4562,
      "step": 6432
    },
    {
      "epoch": 1.1160652324774463,
      "grad_norm": 0.7765712141990662,
      "learning_rate": 2.3282695203463022e-05,
      "loss": 0.5546,
      "step": 6433
    },
    {
      "epoch": 1.1162387231089521,
      "grad_norm": 0.5867044925689697,
      "learning_rate": 2.3275974158331977e-05,
      "loss": 0.5376,
      "step": 6434
    },
    {
      "epoch": 1.116412213740458,
      "grad_norm": 0.7233651876449585,
      "learning_rate": 2.326925273302032e-05,
      "loss": 0.5057,
      "step": 6435
    },
    {
      "epoch": 1.116585704371964,
      "grad_norm": 0.7924695014953613,
      "learning_rate": 2.3262530928308068e-05,
      "loss": 0.4043,
      "step": 6436
    },
    {
      "epoch": 1.1167591950034699,
      "grad_norm": 0.843597948551178,
      "learning_rate": 2.32558087449753e-05,
      "loss": 0.5204,
      "step": 6437
    },
    {
      "epoch": 1.1169326856349757,
      "grad_norm": 0.6833871603012085,
      "learning_rate": 2.3249086183802137e-05,
      "loss": 0.4555,
      "step": 6438
    },
    {
      "epoch": 1.1171061762664816,
      "grad_norm": 0.8030936121940613,
      "learning_rate": 2.324236324556873e-05,
      "loss": 0.5671,
      "step": 6439
    },
    {
      "epoch": 1.1172796668979874,
      "grad_norm": 1.6954410076141357,
      "learning_rate": 2.32356399310553e-05,
      "loss": 0.5236,
      "step": 6440
    },
    {
      "epoch": 1.1174531575294935,
      "grad_norm": 0.8243252635002136,
      "learning_rate": 2.3228916241042078e-05,
      "loss": 0.4556,
      "step": 6441
    },
    {
      "epoch": 1.1176266481609993,
      "grad_norm": 0.6916935443878174,
      "learning_rate": 2.3222192176309367e-05,
      "loss": 0.5424,
      "step": 6442
    },
    {
      "epoch": 1.1178001387925052,
      "grad_norm": 0.9195966124534607,
      "learning_rate": 2.3215467737637498e-05,
      "loss": 0.3984,
      "step": 6443
    },
    {
      "epoch": 1.117973629424011,
      "grad_norm": 0.6890717148780823,
      "learning_rate": 2.320874292580685e-05,
      "loss": 0.6116,
      "step": 6444
    },
    {
      "epoch": 1.118147120055517,
      "grad_norm": 1.1159943342208862,
      "learning_rate": 2.320201774159785e-05,
      "loss": 0.5779,
      "step": 6445
    },
    {
      "epoch": 1.118320610687023,
      "grad_norm": 0.8788090348243713,
      "learning_rate": 2.3195292185790957e-05,
      "loss": 0.4939,
      "step": 6446
    },
    {
      "epoch": 1.1184941013185288,
      "grad_norm": 0.8379570841789246,
      "learning_rate": 2.318856625916668e-05,
      "loss": 0.509,
      "step": 6447
    },
    {
      "epoch": 1.1186675919500346,
      "grad_norm": 0.9516486525535583,
      "learning_rate": 2.318183996250558e-05,
      "loss": 0.3376,
      "step": 6448
    },
    {
      "epoch": 1.1188410825815407,
      "grad_norm": 0.7245095372200012,
      "learning_rate": 2.3175113296588244e-05,
      "loss": 0.6335,
      "step": 6449
    },
    {
      "epoch": 1.1190145732130465,
      "grad_norm": 0.6819574236869812,
      "learning_rate": 2.3168386262195307e-05,
      "loss": 0.5155,
      "step": 6450
    },
    {
      "epoch": 1.1191880638445524,
      "grad_norm": 0.7670549154281616,
      "learning_rate": 2.3161658860107457e-05,
      "loss": 0.5471,
      "step": 6451
    },
    {
      "epoch": 1.1193615544760582,
      "grad_norm": 0.7976487874984741,
      "learning_rate": 2.315493109110541e-05,
      "loss": 0.4602,
      "step": 6452
    },
    {
      "epoch": 1.1195350451075643,
      "grad_norm": 0.9008978009223938,
      "learning_rate": 2.314820295596993e-05,
      "loss": 0.4371,
      "step": 6453
    },
    {
      "epoch": 1.1197085357390701,
      "grad_norm": 0.8811214566230774,
      "learning_rate": 2.314147445548183e-05,
      "loss": 0.52,
      "step": 6454
    },
    {
      "epoch": 1.119882026370576,
      "grad_norm": 0.7722797393798828,
      "learning_rate": 2.313474559042196e-05,
      "loss": 0.5618,
      "step": 6455
    },
    {
      "epoch": 1.1200555170020818,
      "grad_norm": 0.7748006582260132,
      "learning_rate": 2.3128016361571213e-05,
      "loss": 0.5454,
      "step": 6456
    },
    {
      "epoch": 1.1202290076335877,
      "grad_norm": 0.789991021156311,
      "learning_rate": 2.312128676971052e-05,
      "loss": 0.4666,
      "step": 6457
    },
    {
      "epoch": 1.1204024982650937,
      "grad_norm": 0.7222610712051392,
      "learning_rate": 2.3114556815620863e-05,
      "loss": 0.562,
      "step": 6458
    },
    {
      "epoch": 1.1205759888965996,
      "grad_norm": 0.8120276927947998,
      "learning_rate": 2.310782650008326e-05,
      "loss": 0.5955,
      "step": 6459
    },
    {
      "epoch": 1.1207494795281054,
      "grad_norm": 0.70384281873703,
      "learning_rate": 2.3101095823878764e-05,
      "loss": 0.5282,
      "step": 6460
    },
    {
      "epoch": 1.1209229701596113,
      "grad_norm": 0.7047963738441467,
      "learning_rate": 2.3094364787788487e-05,
      "loss": 0.6172,
      "step": 6461
    },
    {
      "epoch": 1.1210964607911174,
      "grad_norm": 0.7741065621376038,
      "learning_rate": 2.308763339259357e-05,
      "loss": 0.6201,
      "step": 6462
    },
    {
      "epoch": 1.1212699514226232,
      "grad_norm": 0.6245689392089844,
      "learning_rate": 2.30809016390752e-05,
      "loss": 0.6821,
      "step": 6463
    },
    {
      "epoch": 1.121443442054129,
      "grad_norm": 1.0588994026184082,
      "learning_rate": 2.3074169528014605e-05,
      "loss": 0.5713,
      "step": 6464
    },
    {
      "epoch": 1.121616932685635,
      "grad_norm": 0.6639510989189148,
      "learning_rate": 2.3067437060193055e-05,
      "loss": 0.5079,
      "step": 6465
    },
    {
      "epoch": 1.121790423317141,
      "grad_norm": 0.6362642645835876,
      "learning_rate": 2.306070423639186e-05,
      "loss": 0.4519,
      "step": 6466
    },
    {
      "epoch": 1.1219639139486468,
      "grad_norm": 0.654573917388916,
      "learning_rate": 2.3053971057392368e-05,
      "loss": 0.5739,
      "step": 6467
    },
    {
      "epoch": 1.1221374045801527,
      "grad_norm": 0.9201018810272217,
      "learning_rate": 2.3047237523975984e-05,
      "loss": 0.5176,
      "step": 6468
    },
    {
      "epoch": 1.1223108952116585,
      "grad_norm": 0.6924317479133606,
      "learning_rate": 2.3040503636924126e-05,
      "loss": 0.5597,
      "step": 6469
    },
    {
      "epoch": 1.1224843858431646,
      "grad_norm": 0.6017991900444031,
      "learning_rate": 2.3033769397018286e-05,
      "loss": 0.6281,
      "step": 6470
    },
    {
      "epoch": 1.1226578764746704,
      "grad_norm": 0.8645722270011902,
      "learning_rate": 2.3027034805039965e-05,
      "loss": 0.4135,
      "step": 6471
    },
    {
      "epoch": 1.1228313671061763,
      "grad_norm": 0.7092148661613464,
      "learning_rate": 2.3020299861770732e-05,
      "loss": 0.4896,
      "step": 6472
    },
    {
      "epoch": 1.123004857737682,
      "grad_norm": 1.2765874862670898,
      "learning_rate": 2.3013564567992184e-05,
      "loss": 0.4609,
      "step": 6473
    },
    {
      "epoch": 1.123178348369188,
      "grad_norm": 0.8498753905296326,
      "learning_rate": 2.300682892448595e-05,
      "loss": 0.5022,
      "step": 6474
    },
    {
      "epoch": 1.123351839000694,
      "grad_norm": 0.6794124245643616,
      "learning_rate": 2.3000092932033718e-05,
      "loss": 0.5402,
      "step": 6475
    },
    {
      "epoch": 1.1235253296321999,
      "grad_norm": 0.8482364416122437,
      "learning_rate": 2.2993356591417203e-05,
      "loss": 0.5017,
      "step": 6476
    },
    {
      "epoch": 1.1236988202637057,
      "grad_norm": 0.9400708675384521,
      "learning_rate": 2.2986619903418172e-05,
      "loss": 0.373,
      "step": 6477
    },
    {
      "epoch": 1.1238723108952116,
      "grad_norm": 0.742933452129364,
      "learning_rate": 2.2979882868818422e-05,
      "loss": 0.5658,
      "step": 6478
    },
    {
      "epoch": 1.1240458015267176,
      "grad_norm": 1.7342190742492676,
      "learning_rate": 2.2973145488399792e-05,
      "loss": 0.437,
      "step": 6479
    },
    {
      "epoch": 1.1242192921582235,
      "grad_norm": 0.7767460942268372,
      "learning_rate": 2.296640776294416e-05,
      "loss": 0.5402,
      "step": 6480
    },
    {
      "epoch": 1.1243927827897293,
      "grad_norm": 0.9296532273292542,
      "learning_rate": 2.2959669693233453e-05,
      "loss": 0.6931,
      "step": 6481
    },
    {
      "epoch": 1.1245662734212352,
      "grad_norm": 0.8746662735939026,
      "learning_rate": 2.2952931280049628e-05,
      "loss": 0.5812,
      "step": 6482
    },
    {
      "epoch": 1.1247397640527412,
      "grad_norm": 1.2094800472259521,
      "learning_rate": 2.294619252417469e-05,
      "loss": 0.4949,
      "step": 6483
    },
    {
      "epoch": 1.124913254684247,
      "grad_norm": 0.6965013742446899,
      "learning_rate": 2.293945342639067e-05,
      "loss": 0.4502,
      "step": 6484
    },
    {
      "epoch": 1.125086745315753,
      "grad_norm": 1.1303836107254028,
      "learning_rate": 2.2932713987479664e-05,
      "loss": 0.5649,
      "step": 6485
    },
    {
      "epoch": 1.1252602359472588,
      "grad_norm": 0.8138877749443054,
      "learning_rate": 2.2925974208223778e-05,
      "loss": 0.5535,
      "step": 6486
    },
    {
      "epoch": 1.1254337265787648,
      "grad_norm": 0.7776005268096924,
      "learning_rate": 2.2919234089405173e-05,
      "loss": 0.5015,
      "step": 6487
    },
    {
      "epoch": 1.1256072172102707,
      "grad_norm": 0.7546865940093994,
      "learning_rate": 2.2912493631806055e-05,
      "loss": 0.4994,
      "step": 6488
    },
    {
      "epoch": 1.1257807078417765,
      "grad_norm": 0.8697831034660339,
      "learning_rate": 2.290575283620865e-05,
      "loss": 0.3845,
      "step": 6489
    },
    {
      "epoch": 1.1259541984732824,
      "grad_norm": 0.6860870122909546,
      "learning_rate": 2.2899011703395254e-05,
      "loss": 0.51,
      "step": 6490
    },
    {
      "epoch": 1.1261276891047882,
      "grad_norm": 1.552304983139038,
      "learning_rate": 2.289227023414816e-05,
      "loss": 0.5311,
      "step": 6491
    },
    {
      "epoch": 1.1263011797362943,
      "grad_norm": 1.3038265705108643,
      "learning_rate": 2.288552842924974e-05,
      "loss": 0.5414,
      "step": 6492
    },
    {
      "epoch": 1.1264746703678001,
      "grad_norm": 0.7931041121482849,
      "learning_rate": 2.2878786289482384e-05,
      "loss": 0.4524,
      "step": 6493
    },
    {
      "epoch": 1.126648160999306,
      "grad_norm": 0.7245680093765259,
      "learning_rate": 2.2872043815628525e-05,
      "loss": 0.5699,
      "step": 6494
    },
    {
      "epoch": 1.126821651630812,
      "grad_norm": 1.0535913705825806,
      "learning_rate": 2.2865301008470633e-05,
      "loss": 0.4155,
      "step": 6495
    },
    {
      "epoch": 1.126995142262318,
      "grad_norm": 0.80494225025177,
      "learning_rate": 2.2858557868791222e-05,
      "loss": 0.45,
      "step": 6496
    },
    {
      "epoch": 1.1271686328938237,
      "grad_norm": 0.7536813020706177,
      "learning_rate": 2.2851814397372838e-05,
      "loss": 0.5144,
      "step": 6497
    },
    {
      "epoch": 1.1273421235253296,
      "grad_norm": 1.258178949356079,
      "learning_rate": 2.284507059499807e-05,
      "loss": 0.5242,
      "step": 6498
    },
    {
      "epoch": 1.1275156141568354,
      "grad_norm": 0.7766262292861938,
      "learning_rate": 2.283832646244955e-05,
      "loss": 0.5414,
      "step": 6499
    },
    {
      "epoch": 1.1276891047883415,
      "grad_norm": 0.6833403706550598,
      "learning_rate": 2.283158200050993e-05,
      "loss": 0.5,
      "step": 6500
    },
    {
      "epoch": 1.1278625954198473,
      "grad_norm": 0.7784548401832581,
      "learning_rate": 2.2824837209961924e-05,
      "loss": 0.4131,
      "step": 6501
    },
    {
      "epoch": 1.1280360860513532,
      "grad_norm": 0.8495446443557739,
      "learning_rate": 2.2818092091588266e-05,
      "loss": 0.4895,
      "step": 6502
    },
    {
      "epoch": 1.128209576682859,
      "grad_norm": 0.7297409176826477,
      "learning_rate": 2.2811346646171734e-05,
      "loss": 0.5499,
      "step": 6503
    },
    {
      "epoch": 1.128383067314365,
      "grad_norm": 1.2626932859420776,
      "learning_rate": 2.280460087449515e-05,
      "loss": 0.6128,
      "step": 6504
    },
    {
      "epoch": 1.128556557945871,
      "grad_norm": 0.6915184259414673,
      "learning_rate": 2.2797854777341368e-05,
      "loss": 0.5267,
      "step": 6505
    },
    {
      "epoch": 1.1287300485773768,
      "grad_norm": 0.9704550504684448,
      "learning_rate": 2.2791108355493278e-05,
      "loss": 0.3778,
      "step": 6506
    },
    {
      "epoch": 1.1289035392088826,
      "grad_norm": 0.6383293271064758,
      "learning_rate": 2.2784361609733812e-05,
      "loss": 0.6055,
      "step": 6507
    },
    {
      "epoch": 1.1290770298403887,
      "grad_norm": 0.7976787686347961,
      "learning_rate": 2.2777614540845934e-05,
      "loss": 0.4866,
      "step": 6508
    },
    {
      "epoch": 1.1292505204718946,
      "grad_norm": 1.0124448537826538,
      "learning_rate": 2.2770867149612658e-05,
      "loss": 0.4137,
      "step": 6509
    },
    {
      "epoch": 1.1294240111034004,
      "grad_norm": 1.5829224586486816,
      "learning_rate": 2.2764119436817015e-05,
      "loss": 0.3836,
      "step": 6510
    },
    {
      "epoch": 1.1295975017349063,
      "grad_norm": 0.8856865167617798,
      "learning_rate": 2.2757371403242094e-05,
      "loss": 0.5831,
      "step": 6511
    },
    {
      "epoch": 1.1297709923664123,
      "grad_norm": 0.6782200932502747,
      "learning_rate": 2.2750623049671003e-05,
      "loss": 0.5133,
      "step": 6512
    },
    {
      "epoch": 1.1299444829979182,
      "grad_norm": 0.9485509991645813,
      "learning_rate": 2.2743874376886903e-05,
      "loss": 0.5314,
      "step": 6513
    },
    {
      "epoch": 1.130117973629424,
      "grad_norm": 1.1593315601348877,
      "learning_rate": 2.273712538567299e-05,
      "loss": 0.4221,
      "step": 6514
    },
    {
      "epoch": 1.1302914642609299,
      "grad_norm": 0.564075767993927,
      "learning_rate": 2.273037607681248e-05,
      "loss": 0.6469,
      "step": 6515
    },
    {
      "epoch": 1.1304649548924357,
      "grad_norm": 1.0049749612808228,
      "learning_rate": 2.2723626451088644e-05,
      "loss": 0.462,
      "step": 6516
    },
    {
      "epoch": 1.1306384455239418,
      "grad_norm": 3.31482195854187,
      "learning_rate": 2.2716876509284794e-05,
      "loss": 0.5275,
      "step": 6517
    },
    {
      "epoch": 1.1308119361554476,
      "grad_norm": 0.814273476600647,
      "learning_rate": 2.2710126252184255e-05,
      "loss": 0.4277,
      "step": 6518
    },
    {
      "epoch": 1.1309854267869535,
      "grad_norm": 3.442901134490967,
      "learning_rate": 2.2703375680570402e-05,
      "loss": 0.5223,
      "step": 6519
    },
    {
      "epoch": 1.1311589174184593,
      "grad_norm": 1.0197612047195435,
      "learning_rate": 2.2696624795226662e-05,
      "loss": 0.4286,
      "step": 6520
    },
    {
      "epoch": 1.1313324080499654,
      "grad_norm": 0.8209486603736877,
      "learning_rate": 2.2689873596936458e-05,
      "loss": 0.4031,
      "step": 6521
    },
    {
      "epoch": 1.1315058986814712,
      "grad_norm": 0.786411464214325,
      "learning_rate": 2.2683122086483297e-05,
      "loss": 0.4977,
      "step": 6522
    },
    {
      "epoch": 1.131679389312977,
      "grad_norm": 0.7514742016792297,
      "learning_rate": 2.2676370264650694e-05,
      "loss": 0.4406,
      "step": 6523
    },
    {
      "epoch": 1.131852879944483,
      "grad_norm": 0.9746724367141724,
      "learning_rate": 2.26696181322222e-05,
      "loss": 0.4857,
      "step": 6524
    },
    {
      "epoch": 1.132026370575989,
      "grad_norm": 0.7940140962600708,
      "learning_rate": 2.266286568998141e-05,
      "loss": 0.4613,
      "step": 6525
    },
    {
      "epoch": 1.1321998612074948,
      "grad_norm": 1.2118453979492188,
      "learning_rate": 2.2656112938711952e-05,
      "loss": 0.4551,
      "step": 6526
    },
    {
      "epoch": 1.1323733518390007,
      "grad_norm": 1.165552020072937,
      "learning_rate": 2.2649359879197497e-05,
      "loss": 0.5624,
      "step": 6527
    },
    {
      "epoch": 1.1325468424705065,
      "grad_norm": 0.8620389699935913,
      "learning_rate": 2.264260651222174e-05,
      "loss": 0.4746,
      "step": 6528
    },
    {
      "epoch": 1.1327203331020126,
      "grad_norm": 0.7156081795692444,
      "learning_rate": 2.263585283856841e-05,
      "loss": 0.5294,
      "step": 6529
    },
    {
      "epoch": 1.1328938237335184,
      "grad_norm": 0.8914456963539124,
      "learning_rate": 2.26290988590213e-05,
      "loss": 0.5168,
      "step": 6530
    },
    {
      "epoch": 1.1330673143650243,
      "grad_norm": 1.0415455102920532,
      "learning_rate": 2.2622344574364197e-05,
      "loss": 0.4011,
      "step": 6531
    },
    {
      "epoch": 1.1332408049965301,
      "grad_norm": 0.8060207366943359,
      "learning_rate": 2.261558998538095e-05,
      "loss": 0.4276,
      "step": 6532
    },
    {
      "epoch": 1.133414295628036,
      "grad_norm": 0.8099369406700134,
      "learning_rate": 2.2608835092855443e-05,
      "loss": 0.5209,
      "step": 6533
    },
    {
      "epoch": 1.133587786259542,
      "grad_norm": 0.6553003191947937,
      "learning_rate": 2.2602079897571576e-05,
      "loss": 0.5065,
      "step": 6534
    },
    {
      "epoch": 1.1337612768910479,
      "grad_norm": 1.8057334423065186,
      "learning_rate": 2.2595324400313306e-05,
      "loss": 0.4036,
      "step": 6535
    },
    {
      "epoch": 1.1339347675225537,
      "grad_norm": 0.6745595335960388,
      "learning_rate": 2.258856860186462e-05,
      "loss": 0.675,
      "step": 6536
    },
    {
      "epoch": 1.1341082581540598,
      "grad_norm": 0.9904066920280457,
      "learning_rate": 2.2581812503009527e-05,
      "loss": 0.5397,
      "step": 6537
    },
    {
      "epoch": 1.1342817487855656,
      "grad_norm": 0.7607759833335876,
      "learning_rate": 2.257505610453209e-05,
      "loss": 0.5304,
      "step": 6538
    },
    {
      "epoch": 1.1344552394170715,
      "grad_norm": 1.4457356929779053,
      "learning_rate": 2.2568299407216384e-05,
      "loss": 0.5128,
      "step": 6539
    },
    {
      "epoch": 1.1346287300485773,
      "grad_norm": 0.9140667915344238,
      "learning_rate": 2.2561542411846537e-05,
      "loss": 0.4609,
      "step": 6540
    },
    {
      "epoch": 1.1348022206800832,
      "grad_norm": 0.7092764377593994,
      "learning_rate": 2.255478511920672e-05,
      "loss": 0.49,
      "step": 6541
    },
    {
      "epoch": 1.1349757113115893,
      "grad_norm": 0.6851636171340942,
      "learning_rate": 2.25480275300811e-05,
      "loss": 0.4692,
      "step": 6542
    },
    {
      "epoch": 1.135149201943095,
      "grad_norm": 1.046521544456482,
      "learning_rate": 2.254126964525393e-05,
      "loss": 0.4111,
      "step": 6543
    },
    {
      "epoch": 1.135322692574601,
      "grad_norm": 0.7946211695671082,
      "learning_rate": 2.253451146550945e-05,
      "loss": 0.5338,
      "step": 6544
    },
    {
      "epoch": 1.1354961832061068,
      "grad_norm": 0.7515681982040405,
      "learning_rate": 2.2527752991631958e-05,
      "loss": 0.5461,
      "step": 6545
    },
    {
      "epoch": 1.1356696738376129,
      "grad_norm": 0.5754083395004272,
      "learning_rate": 2.252099422440579e-05,
      "loss": 0.5704,
      "step": 6546
    },
    {
      "epoch": 1.1358431644691187,
      "grad_norm": 0.8206495046615601,
      "learning_rate": 2.2514235164615305e-05,
      "loss": 0.6459,
      "step": 6547
    },
    {
      "epoch": 1.1360166551006246,
      "grad_norm": 0.8565614819526672,
      "learning_rate": 2.2507475813044896e-05,
      "loss": 0.4427,
      "step": 6548
    },
    {
      "epoch": 1.1361901457321304,
      "grad_norm": 0.9310941696166992,
      "learning_rate": 2.2500716170479e-05,
      "loss": 0.5603,
      "step": 6549
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 0.9128825068473816,
      "learning_rate": 2.2493956237702075e-05,
      "loss": 0.4984,
      "step": 6550
    },
    {
      "epoch": 1.1365371269951423,
      "grad_norm": 0.7900086045265198,
      "learning_rate": 2.248719601549863e-05,
      "loss": 0.379,
      "step": 6551
    },
    {
      "epoch": 1.1367106176266482,
      "grad_norm": 0.7916425466537476,
      "learning_rate": 2.2480435504653185e-05,
      "loss": 0.468,
      "step": 6552
    },
    {
      "epoch": 1.136884108258154,
      "grad_norm": 0.6726658940315247,
      "learning_rate": 2.2473674705950303e-05,
      "loss": 0.4086,
      "step": 6553
    },
    {
      "epoch": 1.13705759888966,
      "grad_norm": 0.7393019199371338,
      "learning_rate": 2.24669136201746e-05,
      "loss": 0.4841,
      "step": 6554
    },
    {
      "epoch": 1.137231089521166,
      "grad_norm": 0.7832701802253723,
      "learning_rate": 2.246015224811069e-05,
      "loss": 0.6074,
      "step": 6555
    },
    {
      "epoch": 1.1374045801526718,
      "grad_norm": 0.8238747715950012,
      "learning_rate": 2.2453390590543246e-05,
      "loss": 0.5914,
      "step": 6556
    },
    {
      "epoch": 1.1375780707841776,
      "grad_norm": 1.0960816144943237,
      "learning_rate": 2.2446628648256964e-05,
      "loss": 0.6648,
      "step": 6557
    },
    {
      "epoch": 1.1377515614156835,
      "grad_norm": 0.8119200468063354,
      "learning_rate": 2.243986642203658e-05,
      "loss": 0.4757,
      "step": 6558
    },
    {
      "epoch": 1.1379250520471895,
      "grad_norm": 1.2356538772583008,
      "learning_rate": 2.2433103912666852e-05,
      "loss": 0.433,
      "step": 6559
    },
    {
      "epoch": 1.1380985426786954,
      "grad_norm": 0.7967992424964905,
      "learning_rate": 2.2426341120932582e-05,
      "loss": 0.5477,
      "step": 6560
    },
    {
      "epoch": 1.1382720333102012,
      "grad_norm": 0.8230440616607666,
      "learning_rate": 2.2419578047618598e-05,
      "loss": 0.4103,
      "step": 6561
    },
    {
      "epoch": 1.138445523941707,
      "grad_norm": 1.0610239505767822,
      "learning_rate": 2.241281469350976e-05,
      "loss": 0.4462,
      "step": 6562
    },
    {
      "epoch": 1.1386190145732131,
      "grad_norm": 0.8043401837348938,
      "learning_rate": 2.2406051059390963e-05,
      "loss": 0.4216,
      "step": 6563
    },
    {
      "epoch": 1.138792505204719,
      "grad_norm": 0.8361603021621704,
      "learning_rate": 2.2399287146047137e-05,
      "loss": 0.4207,
      "step": 6564
    },
    {
      "epoch": 1.1389659958362248,
      "grad_norm": 0.9205870628356934,
      "learning_rate": 2.239252295426325e-05,
      "loss": 0.6172,
      "step": 6565
    },
    {
      "epoch": 1.1391394864677307,
      "grad_norm": 0.8399495482444763,
      "learning_rate": 2.2385758484824275e-05,
      "loss": 0.4995,
      "step": 6566
    },
    {
      "epoch": 1.1393129770992367,
      "grad_norm": 0.5843160152435303,
      "learning_rate": 2.2378993738515255e-05,
      "loss": 0.6514,
      "step": 6567
    },
    {
      "epoch": 1.1394864677307426,
      "grad_norm": 0.7770143747329712,
      "learning_rate": 2.2372228716121246e-05,
      "loss": 0.3884,
      "step": 6568
    },
    {
      "epoch": 1.1396599583622484,
      "grad_norm": 0.6687589883804321,
      "learning_rate": 2.2365463418427317e-05,
      "loss": 0.5896,
      "step": 6569
    },
    {
      "epoch": 1.1398334489937543,
      "grad_norm": 0.7951526641845703,
      "learning_rate": 2.235869784621861e-05,
      "loss": 0.5137,
      "step": 6570
    },
    {
      "epoch": 1.1400069396252603,
      "grad_norm": 0.7361968755722046,
      "learning_rate": 2.2351932000280266e-05,
      "loss": 0.5129,
      "step": 6571
    },
    {
      "epoch": 1.1401804302567662,
      "grad_norm": 0.8000061511993408,
      "learning_rate": 2.2345165881397475e-05,
      "loss": 0.3605,
      "step": 6572
    },
    {
      "epoch": 1.140353920888272,
      "grad_norm": 0.7454240322113037,
      "learning_rate": 2.233839949035545e-05,
      "loss": 0.4493,
      "step": 6573
    },
    {
      "epoch": 1.1405274115197779,
      "grad_norm": 0.924987256526947,
      "learning_rate": 2.2331632827939438e-05,
      "loss": 0.4083,
      "step": 6574
    },
    {
      "epoch": 1.1407009021512837,
      "grad_norm": 0.7753768563270569,
      "learning_rate": 2.232486589493472e-05,
      "loss": 0.5869,
      "step": 6575
    },
    {
      "epoch": 1.1408743927827898,
      "grad_norm": 0.604107141494751,
      "learning_rate": 2.231809869212661e-05,
      "loss": 0.4639,
      "step": 6576
    },
    {
      "epoch": 1.1410478834142956,
      "grad_norm": 2.0864806175231934,
      "learning_rate": 2.231133122030044e-05,
      "loss": 0.417,
      "step": 6577
    },
    {
      "epoch": 1.1412213740458015,
      "grad_norm": 0.7135872840881348,
      "learning_rate": 2.230456348024159e-05,
      "loss": 0.5907,
      "step": 6578
    },
    {
      "epoch": 1.1413948646773073,
      "grad_norm": 0.9648030996322632,
      "learning_rate": 2.2297795472735462e-05,
      "loss": 0.496,
      "step": 6579
    },
    {
      "epoch": 1.1415683553088134,
      "grad_norm": 0.7891243100166321,
      "learning_rate": 2.2291027198567502e-05,
      "loss": 0.4845,
      "step": 6580
    },
    {
      "epoch": 1.1417418459403192,
      "grad_norm": 1.6100218296051025,
      "learning_rate": 2.228425865852316e-05,
      "loss": 0.4495,
      "step": 6581
    },
    {
      "epoch": 1.141915336571825,
      "grad_norm": 0.6381610631942749,
      "learning_rate": 2.2277489853387932e-05,
      "loss": 0.5763,
      "step": 6582
    },
    {
      "epoch": 1.142088827203331,
      "grad_norm": 0.6607668995857239,
      "learning_rate": 2.2270720783947358e-05,
      "loss": 0.5686,
      "step": 6583
    },
    {
      "epoch": 1.142262317834837,
      "grad_norm": 0.9227593541145325,
      "learning_rate": 2.2263951450986987e-05,
      "loss": 0.5004,
      "step": 6584
    },
    {
      "epoch": 1.1424358084663429,
      "grad_norm": 0.7867388129234314,
      "learning_rate": 2.225718185529242e-05,
      "loss": 0.4375,
      "step": 6585
    },
    {
      "epoch": 1.1426092990978487,
      "grad_norm": 1.0895323753356934,
      "learning_rate": 2.2250411997649266e-05,
      "loss": 0.4457,
      "step": 6586
    },
    {
      "epoch": 1.1427827897293545,
      "grad_norm": 0.8809779286384583,
      "learning_rate": 2.2243641878843172e-05,
      "loss": 0.4347,
      "step": 6587
    },
    {
      "epoch": 1.1429562803608606,
      "grad_norm": 0.9329822659492493,
      "learning_rate": 2.2236871499659824e-05,
      "loss": 0.3826,
      "step": 6588
    },
    {
      "epoch": 1.1431297709923665,
      "grad_norm": 0.9002099633216858,
      "learning_rate": 2.2230100860884937e-05,
      "loss": 0.6514,
      "step": 6589
    },
    {
      "epoch": 1.1433032616238723,
      "grad_norm": 0.7116411328315735,
      "learning_rate": 2.2223329963304242e-05,
      "loss": 0.5087,
      "step": 6590
    },
    {
      "epoch": 1.1434767522553781,
      "grad_norm": 0.8289925456047058,
      "learning_rate": 2.2216558807703522e-05,
      "loss": 0.418,
      "step": 6591
    },
    {
      "epoch": 1.143650242886884,
      "grad_norm": 1.378301978111267,
      "learning_rate": 2.2209787394868562e-05,
      "loss": 0.3944,
      "step": 6592
    },
    {
      "epoch": 1.14382373351839,
      "grad_norm": 0.5482950210571289,
      "learning_rate": 2.22030157255852e-05,
      "loss": 0.486,
      "step": 6593
    },
    {
      "epoch": 1.143997224149896,
      "grad_norm": 1.0869117975234985,
      "learning_rate": 2.2196243800639303e-05,
      "loss": 0.4205,
      "step": 6594
    },
    {
      "epoch": 1.1441707147814018,
      "grad_norm": 0.9532647132873535,
      "learning_rate": 2.2189471620816745e-05,
      "loss": 0.4965,
      "step": 6595
    },
    {
      "epoch": 1.1443442054129078,
      "grad_norm": 0.9237571358680725,
      "learning_rate": 2.2182699186903462e-05,
      "loss": 0.4149,
      "step": 6596
    },
    {
      "epoch": 1.1445176960444137,
      "grad_norm": 0.6246567964553833,
      "learning_rate": 2.217592649968539e-05,
      "loss": 0.5579,
      "step": 6597
    },
    {
      "epoch": 1.1446911866759195,
      "grad_norm": 0.9275211095809937,
      "learning_rate": 2.2169153559948513e-05,
      "loss": 0.5593,
      "step": 6598
    },
    {
      "epoch": 1.1448646773074254,
      "grad_norm": 0.7967746257781982,
      "learning_rate": 2.2162380368478836e-05,
      "loss": 0.5905,
      "step": 6599
    },
    {
      "epoch": 1.1450381679389312,
      "grad_norm": 0.6600043177604675,
      "learning_rate": 2.21556069260624e-05,
      "loss": 0.4712,
      "step": 6600
    },
    {
      "epoch": 1.1452116585704373,
      "grad_norm": 0.8281075954437256,
      "learning_rate": 2.2148833233485273e-05,
      "loss": 0.5726,
      "step": 6601
    },
    {
      "epoch": 1.1453851492019431,
      "grad_norm": 0.8647617697715759,
      "learning_rate": 2.2142059291533542e-05,
      "loss": 0.4822,
      "step": 6602
    },
    {
      "epoch": 1.145558639833449,
      "grad_norm": 0.9288738965988159,
      "learning_rate": 2.2135285100993328e-05,
      "loss": 0.5446,
      "step": 6603
    },
    {
      "epoch": 1.1457321304649548,
      "grad_norm": 0.8471190929412842,
      "learning_rate": 2.2128510662650796e-05,
      "loss": 0.476,
      "step": 6604
    },
    {
      "epoch": 1.1459056210964609,
      "grad_norm": 1.2163231372833252,
      "learning_rate": 2.212173597729212e-05,
      "loss": 0.4095,
      "step": 6605
    },
    {
      "epoch": 1.1460791117279667,
      "grad_norm": 0.9635229706764221,
      "learning_rate": 2.211496104570351e-05,
      "loss": 0.5013,
      "step": 6606
    },
    {
      "epoch": 1.1462526023594726,
      "grad_norm": 0.6926249861717224,
      "learning_rate": 2.210818586867121e-05,
      "loss": 0.5568,
      "step": 6607
    },
    {
      "epoch": 1.1464260929909784,
      "grad_norm": 1.0889391899108887,
      "learning_rate": 2.210141044698148e-05,
      "loss": 0.4818,
      "step": 6608
    },
    {
      "epoch": 1.1465995836224843,
      "grad_norm": 0.873817503452301,
      "learning_rate": 2.2094634781420626e-05,
      "loss": 0.4248,
      "step": 6609
    },
    {
      "epoch": 1.1467730742539903,
      "grad_norm": 0.7545918822288513,
      "learning_rate": 2.2087858872774954e-05,
      "loss": 0.4852,
      "step": 6610
    },
    {
      "epoch": 1.1469465648854962,
      "grad_norm": 0.6246511936187744,
      "learning_rate": 2.2081082721830834e-05,
      "loss": 0.5958,
      "step": 6611
    },
    {
      "epoch": 1.147120055517002,
      "grad_norm": 0.6936851739883423,
      "learning_rate": 2.2074306329374636e-05,
      "loss": 0.5459,
      "step": 6612
    },
    {
      "epoch": 1.147293546148508,
      "grad_norm": 0.6763320565223694,
      "learning_rate": 2.2067529696192772e-05,
      "loss": 0.5985,
      "step": 6613
    },
    {
      "epoch": 1.147467036780014,
      "grad_norm": 0.9481201767921448,
      "learning_rate": 2.206075282307168e-05,
      "loss": 0.543,
      "step": 6614
    },
    {
      "epoch": 1.1476405274115198,
      "grad_norm": 1.1455981731414795,
      "learning_rate": 2.205397571079782e-05,
      "loss": 0.4767,
      "step": 6615
    },
    {
      "epoch": 1.1478140180430256,
      "grad_norm": 0.8845289945602417,
      "learning_rate": 2.2047198360157683e-05,
      "loss": 0.4875,
      "step": 6616
    },
    {
      "epoch": 1.1479875086745315,
      "grad_norm": 1.9621926546096802,
      "learning_rate": 2.2040420771937793e-05,
      "loss": 0.4832,
      "step": 6617
    },
    {
      "epoch": 1.1481609993060375,
      "grad_norm": 0.9349886775016785,
      "learning_rate": 2.2033642946924698e-05,
      "loss": 0.4557,
      "step": 6618
    },
    {
      "epoch": 1.1483344899375434,
      "grad_norm": 0.5708053708076477,
      "learning_rate": 2.2026864885904965e-05,
      "loss": 0.5789,
      "step": 6619
    },
    {
      "epoch": 1.1485079805690492,
      "grad_norm": 0.6791675686836243,
      "learning_rate": 2.2020086589665203e-05,
      "loss": 0.464,
      "step": 6620
    },
    {
      "epoch": 1.148681471200555,
      "grad_norm": 0.9479997754096985,
      "learning_rate": 2.2013308058992037e-05,
      "loss": 0.4625,
      "step": 6621
    },
    {
      "epoch": 1.1488549618320612,
      "grad_norm": 0.6508433222770691,
      "learning_rate": 2.2006529294672126e-05,
      "loss": 0.7183,
      "step": 6622
    },
    {
      "epoch": 1.149028452463567,
      "grad_norm": 0.7447717785835266,
      "learning_rate": 2.199975029749215e-05,
      "loss": 0.4012,
      "step": 6623
    },
    {
      "epoch": 1.1492019430950728,
      "grad_norm": 2.0179295539855957,
      "learning_rate": 2.1992971068238826e-05,
      "loss": 0.4464,
      "step": 6624
    },
    {
      "epoch": 1.1493754337265787,
      "grad_norm": 0.6852936148643494,
      "learning_rate": 2.198619160769888e-05,
      "loss": 0.5293,
      "step": 6625
    },
    {
      "epoch": 1.1495489243580848,
      "grad_norm": 0.7375458478927612,
      "learning_rate": 2.197941191665909e-05,
      "loss": 0.5673,
      "step": 6626
    },
    {
      "epoch": 1.1497224149895906,
      "grad_norm": 1.103060245513916,
      "learning_rate": 2.1972631995906237e-05,
      "loss": 0.4238,
      "step": 6627
    },
    {
      "epoch": 1.1498959056210964,
      "grad_norm": 0.7563209533691406,
      "learning_rate": 2.196585184622715e-05,
      "loss": 0.5815,
      "step": 6628
    },
    {
      "epoch": 1.1500693962526023,
      "grad_norm": 0.9774815440177917,
      "learning_rate": 2.1959071468408656e-05,
      "loss": 0.4817,
      "step": 6629
    },
    {
      "epoch": 1.1502428868841084,
      "grad_norm": 1.1103121042251587,
      "learning_rate": 2.195229086323764e-05,
      "loss": 0.6536,
      "step": 6630
    },
    {
      "epoch": 1.1504163775156142,
      "grad_norm": 0.6559248566627502,
      "learning_rate": 2.1945510031500992e-05,
      "loss": 0.589,
      "step": 6631
    },
    {
      "epoch": 1.15058986814712,
      "grad_norm": 0.7209063172340393,
      "learning_rate": 2.193872897398564e-05,
      "loss": 0.5403,
      "step": 6632
    },
    {
      "epoch": 1.150763358778626,
      "grad_norm": 0.6752347946166992,
      "learning_rate": 2.193194769147853e-05,
      "loss": 0.4971,
      "step": 6633
    },
    {
      "epoch": 1.1509368494101317,
      "grad_norm": 0.8687592148780823,
      "learning_rate": 2.1925166184766636e-05,
      "loss": 0.5104,
      "step": 6634
    },
    {
      "epoch": 1.1511103400416378,
      "grad_norm": 0.8973918557167053,
      "learning_rate": 2.191838445463697e-05,
      "loss": 0.5557,
      "step": 6635
    },
    {
      "epoch": 1.1512838306731437,
      "grad_norm": 1.0971548557281494,
      "learning_rate": 2.1911602501876546e-05,
      "loss": 0.5561,
      "step": 6636
    },
    {
      "epoch": 1.1514573213046495,
      "grad_norm": 0.8915349841117859,
      "learning_rate": 2.190482032727243e-05,
      "loss": 0.4652,
      "step": 6637
    },
    {
      "epoch": 1.1516308119361554,
      "grad_norm": 1.8851386308670044,
      "learning_rate": 2.1898037931611688e-05,
      "loss": 0.488,
      "step": 6638
    },
    {
      "epoch": 1.1518043025676614,
      "grad_norm": 0.885982871055603,
      "learning_rate": 2.1891255315681443e-05,
      "loss": 0.5558,
      "step": 6639
    },
    {
      "epoch": 1.1519777931991673,
      "grad_norm": 0.9724069833755493,
      "learning_rate": 2.1884472480268806e-05,
      "loss": 0.4536,
      "step": 6640
    },
    {
      "epoch": 1.1521512838306731,
      "grad_norm": 0.7551723122596741,
      "learning_rate": 2.1877689426160943e-05,
      "loss": 0.538,
      "step": 6641
    },
    {
      "epoch": 1.152324774462179,
      "grad_norm": 0.8001475930213928,
      "learning_rate": 2.1870906154145035e-05,
      "loss": 0.5338,
      "step": 6642
    },
    {
      "epoch": 1.152498265093685,
      "grad_norm": 1.3312873840332031,
      "learning_rate": 2.1864122665008294e-05,
      "loss": 0.5228,
      "step": 6643
    },
    {
      "epoch": 1.1526717557251909,
      "grad_norm": 0.7069399356842041,
      "learning_rate": 2.185733895953794e-05,
      "loss": 0.6171,
      "step": 6644
    },
    {
      "epoch": 1.1528452463566967,
      "grad_norm": 0.9802401661872864,
      "learning_rate": 2.1850555038521236e-05,
      "loss": 0.5704,
      "step": 6645
    },
    {
      "epoch": 1.1530187369882026,
      "grad_norm": 0.6348333954811096,
      "learning_rate": 2.1843770902745462e-05,
      "loss": 0.585,
      "step": 6646
    },
    {
      "epoch": 1.1531922276197086,
      "grad_norm": 0.8841515183448792,
      "learning_rate": 2.183698655299793e-05,
      "loss": 0.5117,
      "step": 6647
    },
    {
      "epoch": 1.1533657182512145,
      "grad_norm": 0.8299418091773987,
      "learning_rate": 2.1830201990065966e-05,
      "loss": 0.5485,
      "step": 6648
    },
    {
      "epoch": 1.1535392088827203,
      "grad_norm": 0.7538990378379822,
      "learning_rate": 2.182341721473693e-05,
      "loss": 0.5186,
      "step": 6649
    },
    {
      "epoch": 1.1537126995142262,
      "grad_norm": 2.092827796936035,
      "learning_rate": 2.1816632227798196e-05,
      "loss": 0.6028,
      "step": 6650
    },
    {
      "epoch": 1.153886190145732,
      "grad_norm": 0.8751448392868042,
      "learning_rate": 2.1809847030037182e-05,
      "loss": 0.4607,
      "step": 6651
    },
    {
      "epoch": 1.154059680777238,
      "grad_norm": 2.1474010944366455,
      "learning_rate": 2.180306162224131e-05,
      "loss": 0.5428,
      "step": 6652
    },
    {
      "epoch": 1.154233171408744,
      "grad_norm": 0.8482744097709656,
      "learning_rate": 2.179627600519803e-05,
      "loss": 0.4346,
      "step": 6653
    },
    {
      "epoch": 1.1544066620402498,
      "grad_norm": 0.900211751461029,
      "learning_rate": 2.1789490179694833e-05,
      "loss": 0.5604,
      "step": 6654
    },
    {
      "epoch": 1.1545801526717558,
      "grad_norm": 0.6769447326660156,
      "learning_rate": 2.1782704146519212e-05,
      "loss": 0.6034,
      "step": 6655
    },
    {
      "epoch": 1.1547536433032617,
      "grad_norm": 0.7091245651245117,
      "learning_rate": 2.1775917906458698e-05,
      "loss": 0.4375,
      "step": 6656
    },
    {
      "epoch": 1.1549271339347675,
      "grad_norm": 0.8113880157470703,
      "learning_rate": 2.1769131460300844e-05,
      "loss": 0.6483,
      "step": 6657
    },
    {
      "epoch": 1.1551006245662734,
      "grad_norm": 0.8065187931060791,
      "learning_rate": 2.176234480883322e-05,
      "loss": 0.5001,
      "step": 6658
    },
    {
      "epoch": 1.1552741151977792,
      "grad_norm": 0.6611795425415039,
      "learning_rate": 2.175555795284343e-05,
      "loss": 0.5692,
      "step": 6659
    },
    {
      "epoch": 1.1554476058292853,
      "grad_norm": 0.6740517020225525,
      "learning_rate": 2.174877089311909e-05,
      "loss": 0.626,
      "step": 6660
    },
    {
      "epoch": 1.1556210964607911,
      "grad_norm": 1.0208956003189087,
      "learning_rate": 2.1741983630447852e-05,
      "loss": 0.5341,
      "step": 6661
    },
    {
      "epoch": 1.155794587092297,
      "grad_norm": 0.5938760042190552,
      "learning_rate": 2.1735196165617385e-05,
      "loss": 0.6029,
      "step": 6662
    },
    {
      "epoch": 1.1559680777238028,
      "grad_norm": 0.6635843515396118,
      "learning_rate": 2.172840849941538e-05,
      "loss": 0.52,
      "step": 6663
    },
    {
      "epoch": 1.156141568355309,
      "grad_norm": 0.7051963210105896,
      "learning_rate": 2.1721620632629552e-05,
      "loss": 0.6143,
      "step": 6664
    },
    {
      "epoch": 1.1563150589868147,
      "grad_norm": 0.934929370880127,
      "learning_rate": 2.171483256604765e-05,
      "loss": 0.5151,
      "step": 6665
    },
    {
      "epoch": 1.1564885496183206,
      "grad_norm": 0.6479410529136658,
      "learning_rate": 2.1708044300457423e-05,
      "loss": 0.5288,
      "step": 6666
    },
    {
      "epoch": 1.1566620402498264,
      "grad_norm": 0.6074390411376953,
      "learning_rate": 2.1701255836646672e-05,
      "loss": 0.5096,
      "step": 6667
    },
    {
      "epoch": 1.1568355308813323,
      "grad_norm": 1.3985155820846558,
      "learning_rate": 2.1694467175403197e-05,
      "loss": 0.5046,
      "step": 6668
    },
    {
      "epoch": 1.1570090215128384,
      "grad_norm": 0.6436254978179932,
      "learning_rate": 2.168767831751483e-05,
      "loss": 0.611,
      "step": 6669
    },
    {
      "epoch": 1.1571825121443442,
      "grad_norm": 0.7783954739570618,
      "learning_rate": 2.1680889263769425e-05,
      "loss": 0.5181,
      "step": 6670
    },
    {
      "epoch": 1.15735600277585,
      "grad_norm": 0.5539189577102661,
      "learning_rate": 2.1674100014954864e-05,
      "loss": 0.686,
      "step": 6671
    },
    {
      "epoch": 1.1575294934073561,
      "grad_norm": 1.0195469856262207,
      "learning_rate": 2.166731057185905e-05,
      "loss": 0.548,
      "step": 6672
    },
    {
      "epoch": 1.157702984038862,
      "grad_norm": 0.6438109278678894,
      "learning_rate": 2.16605209352699e-05,
      "loss": 0.5165,
      "step": 6673
    },
    {
      "epoch": 1.1578764746703678,
      "grad_norm": 1.053868293762207,
      "learning_rate": 2.1653731105975355e-05,
      "loss": 0.549,
      "step": 6674
    },
    {
      "epoch": 1.1580499653018737,
      "grad_norm": 0.8601235747337341,
      "learning_rate": 2.1646941084763397e-05,
      "loss": 0.546,
      "step": 6675
    },
    {
      "epoch": 1.1582234559333795,
      "grad_norm": 1.214060664176941,
      "learning_rate": 2.1640150872421997e-05,
      "loss": 0.5416,
      "step": 6676
    },
    {
      "epoch": 1.1583969465648856,
      "grad_norm": 0.7202631235122681,
      "learning_rate": 2.1633360469739183e-05,
      "loss": 0.5417,
      "step": 6677
    },
    {
      "epoch": 1.1585704371963914,
      "grad_norm": 0.8196054697036743,
      "learning_rate": 2.1626569877502985e-05,
      "loss": 0.5106,
      "step": 6678
    },
    {
      "epoch": 1.1587439278278973,
      "grad_norm": 0.7475643157958984,
      "learning_rate": 2.161977909650145e-05,
      "loss": 0.5076,
      "step": 6679
    },
    {
      "epoch": 1.158917418459403,
      "grad_norm": 0.7019733190536499,
      "learning_rate": 2.161298812752267e-05,
      "loss": 0.4905,
      "step": 6680
    },
    {
      "epoch": 1.1590909090909092,
      "grad_norm": 0.6765922904014587,
      "learning_rate": 2.160619697135474e-05,
      "loss": 0.6139,
      "step": 6681
    },
    {
      "epoch": 1.159264399722415,
      "grad_norm": 0.7036373615264893,
      "learning_rate": 2.1599405628785773e-05,
      "loss": 0.5338,
      "step": 6682
    },
    {
      "epoch": 1.1594378903539209,
      "grad_norm": 0.7694693803787231,
      "learning_rate": 2.1592614100603925e-05,
      "loss": 0.53,
      "step": 6683
    },
    {
      "epoch": 1.1596113809854267,
      "grad_norm": 1.6063934564590454,
      "learning_rate": 2.158582238759735e-05,
      "loss": 0.4412,
      "step": 6684
    },
    {
      "epoch": 1.1597848716169328,
      "grad_norm": 0.7708662152290344,
      "learning_rate": 2.157903049055424e-05,
      "loss": 0.6633,
      "step": 6685
    },
    {
      "epoch": 1.1599583622484386,
      "grad_norm": 0.7185639142990112,
      "learning_rate": 2.15722384102628e-05,
      "loss": 0.4767,
      "step": 6686
    },
    {
      "epoch": 1.1601318528799445,
      "grad_norm": 0.8182277083396912,
      "learning_rate": 2.156544614751127e-05,
      "loss": 0.5151,
      "step": 6687
    },
    {
      "epoch": 1.1603053435114503,
      "grad_norm": 0.7145664691925049,
      "learning_rate": 2.1558653703087876e-05,
      "loss": 0.4597,
      "step": 6688
    },
    {
      "epoch": 1.1604788341429564,
      "grad_norm": 0.8788052201271057,
      "learning_rate": 2.1551861077780914e-05,
      "loss": 0.4357,
      "step": 6689
    },
    {
      "epoch": 1.1606523247744622,
      "grad_norm": 0.693592369556427,
      "learning_rate": 2.1545068272378664e-05,
      "loss": 0.6262,
      "step": 6690
    },
    {
      "epoch": 1.160825815405968,
      "grad_norm": 0.5431566834449768,
      "learning_rate": 2.153827528766944e-05,
      "loss": 0.5736,
      "step": 6691
    },
    {
      "epoch": 1.160999306037474,
      "grad_norm": 0.7231242060661316,
      "learning_rate": 2.1531482124441574e-05,
      "loss": 0.4352,
      "step": 6692
    },
    {
      "epoch": 1.1611727966689798,
      "grad_norm": 0.6315079927444458,
      "learning_rate": 2.1524688783483424e-05,
      "loss": 0.6288,
      "step": 6693
    },
    {
      "epoch": 1.1613462873004858,
      "grad_norm": 0.7496607899665833,
      "learning_rate": 2.151789526558337e-05,
      "loss": 0.551,
      "step": 6694
    },
    {
      "epoch": 1.1615197779319917,
      "grad_norm": 0.9412398934364319,
      "learning_rate": 2.1511101571529793e-05,
      "loss": 0.447,
      "step": 6695
    },
    {
      "epoch": 1.1616932685634975,
      "grad_norm": 0.7098371386528015,
      "learning_rate": 2.1504307702111125e-05,
      "loss": 0.4487,
      "step": 6696
    },
    {
      "epoch": 1.1618667591950034,
      "grad_norm": 0.915355920791626,
      "learning_rate": 2.1497513658115792e-05,
      "loss": 0.6139,
      "step": 6697
    },
    {
      "epoch": 1.1620402498265094,
      "grad_norm": 0.858681857585907,
      "learning_rate": 2.1490719440332252e-05,
      "loss": 0.5276,
      "step": 6698
    },
    {
      "epoch": 1.1622137404580153,
      "grad_norm": 0.674950897693634,
      "learning_rate": 2.148392504954899e-05,
      "loss": 0.5796,
      "step": 6699
    },
    {
      "epoch": 1.1623872310895211,
      "grad_norm": 0.7628039717674255,
      "learning_rate": 2.147713048655449e-05,
      "loss": 0.505,
      "step": 6700
    },
    {
      "epoch": 1.162560721721027,
      "grad_norm": 0.8838892579078674,
      "learning_rate": 2.147033575213728e-05,
      "loss": 0.5066,
      "step": 6701
    },
    {
      "epoch": 1.162734212352533,
      "grad_norm": 0.7454177141189575,
      "learning_rate": 2.1463540847085892e-05,
      "loss": 0.5548,
      "step": 6702
    },
    {
      "epoch": 1.162907702984039,
      "grad_norm": 0.6347874402999878,
      "learning_rate": 2.145674577218888e-05,
      "loss": 0.5387,
      "step": 6703
    },
    {
      "epoch": 1.1630811936155447,
      "grad_norm": 0.9110173583030701,
      "learning_rate": 2.1449950528234828e-05,
      "loss": 0.4803,
      "step": 6704
    },
    {
      "epoch": 1.1632546842470506,
      "grad_norm": 0.7709625363349915,
      "learning_rate": 2.1443155116012328e-05,
      "loss": 0.5662,
      "step": 6705
    },
    {
      "epoch": 1.1634281748785567,
      "grad_norm": 1.081397294998169,
      "learning_rate": 2.143635953630999e-05,
      "loss": 0.5341,
      "step": 6706
    },
    {
      "epoch": 1.1636016655100625,
      "grad_norm": 0.9578310251235962,
      "learning_rate": 2.142956378991646e-05,
      "loss": 0.5344,
      "step": 6707
    },
    {
      "epoch": 1.1637751561415683,
      "grad_norm": 0.7448020577430725,
      "learning_rate": 2.1422767877620382e-05,
      "loss": 0.5674,
      "step": 6708
    },
    {
      "epoch": 1.1639486467730742,
      "grad_norm": 1.105573058128357,
      "learning_rate": 2.1415971800210437e-05,
      "loss": 0.4822,
      "step": 6709
    },
    {
      "epoch": 1.16412213740458,
      "grad_norm": 1.0320541858673096,
      "learning_rate": 2.1409175558475307e-05,
      "loss": 0.6262,
      "step": 6710
    },
    {
      "epoch": 1.164295628036086,
      "grad_norm": 0.6695195436477661,
      "learning_rate": 2.1402379153203716e-05,
      "loss": 0.4943,
      "step": 6711
    },
    {
      "epoch": 1.164469118667592,
      "grad_norm": 0.8333204984664917,
      "learning_rate": 2.1395582585184397e-05,
      "loss": 0.4807,
      "step": 6712
    },
    {
      "epoch": 1.1646426092990978,
      "grad_norm": 0.642285943031311,
      "learning_rate": 2.1388785855206083e-05,
      "loss": 0.538,
      "step": 6713
    },
    {
      "epoch": 1.1648160999306039,
      "grad_norm": 1.5361217260360718,
      "learning_rate": 2.138198896405756e-05,
      "loss": 0.5182,
      "step": 6714
    },
    {
      "epoch": 1.1649895905621097,
      "grad_norm": 0.8022282719612122,
      "learning_rate": 2.1375191912527605e-05,
      "loss": 0.4966,
      "step": 6715
    },
    {
      "epoch": 1.1651630811936156,
      "grad_norm": 0.7354987859725952,
      "learning_rate": 2.1368394701405023e-05,
      "loss": 0.531,
      "step": 6716
    },
    {
      "epoch": 1.1653365718251214,
      "grad_norm": 0.6895592212677002,
      "learning_rate": 2.1361597331478647e-05,
      "loss": 0.6543,
      "step": 6717
    },
    {
      "epoch": 1.1655100624566272,
      "grad_norm": 1.0040481090545654,
      "learning_rate": 2.1354799803537312e-05,
      "loss": 0.504,
      "step": 6718
    },
    {
      "epoch": 1.1656835530881333,
      "grad_norm": 1.0474437475204468,
      "learning_rate": 2.1348002118369878e-05,
      "loss": 0.5848,
      "step": 6719
    },
    {
      "epoch": 1.1658570437196392,
      "grad_norm": 0.9867828488349915,
      "learning_rate": 2.134120427676523e-05,
      "loss": 0.5018,
      "step": 6720
    },
    {
      "epoch": 1.166030534351145,
      "grad_norm": 0.7619190216064453,
      "learning_rate": 2.133440627951226e-05,
      "loss": 0.5389,
      "step": 6721
    },
    {
      "epoch": 1.1662040249826509,
      "grad_norm": 0.8639862537384033,
      "learning_rate": 2.1327608127399895e-05,
      "loss": 0.467,
      "step": 6722
    },
    {
      "epoch": 1.166377515614157,
      "grad_norm": 0.8269357681274414,
      "learning_rate": 2.1320809821217052e-05,
      "loss": 0.5723,
      "step": 6723
    },
    {
      "epoch": 1.1665510062456628,
      "grad_norm": 0.6355014443397522,
      "learning_rate": 2.1314011361752687e-05,
      "loss": 0.52,
      "step": 6724
    },
    {
      "epoch": 1.1667244968771686,
      "grad_norm": 1.0886286497116089,
      "learning_rate": 2.1307212749795782e-05,
      "loss": 0.4779,
      "step": 6725
    },
    {
      "epoch": 1.1668979875086745,
      "grad_norm": 0.6519858837127686,
      "learning_rate": 2.1300413986135313e-05,
      "loss": 0.6099,
      "step": 6726
    },
    {
      "epoch": 1.1670714781401803,
      "grad_norm": 0.614009439945221,
      "learning_rate": 2.1293615071560277e-05,
      "loss": 0.5667,
      "step": 6727
    },
    {
      "epoch": 1.1672449687716864,
      "grad_norm": 1.0716733932495117,
      "learning_rate": 2.128681600685971e-05,
      "loss": 0.4767,
      "step": 6728
    },
    {
      "epoch": 1.1674184594031922,
      "grad_norm": 0.8412541747093201,
      "learning_rate": 2.1280016792822645e-05,
      "loss": 0.5222,
      "step": 6729
    },
    {
      "epoch": 1.167591950034698,
      "grad_norm": 0.8214019536972046,
      "learning_rate": 2.1273217430238146e-05,
      "loss": 0.5557,
      "step": 6730
    },
    {
      "epoch": 1.1677654406662041,
      "grad_norm": 0.6152772903442383,
      "learning_rate": 2.1266417919895274e-05,
      "loss": 0.5397,
      "step": 6731
    },
    {
      "epoch": 1.16793893129771,
      "grad_norm": 0.9900790452957153,
      "learning_rate": 2.1259618262583122e-05,
      "loss": 0.5312,
      "step": 6732
    },
    {
      "epoch": 1.1681124219292158,
      "grad_norm": 0.837279736995697,
      "learning_rate": 2.1252818459090814e-05,
      "loss": 0.4365,
      "step": 6733
    },
    {
      "epoch": 1.1682859125607217,
      "grad_norm": 1.2113510370254517,
      "learning_rate": 2.1246018510207452e-05,
      "loss": 0.5173,
      "step": 6734
    },
    {
      "epoch": 1.1684594031922275,
      "grad_norm": 0.5946635007858276,
      "learning_rate": 2.12392184167222e-05,
      "loss": 0.4924,
      "step": 6735
    },
    {
      "epoch": 1.1686328938237336,
      "grad_norm": 0.6552371382713318,
      "learning_rate": 2.1232418179424204e-05,
      "loss": 0.6875,
      "step": 6736
    },
    {
      "epoch": 1.1688063844552394,
      "grad_norm": 0.6151350736618042,
      "learning_rate": 2.1225617799102638e-05,
      "loss": 0.6597,
      "step": 6737
    },
    {
      "epoch": 1.1689798750867453,
      "grad_norm": 0.6632144451141357,
      "learning_rate": 2.12188172765467e-05,
      "loss": 0.6016,
      "step": 6738
    },
    {
      "epoch": 1.1691533657182511,
      "grad_norm": 1.4138222932815552,
      "learning_rate": 2.1212016612545604e-05,
      "loss": 0.4871,
      "step": 6739
    },
    {
      "epoch": 1.1693268563497572,
      "grad_norm": 0.7806059718132019,
      "learning_rate": 2.120521580788856e-05,
      "loss": 0.5596,
      "step": 6740
    },
    {
      "epoch": 1.169500346981263,
      "grad_norm": 0.7281691431999207,
      "learning_rate": 2.1198414863364822e-05,
      "loss": 0.5558,
      "step": 6741
    },
    {
      "epoch": 1.1696738376127689,
      "grad_norm": 0.6963688135147095,
      "learning_rate": 2.1191613779763635e-05,
      "loss": 0.5559,
      "step": 6742
    },
    {
      "epoch": 1.1698473282442747,
      "grad_norm": 0.8871251940727234,
      "learning_rate": 2.1184812557874287e-05,
      "loss": 0.6001,
      "step": 6743
    },
    {
      "epoch": 1.1700208188757808,
      "grad_norm": 0.7859398722648621,
      "learning_rate": 2.1178011198486064e-05,
      "loss": 0.5193,
      "step": 6744
    },
    {
      "epoch": 1.1701943095072866,
      "grad_norm": 0.736751139163971,
      "learning_rate": 2.117120970238826e-05,
      "loss": 0.4547,
      "step": 6745
    },
    {
      "epoch": 1.1703678001387925,
      "grad_norm": 0.8680247664451599,
      "learning_rate": 2.1164408070370212e-05,
      "loss": 0.4491,
      "step": 6746
    },
    {
      "epoch": 1.1705412907702983,
      "grad_norm": 0.9516239166259766,
      "learning_rate": 2.1157606303221253e-05,
      "loss": 0.4543,
      "step": 6747
    },
    {
      "epoch": 1.1707147814018044,
      "grad_norm": 0.6905824542045593,
      "learning_rate": 2.1150804401730724e-05,
      "loss": 0.6207,
      "step": 6748
    },
    {
      "epoch": 1.1708882720333103,
      "grad_norm": 0.7213749289512634,
      "learning_rate": 2.114400236668801e-05,
      "loss": 0.5787,
      "step": 6749
    },
    {
      "epoch": 1.171061762664816,
      "grad_norm": 0.8580946922302246,
      "learning_rate": 2.1137200198882484e-05,
      "loss": 0.5155,
      "step": 6750
    },
    {
      "epoch": 1.171235253296322,
      "grad_norm": 0.9151411652565002,
      "learning_rate": 2.113039789910355e-05,
      "loss": 0.5774,
      "step": 6751
    },
    {
      "epoch": 1.1714087439278278,
      "grad_norm": 1.038617730140686,
      "learning_rate": 2.112359546814063e-05,
      "loss": 0.373,
      "step": 6752
    },
    {
      "epoch": 1.1715822345593339,
      "grad_norm": 0.8604259490966797,
      "learning_rate": 2.111679290678314e-05,
      "loss": 0.5082,
      "step": 6753
    },
    {
      "epoch": 1.1717557251908397,
      "grad_norm": 0.9970014691352844,
      "learning_rate": 2.110999021582053e-05,
      "loss": 0.4684,
      "step": 6754
    },
    {
      "epoch": 1.1719292158223455,
      "grad_norm": 0.6839314103126526,
      "learning_rate": 2.110318739604227e-05,
      "loss": 0.462,
      "step": 6755
    },
    {
      "epoch": 1.1721027064538514,
      "grad_norm": 0.75113844871521,
      "learning_rate": 2.1096384448237824e-05,
      "loss": 0.6161,
      "step": 6756
    },
    {
      "epoch": 1.1722761970853575,
      "grad_norm": 1.0658999681472778,
      "learning_rate": 2.1089581373196686e-05,
      "loss": 0.448,
      "step": 6757
    },
    {
      "epoch": 1.1724496877168633,
      "grad_norm": 1.0717624425888062,
      "learning_rate": 2.1082778171708355e-05,
      "loss": 0.5045,
      "step": 6758
    },
    {
      "epoch": 1.1726231783483692,
      "grad_norm": 0.7669315338134766,
      "learning_rate": 2.1075974844562354e-05,
      "loss": 0.564,
      "step": 6759
    },
    {
      "epoch": 1.172796668979875,
      "grad_norm": 0.8839965462684631,
      "learning_rate": 2.1069171392548226e-05,
      "loss": 0.5275,
      "step": 6760
    },
    {
      "epoch": 1.172970159611381,
      "grad_norm": 0.8475131988525391,
      "learning_rate": 2.106236781645551e-05,
      "loss": 0.5673,
      "step": 6761
    },
    {
      "epoch": 1.173143650242887,
      "grad_norm": 0.7701882123947144,
      "learning_rate": 2.1055564117073767e-05,
      "loss": 0.4849,
      "step": 6762
    },
    {
      "epoch": 1.1733171408743928,
      "grad_norm": 1.208903193473816,
      "learning_rate": 2.104876029519258e-05,
      "loss": 0.4667,
      "step": 6763
    },
    {
      "epoch": 1.1734906315058986,
      "grad_norm": 1.1139265298843384,
      "learning_rate": 2.1041956351601543e-05,
      "loss": 0.5071,
      "step": 6764
    },
    {
      "epoch": 1.1736641221374047,
      "grad_norm": 0.6981279253959656,
      "learning_rate": 2.1035152287090254e-05,
      "loss": 0.499,
      "step": 6765
    },
    {
      "epoch": 1.1738376127689105,
      "grad_norm": 0.7927271127700806,
      "learning_rate": 2.1028348102448338e-05,
      "loss": 0.4532,
      "step": 6766
    },
    {
      "epoch": 1.1740111034004164,
      "grad_norm": 0.7971651554107666,
      "learning_rate": 2.1021543798465426e-05,
      "loss": 0.5505,
      "step": 6767
    },
    {
      "epoch": 1.1741845940319222,
      "grad_norm": 0.6657699346542358,
      "learning_rate": 2.1014739375931166e-05,
      "loss": 0.5161,
      "step": 6768
    },
    {
      "epoch": 1.174358084663428,
      "grad_norm": 1.4957996606826782,
      "learning_rate": 2.1007934835635213e-05,
      "loss": 0.5529,
      "step": 6769
    },
    {
      "epoch": 1.1745315752949341,
      "grad_norm": 0.5472845435142517,
      "learning_rate": 2.1001130178367256e-05,
      "loss": 0.6241,
      "step": 6770
    },
    {
      "epoch": 1.17470506592644,
      "grad_norm": 0.6698144674301147,
      "learning_rate": 2.0994325404916967e-05,
      "loss": 0.5625,
      "step": 6771
    },
    {
      "epoch": 1.1748785565579458,
      "grad_norm": 1.0643738508224487,
      "learning_rate": 2.098752051607406e-05,
      "loss": 0.3807,
      "step": 6772
    },
    {
      "epoch": 1.1750520471894519,
      "grad_norm": 1.1606651544570923,
      "learning_rate": 2.0980715512628255e-05,
      "loss": 0.5408,
      "step": 6773
    },
    {
      "epoch": 1.1752255378209577,
      "grad_norm": 0.7926031351089478,
      "learning_rate": 2.097391039536926e-05,
      "loss": 0.4792,
      "step": 6774
    },
    {
      "epoch": 1.1753990284524636,
      "grad_norm": 0.6972630620002747,
      "learning_rate": 2.0967105165086835e-05,
      "loss": 0.5724,
      "step": 6775
    },
    {
      "epoch": 1.1755725190839694,
      "grad_norm": 0.8949503302574158,
      "learning_rate": 2.0960299822570728e-05,
      "loss": 0.489,
      "step": 6776
    },
    {
      "epoch": 1.1757460097154753,
      "grad_norm": 0.7007628083229065,
      "learning_rate": 2.0953494368610702e-05,
      "loss": 0.5793,
      "step": 6777
    },
    {
      "epoch": 1.1759195003469813,
      "grad_norm": 0.7773805260658264,
      "learning_rate": 2.094668880399655e-05,
      "loss": 0.5586,
      "step": 6778
    },
    {
      "epoch": 1.1760929909784872,
      "grad_norm": 0.6971990466117859,
      "learning_rate": 2.0939883129518056e-05,
      "loss": 0.5836,
      "step": 6779
    },
    {
      "epoch": 1.176266481609993,
      "grad_norm": 0.7796101570129395,
      "learning_rate": 2.0933077345965032e-05,
      "loss": 0.6204,
      "step": 6780
    },
    {
      "epoch": 1.1764399722414989,
      "grad_norm": 1.1607211828231812,
      "learning_rate": 2.092627145412729e-05,
      "loss": 0.556,
      "step": 6781
    },
    {
      "epoch": 1.176613462873005,
      "grad_norm": 0.8690618276596069,
      "learning_rate": 2.0919465454794672e-05,
      "loss": 0.546,
      "step": 6782
    },
    {
      "epoch": 1.1767869535045108,
      "grad_norm": 0.6725994348526001,
      "learning_rate": 2.091265934875701e-05,
      "loss": 0.5078,
      "step": 6783
    },
    {
      "epoch": 1.1769604441360166,
      "grad_norm": 2.117467164993286,
      "learning_rate": 2.0905853136804173e-05,
      "loss": 0.4205,
      "step": 6784
    },
    {
      "epoch": 1.1771339347675225,
      "grad_norm": 1.204289436340332,
      "learning_rate": 2.0899046819726025e-05,
      "loss": 0.5707,
      "step": 6785
    },
    {
      "epoch": 1.1773074253990283,
      "grad_norm": 0.7071922421455383,
      "learning_rate": 2.089224039831244e-05,
      "loss": 0.5833,
      "step": 6786
    },
    {
      "epoch": 1.1774809160305344,
      "grad_norm": 0.6543411016464233,
      "learning_rate": 2.088543387335332e-05,
      "loss": 0.5482,
      "step": 6787
    },
    {
      "epoch": 1.1776544066620402,
      "grad_norm": 0.9622500538825989,
      "learning_rate": 2.087862724563857e-05,
      "loss": 0.613,
      "step": 6788
    },
    {
      "epoch": 1.177827897293546,
      "grad_norm": 1.0394973754882812,
      "learning_rate": 2.0871820515958102e-05,
      "loss": 0.5042,
      "step": 6789
    },
    {
      "epoch": 1.1780013879250522,
      "grad_norm": 0.8372440338134766,
      "learning_rate": 2.0865013685101844e-05,
      "loss": 0.5587,
      "step": 6790
    },
    {
      "epoch": 1.178174878556558,
      "grad_norm": 0.999835193157196,
      "learning_rate": 2.085820675385975e-05,
      "loss": 0.5917,
      "step": 6791
    },
    {
      "epoch": 1.1783483691880638,
      "grad_norm": 0.7350196242332458,
      "learning_rate": 2.085139972302175e-05,
      "loss": 0.5868,
      "step": 6792
    },
    {
      "epoch": 1.1785218598195697,
      "grad_norm": 1.2748345136642456,
      "learning_rate": 2.0844592593377827e-05,
      "loss": 0.4858,
      "step": 6793
    },
    {
      "epoch": 1.1786953504510755,
      "grad_norm": 0.6886605024337769,
      "learning_rate": 2.083778536571795e-05,
      "loss": 0.5068,
      "step": 6794
    },
    {
      "epoch": 1.1788688410825816,
      "grad_norm": 0.7593858242034912,
      "learning_rate": 2.0830978040832098e-05,
      "loss": 0.5242,
      "step": 6795
    },
    {
      "epoch": 1.1790423317140875,
      "grad_norm": 0.7221627831459045,
      "learning_rate": 2.0824170619510283e-05,
      "loss": 0.4913,
      "step": 6796
    },
    {
      "epoch": 1.1792158223455933,
      "grad_norm": 0.8000275492668152,
      "learning_rate": 2.081736310254251e-05,
      "loss": 0.5968,
      "step": 6797
    },
    {
      "epoch": 1.1793893129770991,
      "grad_norm": 0.7031962871551514,
      "learning_rate": 2.0810555490718787e-05,
      "loss": 0.538,
      "step": 6798
    },
    {
      "epoch": 1.1795628036086052,
      "grad_norm": 0.6669413447380066,
      "learning_rate": 2.0803747784829166e-05,
      "loss": 0.4932,
      "step": 6799
    },
    {
      "epoch": 1.179736294240111,
      "grad_norm": 0.6491285562515259,
      "learning_rate": 2.0796939985663666e-05,
      "loss": 0.454,
      "step": 6800
    },
    {
      "epoch": 1.179909784871617,
      "grad_norm": 0.90521240234375,
      "learning_rate": 2.079013209401236e-05,
      "loss": 0.4735,
      "step": 6801
    },
    {
      "epoch": 1.1800832755031228,
      "grad_norm": 0.7703285217285156,
      "learning_rate": 2.0783324110665306e-05,
      "loss": 0.4625,
      "step": 6802
    },
    {
      "epoch": 1.1802567661346288,
      "grad_norm": 0.7215616703033447,
      "learning_rate": 2.0776516036412565e-05,
      "loss": 0.6068,
      "step": 6803
    },
    {
      "epoch": 1.1804302567661347,
      "grad_norm": 0.9224920272827148,
      "learning_rate": 2.0769707872044242e-05,
      "loss": 0.5234,
      "step": 6804
    },
    {
      "epoch": 1.1806037473976405,
      "grad_norm": 0.664199948310852,
      "learning_rate": 2.076289961835042e-05,
      "loss": 0.5541,
      "step": 6805
    },
    {
      "epoch": 1.1807772380291464,
      "grad_norm": 0.6375593543052673,
      "learning_rate": 2.0756091276121212e-05,
      "loss": 0.4492,
      "step": 6806
    },
    {
      "epoch": 1.1809507286606524,
      "grad_norm": 0.7353429198265076,
      "learning_rate": 2.074928284614673e-05,
      "loss": 0.417,
      "step": 6807
    },
    {
      "epoch": 1.1811242192921583,
      "grad_norm": 0.7317572236061096,
      "learning_rate": 2.0742474329217094e-05,
      "loss": 0.4982,
      "step": 6808
    },
    {
      "epoch": 1.1812977099236641,
      "grad_norm": 0.7748720645904541,
      "learning_rate": 2.0735665726122453e-05,
      "loss": 0.4381,
      "step": 6809
    },
    {
      "epoch": 1.18147120055517,
      "grad_norm": 0.6508405804634094,
      "learning_rate": 2.0728857037652945e-05,
      "loss": 0.5945,
      "step": 6810
    },
    {
      "epoch": 1.1816446911866758,
      "grad_norm": 1.3365366458892822,
      "learning_rate": 2.0722048264598727e-05,
      "loss": 0.4197,
      "step": 6811
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 0.8428879380226135,
      "learning_rate": 2.0715239407749973e-05,
      "loss": 0.5508,
      "step": 6812
    },
    {
      "epoch": 1.1819916724496877,
      "grad_norm": 1.0093331336975098,
      "learning_rate": 2.0708430467896848e-05,
      "loss": 0.5153,
      "step": 6813
    },
    {
      "epoch": 1.1821651630811936,
      "grad_norm": 3.6245226860046387,
      "learning_rate": 2.070162144582954e-05,
      "loss": 0.5547,
      "step": 6814
    },
    {
      "epoch": 1.1823386537126996,
      "grad_norm": 0.6548336744308472,
      "learning_rate": 2.0694812342338252e-05,
      "loss": 0.5793,
      "step": 6815
    },
    {
      "epoch": 1.1825121443442055,
      "grad_norm": 0.7358191013336182,
      "learning_rate": 2.0688003158213172e-05,
      "loss": 0.5156,
      "step": 6816
    },
    {
      "epoch": 1.1826856349757113,
      "grad_norm": 0.8403127193450928,
      "learning_rate": 2.0681193894244536e-05,
      "loss": 0.4742,
      "step": 6817
    },
    {
      "epoch": 1.1828591256072172,
      "grad_norm": 0.7100670337677002,
      "learning_rate": 2.067438455122255e-05,
      "loss": 0.3762,
      "step": 6818
    },
    {
      "epoch": 1.183032616238723,
      "grad_norm": 0.7438713908195496,
      "learning_rate": 2.0667575129937446e-05,
      "loss": 0.5098,
      "step": 6819
    },
    {
      "epoch": 1.183206106870229,
      "grad_norm": 0.8709805607795715,
      "learning_rate": 2.0660765631179474e-05,
      "loss": 0.3984,
      "step": 6820
    },
    {
      "epoch": 1.183379597501735,
      "grad_norm": 1.246625304222107,
      "learning_rate": 2.0653956055738876e-05,
      "loss": 0.495,
      "step": 6821
    },
    {
      "epoch": 1.1835530881332408,
      "grad_norm": 0.9622408747673035,
      "learning_rate": 2.0647146404405923e-05,
      "loss": 0.5278,
      "step": 6822
    },
    {
      "epoch": 1.1837265787647466,
      "grad_norm": 0.6290227770805359,
      "learning_rate": 2.0640336677970875e-05,
      "loss": 0.5858,
      "step": 6823
    },
    {
      "epoch": 1.1839000693962527,
      "grad_norm": 0.7876569032669067,
      "learning_rate": 2.0633526877224006e-05,
      "loss": 0.5173,
      "step": 6824
    },
    {
      "epoch": 1.1840735600277585,
      "grad_norm": 0.7701499462127686,
      "learning_rate": 2.0626717002955606e-05,
      "loss": 0.5896,
      "step": 6825
    },
    {
      "epoch": 1.1842470506592644,
      "grad_norm": 0.605043888092041,
      "learning_rate": 2.061990705595597e-05,
      "loss": 0.6094,
      "step": 6826
    },
    {
      "epoch": 1.1844205412907702,
      "grad_norm": 1.107033610343933,
      "learning_rate": 2.061309703701539e-05,
      "loss": 0.496,
      "step": 6827
    },
    {
      "epoch": 1.184594031922276,
      "grad_norm": 0.8433276414871216,
      "learning_rate": 2.060628694692419e-05,
      "loss": 0.6025,
      "step": 6828
    },
    {
      "epoch": 1.1847675225537821,
      "grad_norm": 0.6223244071006775,
      "learning_rate": 2.0599476786472686e-05,
      "loss": 0.558,
      "step": 6829
    },
    {
      "epoch": 1.184941013185288,
      "grad_norm": 0.5799712538719177,
      "learning_rate": 2.0592666556451197e-05,
      "loss": 0.5922,
      "step": 6830
    },
    {
      "epoch": 1.1851145038167938,
      "grad_norm": 0.9107916951179504,
      "learning_rate": 2.0585856257650067e-05,
      "loss": 0.4255,
      "step": 6831
    },
    {
      "epoch": 1.1852879944483,
      "grad_norm": 3.0029711723327637,
      "learning_rate": 2.0579045890859635e-05,
      "loss": 0.4358,
      "step": 6832
    },
    {
      "epoch": 1.1854614850798058,
      "grad_norm": 0.6497785449028015,
      "learning_rate": 2.057223545687025e-05,
      "loss": 0.5289,
      "step": 6833
    },
    {
      "epoch": 1.1856349757113116,
      "grad_norm": 0.9125445485115051,
      "learning_rate": 2.0565424956472278e-05,
      "loss": 0.5168,
      "step": 6834
    },
    {
      "epoch": 1.1858084663428174,
      "grad_norm": 0.8241649270057678,
      "learning_rate": 2.0558614390456075e-05,
      "loss": 0.5028,
      "step": 6835
    },
    {
      "epoch": 1.1859819569743233,
      "grad_norm": 1.6441798210144043,
      "learning_rate": 2.055180375961203e-05,
      "loss": 0.5607,
      "step": 6836
    },
    {
      "epoch": 1.1861554476058294,
      "grad_norm": 2.1024038791656494,
      "learning_rate": 2.0544993064730507e-05,
      "loss": 0.5106,
      "step": 6837
    },
    {
      "epoch": 1.1863289382373352,
      "grad_norm": 0.6179422736167908,
      "learning_rate": 2.053818230660191e-05,
      "loss": 0.5565,
      "step": 6838
    },
    {
      "epoch": 1.186502428868841,
      "grad_norm": 0.6559538245201111,
      "learning_rate": 2.053137148601662e-05,
      "loss": 0.4733,
      "step": 6839
    },
    {
      "epoch": 1.186675919500347,
      "grad_norm": 0.7132574915885925,
      "learning_rate": 2.052456060376506e-05,
      "loss": 0.5116,
      "step": 6840
    },
    {
      "epoch": 1.186849410131853,
      "grad_norm": 0.9821913838386536,
      "learning_rate": 2.051774966063763e-05,
      "loss": 0.4854,
      "step": 6841
    },
    {
      "epoch": 1.1870229007633588,
      "grad_norm": 0.6002830862998962,
      "learning_rate": 2.051093865742474e-05,
      "loss": 0.6154,
      "step": 6842
    },
    {
      "epoch": 1.1871963913948647,
      "grad_norm": 0.6456484198570251,
      "learning_rate": 2.0504127594916833e-05,
      "loss": 0.4579,
      "step": 6843
    },
    {
      "epoch": 1.1873698820263705,
      "grad_norm": 0.7543825507164001,
      "learning_rate": 2.0497316473904324e-05,
      "loss": 0.516,
      "step": 6844
    },
    {
      "epoch": 1.1875433726578764,
      "grad_norm": 0.8759680986404419,
      "learning_rate": 2.049050529517766e-05,
      "loss": 0.5153,
      "step": 6845
    },
    {
      "epoch": 1.1877168632893824,
      "grad_norm": 0.7334983348846436,
      "learning_rate": 2.048369405952729e-05,
      "loss": 0.5588,
      "step": 6846
    },
    {
      "epoch": 1.1878903539208883,
      "grad_norm": 0.7166071534156799,
      "learning_rate": 2.047688276774366e-05,
      "loss": 0.5132,
      "step": 6847
    },
    {
      "epoch": 1.188063844552394,
      "grad_norm": 0.6016421318054199,
      "learning_rate": 2.0470071420617222e-05,
      "loss": 0.59,
      "step": 6848
    },
    {
      "epoch": 1.1882373351839002,
      "grad_norm": 0.6861176490783691,
      "learning_rate": 2.046326001893846e-05,
      "loss": 0.6113,
      "step": 6849
    },
    {
      "epoch": 1.188410825815406,
      "grad_norm": 0.6174821257591248,
      "learning_rate": 2.045644856349782e-05,
      "loss": 0.5931,
      "step": 6850
    },
    {
      "epoch": 1.1885843164469119,
      "grad_norm": 1.0377147197723389,
      "learning_rate": 2.0449637055085798e-05,
      "loss": 0.5638,
      "step": 6851
    },
    {
      "epoch": 1.1887578070784177,
      "grad_norm": 0.6978501081466675,
      "learning_rate": 2.0442825494492876e-05,
      "loss": 0.5193,
      "step": 6852
    },
    {
      "epoch": 1.1889312977099236,
      "grad_norm": 0.7452558875083923,
      "learning_rate": 2.043601388250953e-05,
      "loss": 0.4707,
      "step": 6853
    },
    {
      "epoch": 1.1891047883414296,
      "grad_norm": 1.0568429231643677,
      "learning_rate": 2.0429202219926273e-05,
      "loss": 0.4493,
      "step": 6854
    },
    {
      "epoch": 1.1892782789729355,
      "grad_norm": 0.8652346134185791,
      "learning_rate": 2.0422390507533593e-05,
      "loss": 0.5179,
      "step": 6855
    },
    {
      "epoch": 1.1894517696044413,
      "grad_norm": 0.9493082761764526,
      "learning_rate": 2.0415578746122007e-05,
      "loss": 0.4335,
      "step": 6856
    },
    {
      "epoch": 1.1896252602359472,
      "grad_norm": 0.6517853140830994,
      "learning_rate": 2.0408766936482016e-05,
      "loss": 0.5033,
      "step": 6857
    },
    {
      "epoch": 1.1897987508674532,
      "grad_norm": 0.8095531463623047,
      "learning_rate": 2.0401955079404154e-05,
      "loss": 0.5819,
      "step": 6858
    },
    {
      "epoch": 1.189972241498959,
      "grad_norm": 0.5268284678459167,
      "learning_rate": 2.039514317567893e-05,
      "loss": 0.5623,
      "step": 6859
    },
    {
      "epoch": 1.190145732130465,
      "grad_norm": 0.6493697762489319,
      "learning_rate": 2.0388331226096886e-05,
      "loss": 0.5021,
      "step": 6860
    },
    {
      "epoch": 1.1903192227619708,
      "grad_norm": 0.6705477833747864,
      "learning_rate": 2.0381519231448544e-05,
      "loss": 0.6041,
      "step": 6861
    },
    {
      "epoch": 1.1904927133934768,
      "grad_norm": 0.8729289770126343,
      "learning_rate": 2.0374707192524455e-05,
      "loss": 0.4406,
      "step": 6862
    },
    {
      "epoch": 1.1906662040249827,
      "grad_norm": 0.8540909886360168,
      "learning_rate": 2.0367895110115166e-05,
      "loss": 0.4896,
      "step": 6863
    },
    {
      "epoch": 1.1908396946564885,
      "grad_norm": 0.8730484843254089,
      "learning_rate": 2.036108298501121e-05,
      "loss": 0.4495,
      "step": 6864
    },
    {
      "epoch": 1.1910131852879944,
      "grad_norm": 1.2369223833084106,
      "learning_rate": 2.035427081800316e-05,
      "loss": 0.6508,
      "step": 6865
    },
    {
      "epoch": 1.1911866759195004,
      "grad_norm": 0.7968800067901611,
      "learning_rate": 2.034745860988156e-05,
      "loss": 0.514,
      "step": 6866
    },
    {
      "epoch": 1.1913601665510063,
      "grad_norm": 0.8077332973480225,
      "learning_rate": 2.0340646361436994e-05,
      "loss": 0.4382,
      "step": 6867
    },
    {
      "epoch": 1.1915336571825121,
      "grad_norm": 0.9408939480781555,
      "learning_rate": 2.0333834073460018e-05,
      "loss": 0.4429,
      "step": 6868
    },
    {
      "epoch": 1.191707147814018,
      "grad_norm": 0.7798374891281128,
      "learning_rate": 2.03270217467412e-05,
      "loss": 0.4854,
      "step": 6869
    },
    {
      "epoch": 1.1918806384455238,
      "grad_norm": 0.5698572397232056,
      "learning_rate": 2.032020938207114e-05,
      "loss": 0.573,
      "step": 6870
    },
    {
      "epoch": 1.19205412907703,
      "grad_norm": 0.6410461068153381,
      "learning_rate": 2.03133969802404e-05,
      "loss": 0.5502,
      "step": 6871
    },
    {
      "epoch": 1.1922276197085357,
      "grad_norm": 0.6756616234779358,
      "learning_rate": 2.030658454203958e-05,
      "loss": 0.4795,
      "step": 6872
    },
    {
      "epoch": 1.1924011103400416,
      "grad_norm": 0.9079252481460571,
      "learning_rate": 2.0299772068259263e-05,
      "loss": 0.4208,
      "step": 6873
    },
    {
      "epoch": 1.1925746009715477,
      "grad_norm": 0.7513494491577148,
      "learning_rate": 2.029295955969005e-05,
      "loss": 0.5061,
      "step": 6874
    },
    {
      "epoch": 1.1927480916030535,
      "grad_norm": 0.7170233726501465,
      "learning_rate": 2.0286147017122538e-05,
      "loss": 0.5616,
      "step": 6875
    },
    {
      "epoch": 1.1929215822345594,
      "grad_norm": 0.675975501537323,
      "learning_rate": 2.027933444134733e-05,
      "loss": 0.5427,
      "step": 6876
    },
    {
      "epoch": 1.1930950728660652,
      "grad_norm": 2.301645278930664,
      "learning_rate": 2.0272521833155038e-05,
      "loss": 0.5244,
      "step": 6877
    },
    {
      "epoch": 1.193268563497571,
      "grad_norm": 0.7549624443054199,
      "learning_rate": 2.0265709193336266e-05,
      "loss": 0.6567,
      "step": 6878
    },
    {
      "epoch": 1.1934420541290771,
      "grad_norm": 0.7844575047492981,
      "learning_rate": 2.0258896522681635e-05,
      "loss": 0.5343,
      "step": 6879
    },
    {
      "epoch": 1.193615544760583,
      "grad_norm": 0.9245356321334839,
      "learning_rate": 2.025208382198176e-05,
      "loss": 0.4023,
      "step": 6880
    },
    {
      "epoch": 1.1937890353920888,
      "grad_norm": 1.1952787637710571,
      "learning_rate": 2.024527109202726e-05,
      "loss": 0.4994,
      "step": 6881
    },
    {
      "epoch": 1.1939625260235947,
      "grad_norm": 0.7105661034584045,
      "learning_rate": 2.0238458333608766e-05,
      "loss": 0.5658,
      "step": 6882
    },
    {
      "epoch": 1.1941360166551007,
      "grad_norm": 0.8205832839012146,
      "learning_rate": 2.023164554751691e-05,
      "loss": 0.4775,
      "step": 6883
    },
    {
      "epoch": 1.1943095072866066,
      "grad_norm": 0.6414397358894348,
      "learning_rate": 2.0224832734542314e-05,
      "loss": 0.5118,
      "step": 6884
    },
    {
      "epoch": 1.1944829979181124,
      "grad_norm": 0.6226764917373657,
      "learning_rate": 2.0218019895475612e-05,
      "loss": 0.5337,
      "step": 6885
    },
    {
      "epoch": 1.1946564885496183,
      "grad_norm": 0.8643163442611694,
      "learning_rate": 2.0211207031107457e-05,
      "loss": 0.5604,
      "step": 6886
    },
    {
      "epoch": 1.194829979181124,
      "grad_norm": 0.5629163980484009,
      "learning_rate": 2.0204394142228473e-05,
      "loss": 0.5669,
      "step": 6887
    },
    {
      "epoch": 1.1950034698126302,
      "grad_norm": 2.306591033935547,
      "learning_rate": 2.0197581229629317e-05,
      "loss": 0.4446,
      "step": 6888
    },
    {
      "epoch": 1.195176960444136,
      "grad_norm": 0.6565482020378113,
      "learning_rate": 2.0190768294100626e-05,
      "loss": 0.6115,
      "step": 6889
    },
    {
      "epoch": 1.1953504510756419,
      "grad_norm": 0.9194337129592896,
      "learning_rate": 2.018395533643305e-05,
      "loss": 0.5658,
      "step": 6890
    },
    {
      "epoch": 1.195523941707148,
      "grad_norm": 0.6584299802780151,
      "learning_rate": 2.0177142357417243e-05,
      "loss": 0.6061,
      "step": 6891
    },
    {
      "epoch": 1.1956974323386538,
      "grad_norm": 3.2561211585998535,
      "learning_rate": 2.017032935784386e-05,
      "loss": 0.5623,
      "step": 6892
    },
    {
      "epoch": 1.1958709229701596,
      "grad_norm": 0.9865792393684387,
      "learning_rate": 2.0163516338503556e-05,
      "loss": 0.4666,
      "step": 6893
    },
    {
      "epoch": 1.1960444136016655,
      "grad_norm": 0.7090287804603577,
      "learning_rate": 2.0156703300186997e-05,
      "loss": 0.5883,
      "step": 6894
    },
    {
      "epoch": 1.1962179042331713,
      "grad_norm": 1.2316094636917114,
      "learning_rate": 2.0149890243684827e-05,
      "loss": 0.4607,
      "step": 6895
    },
    {
      "epoch": 1.1963913948646774,
      "grad_norm": 1.4981300830841064,
      "learning_rate": 2.0143077169787725e-05,
      "loss": 0.3534,
      "step": 6896
    },
    {
      "epoch": 1.1965648854961832,
      "grad_norm": 0.7626928687095642,
      "learning_rate": 2.0136264079286354e-05,
      "loss": 0.4762,
      "step": 6897
    },
    {
      "epoch": 1.196738376127689,
      "grad_norm": 0.8046074509620667,
      "learning_rate": 2.012945097297137e-05,
      "loss": 0.4624,
      "step": 6898
    },
    {
      "epoch": 1.196911866759195,
      "grad_norm": 1.0258681774139404,
      "learning_rate": 2.0122637851633455e-05,
      "loss": 0.4755,
      "step": 6899
    },
    {
      "epoch": 1.197085357390701,
      "grad_norm": 0.85794997215271,
      "learning_rate": 2.0115824716063273e-05,
      "loss": 0.4977,
      "step": 6900
    },
    {
      "epoch": 1.1972588480222068,
      "grad_norm": 0.6790227890014648,
      "learning_rate": 2.01090115670515e-05,
      "loss": 0.5133,
      "step": 6901
    },
    {
      "epoch": 1.1974323386537127,
      "grad_norm": 1.0489752292633057,
      "learning_rate": 2.0102198405388806e-05,
      "loss": 0.42,
      "step": 6902
    },
    {
      "epoch": 1.1976058292852185,
      "grad_norm": 0.8256232142448425,
      "learning_rate": 2.0095385231865864e-05,
      "loss": 0.4509,
      "step": 6903
    },
    {
      "epoch": 1.1977793199167244,
      "grad_norm": 0.8152967691421509,
      "learning_rate": 2.008857204727336e-05,
      "loss": 0.4639,
      "step": 6904
    },
    {
      "epoch": 1.1979528105482304,
      "grad_norm": 0.693350076675415,
      "learning_rate": 2.0081758852401964e-05,
      "loss": 0.5236,
      "step": 6905
    },
    {
      "epoch": 1.1981263011797363,
      "grad_norm": 1.2258223295211792,
      "learning_rate": 2.0074945648042353e-05,
      "loss": 0.424,
      "step": 6906
    },
    {
      "epoch": 1.1982997918112421,
      "grad_norm": 0.9876985549926758,
      "learning_rate": 2.006813243498522e-05,
      "loss": 0.537,
      "step": 6907
    },
    {
      "epoch": 1.1984732824427482,
      "grad_norm": 0.880332887172699,
      "learning_rate": 2.0061319214021237e-05,
      "loss": 0.5062,
      "step": 6908
    },
    {
      "epoch": 1.198646773074254,
      "grad_norm": 0.7460529804229736,
      "learning_rate": 2.005450598594109e-05,
      "loss": 0.5757,
      "step": 6909
    },
    {
      "epoch": 1.19882026370576,
      "grad_norm": 0.6492012143135071,
      "learning_rate": 2.0047692751535454e-05,
      "loss": 0.598,
      "step": 6910
    },
    {
      "epoch": 1.1989937543372657,
      "grad_norm": 0.9850520491600037,
      "learning_rate": 2.004087951159502e-05,
      "loss": 0.4226,
      "step": 6911
    },
    {
      "epoch": 1.1991672449687716,
      "grad_norm": 1.0020174980163574,
      "learning_rate": 2.0034066266910475e-05,
      "loss": 0.5101,
      "step": 6912
    },
    {
      "epoch": 1.1993407356002777,
      "grad_norm": 0.7792971730232239,
      "learning_rate": 2.0027253018272498e-05,
      "loss": 0.5876,
      "step": 6913
    },
    {
      "epoch": 1.1995142262317835,
      "grad_norm": 2.3527796268463135,
      "learning_rate": 2.0020439766471775e-05,
      "loss": 0.486,
      "step": 6914
    },
    {
      "epoch": 1.1996877168632893,
      "grad_norm": 0.6588824391365051,
      "learning_rate": 2.0013626512298996e-05,
      "loss": 0.6024,
      "step": 6915
    },
    {
      "epoch": 1.1998612074947952,
      "grad_norm": 0.7276008129119873,
      "learning_rate": 2.000681325654484e-05,
      "loss": 0.4839,
      "step": 6916
    },
    {
      "epoch": 1.2000346981263013,
      "grad_norm": 0.8354933261871338,
      "learning_rate": 2e-05,
      "loss": 0.5,
      "step": 6917
    },
    {
      "epoch": 1.200208188757807,
      "grad_norm": 0.727831244468689,
      "learning_rate": 1.999318674345516e-05,
      "loss": 0.5079,
      "step": 6918
    },
    {
      "epoch": 1.200381679389313,
      "grad_norm": 0.8418210744857788,
      "learning_rate": 1.9986373487701014e-05,
      "loss": 0.526,
      "step": 6919
    },
    {
      "epoch": 1.2005551700208188,
      "grad_norm": 1.2001993656158447,
      "learning_rate": 1.997956023352823e-05,
      "loss": 0.4738,
      "step": 6920
    },
    {
      "epoch": 1.2007286606523249,
      "grad_norm": 0.7162483930587769,
      "learning_rate": 1.9972746981727505e-05,
      "loss": 0.4237,
      "step": 6921
    },
    {
      "epoch": 1.2009021512838307,
      "grad_norm": 0.7253560423851013,
      "learning_rate": 1.9965933733089535e-05,
      "loss": 0.5894,
      "step": 6922
    },
    {
      "epoch": 1.2010756419153366,
      "grad_norm": 1.0020893812179565,
      "learning_rate": 1.9959120488404986e-05,
      "loss": 0.6227,
      "step": 6923
    },
    {
      "epoch": 1.2012491325468424,
      "grad_norm": 0.7527416348457336,
      "learning_rate": 1.995230724846455e-05,
      "loss": 0.4722,
      "step": 6924
    },
    {
      "epoch": 1.2014226231783485,
      "grad_norm": 0.6637478470802307,
      "learning_rate": 1.9945494014058915e-05,
      "loss": 0.5936,
      "step": 6925
    },
    {
      "epoch": 1.2015961138098543,
      "grad_norm": 0.8485209941864014,
      "learning_rate": 1.993868078597877e-05,
      "loss": 0.4106,
      "step": 6926
    },
    {
      "epoch": 1.2017696044413602,
      "grad_norm": 0.8287883400917053,
      "learning_rate": 1.9931867565014785e-05,
      "loss": 0.5531,
      "step": 6927
    },
    {
      "epoch": 1.201943095072866,
      "grad_norm": 0.9901884198188782,
      "learning_rate": 1.9925054351957647e-05,
      "loss": 0.5551,
      "step": 6928
    },
    {
      "epoch": 1.2021165857043719,
      "grad_norm": 0.8183797001838684,
      "learning_rate": 1.9918241147598043e-05,
      "loss": 0.5022,
      "step": 6929
    },
    {
      "epoch": 1.202290076335878,
      "grad_norm": 0.7539851069450378,
      "learning_rate": 1.9911427952726644e-05,
      "loss": 0.4634,
      "step": 6930
    },
    {
      "epoch": 1.2024635669673838,
      "grad_norm": 0.6804836988449097,
      "learning_rate": 1.990461476813414e-05,
      "loss": 0.5204,
      "step": 6931
    },
    {
      "epoch": 1.2026370575988896,
      "grad_norm": 0.7465506792068481,
      "learning_rate": 1.9897801594611204e-05,
      "loss": 0.4366,
      "step": 6932
    },
    {
      "epoch": 1.2028105482303957,
      "grad_norm": 1.096960186958313,
      "learning_rate": 1.9890988432948508e-05,
      "loss": 0.4982,
      "step": 6933
    },
    {
      "epoch": 1.2029840388619015,
      "grad_norm": 1.396376371383667,
      "learning_rate": 1.988417528393673e-05,
      "loss": 0.5154,
      "step": 6934
    },
    {
      "epoch": 1.2031575294934074,
      "grad_norm": 0.8130907416343689,
      "learning_rate": 1.9877362148366555e-05,
      "loss": 0.5557,
      "step": 6935
    },
    {
      "epoch": 1.2033310201249132,
      "grad_norm": 0.7420282959938049,
      "learning_rate": 1.9870549027028635e-05,
      "loss": 0.4702,
      "step": 6936
    },
    {
      "epoch": 1.203504510756419,
      "grad_norm": 0.7370325922966003,
      "learning_rate": 1.9863735920713653e-05,
      "loss": 0.4708,
      "step": 6937
    },
    {
      "epoch": 1.2036780013879251,
      "grad_norm": 0.9016007781028748,
      "learning_rate": 1.9856922830212286e-05,
      "loss": 0.5317,
      "step": 6938
    },
    {
      "epoch": 1.203851492019431,
      "grad_norm": 0.8386825919151306,
      "learning_rate": 1.9850109756315176e-05,
      "loss": 0.4097,
      "step": 6939
    },
    {
      "epoch": 1.2040249826509368,
      "grad_norm": 0.7259785532951355,
      "learning_rate": 1.984329669981301e-05,
      "loss": 0.4994,
      "step": 6940
    },
    {
      "epoch": 1.2041984732824427,
      "grad_norm": 0.8297507762908936,
      "learning_rate": 1.983648366149644e-05,
      "loss": 0.5159,
      "step": 6941
    },
    {
      "epoch": 1.2043719639139487,
      "grad_norm": 1.1169674396514893,
      "learning_rate": 1.9829670642156147e-05,
      "loss": 0.4608,
      "step": 6942
    },
    {
      "epoch": 1.2045454545454546,
      "grad_norm": 0.5992581844329834,
      "learning_rate": 1.982285764258276e-05,
      "loss": 0.5321,
      "step": 6943
    },
    {
      "epoch": 1.2047189451769604,
      "grad_norm": 0.663050651550293,
      "learning_rate": 1.981604466356695e-05,
      "loss": 0.5244,
      "step": 6944
    },
    {
      "epoch": 1.2048924358084663,
      "grad_norm": 0.8010140657424927,
      "learning_rate": 1.9809231705899384e-05,
      "loss": 0.4982,
      "step": 6945
    },
    {
      "epoch": 1.2050659264399721,
      "grad_norm": 1.018222689628601,
      "learning_rate": 1.980241877037069e-05,
      "loss": 0.5137,
      "step": 6946
    },
    {
      "epoch": 1.2052394170714782,
      "grad_norm": 0.7997981905937195,
      "learning_rate": 1.9795605857771527e-05,
      "loss": 0.5393,
      "step": 6947
    },
    {
      "epoch": 1.205412907702984,
      "grad_norm": 0.842365562915802,
      "learning_rate": 1.9788792968892553e-05,
      "loss": 0.5146,
      "step": 6948
    },
    {
      "epoch": 1.2055863983344899,
      "grad_norm": 1.6468077898025513,
      "learning_rate": 1.978198010452439e-05,
      "loss": 0.66,
      "step": 6949
    },
    {
      "epoch": 1.205759888965996,
      "grad_norm": 0.8329322338104248,
      "learning_rate": 1.977516726545769e-05,
      "loss": 0.5454,
      "step": 6950
    },
    {
      "epoch": 1.2059333795975018,
      "grad_norm": 0.5599431395530701,
      "learning_rate": 1.97683544524831e-05,
      "loss": 0.6841,
      "step": 6951
    },
    {
      "epoch": 1.2061068702290076,
      "grad_norm": 0.7028522491455078,
      "learning_rate": 1.976154166639124e-05,
      "loss": 0.5775,
      "step": 6952
    },
    {
      "epoch": 1.2062803608605135,
      "grad_norm": 0.8986274003982544,
      "learning_rate": 1.9754728907972745e-05,
      "loss": 0.4996,
      "step": 6953
    },
    {
      "epoch": 1.2064538514920193,
      "grad_norm": 1.0744249820709229,
      "learning_rate": 1.9747916178018246e-05,
      "loss": 0.5764,
      "step": 6954
    },
    {
      "epoch": 1.2066273421235254,
      "grad_norm": 1.011263132095337,
      "learning_rate": 1.974110347731837e-05,
      "loss": 0.5446,
      "step": 6955
    },
    {
      "epoch": 1.2068008327550312,
      "grad_norm": 0.8077258467674255,
      "learning_rate": 1.9734290806663738e-05,
      "loss": 0.5444,
      "step": 6956
    },
    {
      "epoch": 1.206974323386537,
      "grad_norm": 0.702093243598938,
      "learning_rate": 1.972747816684497e-05,
      "loss": 0.3857,
      "step": 6957
    },
    {
      "epoch": 1.207147814018043,
      "grad_norm": 0.7129010558128357,
      "learning_rate": 1.9720665558652676e-05,
      "loss": 0.543,
      "step": 6958
    },
    {
      "epoch": 1.207321304649549,
      "grad_norm": 0.7714444398880005,
      "learning_rate": 1.971385298287747e-05,
      "loss": 0.4484,
      "step": 6959
    },
    {
      "epoch": 1.2074947952810549,
      "grad_norm": 0.7922438979148865,
      "learning_rate": 1.970704044030995e-05,
      "loss": 0.4916,
      "step": 6960
    },
    {
      "epoch": 1.2076682859125607,
      "grad_norm": 1.1409392356872559,
      "learning_rate": 1.9700227931740747e-05,
      "loss": 0.5165,
      "step": 6961
    },
    {
      "epoch": 1.2078417765440665,
      "grad_norm": 0.777527391910553,
      "learning_rate": 1.9693415457960426e-05,
      "loss": 0.484,
      "step": 6962
    },
    {
      "epoch": 1.2080152671755726,
      "grad_norm": 0.7120514512062073,
      "learning_rate": 1.9686603019759602e-05,
      "loss": 0.5157,
      "step": 6963
    },
    {
      "epoch": 1.2081887578070785,
      "grad_norm": 0.5948834419250488,
      "learning_rate": 1.9679790617928872e-05,
      "loss": 0.7015,
      "step": 6964
    },
    {
      "epoch": 1.2083622484385843,
      "grad_norm": 1.2719899415969849,
      "learning_rate": 1.9672978253258802e-05,
      "loss": 0.5376,
      "step": 6965
    },
    {
      "epoch": 1.2085357390700902,
      "grad_norm": 0.8294084072113037,
      "learning_rate": 1.966616592653999e-05,
      "loss": 0.4955,
      "step": 6966
    },
    {
      "epoch": 1.2087092297015962,
      "grad_norm": 0.8743494153022766,
      "learning_rate": 1.9659353638563016e-05,
      "loss": 0.4963,
      "step": 6967
    },
    {
      "epoch": 1.208882720333102,
      "grad_norm": 1.3656015396118164,
      "learning_rate": 1.9652541390118443e-05,
      "loss": 0.4631,
      "step": 6968
    },
    {
      "epoch": 1.209056210964608,
      "grad_norm": 1.0569608211517334,
      "learning_rate": 1.9645729181996846e-05,
      "loss": 0.5775,
      "step": 6969
    },
    {
      "epoch": 1.2092297015961138,
      "grad_norm": 0.7379013299942017,
      "learning_rate": 1.963891701498879e-05,
      "loss": 0.5563,
      "step": 6970
    },
    {
      "epoch": 1.2094031922276196,
      "grad_norm": 0.7099375128746033,
      "learning_rate": 1.9632104889884844e-05,
      "loss": 0.6627,
      "step": 6971
    },
    {
      "epoch": 1.2095766828591257,
      "grad_norm": 0.7381976246833801,
      "learning_rate": 1.9625292807475548e-05,
      "loss": 0.6183,
      "step": 6972
    },
    {
      "epoch": 1.2097501734906315,
      "grad_norm": 0.9388188719749451,
      "learning_rate": 1.9618480768551456e-05,
      "loss": 0.4502,
      "step": 6973
    },
    {
      "epoch": 1.2099236641221374,
      "grad_norm": 0.6752588748931885,
      "learning_rate": 1.9611668773903124e-05,
      "loss": 0.5327,
      "step": 6974
    },
    {
      "epoch": 1.2100971547536432,
      "grad_norm": 1.0585280656814575,
      "learning_rate": 1.9604856824321076e-05,
      "loss": 0.4553,
      "step": 6975
    },
    {
      "epoch": 1.2102706453851493,
      "grad_norm": 0.8340199589729309,
      "learning_rate": 1.9598044920595853e-05,
      "loss": 0.5798,
      "step": 6976
    },
    {
      "epoch": 1.2104441360166551,
      "grad_norm": 0.7418455481529236,
      "learning_rate": 1.9591233063517988e-05,
      "loss": 0.4629,
      "step": 6977
    },
    {
      "epoch": 1.210617626648161,
      "grad_norm": 0.8385781049728394,
      "learning_rate": 1.9584421253878e-05,
      "loss": 0.4479,
      "step": 6978
    },
    {
      "epoch": 1.2107911172796668,
      "grad_norm": 0.6666460633277893,
      "learning_rate": 1.957760949246641e-05,
      "loss": 0.4623,
      "step": 6979
    },
    {
      "epoch": 1.2109646079111729,
      "grad_norm": 0.6215448975563049,
      "learning_rate": 1.9570797780073737e-05,
      "loss": 0.4785,
      "step": 6980
    },
    {
      "epoch": 1.2111380985426787,
      "grad_norm": 1.3658274412155151,
      "learning_rate": 1.9563986117490476e-05,
      "loss": 0.4884,
      "step": 6981
    },
    {
      "epoch": 1.2113115891741846,
      "grad_norm": 0.9017288088798523,
      "learning_rate": 1.955717450550713e-05,
      "loss": 0.4233,
      "step": 6982
    },
    {
      "epoch": 1.2114850798056904,
      "grad_norm": 0.8542715311050415,
      "learning_rate": 1.9550362944914202e-05,
      "loss": 0.4851,
      "step": 6983
    },
    {
      "epoch": 1.2116585704371965,
      "grad_norm": 0.8546389937400818,
      "learning_rate": 1.9543551436502186e-05,
      "loss": 0.4491,
      "step": 6984
    },
    {
      "epoch": 1.2118320610687023,
      "grad_norm": 0.6703542470932007,
      "learning_rate": 1.953673998106155e-05,
      "loss": 0.5707,
      "step": 6985
    },
    {
      "epoch": 1.2120055517002082,
      "grad_norm": 0.6389185786247253,
      "learning_rate": 1.9529928579382778e-05,
      "loss": 0.6223,
      "step": 6986
    },
    {
      "epoch": 1.212179042331714,
      "grad_norm": 0.7991594672203064,
      "learning_rate": 1.952311723225635e-05,
      "loss": 0.5618,
      "step": 6987
    },
    {
      "epoch": 1.2123525329632199,
      "grad_norm": 0.7087690234184265,
      "learning_rate": 1.9516305940472714e-05,
      "loss": 0.5809,
      "step": 6988
    },
    {
      "epoch": 1.212526023594726,
      "grad_norm": 1.2076226472854614,
      "learning_rate": 1.950949470482234e-05,
      "loss": 0.5221,
      "step": 6989
    },
    {
      "epoch": 1.2126995142262318,
      "grad_norm": 0.9708346724510193,
      "learning_rate": 1.9502683526095683e-05,
      "loss": 0.4517,
      "step": 6990
    },
    {
      "epoch": 1.2128730048577376,
      "grad_norm": 1.456707239151001,
      "learning_rate": 1.9495872405083178e-05,
      "loss": 0.5212,
      "step": 6991
    },
    {
      "epoch": 1.2130464954892437,
      "grad_norm": 1.0726432800292969,
      "learning_rate": 1.948906134257526e-05,
      "loss": 0.4384,
      "step": 6992
    },
    {
      "epoch": 1.2132199861207495,
      "grad_norm": 1.0962409973144531,
      "learning_rate": 1.9482250339362383e-05,
      "loss": 0.4515,
      "step": 6993
    },
    {
      "epoch": 1.2133934767522554,
      "grad_norm": 0.7538032531738281,
      "learning_rate": 1.947543939623495e-05,
      "loss": 0.4922,
      "step": 6994
    },
    {
      "epoch": 1.2135669673837612,
      "grad_norm": 0.6666850447654724,
      "learning_rate": 1.9468628513983382e-05,
      "loss": 0.5846,
      "step": 6995
    },
    {
      "epoch": 1.213740458015267,
      "grad_norm": 1.1026028394699097,
      "learning_rate": 1.9461817693398105e-05,
      "loss": 0.4419,
      "step": 6996
    },
    {
      "epoch": 1.2139139486467732,
      "grad_norm": 0.9549009799957275,
      "learning_rate": 1.94550069352695e-05,
      "loss": 0.3739,
      "step": 6997
    },
    {
      "epoch": 1.214087439278279,
      "grad_norm": 0.7214818596839905,
      "learning_rate": 1.944819624038798e-05,
      "loss": 0.4482,
      "step": 6998
    },
    {
      "epoch": 1.2142609299097848,
      "grad_norm": 0.8108447194099426,
      "learning_rate": 1.9441385609543925e-05,
      "loss": 0.4645,
      "step": 6999
    },
    {
      "epoch": 1.2144344205412907,
      "grad_norm": 1.1380990743637085,
      "learning_rate": 1.943457504352773e-05,
      "loss": 0.4808,
      "step": 7000
    },
    {
      "epoch": 1.2146079111727968,
      "grad_norm": 1.3156373500823975,
      "learning_rate": 1.9427764543129756e-05,
      "loss": 0.4548,
      "step": 7001
    },
    {
      "epoch": 1.2147814018043026,
      "grad_norm": 0.7099040746688843,
      "learning_rate": 1.942095410914037e-05,
      "loss": 0.6019,
      "step": 7002
    },
    {
      "epoch": 1.2149548924358085,
      "grad_norm": 1.1159850358963013,
      "learning_rate": 1.941414374234994e-05,
      "loss": 0.6477,
      "step": 7003
    },
    {
      "epoch": 1.2151283830673143,
      "grad_norm": 1.122902512550354,
      "learning_rate": 1.9407333443548806e-05,
      "loss": 0.481,
      "step": 7004
    },
    {
      "epoch": 1.2153018736988201,
      "grad_norm": 0.8170135021209717,
      "learning_rate": 1.9400523213527324e-05,
      "loss": 0.6008,
      "step": 7005
    },
    {
      "epoch": 1.2154753643303262,
      "grad_norm": 0.811800479888916,
      "learning_rate": 1.9393713053075816e-05,
      "loss": 0.6046,
      "step": 7006
    },
    {
      "epoch": 1.215648854961832,
      "grad_norm": 0.790324866771698,
      "learning_rate": 1.9386902962984613e-05,
      "loss": 0.4691,
      "step": 7007
    },
    {
      "epoch": 1.215822345593338,
      "grad_norm": 0.954623281955719,
      "learning_rate": 1.9380092944044036e-05,
      "loss": 0.5197,
      "step": 7008
    },
    {
      "epoch": 1.215995836224844,
      "grad_norm": 0.7938142418861389,
      "learning_rate": 1.9373282997044404e-05,
      "loss": 0.4347,
      "step": 7009
    },
    {
      "epoch": 1.2161693268563498,
      "grad_norm": 0.6294236183166504,
      "learning_rate": 1.9366473122776e-05,
      "loss": 0.5979,
      "step": 7010
    },
    {
      "epoch": 1.2163428174878557,
      "grad_norm": 0.7136385440826416,
      "learning_rate": 1.935966332202913e-05,
      "loss": 0.5465,
      "step": 7011
    },
    {
      "epoch": 1.2165163081193615,
      "grad_norm": 2.5835301876068115,
      "learning_rate": 1.9352853595594077e-05,
      "loss": 0.4447,
      "step": 7012
    },
    {
      "epoch": 1.2166897987508674,
      "grad_norm": 0.7536041140556335,
      "learning_rate": 1.9346043944261127e-05,
      "loss": 0.6012,
      "step": 7013
    },
    {
      "epoch": 1.2168632893823734,
      "grad_norm": 0.5537289381027222,
      "learning_rate": 1.9339234368820533e-05,
      "loss": 0.6119,
      "step": 7014
    },
    {
      "epoch": 1.2170367800138793,
      "grad_norm": 0.7403371334075928,
      "learning_rate": 1.9332424870062557e-05,
      "loss": 0.5985,
      "step": 7015
    },
    {
      "epoch": 1.2172102706453851,
      "grad_norm": 1.8811901807785034,
      "learning_rate": 1.932561544877746e-05,
      "loss": 0.4784,
      "step": 7016
    },
    {
      "epoch": 1.217383761276891,
      "grad_norm": 0.7084232568740845,
      "learning_rate": 1.9318806105755474e-05,
      "loss": 0.4897,
      "step": 7017
    },
    {
      "epoch": 1.217557251908397,
      "grad_norm": 0.8779373168945312,
      "learning_rate": 1.9311996841786825e-05,
      "loss": 0.4393,
      "step": 7018
    },
    {
      "epoch": 1.2177307425399029,
      "grad_norm": 0.6649523973464966,
      "learning_rate": 1.9305187657661758e-05,
      "loss": 0.4427,
      "step": 7019
    },
    {
      "epoch": 1.2179042331714087,
      "grad_norm": 0.7244396805763245,
      "learning_rate": 1.9298378554170463e-05,
      "loss": 0.5028,
      "step": 7020
    },
    {
      "epoch": 1.2180777238029146,
      "grad_norm": 1.0198639631271362,
      "learning_rate": 1.9291569532103155e-05,
      "loss": 0.4269,
      "step": 7021
    },
    {
      "epoch": 1.2182512144344206,
      "grad_norm": 0.6823749542236328,
      "learning_rate": 1.9284760592250037e-05,
      "loss": 0.5574,
      "step": 7022
    },
    {
      "epoch": 1.2184247050659265,
      "grad_norm": 1.2214984893798828,
      "learning_rate": 1.9277951735401276e-05,
      "loss": 0.5712,
      "step": 7023
    },
    {
      "epoch": 1.2185981956974323,
      "grad_norm": 0.7772981524467468,
      "learning_rate": 1.9271142962347058e-05,
      "loss": 0.5751,
      "step": 7024
    },
    {
      "epoch": 1.2187716863289382,
      "grad_norm": 1.054054617881775,
      "learning_rate": 1.926433427387755e-05,
      "loss": 0.5153,
      "step": 7025
    },
    {
      "epoch": 1.2189451769604442,
      "grad_norm": 0.9166788458824158,
      "learning_rate": 1.925752567078291e-05,
      "loss": 0.5312,
      "step": 7026
    },
    {
      "epoch": 1.21911866759195,
      "grad_norm": 0.7016672492027283,
      "learning_rate": 1.925071715385328e-05,
      "loss": 0.5461,
      "step": 7027
    },
    {
      "epoch": 1.219292158223456,
      "grad_norm": 0.7995874285697937,
      "learning_rate": 1.924390872387879e-05,
      "loss": 0.614,
      "step": 7028
    },
    {
      "epoch": 1.2194656488549618,
      "grad_norm": 0.8818855285644531,
      "learning_rate": 1.9237100381649586e-05,
      "loss": 0.6071,
      "step": 7029
    },
    {
      "epoch": 1.2196391394864676,
      "grad_norm": 0.7914130687713623,
      "learning_rate": 1.923029212795576e-05,
      "loss": 0.4543,
      "step": 7030
    },
    {
      "epoch": 1.2198126301179737,
      "grad_norm": 0.7113505005836487,
      "learning_rate": 1.9223483963587435e-05,
      "loss": 0.4501,
      "step": 7031
    },
    {
      "epoch": 1.2199861207494795,
      "grad_norm": 0.7274286150932312,
      "learning_rate": 1.9216675889334704e-05,
      "loss": 0.5541,
      "step": 7032
    },
    {
      "epoch": 1.2201596113809854,
      "grad_norm": 0.7852982878684998,
      "learning_rate": 1.9209867905987643e-05,
      "loss": 0.5742,
      "step": 7033
    },
    {
      "epoch": 1.2203331020124912,
      "grad_norm": 1.0195103883743286,
      "learning_rate": 1.9203060014336334e-05,
      "loss": 0.5359,
      "step": 7034
    },
    {
      "epoch": 1.2205065926439973,
      "grad_norm": 0.9056409597396851,
      "learning_rate": 1.9196252215170848e-05,
      "loss": 0.5608,
      "step": 7035
    },
    {
      "epoch": 1.2206800832755031,
      "grad_norm": 0.7680935263633728,
      "learning_rate": 1.9189444509281216e-05,
      "loss": 0.4695,
      "step": 7036
    },
    {
      "epoch": 1.220853573907009,
      "grad_norm": 0.6084681153297424,
      "learning_rate": 1.9182636897457496e-05,
      "loss": 0.5288,
      "step": 7037
    },
    {
      "epoch": 1.2210270645385148,
      "grad_norm": 0.9058395028114319,
      "learning_rate": 1.9175829380489727e-05,
      "loss": 0.4277,
      "step": 7038
    },
    {
      "epoch": 1.221200555170021,
      "grad_norm": 0.7618827819824219,
      "learning_rate": 1.9169021959167905e-05,
      "loss": 0.4824,
      "step": 7039
    },
    {
      "epoch": 1.2213740458015268,
      "grad_norm": 1.0226870775222778,
      "learning_rate": 1.9162214634282055e-05,
      "loss": 0.4717,
      "step": 7040
    },
    {
      "epoch": 1.2215475364330326,
      "grad_norm": 0.8193103671073914,
      "learning_rate": 1.9155407406622177e-05,
      "loss": 0.5593,
      "step": 7041
    },
    {
      "epoch": 1.2217210270645384,
      "grad_norm": 0.6056613326072693,
      "learning_rate": 1.9148600276978254e-05,
      "loss": 0.6573,
      "step": 7042
    },
    {
      "epoch": 1.2218945176960445,
      "grad_norm": 0.8092714548110962,
      "learning_rate": 1.9141793246140257e-05,
      "loss": 0.5009,
      "step": 7043
    },
    {
      "epoch": 1.2220680083275504,
      "grad_norm": 0.8361021876335144,
      "learning_rate": 1.9134986314898156e-05,
      "loss": 0.4344,
      "step": 7044
    },
    {
      "epoch": 1.2222414989590562,
      "grad_norm": 0.816299557685852,
      "learning_rate": 1.9128179484041908e-05,
      "loss": 0.5144,
      "step": 7045
    },
    {
      "epoch": 1.222414989590562,
      "grad_norm": 0.8609830141067505,
      "learning_rate": 1.9121372754361437e-05,
      "loss": 0.5228,
      "step": 7046
    },
    {
      "epoch": 1.222588480222068,
      "grad_norm": 0.8750868439674377,
      "learning_rate": 1.911456612664668e-05,
      "loss": 0.3973,
      "step": 7047
    },
    {
      "epoch": 1.222761970853574,
      "grad_norm": 0.809299111366272,
      "learning_rate": 1.9107759601687562e-05,
      "loss": 0.5521,
      "step": 7048
    },
    {
      "epoch": 1.2229354614850798,
      "grad_norm": 0.811198890209198,
      "learning_rate": 1.9100953180273985e-05,
      "loss": 0.4379,
      "step": 7049
    },
    {
      "epoch": 1.2231089521165857,
      "grad_norm": 0.823857843875885,
      "learning_rate": 1.909414686319583e-05,
      "loss": 0.4657,
      "step": 7050
    },
    {
      "epoch": 1.2232824427480917,
      "grad_norm": 0.8796674609184265,
      "learning_rate": 1.9087340651242995e-05,
      "loss": 0.4768,
      "step": 7051
    },
    {
      "epoch": 1.2234559333795976,
      "grad_norm": 1.166199803352356,
      "learning_rate": 1.9080534545205334e-05,
      "loss": 0.4156,
      "step": 7052
    },
    {
      "epoch": 1.2236294240111034,
      "grad_norm": 1.3353817462921143,
      "learning_rate": 1.9073728545872717e-05,
      "loss": 0.5551,
      "step": 7053
    },
    {
      "epoch": 1.2238029146426093,
      "grad_norm": 0.8513011932373047,
      "learning_rate": 1.9066922654034975e-05,
      "loss": 0.5559,
      "step": 7054
    },
    {
      "epoch": 1.223976405274115,
      "grad_norm": 1.4038827419281006,
      "learning_rate": 1.906011687048195e-05,
      "loss": 0.4769,
      "step": 7055
    },
    {
      "epoch": 1.2241498959056212,
      "grad_norm": 0.9031059741973877,
      "learning_rate": 1.9053311196003457e-05,
      "loss": 0.4774,
      "step": 7056
    },
    {
      "epoch": 1.224323386537127,
      "grad_norm": 1.3607302904129028,
      "learning_rate": 1.9046505631389297e-05,
      "loss": 0.5842,
      "step": 7057
    },
    {
      "epoch": 1.2244968771686329,
      "grad_norm": 0.5526803731918335,
      "learning_rate": 1.9039700177429282e-05,
      "loss": 0.6354,
      "step": 7058
    },
    {
      "epoch": 1.2246703678001387,
      "grad_norm": 0.6610570549964905,
      "learning_rate": 1.9032894834913172e-05,
      "loss": 0.4768,
      "step": 7059
    },
    {
      "epoch": 1.2248438584316448,
      "grad_norm": 1.3063814640045166,
      "learning_rate": 1.9026089604630743e-05,
      "loss": 0.5011,
      "step": 7060
    },
    {
      "epoch": 1.2250173490631506,
      "grad_norm": 1.0547101497650146,
      "learning_rate": 1.901928448737176e-05,
      "loss": 0.4799,
      "step": 7061
    },
    {
      "epoch": 1.2251908396946565,
      "grad_norm": 0.7201014757156372,
      "learning_rate": 1.9012479483925942e-05,
      "loss": 0.5045,
      "step": 7062
    },
    {
      "epoch": 1.2253643303261623,
      "grad_norm": 0.7078681588172913,
      "learning_rate": 1.9005674595083033e-05,
      "loss": 0.4971,
      "step": 7063
    },
    {
      "epoch": 1.2255378209576682,
      "grad_norm": 0.8380757570266724,
      "learning_rate": 1.8998869821632757e-05,
      "loss": 0.4874,
      "step": 7064
    },
    {
      "epoch": 1.2257113115891742,
      "grad_norm": 0.7439054846763611,
      "learning_rate": 1.8992065164364793e-05,
      "loss": 0.5192,
      "step": 7065
    },
    {
      "epoch": 1.22588480222068,
      "grad_norm": 0.697297990322113,
      "learning_rate": 1.898526062406884e-05,
      "loss": 0.4739,
      "step": 7066
    },
    {
      "epoch": 1.226058292852186,
      "grad_norm": 0.8759002089500427,
      "learning_rate": 1.8978456201534587e-05,
      "loss": 0.5293,
      "step": 7067
    },
    {
      "epoch": 1.226231783483692,
      "grad_norm": 0.665690541267395,
      "learning_rate": 1.8971651897551672e-05,
      "loss": 0.6355,
      "step": 7068
    },
    {
      "epoch": 1.2264052741151978,
      "grad_norm": 0.8324854969978333,
      "learning_rate": 1.896484771290975e-05,
      "loss": 0.53,
      "step": 7069
    },
    {
      "epoch": 1.2265787647467037,
      "grad_norm": 0.949212908744812,
      "learning_rate": 1.8958043648398457e-05,
      "loss": 0.5593,
      "step": 7070
    },
    {
      "epoch": 1.2267522553782095,
      "grad_norm": 1.048187017440796,
      "learning_rate": 1.8951239704807424e-05,
      "loss": 0.4734,
      "step": 7071
    },
    {
      "epoch": 1.2269257460097154,
      "grad_norm": 0.7703348994255066,
      "learning_rate": 1.8944435882926236e-05,
      "loss": 0.5366,
      "step": 7072
    },
    {
      "epoch": 1.2270992366412214,
      "grad_norm": 0.6312597990036011,
      "learning_rate": 1.8937632183544495e-05,
      "loss": 0.4113,
      "step": 7073
    },
    {
      "epoch": 1.2272727272727273,
      "grad_norm": 1.056952953338623,
      "learning_rate": 1.893082860745178e-05,
      "loss": 0.5793,
      "step": 7074
    },
    {
      "epoch": 1.2274462179042331,
      "grad_norm": 0.8766586780548096,
      "learning_rate": 1.892402515543765e-05,
      "loss": 0.5913,
      "step": 7075
    },
    {
      "epoch": 1.227619708535739,
      "grad_norm": 0.6282052397727966,
      "learning_rate": 1.8917221828291652e-05,
      "loss": 0.6182,
      "step": 7076
    },
    {
      "epoch": 1.227793199167245,
      "grad_norm": 0.7187272310256958,
      "learning_rate": 1.8910418626803328e-05,
      "loss": 0.5498,
      "step": 7077
    },
    {
      "epoch": 1.227966689798751,
      "grad_norm": 0.7514996528625488,
      "learning_rate": 1.8903615551762182e-05,
      "loss": 0.5405,
      "step": 7078
    },
    {
      "epoch": 1.2281401804302567,
      "grad_norm": 0.8368105888366699,
      "learning_rate": 1.8896812603957732e-05,
      "loss": 0.4512,
      "step": 7079
    },
    {
      "epoch": 1.2283136710617626,
      "grad_norm": 1.9764368534088135,
      "learning_rate": 1.8890009784179476e-05,
      "loss": 0.4839,
      "step": 7080
    },
    {
      "epoch": 1.2284871616932687,
      "grad_norm": 0.899013340473175,
      "learning_rate": 1.8883207093216865e-05,
      "loss": 0.4847,
      "step": 7081
    },
    {
      "epoch": 1.2286606523247745,
      "grad_norm": 1.7209386825561523,
      "learning_rate": 1.8876404531859376e-05,
      "loss": 0.4948,
      "step": 7082
    },
    {
      "epoch": 1.2288341429562804,
      "grad_norm": 0.5783351063728333,
      "learning_rate": 1.8869602100896446e-05,
      "loss": 0.5078,
      "step": 7083
    },
    {
      "epoch": 1.2290076335877862,
      "grad_norm": 0.771289050579071,
      "learning_rate": 1.8862799801117523e-05,
      "loss": 0.5085,
      "step": 7084
    },
    {
      "epoch": 1.2291811242192923,
      "grad_norm": 0.8002333641052246,
      "learning_rate": 1.8855997633311995e-05,
      "loss": 0.4974,
      "step": 7085
    },
    {
      "epoch": 1.229354614850798,
      "grad_norm": 0.791644275188446,
      "learning_rate": 1.884919559826928e-05,
      "loss": 0.464,
      "step": 7086
    },
    {
      "epoch": 1.229528105482304,
      "grad_norm": 0.8913290500640869,
      "learning_rate": 1.884239369677876e-05,
      "loss": 0.4539,
      "step": 7087
    },
    {
      "epoch": 1.2297015961138098,
      "grad_norm": 0.8145409226417542,
      "learning_rate": 1.8835591929629795e-05,
      "loss": 0.4933,
      "step": 7088
    },
    {
      "epoch": 1.2298750867453156,
      "grad_norm": 0.63204425573349,
      "learning_rate": 1.8828790297611742e-05,
      "loss": 0.5541,
      "step": 7089
    },
    {
      "epoch": 1.2300485773768217,
      "grad_norm": 0.7683725953102112,
      "learning_rate": 1.882198880151395e-05,
      "loss": 0.5432,
      "step": 7090
    },
    {
      "epoch": 1.2302220680083276,
      "grad_norm": 0.8539783954620361,
      "learning_rate": 1.8815187442125716e-05,
      "loss": 0.4684,
      "step": 7091
    },
    {
      "epoch": 1.2303955586398334,
      "grad_norm": 0.7257925868034363,
      "learning_rate": 1.8808386220236365e-05,
      "loss": 0.4198,
      "step": 7092
    },
    {
      "epoch": 1.2305690492713393,
      "grad_norm": 0.622333288192749,
      "learning_rate": 1.8801585136635188e-05,
      "loss": 0.5259,
      "step": 7093
    },
    {
      "epoch": 1.2307425399028453,
      "grad_norm": 0.7729082107543945,
      "learning_rate": 1.8794784192111448e-05,
      "loss": 0.5856,
      "step": 7094
    },
    {
      "epoch": 1.2309160305343512,
      "grad_norm": 0.7876334190368652,
      "learning_rate": 1.8787983387454402e-05,
      "loss": 0.5573,
      "step": 7095
    },
    {
      "epoch": 1.231089521165857,
      "grad_norm": 0.693033754825592,
      "learning_rate": 1.8781182723453303e-05,
      "loss": 0.6444,
      "step": 7096
    },
    {
      "epoch": 1.2312630117973629,
      "grad_norm": 0.5837717652320862,
      "learning_rate": 1.877438220089737e-05,
      "loss": 0.5847,
      "step": 7097
    },
    {
      "epoch": 1.231436502428869,
      "grad_norm": 0.7758795022964478,
      "learning_rate": 1.8767581820575803e-05,
      "loss": 0.6039,
      "step": 7098
    },
    {
      "epoch": 1.2316099930603748,
      "grad_norm": 1.2334827184677124,
      "learning_rate": 1.8760781583277804e-05,
      "loss": 0.4965,
      "step": 7099
    },
    {
      "epoch": 1.2317834836918806,
      "grad_norm": 0.7810057401657104,
      "learning_rate": 1.875398148979255e-05,
      "loss": 0.5089,
      "step": 7100
    },
    {
      "epoch": 1.2319569743233865,
      "grad_norm": 0.6216914653778076,
      "learning_rate": 1.8747181540909193e-05,
      "loss": 0.4302,
      "step": 7101
    },
    {
      "epoch": 1.2321304649548925,
      "grad_norm": 1.337182641029358,
      "learning_rate": 1.874038173741688e-05,
      "loss": 0.512,
      "step": 7102
    },
    {
      "epoch": 1.2323039555863984,
      "grad_norm": 0.7715322971343994,
      "learning_rate": 1.8733582080104736e-05,
      "loss": 0.5513,
      "step": 7103
    },
    {
      "epoch": 1.2324774462179042,
      "grad_norm": 1.0621769428253174,
      "learning_rate": 1.8726782569761864e-05,
      "loss": 0.5614,
      "step": 7104
    },
    {
      "epoch": 1.23265093684941,
      "grad_norm": 1.0144551992416382,
      "learning_rate": 1.8719983207177358e-05,
      "loss": 0.5898,
      "step": 7105
    },
    {
      "epoch": 1.232824427480916,
      "grad_norm": 0.8312714695930481,
      "learning_rate": 1.87131839931403e-05,
      "loss": 0.4198,
      "step": 7106
    },
    {
      "epoch": 1.232997918112422,
      "grad_norm": 0.7275205850601196,
      "learning_rate": 1.8706384928439726e-05,
      "loss": 0.3987,
      "step": 7107
    },
    {
      "epoch": 1.2331714087439278,
      "grad_norm": 1.1540769338607788,
      "learning_rate": 1.8699586013864694e-05,
      "loss": 0.5364,
      "step": 7108
    },
    {
      "epoch": 1.2333448993754337,
      "grad_norm": 0.7301573753356934,
      "learning_rate": 1.8692787250204228e-05,
      "loss": 0.4919,
      "step": 7109
    },
    {
      "epoch": 1.2335183900069397,
      "grad_norm": 0.9503788948059082,
      "learning_rate": 1.8685988638247316e-05,
      "loss": 0.4124,
      "step": 7110
    },
    {
      "epoch": 1.2336918806384456,
      "grad_norm": 1.893370270729065,
      "learning_rate": 1.8679190178782954e-05,
      "loss": 0.6019,
      "step": 7111
    },
    {
      "epoch": 1.2338653712699514,
      "grad_norm": 0.7545531392097473,
      "learning_rate": 1.8672391872600108e-05,
      "loss": 0.4888,
      "step": 7112
    },
    {
      "epoch": 1.2340388619014573,
      "grad_norm": 0.8170567750930786,
      "learning_rate": 1.8665593720487743e-05,
      "loss": 0.4727,
      "step": 7113
    },
    {
      "epoch": 1.2342123525329631,
      "grad_norm": 0.7147437930107117,
      "learning_rate": 1.8658795723234774e-05,
      "loss": 0.5671,
      "step": 7114
    },
    {
      "epoch": 1.2343858431644692,
      "grad_norm": 1.3411918878555298,
      "learning_rate": 1.8651997881630125e-05,
      "loss": 0.4481,
      "step": 7115
    },
    {
      "epoch": 1.234559333795975,
      "grad_norm": 0.723257303237915,
      "learning_rate": 1.8645200196462698e-05,
      "loss": 0.4497,
      "step": 7116
    },
    {
      "epoch": 1.234732824427481,
      "grad_norm": 1.2285536527633667,
      "learning_rate": 1.8638402668521356e-05,
      "loss": 0.4507,
      "step": 7117
    },
    {
      "epoch": 1.2349063150589867,
      "grad_norm": 0.7114037275314331,
      "learning_rate": 1.8631605298594977e-05,
      "loss": 0.5894,
      "step": 7118
    },
    {
      "epoch": 1.2350798056904928,
      "grad_norm": 0.6353415846824646,
      "learning_rate": 1.8624808087472405e-05,
      "loss": 0.5189,
      "step": 7119
    },
    {
      "epoch": 1.2352532963219987,
      "grad_norm": 0.8124098777770996,
      "learning_rate": 1.8618011035942444e-05,
      "loss": 0.4941,
      "step": 7120
    },
    {
      "epoch": 1.2354267869535045,
      "grad_norm": 0.5769098401069641,
      "learning_rate": 1.8611214144793917e-05,
      "loss": 0.504,
      "step": 7121
    },
    {
      "epoch": 1.2356002775850103,
      "grad_norm": 1.2070446014404297,
      "learning_rate": 1.860441741481561e-05,
      "loss": 0.5132,
      "step": 7122
    },
    {
      "epoch": 1.2357737682165162,
      "grad_norm": 1.2472407817840576,
      "learning_rate": 1.8597620846796287e-05,
      "loss": 0.4606,
      "step": 7123
    },
    {
      "epoch": 1.2359472588480223,
      "grad_norm": 0.790421724319458,
      "learning_rate": 1.8590824441524696e-05,
      "loss": 0.3658,
      "step": 7124
    },
    {
      "epoch": 1.236120749479528,
      "grad_norm": 0.7804902791976929,
      "learning_rate": 1.8584028199789573e-05,
      "loss": 0.4025,
      "step": 7125
    },
    {
      "epoch": 1.236294240111034,
      "grad_norm": 1.3904036283493042,
      "learning_rate": 1.8577232122379625e-05,
      "loss": 0.4933,
      "step": 7126
    },
    {
      "epoch": 1.23646773074254,
      "grad_norm": 0.6702448725700378,
      "learning_rate": 1.8570436210083547e-05,
      "loss": 0.6199,
      "step": 7127
    },
    {
      "epoch": 1.2366412213740459,
      "grad_norm": 1.8516900539398193,
      "learning_rate": 1.8563640463690015e-05,
      "loss": 0.5674,
      "step": 7128
    },
    {
      "epoch": 1.2368147120055517,
      "grad_norm": 0.7270691394805908,
      "learning_rate": 1.8556844883987682e-05,
      "loss": 0.5043,
      "step": 7129
    },
    {
      "epoch": 1.2369882026370576,
      "grad_norm": 0.6551758050918579,
      "learning_rate": 1.8550049471765176e-05,
      "loss": 0.6722,
      "step": 7130
    },
    {
      "epoch": 1.2371616932685634,
      "grad_norm": 0.953046977519989,
      "learning_rate": 1.854325422781112e-05,
      "loss": 0.63,
      "step": 7131
    },
    {
      "epoch": 1.2373351839000695,
      "grad_norm": 0.7443362474441528,
      "learning_rate": 1.853645915291412e-05,
      "loss": 0.4866,
      "step": 7132
    },
    {
      "epoch": 1.2375086745315753,
      "grad_norm": 0.6354418992996216,
      "learning_rate": 1.8529664247862726e-05,
      "loss": 0.5452,
      "step": 7133
    },
    {
      "epoch": 1.2376821651630812,
      "grad_norm": 0.6309183239936829,
      "learning_rate": 1.8522869513445515e-05,
      "loss": 0.5303,
      "step": 7134
    },
    {
      "epoch": 1.237855655794587,
      "grad_norm": 1.0496909618377686,
      "learning_rate": 1.851607495045102e-05,
      "loss": 0.5178,
      "step": 7135
    },
    {
      "epoch": 1.238029146426093,
      "grad_norm": 1.0238380432128906,
      "learning_rate": 1.850928055966775e-05,
      "loss": 0.5168,
      "step": 7136
    },
    {
      "epoch": 1.238202637057599,
      "grad_norm": 1.1535464525222778,
      "learning_rate": 1.8502486341884215e-05,
      "loss": 0.4329,
      "step": 7137
    },
    {
      "epoch": 1.2383761276891048,
      "grad_norm": 1.152807593345642,
      "learning_rate": 1.8495692297888885e-05,
      "loss": 0.4663,
      "step": 7138
    },
    {
      "epoch": 1.2385496183206106,
      "grad_norm": 0.844763457775116,
      "learning_rate": 1.8488898428470213e-05,
      "loss": 0.4979,
      "step": 7139
    },
    {
      "epoch": 1.2387231089521167,
      "grad_norm": 0.8161091208457947,
      "learning_rate": 1.848210473441664e-05,
      "loss": 0.5482,
      "step": 7140
    },
    {
      "epoch": 1.2388965995836225,
      "grad_norm": 0.9885815382003784,
      "learning_rate": 1.8475311216516576e-05,
      "loss": 0.5081,
      "step": 7141
    },
    {
      "epoch": 1.2390700902151284,
      "grad_norm": 0.8437579274177551,
      "learning_rate": 1.8468517875558433e-05,
      "loss": 0.5328,
      "step": 7142
    },
    {
      "epoch": 1.2392435808466342,
      "grad_norm": 1.621742606163025,
      "learning_rate": 1.8461724712330567e-05,
      "loss": 0.3857,
      "step": 7143
    },
    {
      "epoch": 1.2394170714781403,
      "grad_norm": 0.7981905341148376,
      "learning_rate": 1.845493172762134e-05,
      "loss": 0.5271,
      "step": 7144
    },
    {
      "epoch": 1.2395905621096461,
      "grad_norm": 0.6473326086997986,
      "learning_rate": 1.8448138922219093e-05,
      "loss": 0.5758,
      "step": 7145
    },
    {
      "epoch": 1.239764052741152,
      "grad_norm": 0.7194153070449829,
      "learning_rate": 1.8441346296912128e-05,
      "loss": 0.5077,
      "step": 7146
    },
    {
      "epoch": 1.2399375433726578,
      "grad_norm": 0.6089683175086975,
      "learning_rate": 1.843455385248874e-05,
      "loss": 0.5817,
      "step": 7147
    },
    {
      "epoch": 1.2401110340041637,
      "grad_norm": 0.868064284324646,
      "learning_rate": 1.8427761589737203e-05,
      "loss": 0.6823,
      "step": 7148
    },
    {
      "epoch": 1.2402845246356697,
      "grad_norm": 0.8100531697273254,
      "learning_rate": 1.8420969509445764e-05,
      "loss": 0.4725,
      "step": 7149
    },
    {
      "epoch": 1.2404580152671756,
      "grad_norm": 1.2863043546676636,
      "learning_rate": 1.8414177612402657e-05,
      "loss": 0.4534,
      "step": 7150
    },
    {
      "epoch": 1.2406315058986814,
      "grad_norm": 0.734415590763092,
      "learning_rate": 1.8407385899396085e-05,
      "loss": 0.4753,
      "step": 7151
    },
    {
      "epoch": 1.2408049965301873,
      "grad_norm": 0.9520745277404785,
      "learning_rate": 1.8400594371214234e-05,
      "loss": 0.4424,
      "step": 7152
    },
    {
      "epoch": 1.2409784871616933,
      "grad_norm": 1.0317492485046387,
      "learning_rate": 1.8393803028645267e-05,
      "loss": 0.5057,
      "step": 7153
    },
    {
      "epoch": 1.2411519777931992,
      "grad_norm": 0.8807188272476196,
      "learning_rate": 1.8387011872477338e-05,
      "loss": 0.4983,
      "step": 7154
    },
    {
      "epoch": 1.241325468424705,
      "grad_norm": 1.0905085802078247,
      "learning_rate": 1.8380220903498554e-05,
      "loss": 0.432,
      "step": 7155
    },
    {
      "epoch": 1.2414989590562109,
      "grad_norm": 0.6536372900009155,
      "learning_rate": 1.8373430122497022e-05,
      "loss": 0.5674,
      "step": 7156
    },
    {
      "epoch": 1.241672449687717,
      "grad_norm": 0.8610592484474182,
      "learning_rate": 1.836663953026082e-05,
      "loss": 0.5138,
      "step": 7157
    },
    {
      "epoch": 1.2418459403192228,
      "grad_norm": 0.6895486116409302,
      "learning_rate": 1.835984912757801e-05,
      "loss": 0.5752,
      "step": 7158
    },
    {
      "epoch": 1.2420194309507286,
      "grad_norm": 0.7571020722389221,
      "learning_rate": 1.8353058915236613e-05,
      "loss": 0.4442,
      "step": 7159
    },
    {
      "epoch": 1.2421929215822345,
      "grad_norm": 0.7846590280532837,
      "learning_rate": 1.8346268894024644e-05,
      "loss": 0.505,
      "step": 7160
    },
    {
      "epoch": 1.2423664122137406,
      "grad_norm": 0.7504282593727112,
      "learning_rate": 1.833947906473011e-05,
      "loss": 0.4613,
      "step": 7161
    },
    {
      "epoch": 1.2425399028452464,
      "grad_norm": 0.6968479156494141,
      "learning_rate": 1.8332689428140956e-05,
      "loss": 0.5884,
      "step": 7162
    },
    {
      "epoch": 1.2427133934767522,
      "grad_norm": 0.6622795462608337,
      "learning_rate": 1.8325899985045135e-05,
      "loss": 0.4684,
      "step": 7163
    },
    {
      "epoch": 1.242886884108258,
      "grad_norm": 0.7619268298149109,
      "learning_rate": 1.831911073623058e-05,
      "loss": 0.5044,
      "step": 7164
    },
    {
      "epoch": 1.243060374739764,
      "grad_norm": 0.7499359846115112,
      "learning_rate": 1.8312321682485178e-05,
      "loss": 0.6053,
      "step": 7165
    },
    {
      "epoch": 1.24323386537127,
      "grad_norm": 1.0497268438339233,
      "learning_rate": 1.830553282459681e-05,
      "loss": 0.5619,
      "step": 7166
    },
    {
      "epoch": 1.2434073560027759,
      "grad_norm": 0.6962741017341614,
      "learning_rate": 1.8298744163353338e-05,
      "loss": 0.4986,
      "step": 7167
    },
    {
      "epoch": 1.2435808466342817,
      "grad_norm": 0.6543954610824585,
      "learning_rate": 1.8291955699542584e-05,
      "loss": 0.557,
      "step": 7168
    },
    {
      "epoch": 1.2437543372657878,
      "grad_norm": 0.6712220311164856,
      "learning_rate": 1.828516743395236e-05,
      "loss": 0.5276,
      "step": 7169
    },
    {
      "epoch": 1.2439278278972936,
      "grad_norm": 0.9903103113174438,
      "learning_rate": 1.8278379367370448e-05,
      "loss": 0.4713,
      "step": 7170
    },
    {
      "epoch": 1.2441013185287995,
      "grad_norm": 1.084958791732788,
      "learning_rate": 1.8271591500584625e-05,
      "loss": 0.4508,
      "step": 7171
    },
    {
      "epoch": 1.2442748091603053,
      "grad_norm": 1.1252861022949219,
      "learning_rate": 1.8264803834382622e-05,
      "loss": 0.5098,
      "step": 7172
    },
    {
      "epoch": 1.2444482997918112,
      "grad_norm": 1.4772077798843384,
      "learning_rate": 1.825801636955215e-05,
      "loss": 0.4685,
      "step": 7173
    },
    {
      "epoch": 1.2446217904233172,
      "grad_norm": 0.9101685881614685,
      "learning_rate": 1.8251229106880916e-05,
      "loss": 0.5321,
      "step": 7174
    },
    {
      "epoch": 1.244795281054823,
      "grad_norm": 0.6553255319595337,
      "learning_rate": 1.8244442047156577e-05,
      "loss": 0.5962,
      "step": 7175
    },
    {
      "epoch": 1.244968771686329,
      "grad_norm": 1.0963441133499146,
      "learning_rate": 1.8237655191166785e-05,
      "loss": 0.3542,
      "step": 7176
    },
    {
      "epoch": 1.2451422623178348,
      "grad_norm": 0.8238844275474548,
      "learning_rate": 1.8230868539699166e-05,
      "loss": 0.4502,
      "step": 7177
    },
    {
      "epoch": 1.2453157529493408,
      "grad_norm": 0.8224257826805115,
      "learning_rate": 1.8224082093541306e-05,
      "loss": 0.5392,
      "step": 7178
    },
    {
      "epoch": 1.2454892435808467,
      "grad_norm": 0.69416344165802,
      "learning_rate": 1.821729585348079e-05,
      "loss": 0.6357,
      "step": 7179
    },
    {
      "epoch": 1.2456627342123525,
      "grad_norm": 0.916454553604126,
      "learning_rate": 1.8210509820305174e-05,
      "loss": 0.5184,
      "step": 7180
    },
    {
      "epoch": 1.2458362248438584,
      "grad_norm": 0.7400712370872498,
      "learning_rate": 1.8203723994801974e-05,
      "loss": 0.5072,
      "step": 7181
    },
    {
      "epoch": 1.2460097154753642,
      "grad_norm": 0.8262057304382324,
      "learning_rate": 1.8196938377758696e-05,
      "loss": 0.6155,
      "step": 7182
    },
    {
      "epoch": 1.2461832061068703,
      "grad_norm": 0.7413461208343506,
      "learning_rate": 1.819015296996282e-05,
      "loss": 0.5012,
      "step": 7183
    },
    {
      "epoch": 1.2463566967383761,
      "grad_norm": 0.7607156038284302,
      "learning_rate": 1.818336777220181e-05,
      "loss": 0.4351,
      "step": 7184
    },
    {
      "epoch": 1.246530187369882,
      "grad_norm": 1.2906935214996338,
      "learning_rate": 1.817658278526308e-05,
      "loss": 0.4164,
      "step": 7185
    },
    {
      "epoch": 1.246703678001388,
      "grad_norm": 0.6653996109962463,
      "learning_rate": 1.8169798009934038e-05,
      "loss": 0.5944,
      "step": 7186
    },
    {
      "epoch": 1.2468771686328939,
      "grad_norm": 0.8081101179122925,
      "learning_rate": 1.816301344700208e-05,
      "loss": 0.5083,
      "step": 7187
    },
    {
      "epoch": 1.2470506592643997,
      "grad_norm": 0.8156636953353882,
      "learning_rate": 1.815622909725454e-05,
      "loss": 0.5066,
      "step": 7188
    },
    {
      "epoch": 1.2472241498959056,
      "grad_norm": 0.6418240666389465,
      "learning_rate": 1.8149444961478767e-05,
      "loss": 0.5635,
      "step": 7189
    },
    {
      "epoch": 1.2473976405274114,
      "grad_norm": 0.6648577451705933,
      "learning_rate": 1.8142661040462068e-05,
      "loss": 0.4795,
      "step": 7190
    },
    {
      "epoch": 1.2475711311589175,
      "grad_norm": 1.0434131622314453,
      "learning_rate": 1.8135877334991713e-05,
      "loss": 0.5547,
      "step": 7191
    },
    {
      "epoch": 1.2477446217904233,
      "grad_norm": 0.9893107414245605,
      "learning_rate": 1.8129093845854965e-05,
      "loss": 0.3964,
      "step": 7192
    },
    {
      "epoch": 1.2479181124219292,
      "grad_norm": 1.225264310836792,
      "learning_rate": 1.8122310573839063e-05,
      "loss": 0.3656,
      "step": 7193
    },
    {
      "epoch": 1.248091603053435,
      "grad_norm": 1.0399143695831299,
      "learning_rate": 1.81155275197312e-05,
      "loss": 0.5874,
      "step": 7194
    },
    {
      "epoch": 1.248265093684941,
      "grad_norm": 0.6252385377883911,
      "learning_rate": 1.8108744684318564e-05,
      "loss": 0.6443,
      "step": 7195
    },
    {
      "epoch": 1.248438584316447,
      "grad_norm": 0.6317516565322876,
      "learning_rate": 1.8101962068388315e-05,
      "loss": 0.6039,
      "step": 7196
    },
    {
      "epoch": 1.2486120749479528,
      "grad_norm": 0.7467001676559448,
      "learning_rate": 1.8095179672727575e-05,
      "loss": 0.524,
      "step": 7197
    },
    {
      "epoch": 1.2487855655794586,
      "grad_norm": 0.682138979434967,
      "learning_rate": 1.808839749812346e-05,
      "loss": 0.4747,
      "step": 7198
    },
    {
      "epoch": 1.2489590562109647,
      "grad_norm": 1.088470458984375,
      "learning_rate": 1.8081615545363035e-05,
      "loss": 0.4355,
      "step": 7199
    },
    {
      "epoch": 1.2491325468424705,
      "grad_norm": 0.7856414318084717,
      "learning_rate": 1.807483381523337e-05,
      "loss": 0.441,
      "step": 7200
    },
    {
      "epoch": 1.2493060374739764,
      "grad_norm": 0.794278621673584,
      "learning_rate": 1.8068052308521474e-05,
      "loss": 0.5021,
      "step": 7201
    },
    {
      "epoch": 1.2494795281054822,
      "grad_norm": 0.9811384081840515,
      "learning_rate": 1.8061271026014364e-05,
      "loss": 0.3582,
      "step": 7202
    },
    {
      "epoch": 1.2496530187369883,
      "grad_norm": 0.8610075116157532,
      "learning_rate": 1.8054489968499018e-05,
      "loss": 0.4745,
      "step": 7203
    },
    {
      "epoch": 1.2498265093684942,
      "grad_norm": 0.6773804426193237,
      "learning_rate": 1.8047709136762368e-05,
      "loss": 0.5815,
      "step": 7204
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9444378018379211,
      "learning_rate": 1.8040928531591347e-05,
      "loss": 0.4127,
      "step": 7205
    },
    {
      "epoch": 1.2501734906315058,
      "grad_norm": 0.7318785786628723,
      "learning_rate": 1.8034148153772864e-05,
      "loss": 0.5726,
      "step": 7206
    },
    {
      "epoch": 1.2503469812630117,
      "grad_norm": 0.863391637802124,
      "learning_rate": 1.8027368004093766e-05,
      "loss": 0.4684,
      "step": 7207
    },
    {
      "epoch": 1.2505204718945178,
      "grad_norm": 0.8419562578201294,
      "learning_rate": 1.8020588083340912e-05,
      "loss": 0.4878,
      "step": 7208
    },
    {
      "epoch": 1.2506939625260236,
      "grad_norm": 0.8791180849075317,
      "learning_rate": 1.801380839230113e-05,
      "loss": 0.532,
      "step": 7209
    },
    {
      "epoch": 1.2508674531575295,
      "grad_norm": 0.8953050971031189,
      "learning_rate": 1.8007028931761184e-05,
      "loss": 0.4703,
      "step": 7210
    },
    {
      "epoch": 1.2510409437890355,
      "grad_norm": 1.189442753791809,
      "learning_rate": 1.8000249702507854e-05,
      "loss": 0.4456,
      "step": 7211
    },
    {
      "epoch": 1.2512144344205414,
      "grad_norm": 1.0011205673217773,
      "learning_rate": 1.7993470705327877e-05,
      "loss": 0.495,
      "step": 7212
    },
    {
      "epoch": 1.2513879250520472,
      "grad_norm": 0.903160035610199,
      "learning_rate": 1.798669194100797e-05,
      "loss": 0.4813,
      "step": 7213
    },
    {
      "epoch": 1.251561415683553,
      "grad_norm": 0.7249449491500854,
      "learning_rate": 1.79799134103348e-05,
      "loss": 0.4936,
      "step": 7214
    },
    {
      "epoch": 1.251734906315059,
      "grad_norm": 2.5167765617370605,
      "learning_rate": 1.7973135114095038e-05,
      "loss": 0.5417,
      "step": 7215
    },
    {
      "epoch": 1.2519083969465647,
      "grad_norm": 0.715190589427948,
      "learning_rate": 1.7966357053075312e-05,
      "loss": 0.6365,
      "step": 7216
    },
    {
      "epoch": 1.2520818875780708,
      "grad_norm": 0.7164963483810425,
      "learning_rate": 1.795957922806221e-05,
      "loss": 0.6732,
      "step": 7217
    },
    {
      "epoch": 1.2522553782095767,
      "grad_norm": 1.0228049755096436,
      "learning_rate": 1.795280163984232e-05,
      "loss": 0.5239,
      "step": 7218
    },
    {
      "epoch": 1.2524288688410825,
      "grad_norm": 0.6803464889526367,
      "learning_rate": 1.7946024289202188e-05,
      "loss": 0.4541,
      "step": 7219
    },
    {
      "epoch": 1.2526023594725886,
      "grad_norm": 0.797497034072876,
      "learning_rate": 1.7939247176928328e-05,
      "loss": 0.4326,
      "step": 7220
    },
    {
      "epoch": 1.2527758501040944,
      "grad_norm": 0.9367165565490723,
      "learning_rate": 1.793247030380723e-05,
      "loss": 0.6234,
      "step": 7221
    },
    {
      "epoch": 1.2529493407356003,
      "grad_norm": 0.7852381467819214,
      "learning_rate": 1.792569367062537e-05,
      "loss": 0.5708,
      "step": 7222
    },
    {
      "epoch": 1.2531228313671061,
      "grad_norm": 0.7588016390800476,
      "learning_rate": 1.7918917278169173e-05,
      "loss": 0.4335,
      "step": 7223
    },
    {
      "epoch": 1.253296321998612,
      "grad_norm": 0.87742680311203,
      "learning_rate": 1.791214112722505e-05,
      "loss": 0.5884,
      "step": 7224
    },
    {
      "epoch": 1.253469812630118,
      "grad_norm": 1.2502604722976685,
      "learning_rate": 1.7905365218579387e-05,
      "loss": 0.5402,
      "step": 7225
    },
    {
      "epoch": 1.2536433032616239,
      "grad_norm": 1.3978842496871948,
      "learning_rate": 1.7898589553018523e-05,
      "loss": 0.5352,
      "step": 7226
    },
    {
      "epoch": 1.2538167938931297,
      "grad_norm": 0.7687990069389343,
      "learning_rate": 1.7891814131328795e-05,
      "loss": 0.3994,
      "step": 7227
    },
    {
      "epoch": 1.2539902845246358,
      "grad_norm": 1.105607032775879,
      "learning_rate": 1.788503895429649e-05,
      "loss": 0.4652,
      "step": 7228
    },
    {
      "epoch": 1.2541637751561416,
      "grad_norm": 0.8369463682174683,
      "learning_rate": 1.787826402270789e-05,
      "loss": 0.4209,
      "step": 7229
    },
    {
      "epoch": 1.2543372657876475,
      "grad_norm": 0.6711232662200928,
      "learning_rate": 1.7871489337349208e-05,
      "loss": 0.4785,
      "step": 7230
    },
    {
      "epoch": 1.2545107564191533,
      "grad_norm": 10.019854545593262,
      "learning_rate": 1.7864714899006672e-05,
      "loss": 0.5475,
      "step": 7231
    },
    {
      "epoch": 1.2546842470506592,
      "grad_norm": 0.9782271385192871,
      "learning_rate": 1.785794070846647e-05,
      "loss": 0.4442,
      "step": 7232
    },
    {
      "epoch": 1.2548577376821652,
      "grad_norm": 0.7232918739318848,
      "learning_rate": 1.7851166766514737e-05,
      "loss": 0.5372,
      "step": 7233
    },
    {
      "epoch": 1.255031228313671,
      "grad_norm": 0.8433080315589905,
      "learning_rate": 1.78443930739376e-05,
      "loss": 0.386,
      "step": 7234
    },
    {
      "epoch": 1.255204718945177,
      "grad_norm": 0.7202969193458557,
      "learning_rate": 1.783761963152117e-05,
      "loss": 0.4719,
      "step": 7235
    },
    {
      "epoch": 1.2553782095766828,
      "grad_norm": 1.5871689319610596,
      "learning_rate": 1.7830846440051493e-05,
      "loss": 0.5349,
      "step": 7236
    },
    {
      "epoch": 1.2555517002081888,
      "grad_norm": 0.8011099100112915,
      "learning_rate": 1.7824073500314614e-05,
      "loss": 0.5406,
      "step": 7237
    },
    {
      "epoch": 1.2557251908396947,
      "grad_norm": 0.8340924978256226,
      "learning_rate": 1.7817300813096548e-05,
      "loss": 0.4517,
      "step": 7238
    },
    {
      "epoch": 1.2558986814712005,
      "grad_norm": 0.8279672861099243,
      "learning_rate": 1.7810528379183262e-05,
      "loss": 0.4304,
      "step": 7239
    },
    {
      "epoch": 1.2560721721027064,
      "grad_norm": 0.7791277170181274,
      "learning_rate": 1.7803756199360704e-05,
      "loss": 0.4708,
      "step": 7240
    },
    {
      "epoch": 1.2562456627342122,
      "grad_norm": 1.3934314250946045,
      "learning_rate": 1.7796984274414797e-05,
      "loss": 0.5194,
      "step": 7241
    },
    {
      "epoch": 1.2564191533657183,
      "grad_norm": 0.8346148729324341,
      "learning_rate": 1.7790212605131448e-05,
      "loss": 0.4475,
      "step": 7242
    },
    {
      "epoch": 1.2565926439972241,
      "grad_norm": 0.8930989503860474,
      "learning_rate": 1.7783441192296488e-05,
      "loss": 0.4553,
      "step": 7243
    },
    {
      "epoch": 1.25676613462873,
      "grad_norm": 0.7225217819213867,
      "learning_rate": 1.7776670036695758e-05,
      "loss": 0.4896,
      "step": 7244
    },
    {
      "epoch": 1.256939625260236,
      "grad_norm": 0.9962308406829834,
      "learning_rate": 1.7769899139115066e-05,
      "loss": 0.52,
      "step": 7245
    },
    {
      "epoch": 1.257113115891742,
      "grad_norm": 1.6759709119796753,
      "learning_rate": 1.776312850034018e-05,
      "loss": 0.5365,
      "step": 7246
    },
    {
      "epoch": 1.2572866065232478,
      "grad_norm": 1.0960193872451782,
      "learning_rate": 1.7756358121156835e-05,
      "loss": 0.48,
      "step": 7247
    },
    {
      "epoch": 1.2574600971547536,
      "grad_norm": 0.6902223229408264,
      "learning_rate": 1.7749588002350748e-05,
      "loss": 0.5099,
      "step": 7248
    },
    {
      "epoch": 1.2576335877862594,
      "grad_norm": 1.0590311288833618,
      "learning_rate": 1.7742818144707588e-05,
      "loss": 0.3909,
      "step": 7249
    },
    {
      "epoch": 1.2578070784177655,
      "grad_norm": 0.7316137552261353,
      "learning_rate": 1.7736048549013013e-05,
      "loss": 0.463,
      "step": 7250
    },
    {
      "epoch": 1.2579805690492714,
      "grad_norm": 0.8568545579910278,
      "learning_rate": 1.7729279216052652e-05,
      "loss": 0.5719,
      "step": 7251
    },
    {
      "epoch": 1.2581540596807772,
      "grad_norm": 0.799411416053772,
      "learning_rate": 1.7722510146612075e-05,
      "loss": 0.4921,
      "step": 7252
    },
    {
      "epoch": 1.2583275503122833,
      "grad_norm": 0.6816998720169067,
      "learning_rate": 1.771574134147685e-05,
      "loss": 0.5752,
      "step": 7253
    },
    {
      "epoch": 1.2585010409437891,
      "grad_norm": 0.9275938272476196,
      "learning_rate": 1.770897280143251e-05,
      "loss": 0.4592,
      "step": 7254
    },
    {
      "epoch": 1.258674531575295,
      "grad_norm": 0.764086902141571,
      "learning_rate": 1.770220452726454e-05,
      "loss": 0.4797,
      "step": 7255
    },
    {
      "epoch": 1.2588480222068008,
      "grad_norm": 0.8848465085029602,
      "learning_rate": 1.7695436519758412e-05,
      "loss": 0.5416,
      "step": 7256
    },
    {
      "epoch": 1.2590215128383067,
      "grad_norm": 0.7068691849708557,
      "learning_rate": 1.7688668779699562e-05,
      "loss": 0.4258,
      "step": 7257
    },
    {
      "epoch": 1.2591950034698125,
      "grad_norm": 2.146357297897339,
      "learning_rate": 1.76819013078734e-05,
      "loss": 0.3964,
      "step": 7258
    },
    {
      "epoch": 1.2593684941013186,
      "grad_norm": 0.7820359468460083,
      "learning_rate": 1.767513410506528e-05,
      "loss": 0.4858,
      "step": 7259
    },
    {
      "epoch": 1.2595419847328244,
      "grad_norm": 1.0822325944900513,
      "learning_rate": 1.7668367172060562e-05,
      "loss": 0.4261,
      "step": 7260
    },
    {
      "epoch": 1.2597154753643303,
      "grad_norm": 0.7217856049537659,
      "learning_rate": 1.766160050964456e-05,
      "loss": 0.5461,
      "step": 7261
    },
    {
      "epoch": 1.2598889659958363,
      "grad_norm": 2.3959295749664307,
      "learning_rate": 1.765483411860253e-05,
      "loss": 0.574,
      "step": 7262
    },
    {
      "epoch": 1.2600624566273422,
      "grad_norm": 0.7436891794204712,
      "learning_rate": 1.7648067999719734e-05,
      "loss": 0.4056,
      "step": 7263
    },
    {
      "epoch": 1.260235947258848,
      "grad_norm": 0.707612931728363,
      "learning_rate": 1.7641302153781402e-05,
      "loss": 0.374,
      "step": 7264
    },
    {
      "epoch": 1.2604094378903539,
      "grad_norm": 0.7460191249847412,
      "learning_rate": 1.7634536581572687e-05,
      "loss": 0.4739,
      "step": 7265
    },
    {
      "epoch": 1.2605829285218597,
      "grad_norm": 3.197263479232788,
      "learning_rate": 1.7627771283878764e-05,
      "loss": 0.4283,
      "step": 7266
    },
    {
      "epoch": 1.2607564191533658,
      "grad_norm": 0.7251825928688049,
      "learning_rate": 1.762100626148475e-05,
      "loss": 0.5815,
      "step": 7267
    },
    {
      "epoch": 1.2609299097848716,
      "grad_norm": 0.7742326855659485,
      "learning_rate": 1.761424151517573e-05,
      "loss": 0.4871,
      "step": 7268
    },
    {
      "epoch": 1.2611034004163775,
      "grad_norm": 0.7075135707855225,
      "learning_rate": 1.7607477045736758e-05,
      "loss": 0.4977,
      "step": 7269
    },
    {
      "epoch": 1.2612768910478835,
      "grad_norm": 1.1125656366348267,
      "learning_rate": 1.7600712853952863e-05,
      "loss": 0.4691,
      "step": 7270
    },
    {
      "epoch": 1.2614503816793894,
      "grad_norm": 0.8418900966644287,
      "learning_rate": 1.7593948940609043e-05,
      "loss": 0.5305,
      "step": 7271
    },
    {
      "epoch": 1.2616238723108952,
      "grad_norm": 0.6288782954216003,
      "learning_rate": 1.7587185306490245e-05,
      "loss": 0.4772,
      "step": 7272
    },
    {
      "epoch": 1.261797362942401,
      "grad_norm": 0.7079287171363831,
      "learning_rate": 1.758042195238141e-05,
      "loss": 0.5668,
      "step": 7273
    },
    {
      "epoch": 1.261970853573907,
      "grad_norm": 0.6741266846656799,
      "learning_rate": 1.7573658879067424e-05,
      "loss": 0.4419,
      "step": 7274
    },
    {
      "epoch": 1.2621443442054128,
      "grad_norm": 0.6497853398323059,
      "learning_rate": 1.756689608733315e-05,
      "loss": 0.6057,
      "step": 7275
    },
    {
      "epoch": 1.2623178348369188,
      "grad_norm": 0.6376200318336487,
      "learning_rate": 1.7560133577963423e-05,
      "loss": 0.5757,
      "step": 7276
    },
    {
      "epoch": 1.2624913254684247,
      "grad_norm": 1.0629897117614746,
      "learning_rate": 1.7553371351743043e-05,
      "loss": 0.4725,
      "step": 7277
    },
    {
      "epoch": 1.2626648160999305,
      "grad_norm": 0.839389979839325,
      "learning_rate": 1.754660940945676e-05,
      "loss": 0.4766,
      "step": 7278
    },
    {
      "epoch": 1.2628383067314366,
      "grad_norm": 0.9092323780059814,
      "learning_rate": 1.7539847751889314e-05,
      "loss": 0.5048,
      "step": 7279
    },
    {
      "epoch": 1.2630117973629424,
      "grad_norm": 0.790918231010437,
      "learning_rate": 1.753308637982541e-05,
      "loss": 0.3907,
      "step": 7280
    },
    {
      "epoch": 1.2631852879944483,
      "grad_norm": 1.2072266340255737,
      "learning_rate": 1.75263252940497e-05,
      "loss": 0.3888,
      "step": 7281
    },
    {
      "epoch": 1.2633587786259541,
      "grad_norm": 0.7060717344284058,
      "learning_rate": 1.751956449534682e-05,
      "loss": 0.4888,
      "step": 7282
    },
    {
      "epoch": 1.26353226925746,
      "grad_norm": 0.9373074173927307,
      "learning_rate": 1.7512803984501385e-05,
      "loss": 0.4214,
      "step": 7283
    },
    {
      "epoch": 1.263705759888966,
      "grad_norm": 0.7018997669219971,
      "learning_rate": 1.7506043762297932e-05,
      "loss": 0.4296,
      "step": 7284
    },
    {
      "epoch": 1.263879250520472,
      "grad_norm": 1.4823564291000366,
      "learning_rate": 1.7499283829521006e-05,
      "loss": 0.4584,
      "step": 7285
    },
    {
      "epoch": 1.2640527411519777,
      "grad_norm": 0.8001006245613098,
      "learning_rate": 1.7492524186955108e-05,
      "loss": 0.5175,
      "step": 7286
    },
    {
      "epoch": 1.2642262317834838,
      "grad_norm": 0.9118646383285522,
      "learning_rate": 1.7485764835384705e-05,
      "loss": 0.4099,
      "step": 7287
    },
    {
      "epoch": 1.2643997224149897,
      "grad_norm": 1.0971133708953857,
      "learning_rate": 1.7479005775594216e-05,
      "loss": 0.4161,
      "step": 7288
    },
    {
      "epoch": 1.2645732130464955,
      "grad_norm": 0.7593318819999695,
      "learning_rate": 1.7472247008368046e-05,
      "loss": 0.4926,
      "step": 7289
    },
    {
      "epoch": 1.2647467036780013,
      "grad_norm": 0.7047328352928162,
      "learning_rate": 1.746548853449056e-05,
      "loss": 0.506,
      "step": 7290
    },
    {
      "epoch": 1.2649201943095072,
      "grad_norm": 0.8287725448608398,
      "learning_rate": 1.7458730354746077e-05,
      "loss": 0.4471,
      "step": 7291
    },
    {
      "epoch": 1.2650936849410133,
      "grad_norm": 0.8523426055908203,
      "learning_rate": 1.74519724699189e-05,
      "loss": 0.4932,
      "step": 7292
    },
    {
      "epoch": 1.265267175572519,
      "grad_norm": 1.0193105936050415,
      "learning_rate": 1.7445214880793287e-05,
      "loss": 0.624,
      "step": 7293
    },
    {
      "epoch": 1.265440666204025,
      "grad_norm": 0.7864932417869568,
      "learning_rate": 1.7438457588153466e-05,
      "loss": 0.4742,
      "step": 7294
    },
    {
      "epoch": 1.2656141568355308,
      "grad_norm": 0.7212458252906799,
      "learning_rate": 1.7431700592783622e-05,
      "loss": 0.6436,
      "step": 7295
    },
    {
      "epoch": 1.2657876474670369,
      "grad_norm": 0.7995871901512146,
      "learning_rate": 1.742494389546792e-05,
      "loss": 0.4543,
      "step": 7296
    },
    {
      "epoch": 1.2659611380985427,
      "grad_norm": 0.747856080532074,
      "learning_rate": 1.7418187496990476e-05,
      "loss": 0.4777,
      "step": 7297
    },
    {
      "epoch": 1.2661346287300486,
      "grad_norm": 0.7125510573387146,
      "learning_rate": 1.7411431398135384e-05,
      "loss": 0.5,
      "step": 7298
    },
    {
      "epoch": 1.2663081193615544,
      "grad_norm": 0.7325502634048462,
      "learning_rate": 1.7404675599686697e-05,
      "loss": 0.5443,
      "step": 7299
    },
    {
      "epoch": 1.2664816099930603,
      "grad_norm": 0.9522684812545776,
      "learning_rate": 1.739792010242843e-05,
      "loss": 0.4669,
      "step": 7300
    },
    {
      "epoch": 1.2666551006245663,
      "grad_norm": 0.9179607033729553,
      "learning_rate": 1.7391164907144563e-05,
      "loss": 0.5337,
      "step": 7301
    },
    {
      "epoch": 1.2668285912560722,
      "grad_norm": 1.179361343383789,
      "learning_rate": 1.738441001461905e-05,
      "loss": 0.4153,
      "step": 7302
    },
    {
      "epoch": 1.267002081887578,
      "grad_norm": 0.8450411558151245,
      "learning_rate": 1.7377655425635813e-05,
      "loss": 0.4528,
      "step": 7303
    },
    {
      "epoch": 1.267175572519084,
      "grad_norm": 0.9140102863311768,
      "learning_rate": 1.7370901140978706e-05,
      "loss": 0.399,
      "step": 7304
    },
    {
      "epoch": 1.26734906315059,
      "grad_norm": 0.9227039813995361,
      "learning_rate": 1.7364147161431585e-05,
      "loss": 0.3571,
      "step": 7305
    },
    {
      "epoch": 1.2675225537820958,
      "grad_norm": 0.9356099963188171,
      "learning_rate": 1.735739348777827e-05,
      "loss": 0.3888,
      "step": 7306
    },
    {
      "epoch": 1.2676960444136016,
      "grad_norm": 0.7274833917617798,
      "learning_rate": 1.735064012080251e-05,
      "loss": 0.4417,
      "step": 7307
    },
    {
      "epoch": 1.2678695350451075,
      "grad_norm": 0.8161334991455078,
      "learning_rate": 1.734388706128805e-05,
      "loss": 0.5983,
      "step": 7308
    },
    {
      "epoch": 1.2680430256766135,
      "grad_norm": 0.866718053817749,
      "learning_rate": 1.73371343100186e-05,
      "loss": 0.5098,
      "step": 7309
    },
    {
      "epoch": 1.2682165163081194,
      "grad_norm": 0.5527801513671875,
      "learning_rate": 1.7330381867777808e-05,
      "loss": 0.4913,
      "step": 7310
    },
    {
      "epoch": 1.2683900069396252,
      "grad_norm": 3.248582124710083,
      "learning_rate": 1.7323629735349313e-05,
      "loss": 0.5182,
      "step": 7311
    },
    {
      "epoch": 1.2685634975711313,
      "grad_norm": 0.7193784117698669,
      "learning_rate": 1.731687791351671e-05,
      "loss": 0.4734,
      "step": 7312
    },
    {
      "epoch": 1.2687369882026371,
      "grad_norm": 0.6383418440818787,
      "learning_rate": 1.7310126403063545e-05,
      "loss": 0.6099,
      "step": 7313
    },
    {
      "epoch": 1.268910478834143,
      "grad_norm": 0.6860360503196716,
      "learning_rate": 1.730337520477335e-05,
      "loss": 0.6025,
      "step": 7314
    },
    {
      "epoch": 1.2690839694656488,
      "grad_norm": 0.7680951356887817,
      "learning_rate": 1.7296624319429598e-05,
      "loss": 0.498,
      "step": 7315
    },
    {
      "epoch": 1.2692574600971547,
      "grad_norm": 0.7124847769737244,
      "learning_rate": 1.7289873747815755e-05,
      "loss": 0.6033,
      "step": 7316
    },
    {
      "epoch": 1.2694309507286605,
      "grad_norm": 0.7696164846420288,
      "learning_rate": 1.7283123490715213e-05,
      "loss": 0.535,
      "step": 7317
    },
    {
      "epoch": 1.2696044413601666,
      "grad_norm": 0.8060824871063232,
      "learning_rate": 1.7276373548911355e-05,
      "loss": 0.427,
      "step": 7318
    },
    {
      "epoch": 1.2697779319916724,
      "grad_norm": 0.8457629084587097,
      "learning_rate": 1.7269623923187522e-05,
      "loss": 0.5208,
      "step": 7319
    },
    {
      "epoch": 1.2699514226231783,
      "grad_norm": 0.7939236164093018,
      "learning_rate": 1.7262874614327016e-05,
      "loss": 0.426,
      "step": 7320
    },
    {
      "epoch": 1.2701249132546844,
      "grad_norm": 1.6990852355957031,
      "learning_rate": 1.72561256231131e-05,
      "loss": 0.4905,
      "step": 7321
    },
    {
      "epoch": 1.2702984038861902,
      "grad_norm": 0.6765450835227966,
      "learning_rate": 1.7249376950329004e-05,
      "loss": 0.5074,
      "step": 7322
    },
    {
      "epoch": 1.270471894517696,
      "grad_norm": 0.8751128315925598,
      "learning_rate": 1.7242628596757916e-05,
      "loss": 0.4418,
      "step": 7323
    },
    {
      "epoch": 1.270645385149202,
      "grad_norm": 0.7906344532966614,
      "learning_rate": 1.7235880563182988e-05,
      "loss": 0.5089,
      "step": 7324
    },
    {
      "epoch": 1.2708188757807077,
      "grad_norm": 0.604007363319397,
      "learning_rate": 1.7229132850387352e-05,
      "loss": 0.5376,
      "step": 7325
    },
    {
      "epoch": 1.2709923664122138,
      "grad_norm": 1.5954300165176392,
      "learning_rate": 1.7222385459154072e-05,
      "loss": 0.4464,
      "step": 7326
    },
    {
      "epoch": 1.2711658570437196,
      "grad_norm": 0.7143296003341675,
      "learning_rate": 1.721563839026619e-05,
      "loss": 0.5603,
      "step": 7327
    },
    {
      "epoch": 1.2713393476752255,
      "grad_norm": 0.9084748029708862,
      "learning_rate": 1.720889164450672e-05,
      "loss": 0.4924,
      "step": 7328
    },
    {
      "epoch": 1.2715128383067316,
      "grad_norm": 0.9519553184509277,
      "learning_rate": 1.720214522265864e-05,
      "loss": 0.4237,
      "step": 7329
    },
    {
      "epoch": 1.2716863289382374,
      "grad_norm": 0.8374545574188232,
      "learning_rate": 1.7195399125504853e-05,
      "loss": 0.4375,
      "step": 7330
    },
    {
      "epoch": 1.2718598195697433,
      "grad_norm": 1.1499824523925781,
      "learning_rate": 1.7188653353828266e-05,
      "loss": 0.4177,
      "step": 7331
    },
    {
      "epoch": 1.272033310201249,
      "grad_norm": 0.8614972829818726,
      "learning_rate": 1.7181907908411744e-05,
      "loss": 0.4067,
      "step": 7332
    },
    {
      "epoch": 1.272206800832755,
      "grad_norm": 0.6160062551498413,
      "learning_rate": 1.7175162790038086e-05,
      "loss": 0.4989,
      "step": 7333
    },
    {
      "epoch": 1.2723802914642608,
      "grad_norm": 0.630744993686676,
      "learning_rate": 1.716841799949007e-05,
      "loss": 0.5487,
      "step": 7334
    },
    {
      "epoch": 1.2725537820957669,
      "grad_norm": 2.016706943511963,
      "learning_rate": 1.716167353755046e-05,
      "loss": 0.4154,
      "step": 7335
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.9351750612258911,
      "learning_rate": 1.7154929405001936e-05,
      "loss": 0.5192,
      "step": 7336
    },
    {
      "epoch": 1.2729007633587786,
      "grad_norm": 0.7126862406730652,
      "learning_rate": 1.7148185602627166e-05,
      "loss": 0.5167,
      "step": 7337
    },
    {
      "epoch": 1.2730742539902846,
      "grad_norm": 1.0575381517410278,
      "learning_rate": 1.7141442131208788e-05,
      "loss": 0.4878,
      "step": 7338
    },
    {
      "epoch": 1.2732477446217905,
      "grad_norm": 0.7802735567092896,
      "learning_rate": 1.7134698991529373e-05,
      "loss": 0.5439,
      "step": 7339
    },
    {
      "epoch": 1.2734212352532963,
      "grad_norm": 0.680596649646759,
      "learning_rate": 1.712795618437148e-05,
      "loss": 0.479,
      "step": 7340
    },
    {
      "epoch": 1.2735947258848022,
      "grad_norm": 0.6597265601158142,
      "learning_rate": 1.7121213710517616e-05,
      "loss": 0.4742,
      "step": 7341
    },
    {
      "epoch": 1.273768216516308,
      "grad_norm": 0.8055812120437622,
      "learning_rate": 1.7114471570750266e-05,
      "loss": 0.5016,
      "step": 7342
    },
    {
      "epoch": 1.273941707147814,
      "grad_norm": 0.9792829751968384,
      "learning_rate": 1.7107729765851847e-05,
      "loss": 0.609,
      "step": 7343
    },
    {
      "epoch": 1.27411519777932,
      "grad_norm": 0.7375954985618591,
      "learning_rate": 1.7100988296604756e-05,
      "loss": 0.4707,
      "step": 7344
    },
    {
      "epoch": 1.2742886884108258,
      "grad_norm": 0.9349376559257507,
      "learning_rate": 1.7094247163791352e-05,
      "loss": 0.6333,
      "step": 7345
    },
    {
      "epoch": 1.2744621790423318,
      "grad_norm": 0.82657390832901,
      "learning_rate": 1.708750636819395e-05,
      "loss": 0.5251,
      "step": 7346
    },
    {
      "epoch": 1.2746356696738377,
      "grad_norm": 0.9187333583831787,
      "learning_rate": 1.7080765910594833e-05,
      "loss": 0.4808,
      "step": 7347
    },
    {
      "epoch": 1.2748091603053435,
      "grad_norm": 1.084909200668335,
      "learning_rate": 1.7074025791776232e-05,
      "loss": 0.4784,
      "step": 7348
    },
    {
      "epoch": 1.2749826509368494,
      "grad_norm": 0.8411982655525208,
      "learning_rate": 1.7067286012520343e-05,
      "loss": 0.4703,
      "step": 7349
    },
    {
      "epoch": 1.2751561415683552,
      "grad_norm": 0.5996063351631165,
      "learning_rate": 1.706054657360933e-05,
      "loss": 0.5082,
      "step": 7350
    },
    {
      "epoch": 1.2753296321998613,
      "grad_norm": 0.7579880952835083,
      "learning_rate": 1.705380747582532e-05,
      "loss": 0.4292,
      "step": 7351
    },
    {
      "epoch": 1.2755031228313671,
      "grad_norm": 1.000863790512085,
      "learning_rate": 1.704706871995038e-05,
      "loss": 0.4073,
      "step": 7352
    },
    {
      "epoch": 1.275676613462873,
      "grad_norm": 1.7665989398956299,
      "learning_rate": 1.704033030676655e-05,
      "loss": 0.4327,
      "step": 7353
    },
    {
      "epoch": 1.2758501040943788,
      "grad_norm": 0.6814614534378052,
      "learning_rate": 1.703359223705585e-05,
      "loss": 0.5709,
      "step": 7354
    },
    {
      "epoch": 1.276023594725885,
      "grad_norm": 0.7269407510757446,
      "learning_rate": 1.7026854511600218e-05,
      "loss": 0.5576,
      "step": 7355
    },
    {
      "epoch": 1.2761970853573907,
      "grad_norm": 0.7705087661743164,
      "learning_rate": 1.7020117131181585e-05,
      "loss": 0.3433,
      "step": 7356
    },
    {
      "epoch": 1.2763705759888966,
      "grad_norm": 0.8691286444664001,
      "learning_rate": 1.7013380096581828e-05,
      "loss": 0.4166,
      "step": 7357
    },
    {
      "epoch": 1.2765440666204024,
      "grad_norm": 0.867172360420227,
      "learning_rate": 1.70066434085828e-05,
      "loss": 0.463,
      "step": 7358
    },
    {
      "epoch": 1.2767175572519083,
      "grad_norm": 0.7234309315681458,
      "learning_rate": 1.6999907067966285e-05,
      "loss": 0.5438,
      "step": 7359
    },
    {
      "epoch": 1.2768910478834143,
      "grad_norm": 0.6743608117103577,
      "learning_rate": 1.6993171075514054e-05,
      "loss": 0.5529,
      "step": 7360
    },
    {
      "epoch": 1.2770645385149202,
      "grad_norm": 0.7440758347511292,
      "learning_rate": 1.6986435432007826e-05,
      "loss": 0.4937,
      "step": 7361
    },
    {
      "epoch": 1.277238029146426,
      "grad_norm": 1.2034293413162231,
      "learning_rate": 1.697970013822927e-05,
      "loss": 0.6184,
      "step": 7362
    },
    {
      "epoch": 1.277411519777932,
      "grad_norm": 0.7799433469772339,
      "learning_rate": 1.6972965194960034e-05,
      "loss": 0.4814,
      "step": 7363
    },
    {
      "epoch": 1.277585010409438,
      "grad_norm": 0.8648630976676941,
      "learning_rate": 1.6966230602981727e-05,
      "loss": 0.5325,
      "step": 7364
    },
    {
      "epoch": 1.2777585010409438,
      "grad_norm": 0.7705937027931213,
      "learning_rate": 1.695949636307588e-05,
      "loss": 0.4996,
      "step": 7365
    },
    {
      "epoch": 1.2779319916724496,
      "grad_norm": 0.7275580167770386,
      "learning_rate": 1.6952762476024023e-05,
      "loss": 0.4856,
      "step": 7366
    },
    {
      "epoch": 1.2781054823039555,
      "grad_norm": 0.7204875349998474,
      "learning_rate": 1.694602894260764e-05,
      "loss": 0.5568,
      "step": 7367
    },
    {
      "epoch": 1.2782789729354616,
      "grad_norm": 0.8418717980384827,
      "learning_rate": 1.6939295763608146e-05,
      "loss": 0.4301,
      "step": 7368
    },
    {
      "epoch": 1.2784524635669674,
      "grad_norm": 0.7505890727043152,
      "learning_rate": 1.6932562939806952e-05,
      "loss": 0.4567,
      "step": 7369
    },
    {
      "epoch": 1.2786259541984732,
      "grad_norm": 0.7095392346382141,
      "learning_rate": 1.6925830471985398e-05,
      "loss": 0.4727,
      "step": 7370
    },
    {
      "epoch": 1.2787994448299793,
      "grad_norm": 0.9720133543014526,
      "learning_rate": 1.6919098360924804e-05,
      "loss": 0.4852,
      "step": 7371
    },
    {
      "epoch": 1.2789729354614852,
      "grad_norm": 2.4071052074432373,
      "learning_rate": 1.6912366607406433e-05,
      "loss": 0.5085,
      "step": 7372
    },
    {
      "epoch": 1.279146426092991,
      "grad_norm": 0.7946075797080994,
      "learning_rate": 1.6905635212211517e-05,
      "loss": 0.5366,
      "step": 7373
    },
    {
      "epoch": 1.2793199167244969,
      "grad_norm": 0.9184132814407349,
      "learning_rate": 1.6898904176121246e-05,
      "loss": 0.4015,
      "step": 7374
    },
    {
      "epoch": 1.2794934073560027,
      "grad_norm": 0.76685631275177,
      "learning_rate": 1.6892173499916752e-05,
      "loss": 0.4385,
      "step": 7375
    },
    {
      "epoch": 1.2796668979875085,
      "grad_norm": 0.8007482290267944,
      "learning_rate": 1.688544318437914e-05,
      "loss": 0.3929,
      "step": 7376
    },
    {
      "epoch": 1.2798403886190146,
      "grad_norm": 0.9679758548736572,
      "learning_rate": 1.687871323028949e-05,
      "loss": 0.4316,
      "step": 7377
    },
    {
      "epoch": 1.2800138792505205,
      "grad_norm": 0.6715041399002075,
      "learning_rate": 1.6871983638428794e-05,
      "loss": 0.4387,
      "step": 7378
    },
    {
      "epoch": 1.2801873698820263,
      "grad_norm": 0.9745506048202515,
      "learning_rate": 1.6865254409578042e-05,
      "loss": 0.489,
      "step": 7379
    },
    {
      "epoch": 1.2803608605135324,
      "grad_norm": 0.8233616948127747,
      "learning_rate": 1.685852554451818e-05,
      "loss": 0.4985,
      "step": 7380
    },
    {
      "epoch": 1.2805343511450382,
      "grad_norm": 0.6738633513450623,
      "learning_rate": 1.6851797044030076e-05,
      "loss": 0.4419,
      "step": 7381
    },
    {
      "epoch": 1.280707841776544,
      "grad_norm": 0.6327497959136963,
      "learning_rate": 1.6845068908894597e-05,
      "loss": 0.4037,
      "step": 7382
    },
    {
      "epoch": 1.28088133240805,
      "grad_norm": 0.6840953826904297,
      "learning_rate": 1.6838341139892556e-05,
      "loss": 0.4702,
      "step": 7383
    },
    {
      "epoch": 1.2810548230395558,
      "grad_norm": 0.9774636626243591,
      "learning_rate": 1.68316137378047e-05,
      "loss": 0.4967,
      "step": 7384
    },
    {
      "epoch": 1.2812283136710618,
      "grad_norm": 0.9470757246017456,
      "learning_rate": 1.682488670341176e-05,
      "loss": 0.5508,
      "step": 7385
    },
    {
      "epoch": 1.2814018043025677,
      "grad_norm": 1.1846847534179688,
      "learning_rate": 1.681816003749442e-05,
      "loss": 0.4384,
      "step": 7386
    },
    {
      "epoch": 1.2815752949340735,
      "grad_norm": 0.8350231647491455,
      "learning_rate": 1.681143374083332e-05,
      "loss": 0.4119,
      "step": 7387
    },
    {
      "epoch": 1.2817487855655796,
      "grad_norm": 0.8339827060699463,
      "learning_rate": 1.6804707814209046e-05,
      "loss": 0.3962,
      "step": 7388
    },
    {
      "epoch": 1.2819222761970854,
      "grad_norm": 1.056058645248413,
      "learning_rate": 1.6797982258402154e-05,
      "loss": 0.5504,
      "step": 7389
    },
    {
      "epoch": 1.2820957668285913,
      "grad_norm": 0.9062198996543884,
      "learning_rate": 1.6791257074193156e-05,
      "loss": 0.5337,
      "step": 7390
    },
    {
      "epoch": 1.2822692574600971,
      "grad_norm": 1.0762733221054077,
      "learning_rate": 1.678453226236251e-05,
      "loss": 0.4216,
      "step": 7391
    },
    {
      "epoch": 1.282442748091603,
      "grad_norm": 1.0174580812454224,
      "learning_rate": 1.677780782369064e-05,
      "loss": 0.3956,
      "step": 7392
    },
    {
      "epoch": 1.2826162387231088,
      "grad_norm": 0.8656790256500244,
      "learning_rate": 1.677108375895793e-05,
      "loss": 0.5273,
      "step": 7393
    },
    {
      "epoch": 1.2827897293546149,
      "grad_norm": 1.3800997734069824,
      "learning_rate": 1.6764360068944706e-05,
      "loss": 0.3701,
      "step": 7394
    },
    {
      "epoch": 1.2829632199861207,
      "grad_norm": 0.702471911907196,
      "learning_rate": 1.6757636754431272e-05,
      "loss": 0.5071,
      "step": 7395
    },
    {
      "epoch": 1.2831367106176266,
      "grad_norm": 1.2209949493408203,
      "learning_rate": 1.6750913816197873e-05,
      "loss": 0.3948,
      "step": 7396
    },
    {
      "epoch": 1.2833102012491326,
      "grad_norm": 0.8057335019111633,
      "learning_rate": 1.6744191255024707e-05,
      "loss": 0.4244,
      "step": 7397
    },
    {
      "epoch": 1.2834836918806385,
      "grad_norm": 0.7313793897628784,
      "learning_rate": 1.6737469071691936e-05,
      "loss": 0.4855,
      "step": 7398
    },
    {
      "epoch": 1.2836571825121443,
      "grad_norm": 0.7577414512634277,
      "learning_rate": 1.6730747266979683e-05,
      "loss": 0.4327,
      "step": 7399
    },
    {
      "epoch": 1.2838306731436502,
      "grad_norm": 1.755483865737915,
      "learning_rate": 1.6724025841668026e-05,
      "loss": 0.4133,
      "step": 7400
    },
    {
      "epoch": 1.284004163775156,
      "grad_norm": 0.8913657665252686,
      "learning_rate": 1.6717304796536984e-05,
      "loss": 0.4237,
      "step": 7401
    },
    {
      "epoch": 1.284177654406662,
      "grad_norm": 0.9089097380638123,
      "learning_rate": 1.6710584132366542e-05,
      "loss": 0.4487,
      "step": 7402
    },
    {
      "epoch": 1.284351145038168,
      "grad_norm": 0.8932932615280151,
      "learning_rate": 1.6703863849936654e-05,
      "loss": 0.5096,
      "step": 7403
    },
    {
      "epoch": 1.2845246356696738,
      "grad_norm": 0.6808133721351624,
      "learning_rate": 1.6697143950027194e-05,
      "loss": 0.4312,
      "step": 7404
    },
    {
      "epoch": 1.2846981263011799,
      "grad_norm": 0.7399442791938782,
      "learning_rate": 1.6690424433418032e-05,
      "loss": 0.5237,
      "step": 7405
    },
    {
      "epoch": 1.2848716169326857,
      "grad_norm": 1.6645556688308716,
      "learning_rate": 1.6683705300888977e-05,
      "loss": 0.5393,
      "step": 7406
    },
    {
      "epoch": 1.2850451075641915,
      "grad_norm": 1.2402348518371582,
      "learning_rate": 1.6676986553219778e-05,
      "loss": 0.6115,
      "step": 7407
    },
    {
      "epoch": 1.2852185981956974,
      "grad_norm": 0.7507150173187256,
      "learning_rate": 1.667026819119016e-05,
      "loss": 0.5576,
      "step": 7408
    },
    {
      "epoch": 1.2853920888272032,
      "grad_norm": 0.9310218691825867,
      "learning_rate": 1.666355021557981e-05,
      "loss": 0.4143,
      "step": 7409
    },
    {
      "epoch": 1.2855655794587093,
      "grad_norm": 0.7509005665779114,
      "learning_rate": 1.6656832627168338e-05,
      "loss": 0.4741,
      "step": 7410
    },
    {
      "epoch": 1.2857390700902152,
      "grad_norm": 0.6133527755737305,
      "learning_rate": 1.665011542673533e-05,
      "loss": 0.509,
      "step": 7411
    },
    {
      "epoch": 1.285912560721721,
      "grad_norm": 0.9684708714485168,
      "learning_rate": 1.6643398615060346e-05,
      "loss": 0.5171,
      "step": 7412
    },
    {
      "epoch": 1.2860860513532268,
      "grad_norm": 0.9050800204277039,
      "learning_rate": 1.6636682192922847e-05,
      "loss": 0.6178,
      "step": 7413
    },
    {
      "epoch": 1.286259541984733,
      "grad_norm": 0.8545094728469849,
      "learning_rate": 1.6629966161102304e-05,
      "loss": 0.5242,
      "step": 7414
    },
    {
      "epoch": 1.2864330326162388,
      "grad_norm": 0.8504958152770996,
      "learning_rate": 1.6623250520378114e-05,
      "loss": 0.5021,
      "step": 7415
    },
    {
      "epoch": 1.2866065232477446,
      "grad_norm": 1.0944947004318237,
      "learning_rate": 1.661653527152964e-05,
      "loss": 0.5497,
      "step": 7416
    },
    {
      "epoch": 1.2867800138792505,
      "grad_norm": 0.7630634307861328,
      "learning_rate": 1.6609820415336188e-05,
      "loss": 0.5302,
      "step": 7417
    },
    {
      "epoch": 1.2869535045107563,
      "grad_norm": 1.0300921201705933,
      "learning_rate": 1.6603105952577024e-05,
      "loss": 0.3488,
      "step": 7418
    },
    {
      "epoch": 1.2871269951422624,
      "grad_norm": 0.7179005742073059,
      "learning_rate": 1.6596391884031378e-05,
      "loss": 0.4191,
      "step": 7419
    },
    {
      "epoch": 1.2873004857737682,
      "grad_norm": 0.7966880798339844,
      "learning_rate": 1.6589678210478415e-05,
      "loss": 0.5724,
      "step": 7420
    },
    {
      "epoch": 1.287473976405274,
      "grad_norm": 0.8220646381378174,
      "learning_rate": 1.658296493269727e-05,
      "loss": 0.6222,
      "step": 7421
    },
    {
      "epoch": 1.2876474670367801,
      "grad_norm": 1.581972599029541,
      "learning_rate": 1.657625205146703e-05,
      "loss": 0.4386,
      "step": 7422
    },
    {
      "epoch": 1.287820957668286,
      "grad_norm": 0.6322770118713379,
      "learning_rate": 1.6569539567566726e-05,
      "loss": 0.3914,
      "step": 7423
    },
    {
      "epoch": 1.2879944482997918,
      "grad_norm": 0.7599250078201294,
      "learning_rate": 1.6562827481775353e-05,
      "loss": 0.4332,
      "step": 7424
    },
    {
      "epoch": 1.2881679389312977,
      "grad_norm": 1.8352106809616089,
      "learning_rate": 1.6556115794871862e-05,
      "loss": 0.3678,
      "step": 7425
    },
    {
      "epoch": 1.2883414295628035,
      "grad_norm": 0.7707764506340027,
      "learning_rate": 1.6549404507635135e-05,
      "loss": 0.5446,
      "step": 7426
    },
    {
      "epoch": 1.2885149201943096,
      "grad_norm": 0.6282917857170105,
      "learning_rate": 1.654269362084404e-05,
      "loss": 0.485,
      "step": 7427
    },
    {
      "epoch": 1.2886884108258154,
      "grad_norm": 0.9740319848060608,
      "learning_rate": 1.6535983135277378e-05,
      "loss": 0.4186,
      "step": 7428
    },
    {
      "epoch": 1.2888619014573213,
      "grad_norm": 0.7551928758621216,
      "learning_rate": 1.6529273051713917e-05,
      "loss": 0.4285,
      "step": 7429
    },
    {
      "epoch": 1.2890353920888273,
      "grad_norm": 0.8909952044487,
      "learning_rate": 1.6522563370932355e-05,
      "loss": 0.4718,
      "step": 7430
    },
    {
      "epoch": 1.2892088827203332,
      "grad_norm": 0.8301438689231873,
      "learning_rate": 1.6515854093711364e-05,
      "loss": 0.3869,
      "step": 7431
    },
    {
      "epoch": 1.289382373351839,
      "grad_norm": 1.0200198888778687,
      "learning_rate": 1.6509145220829574e-05,
      "loss": 0.4265,
      "step": 7432
    },
    {
      "epoch": 1.2895558639833449,
      "grad_norm": 0.7204471230506897,
      "learning_rate": 1.650243675306554e-05,
      "loss": 0.4183,
      "step": 7433
    },
    {
      "epoch": 1.2897293546148507,
      "grad_norm": 0.8902462124824524,
      "learning_rate": 1.64957286911978e-05,
      "loss": 0.4216,
      "step": 7434
    },
    {
      "epoch": 1.2899028452463566,
      "grad_norm": 0.7522461414337158,
      "learning_rate": 1.6489021036004835e-05,
      "loss": 0.5361,
      "step": 7435
    },
    {
      "epoch": 1.2900763358778626,
      "grad_norm": 0.9403187036514282,
      "learning_rate": 1.6482313788265058e-05,
      "loss": 0.473,
      "step": 7436
    },
    {
      "epoch": 1.2902498265093685,
      "grad_norm": 0.6944003701210022,
      "learning_rate": 1.647560694875687e-05,
      "loss": 0.4741,
      "step": 7437
    },
    {
      "epoch": 1.2904233171408743,
      "grad_norm": 0.7476251721382141,
      "learning_rate": 1.64689005182586e-05,
      "loss": 0.5466,
      "step": 7438
    },
    {
      "epoch": 1.2905968077723804,
      "grad_norm": 0.6289733648300171,
      "learning_rate": 1.6462194497548546e-05,
      "loss": 0.5042,
      "step": 7439
    },
    {
      "epoch": 1.2907702984038862,
      "grad_norm": 0.8400498628616333,
      "learning_rate": 1.6455488887404935e-05,
      "loss": 0.4337,
      "step": 7440
    },
    {
      "epoch": 1.290943789035392,
      "grad_norm": 0.7845512628555298,
      "learning_rate": 1.6448783688605976e-05,
      "loss": 0.4918,
      "step": 7441
    },
    {
      "epoch": 1.291117279666898,
      "grad_norm": 0.6369579434394836,
      "learning_rate": 1.6442078901929803e-05,
      "loss": 0.5485,
      "step": 7442
    },
    {
      "epoch": 1.2912907702984038,
      "grad_norm": 0.7335418462753296,
      "learning_rate": 1.6435374528154517e-05,
      "loss": 0.5928,
      "step": 7443
    },
    {
      "epoch": 1.2914642609299098,
      "grad_norm": 1.2369920015335083,
      "learning_rate": 1.6428670568058176e-05,
      "loss": 0.5278,
      "step": 7444
    },
    {
      "epoch": 1.2916377515614157,
      "grad_norm": 0.6675391793251038,
      "learning_rate": 1.6421967022418776e-05,
      "loss": 0.5138,
      "step": 7445
    },
    {
      "epoch": 1.2918112421929215,
      "grad_norm": 0.626063346862793,
      "learning_rate": 1.641526389201427e-05,
      "loss": 0.6233,
      "step": 7446
    },
    {
      "epoch": 1.2919847328244276,
      "grad_norm": 0.8406535387039185,
      "learning_rate": 1.6408561177622566e-05,
      "loss": 0.5139,
      "step": 7447
    },
    {
      "epoch": 1.2921582234559335,
      "grad_norm": 0.9377653002738953,
      "learning_rate": 1.640185888002153e-05,
      "loss": 0.5919,
      "step": 7448
    },
    {
      "epoch": 1.2923317140874393,
      "grad_norm": 0.6694398522377014,
      "learning_rate": 1.6395156999988956e-05,
      "loss": 0.4327,
      "step": 7449
    },
    {
      "epoch": 1.2925052047189451,
      "grad_norm": 0.6803465485572815,
      "learning_rate": 1.6388455538302612e-05,
      "loss": 0.4812,
      "step": 7450
    },
    {
      "epoch": 1.292678695350451,
      "grad_norm": 0.8606630563735962,
      "learning_rate": 1.638175449574022e-05,
      "loss": 0.4614,
      "step": 7451
    },
    {
      "epoch": 1.2928521859819568,
      "grad_norm": 0.9769142866134644,
      "learning_rate": 1.6375053873079424e-05,
      "loss": 0.4155,
      "step": 7452
    },
    {
      "epoch": 1.293025676613463,
      "grad_norm": 0.761888325214386,
      "learning_rate": 1.6368353671097854e-05,
      "loss": 0.5502,
      "step": 7453
    },
    {
      "epoch": 1.2931991672449688,
      "grad_norm": 0.7804129719734192,
      "learning_rate": 1.6361653890573078e-05,
      "loss": 0.5378,
      "step": 7454
    },
    {
      "epoch": 1.2933726578764746,
      "grad_norm": 0.7592903971672058,
      "learning_rate": 1.6354954532282598e-05,
      "loss": 0.4705,
      "step": 7455
    },
    {
      "epoch": 1.2935461485079807,
      "grad_norm": 0.7587009072303772,
      "learning_rate": 1.6348255597003896e-05,
      "loss": 0.4542,
      "step": 7456
    },
    {
      "epoch": 1.2937196391394865,
      "grad_norm": 0.7308903932571411,
      "learning_rate": 1.6341557085514385e-05,
      "loss": 0.6405,
      "step": 7457
    },
    {
      "epoch": 1.2938931297709924,
      "grad_norm": 0.7195034623146057,
      "learning_rate": 1.633485899859144e-05,
      "loss": 0.4769,
      "step": 7458
    },
    {
      "epoch": 1.2940666204024982,
      "grad_norm": 0.7280550599098206,
      "learning_rate": 1.6328161337012377e-05,
      "loss": 0.4736,
      "step": 7459
    },
    {
      "epoch": 1.294240111034004,
      "grad_norm": 0.9579290747642517,
      "learning_rate": 1.632146410155447e-05,
      "loss": 0.4453,
      "step": 7460
    },
    {
      "epoch": 1.2944136016655101,
      "grad_norm": 0.8503251671791077,
      "learning_rate": 1.6314767292994946e-05,
      "loss": 0.3949,
      "step": 7461
    },
    {
      "epoch": 1.294587092297016,
      "grad_norm": 1.023876428604126,
      "learning_rate": 1.6308070912110965e-05,
      "loss": 0.3865,
      "step": 7462
    },
    {
      "epoch": 1.2947605829285218,
      "grad_norm": 0.7069829106330872,
      "learning_rate": 1.6301374959679654e-05,
      "loss": 0.3804,
      "step": 7463
    },
    {
      "epoch": 1.2949340735600279,
      "grad_norm": 0.9230320453643799,
      "learning_rate": 1.6294679436478095e-05,
      "loss": 0.4145,
      "step": 7464
    },
    {
      "epoch": 1.2951075641915337,
      "grad_norm": 0.9587447047233582,
      "learning_rate": 1.6287984343283304e-05,
      "loss": 0.5261,
      "step": 7465
    },
    {
      "epoch": 1.2952810548230396,
      "grad_norm": 0.9868846535682678,
      "learning_rate": 1.6281289680872252e-05,
      "loss": 0.4058,
      "step": 7466
    },
    {
      "epoch": 1.2954545454545454,
      "grad_norm": 0.6967030167579651,
      "learning_rate": 1.627459545002187e-05,
      "loss": 0.4697,
      "step": 7467
    },
    {
      "epoch": 1.2956280360860513,
      "grad_norm": 1.0543104410171509,
      "learning_rate": 1.6267901651509022e-05,
      "loss": 0.4632,
      "step": 7468
    },
    {
      "epoch": 1.2958015267175573,
      "grad_norm": 0.6506008505821228,
      "learning_rate": 1.6261208286110536e-05,
      "loss": 0.5687,
      "step": 7469
    },
    {
      "epoch": 1.2959750173490632,
      "grad_norm": 0.7393031120300293,
      "learning_rate": 1.6254515354603194e-05,
      "loss": 0.4377,
      "step": 7470
    },
    {
      "epoch": 1.296148507980569,
      "grad_norm": 0.7607579827308655,
      "learning_rate": 1.6247822857763703e-05,
      "loss": 0.5089,
      "step": 7471
    },
    {
      "epoch": 1.2963219986120749,
      "grad_norm": 1.530008316040039,
      "learning_rate": 1.6241130796368737e-05,
      "loss": 0.5399,
      "step": 7472
    },
    {
      "epoch": 1.296495489243581,
      "grad_norm": 1.0021963119506836,
      "learning_rate": 1.6234439171194925e-05,
      "loss": 0.4792,
      "step": 7473
    },
    {
      "epoch": 1.2966689798750868,
      "grad_norm": 0.7233889698982239,
      "learning_rate": 1.6227747983018845e-05,
      "loss": 0.5323,
      "step": 7474
    },
    {
      "epoch": 1.2968424705065926,
      "grad_norm": 1.0551574230194092,
      "learning_rate": 1.6221057232616994e-05,
      "loss": 0.5114,
      "step": 7475
    },
    {
      "epoch": 1.2970159611380985,
      "grad_norm": 0.6602882742881775,
      "learning_rate": 1.6214366920765856e-05,
      "loss": 0.4359,
      "step": 7476
    },
    {
      "epoch": 1.2971894517696043,
      "grad_norm": 0.6236412525177002,
      "learning_rate": 1.6207677048241858e-05,
      "loss": 0.6041,
      "step": 7477
    },
    {
      "epoch": 1.2973629424011104,
      "grad_norm": 0.8288519382476807,
      "learning_rate": 1.620098761582135e-05,
      "loss": 0.4841,
      "step": 7478
    },
    {
      "epoch": 1.2975364330326162,
      "grad_norm": 1.2267411947250366,
      "learning_rate": 1.6194298624280653e-05,
      "loss": 0.6281,
      "step": 7479
    },
    {
      "epoch": 1.297709923664122,
      "grad_norm": 0.6956679224967957,
      "learning_rate": 1.6187610074396044e-05,
      "loss": 0.4227,
      "step": 7480
    },
    {
      "epoch": 1.2978834142956281,
      "grad_norm": 0.9546607136726379,
      "learning_rate": 1.6180921966943722e-05,
      "loss": 0.4219,
      "step": 7481
    },
    {
      "epoch": 1.298056904927134,
      "grad_norm": 0.9601172208786011,
      "learning_rate": 1.6174234302699856e-05,
      "loss": 0.5724,
      "step": 7482
    },
    {
      "epoch": 1.2982303955586398,
      "grad_norm": 0.7844841480255127,
      "learning_rate": 1.616754708244056e-05,
      "loss": 0.4523,
      "step": 7483
    },
    {
      "epoch": 1.2984038861901457,
      "grad_norm": 0.7818117141723633,
      "learning_rate": 1.616086030694189e-05,
      "loss": 0.5029,
      "step": 7484
    },
    {
      "epoch": 1.2985773768216515,
      "grad_norm": 0.8169832229614258,
      "learning_rate": 1.615417397697985e-05,
      "loss": 0.4934,
      "step": 7485
    },
    {
      "epoch": 1.2987508674531576,
      "grad_norm": 0.7963302731513977,
      "learning_rate": 1.6147488093330405e-05,
      "loss": 0.3839,
      "step": 7486
    },
    {
      "epoch": 1.2989243580846634,
      "grad_norm": 1.164228916168213,
      "learning_rate": 1.6140802656769457e-05,
      "loss": 0.3376,
      "step": 7487
    },
    {
      "epoch": 1.2990978487161693,
      "grad_norm": 0.7500046491622925,
      "learning_rate": 1.6134117668072858e-05,
      "loss": 0.4537,
      "step": 7488
    },
    {
      "epoch": 1.2992713393476754,
      "grad_norm": 1.1466459035873413,
      "learning_rate": 1.6127433128016403e-05,
      "loss": 0.4384,
      "step": 7489
    },
    {
      "epoch": 1.2994448299791812,
      "grad_norm": 0.790235698223114,
      "learning_rate": 1.612074903737585e-05,
      "loss": 0.3829,
      "step": 7490
    },
    {
      "epoch": 1.299618320610687,
      "grad_norm": 1.0291216373443604,
      "learning_rate": 1.611406539692689e-05,
      "loss": 0.4874,
      "step": 7491
    },
    {
      "epoch": 1.299791811242193,
      "grad_norm": 0.6450439691543579,
      "learning_rate": 1.610738220744517e-05,
      "loss": 0.6089,
      "step": 7492
    },
    {
      "epoch": 1.2999653018736987,
      "grad_norm": 0.6980123519897461,
      "learning_rate": 1.6100699469706285e-05,
      "loss": 0.4474,
      "step": 7493
    },
    {
      "epoch": 1.3001387925052046,
      "grad_norm": 0.7495476603507996,
      "learning_rate": 1.6094017184485763e-05,
      "loss": 0.5873,
      "step": 7494
    },
    {
      "epoch": 1.3003122831367107,
      "grad_norm": 0.6476255059242249,
      "learning_rate": 1.6087335352559097e-05,
      "loss": 0.5492,
      "step": 7495
    },
    {
      "epoch": 1.3004857737682165,
      "grad_norm": 0.7907722592353821,
      "learning_rate": 1.6080653974701732e-05,
      "loss": 0.4285,
      "step": 7496
    },
    {
      "epoch": 1.3006592643997223,
      "grad_norm": 0.8223806619644165,
      "learning_rate": 1.6073973051689032e-05,
      "loss": 0.4886,
      "step": 7497
    },
    {
      "epoch": 1.3008327550312284,
      "grad_norm": 0.6121790409088135,
      "learning_rate": 1.6067292584296333e-05,
      "loss": 0.4142,
      "step": 7498
    },
    {
      "epoch": 1.3010062456627343,
      "grad_norm": 0.7437500953674316,
      "learning_rate": 1.6060612573298912e-05,
      "loss": 0.4806,
      "step": 7499
    },
    {
      "epoch": 1.30117973629424,
      "grad_norm": 0.9398663640022278,
      "learning_rate": 1.6053933019472003e-05,
      "loss": 0.3938,
      "step": 7500
    },
    {
      "epoch": 1.301353226925746,
      "grad_norm": 0.9563746452331543,
      "learning_rate": 1.6047253923590756e-05,
      "loss": 0.4484,
      "step": 7501
    },
    {
      "epoch": 1.3015267175572518,
      "grad_norm": 0.5969168543815613,
      "learning_rate": 1.6040575286430295e-05,
      "loss": 0.5699,
      "step": 7502
    },
    {
      "epoch": 1.3017002081887579,
      "grad_norm": 0.7994356751441956,
      "learning_rate": 1.6033897108765696e-05,
      "loss": 0.5862,
      "step": 7503
    },
    {
      "epoch": 1.3018736988202637,
      "grad_norm": 0.7313307523727417,
      "learning_rate": 1.602721939137195e-05,
      "loss": 0.5156,
      "step": 7504
    },
    {
      "epoch": 1.3020471894517696,
      "grad_norm": 1.1211521625518799,
      "learning_rate": 1.6020542135024023e-05,
      "loss": 0.3884,
      "step": 7505
    },
    {
      "epoch": 1.3022206800832756,
      "grad_norm": 0.6634950637817383,
      "learning_rate": 1.6013865340496826e-05,
      "loss": 0.495,
      "step": 7506
    },
    {
      "epoch": 1.3023941707147815,
      "grad_norm": 0.8779208064079285,
      "learning_rate": 1.6007189008565195e-05,
      "loss": 0.5415,
      "step": 7507
    },
    {
      "epoch": 1.3025676613462873,
      "grad_norm": 0.7145334482192993,
      "learning_rate": 1.6000513140003927e-05,
      "loss": 0.4457,
      "step": 7508
    },
    {
      "epoch": 1.3027411519777932,
      "grad_norm": 0.724416971206665,
      "learning_rate": 1.5993837735587783e-05,
      "loss": 0.5167,
      "step": 7509
    },
    {
      "epoch": 1.302914642609299,
      "grad_norm": 0.7721180319786072,
      "learning_rate": 1.5987162796091428e-05,
      "loss": 0.5464,
      "step": 7510
    },
    {
      "epoch": 1.3030881332408049,
      "grad_norm": 0.8560741543769836,
      "learning_rate": 1.5980488322289505e-05,
      "loss": 0.4615,
      "step": 7511
    },
    {
      "epoch": 1.303261623872311,
      "grad_norm": 0.9781146049499512,
      "learning_rate": 1.5973814314956602e-05,
      "loss": 0.5835,
      "step": 7512
    },
    {
      "epoch": 1.3034351145038168,
      "grad_norm": 1.5704987049102783,
      "learning_rate": 1.5967140774867235e-05,
      "loss": 0.5801,
      "step": 7513
    },
    {
      "epoch": 1.3036086051353226,
      "grad_norm": 0.7586106657981873,
      "learning_rate": 1.596046770279588e-05,
      "loss": 0.4703,
      "step": 7514
    },
    {
      "epoch": 1.3037820957668287,
      "grad_norm": 0.5656479001045227,
      "learning_rate": 1.5953795099516955e-05,
      "loss": 0.5348,
      "step": 7515
    },
    {
      "epoch": 1.3039555863983345,
      "grad_norm": 1.0700138807296753,
      "learning_rate": 1.5947122965804827e-05,
      "loss": 0.4349,
      "step": 7516
    },
    {
      "epoch": 1.3041290770298404,
      "grad_norm": 0.7711705565452576,
      "learning_rate": 1.59404513024338e-05,
      "loss": 0.4879,
      "step": 7517
    },
    {
      "epoch": 1.3043025676613462,
      "grad_norm": 0.7690720558166504,
      "learning_rate": 1.5933780110178128e-05,
      "loss": 0.4318,
      "step": 7518
    },
    {
      "epoch": 1.304476058292852,
      "grad_norm": 0.8652418255805969,
      "learning_rate": 1.5927109389812013e-05,
      "loss": 0.413,
      "step": 7519
    },
    {
      "epoch": 1.3046495489243581,
      "grad_norm": 0.80824214220047,
      "learning_rate": 1.59204391421096e-05,
      "loss": 0.4968,
      "step": 7520
    },
    {
      "epoch": 1.304823039555864,
      "grad_norm": 0.8984287977218628,
      "learning_rate": 1.5913769367844974e-05,
      "loss": 0.534,
      "step": 7521
    },
    {
      "epoch": 1.3049965301873698,
      "grad_norm": 0.6900107264518738,
      "learning_rate": 1.5907100067792186e-05,
      "loss": 0.5259,
      "step": 7522
    },
    {
      "epoch": 1.305170020818876,
      "grad_norm": 0.6418942213058472,
      "learning_rate": 1.590043124272519e-05,
      "loss": 0.5896,
      "step": 7523
    },
    {
      "epoch": 1.3053435114503817,
      "grad_norm": 0.8585731387138367,
      "learning_rate": 1.589376289341793e-05,
      "loss": 0.4115,
      "step": 7524
    },
    {
      "epoch": 1.3055170020818876,
      "grad_norm": 0.7281717658042908,
      "learning_rate": 1.5887095020644282e-05,
      "loss": 0.5431,
      "step": 7525
    },
    {
      "epoch": 1.3056904927133934,
      "grad_norm": 0.6613842844963074,
      "learning_rate": 1.5880427625178035e-05,
      "loss": 0.5422,
      "step": 7526
    },
    {
      "epoch": 1.3058639833448993,
      "grad_norm": 0.7034533619880676,
      "learning_rate": 1.5873760707792966e-05,
      "loss": 0.5529,
      "step": 7527
    },
    {
      "epoch": 1.3060374739764053,
      "grad_norm": 0.905914843082428,
      "learning_rate": 1.586709426926277e-05,
      "loss": 0.5544,
      "step": 7528
    },
    {
      "epoch": 1.3062109646079112,
      "grad_norm": 0.7205399870872498,
      "learning_rate": 1.5860428310361117e-05,
      "loss": 0.3742,
      "step": 7529
    },
    {
      "epoch": 1.306384455239417,
      "grad_norm": 0.7713692784309387,
      "learning_rate": 1.5853762831861567e-05,
      "loss": 0.5132,
      "step": 7530
    },
    {
      "epoch": 1.3065579458709229,
      "grad_norm": 0.6062445044517517,
      "learning_rate": 1.5847097834537674e-05,
      "loss": 0.5809,
      "step": 7531
    },
    {
      "epoch": 1.306731436502429,
      "grad_norm": 0.7817365527153015,
      "learning_rate": 1.5840433319162925e-05,
      "loss": 0.5175,
      "step": 7532
    },
    {
      "epoch": 1.3069049271339348,
      "grad_norm": 0.6964122653007507,
      "learning_rate": 1.5833769286510727e-05,
      "loss": 0.5847,
      "step": 7533
    },
    {
      "epoch": 1.3070784177654406,
      "grad_norm": 0.9588583111763,
      "learning_rate": 1.5827105737354456e-05,
      "loss": 0.4564,
      "step": 7534
    },
    {
      "epoch": 1.3072519083969465,
      "grad_norm": 0.8326556086540222,
      "learning_rate": 1.5820442672467436e-05,
      "loss": 0.402,
      "step": 7535
    },
    {
      "epoch": 1.3074253990284523,
      "grad_norm": 0.7271561026573181,
      "learning_rate": 1.5813780092622907e-05,
      "loss": 0.476,
      "step": 7536
    },
    {
      "epoch": 1.3075988896599584,
      "grad_norm": 0.7360162734985352,
      "learning_rate": 1.5807117998594077e-05,
      "loss": 0.4076,
      "step": 7537
    },
    {
      "epoch": 1.3077723802914643,
      "grad_norm": 0.7691264152526855,
      "learning_rate": 1.580045639115409e-05,
      "loss": 0.4091,
      "step": 7538
    },
    {
      "epoch": 1.30794587092297,
      "grad_norm": 0.6766218543052673,
      "learning_rate": 1.5793795271076033e-05,
      "loss": 0.5032,
      "step": 7539
    },
    {
      "epoch": 1.3081193615544762,
      "grad_norm": 0.6278243064880371,
      "learning_rate": 1.5787134639132935e-05,
      "loss": 0.5909,
      "step": 7540
    },
    {
      "epoch": 1.308292852185982,
      "grad_norm": 1.2189621925354004,
      "learning_rate": 1.5780474496097773e-05,
      "loss": 0.58,
      "step": 7541
    },
    {
      "epoch": 1.3084663428174879,
      "grad_norm": 0.625394344329834,
      "learning_rate": 1.577381484274346e-05,
      "loss": 0.5812,
      "step": 7542
    },
    {
      "epoch": 1.3086398334489937,
      "grad_norm": 0.769202709197998,
      "learning_rate": 1.5767155679842857e-05,
      "loss": 0.5537,
      "step": 7543
    },
    {
      "epoch": 1.3088133240804996,
      "grad_norm": 0.7040926814079285,
      "learning_rate": 1.576049700816877e-05,
      "loss": 0.5208,
      "step": 7544
    },
    {
      "epoch": 1.3089868147120056,
      "grad_norm": 0.9660540819168091,
      "learning_rate": 1.5753838828493953e-05,
      "loss": 0.4706,
      "step": 7545
    },
    {
      "epoch": 1.3091603053435115,
      "grad_norm": 0.8025970458984375,
      "learning_rate": 1.574718114159108e-05,
      "loss": 0.5432,
      "step": 7546
    },
    {
      "epoch": 1.3093337959750173,
      "grad_norm": 1.054005742073059,
      "learning_rate": 1.574052394823279e-05,
      "loss": 0.4279,
      "step": 7547
    },
    {
      "epoch": 1.3095072866065234,
      "grad_norm": 1.0567641258239746,
      "learning_rate": 1.5733867249191667e-05,
      "loss": 0.4843,
      "step": 7548
    },
    {
      "epoch": 1.3096807772380292,
      "grad_norm": 2.3491523265838623,
      "learning_rate": 1.5727211045240217e-05,
      "loss": 0.4946,
      "step": 7549
    },
    {
      "epoch": 1.309854267869535,
      "grad_norm": 0.8406051397323608,
      "learning_rate": 1.57205553371509e-05,
      "loss": 0.3904,
      "step": 7550
    },
    {
      "epoch": 1.310027758501041,
      "grad_norm": 0.6834543943405151,
      "learning_rate": 1.571390012569613e-05,
      "loss": 0.4987,
      "step": 7551
    },
    {
      "epoch": 1.3102012491325468,
      "grad_norm": 0.5546103119850159,
      "learning_rate": 1.570724541164824e-05,
      "loss": 0.5178,
      "step": 7552
    },
    {
      "epoch": 1.3103747397640526,
      "grad_norm": 1.1311421394348145,
      "learning_rate": 1.570059119577952e-05,
      "loss": 0.5161,
      "step": 7553
    },
    {
      "epoch": 1.3105482303955587,
      "grad_norm": 0.652874231338501,
      "learning_rate": 1.569393747886221e-05,
      "loss": 0.5701,
      "step": 7554
    },
    {
      "epoch": 1.3107217210270645,
      "grad_norm": 0.7094644904136658,
      "learning_rate": 1.5687284261668465e-05,
      "loss": 0.5991,
      "step": 7555
    },
    {
      "epoch": 1.3108952116585704,
      "grad_norm": 0.7584097385406494,
      "learning_rate": 1.5680631544970405e-05,
      "loss": 0.5813,
      "step": 7556
    },
    {
      "epoch": 1.3110687022900764,
      "grad_norm": 0.9776495695114136,
      "learning_rate": 1.567397932954009e-05,
      "loss": 0.4181,
      "step": 7557
    },
    {
      "epoch": 1.3112421929215823,
      "grad_norm": 1.0048600435256958,
      "learning_rate": 1.5667327616149522e-05,
      "loss": 0.4086,
      "step": 7558
    },
    {
      "epoch": 1.3114156835530881,
      "grad_norm": 0.7916238903999329,
      "learning_rate": 1.5660676405570625e-05,
      "loss": 0.4926,
      "step": 7559
    },
    {
      "epoch": 1.311589174184594,
      "grad_norm": 0.6625590920448303,
      "learning_rate": 1.5654025698575286e-05,
      "loss": 0.5209,
      "step": 7560
    },
    {
      "epoch": 1.3117626648160998,
      "grad_norm": 0.8551685214042664,
      "learning_rate": 1.5647375495935334e-05,
      "loss": 0.5686,
      "step": 7561
    },
    {
      "epoch": 1.311936155447606,
      "grad_norm": 0.9185087084770203,
      "learning_rate": 1.5640725798422525e-05,
      "loss": 0.4187,
      "step": 7562
    },
    {
      "epoch": 1.3121096460791117,
      "grad_norm": 1.0313657522201538,
      "learning_rate": 1.5634076606808567e-05,
      "loss": 0.4257,
      "step": 7563
    },
    {
      "epoch": 1.3122831367106176,
      "grad_norm": 0.7563906311988831,
      "learning_rate": 1.5627427921865106e-05,
      "loss": 0.453,
      "step": 7564
    },
    {
      "epoch": 1.3124566273421236,
      "grad_norm": 0.6281924247741699,
      "learning_rate": 1.562077974436373e-05,
      "loss": 0.4663,
      "step": 7565
    },
    {
      "epoch": 1.3126301179736295,
      "grad_norm": 0.7944850325584412,
      "learning_rate": 1.5614132075075967e-05,
      "loss": 0.4352,
      "step": 7566
    },
    {
      "epoch": 1.3128036086051353,
      "grad_norm": 0.8636308908462524,
      "learning_rate": 1.560748491477329e-05,
      "loss": 0.3758,
      "step": 7567
    },
    {
      "epoch": 1.3129770992366412,
      "grad_norm": 1.3377954959869385,
      "learning_rate": 1.5600838264227102e-05,
      "loss": 0.4989,
      "step": 7568
    },
    {
      "epoch": 1.313150589868147,
      "grad_norm": 0.7625263333320618,
      "learning_rate": 1.5594192124208758e-05,
      "loss": 0.4374,
      "step": 7569
    },
    {
      "epoch": 1.313324080499653,
      "grad_norm": 0.6469513177871704,
      "learning_rate": 1.5587546495489563e-05,
      "loss": 0.4775,
      "step": 7570
    },
    {
      "epoch": 1.313497571131159,
      "grad_norm": 0.7678002119064331,
      "learning_rate": 1.558090137884073e-05,
      "loss": 0.4462,
      "step": 7571
    },
    {
      "epoch": 1.3136710617626648,
      "grad_norm": 0.7460236549377441,
      "learning_rate": 1.557425677503344e-05,
      "loss": 0.4767,
      "step": 7572
    },
    {
      "epoch": 1.3138445523941706,
      "grad_norm": 0.751082181930542,
      "learning_rate": 1.5567612684838805e-05,
      "loss": 0.4169,
      "step": 7573
    },
    {
      "epoch": 1.3140180430256767,
      "grad_norm": 1.1679073572158813,
      "learning_rate": 1.5560969109027896e-05,
      "loss": 0.4708,
      "step": 7574
    },
    {
      "epoch": 1.3141915336571826,
      "grad_norm": 0.7063091397285461,
      "learning_rate": 1.5554326048371686e-05,
      "loss": 0.5521,
      "step": 7575
    },
    {
      "epoch": 1.3143650242886884,
      "grad_norm": 0.5069116950035095,
      "learning_rate": 1.5547683503641115e-05,
      "loss": 0.5902,
      "step": 7576
    },
    {
      "epoch": 1.3145385149201942,
      "grad_norm": 0.7836443781852722,
      "learning_rate": 1.5541041475607073e-05,
      "loss": 0.4136,
      "step": 7577
    },
    {
      "epoch": 1.3147120055517,
      "grad_norm": 1.1055240631103516,
      "learning_rate": 1.5534399965040353e-05,
      "loss": 0.4011,
      "step": 7578
    },
    {
      "epoch": 1.3148854961832062,
      "grad_norm": 0.8819207549095154,
      "learning_rate": 1.552775897271172e-05,
      "loss": 0.4116,
      "step": 7579
    },
    {
      "epoch": 1.315058986814712,
      "grad_norm": 0.7257546186447144,
      "learning_rate": 1.552111849939188e-05,
      "loss": 0.5327,
      "step": 7580
    },
    {
      "epoch": 1.3152324774462179,
      "grad_norm": 1.2918899059295654,
      "learning_rate": 1.5514478545851452e-05,
      "loss": 0.4115,
      "step": 7581
    },
    {
      "epoch": 1.315405968077724,
      "grad_norm": 0.5972050428390503,
      "learning_rate": 1.550783911286101e-05,
      "loss": 0.5763,
      "step": 7582
    },
    {
      "epoch": 1.3155794587092298,
      "grad_norm": 0.8291058540344238,
      "learning_rate": 1.550120020119108e-05,
      "loss": 0.4026,
      "step": 7583
    },
    {
      "epoch": 1.3157529493407356,
      "grad_norm": 0.6583446264266968,
      "learning_rate": 1.5494561811612102e-05,
      "loss": 0.5385,
      "step": 7584
    },
    {
      "epoch": 1.3159264399722415,
      "grad_norm": 0.7057300806045532,
      "learning_rate": 1.548792394489448e-05,
      "loss": 0.4528,
      "step": 7585
    },
    {
      "epoch": 1.3160999306037473,
      "grad_norm": 0.6967214941978455,
      "learning_rate": 1.548128660180854e-05,
      "loss": 0.4697,
      "step": 7586
    },
    {
      "epoch": 1.3162734212352534,
      "grad_norm": 0.7139752507209778,
      "learning_rate": 1.5474649783124555e-05,
      "loss": 0.4589,
      "step": 7587
    },
    {
      "epoch": 1.3164469118667592,
      "grad_norm": 0.8640236854553223,
      "learning_rate": 1.5468013489612742e-05,
      "loss": 0.5447,
      "step": 7588
    },
    {
      "epoch": 1.316620402498265,
      "grad_norm": 0.8197435736656189,
      "learning_rate": 1.5461377722043235e-05,
      "loss": 0.5264,
      "step": 7589
    },
    {
      "epoch": 1.316793893129771,
      "grad_norm": 0.5978457927703857,
      "learning_rate": 1.5454742481186137e-05,
      "loss": 0.5288,
      "step": 7590
    },
    {
      "epoch": 1.316967383761277,
      "grad_norm": 0.6455949544906616,
      "learning_rate": 1.5448107767811468e-05,
      "loss": 0.5421,
      "step": 7591
    },
    {
      "epoch": 1.3171408743927828,
      "grad_norm": 1.9688469171524048,
      "learning_rate": 1.5441473582689198e-05,
      "loss": 0.641,
      "step": 7592
    },
    {
      "epoch": 1.3173143650242887,
      "grad_norm": 0.782586395740509,
      "learning_rate": 1.5434839926589236e-05,
      "loss": 0.6558,
      "step": 7593
    },
    {
      "epoch": 1.3174878556557945,
      "grad_norm": 0.7474291920661926,
      "learning_rate": 1.5428206800281413e-05,
      "loss": 0.5875,
      "step": 7594
    },
    {
      "epoch": 1.3176613462873004,
      "grad_norm": 0.7581984400749207,
      "learning_rate": 1.5421574204535516e-05,
      "loss": 0.5179,
      "step": 7595
    },
    {
      "epoch": 1.3178348369188064,
      "grad_norm": 0.6876397132873535,
      "learning_rate": 1.5414942140121278e-05,
      "loss": 0.4359,
      "step": 7596
    },
    {
      "epoch": 1.3180083275503123,
      "grad_norm": 0.6083647608757019,
      "learning_rate": 1.5408310607808336e-05,
      "loss": 0.5122,
      "step": 7597
    },
    {
      "epoch": 1.3181818181818181,
      "grad_norm": 1.8246253728866577,
      "learning_rate": 1.54016796083663e-05,
      "loss": 0.4479,
      "step": 7598
    },
    {
      "epoch": 1.3183553088133242,
      "grad_norm": 0.7424356341362,
      "learning_rate": 1.5395049142564717e-05,
      "loss": 0.449,
      "step": 7599
    },
    {
      "epoch": 1.31852879944483,
      "grad_norm": 0.7722662687301636,
      "learning_rate": 1.538841921117303e-05,
      "loss": 0.4231,
      "step": 7600
    },
    {
      "epoch": 1.3187022900763359,
      "grad_norm": 0.6274324655532837,
      "learning_rate": 1.5381789814960674e-05,
      "loss": 0.3984,
      "step": 7601
    },
    {
      "epoch": 1.3188757807078417,
      "grad_norm": 1.1475446224212646,
      "learning_rate": 1.5375160954696986e-05,
      "loss": 0.4741,
      "step": 7602
    },
    {
      "epoch": 1.3190492713393476,
      "grad_norm": 0.7600390315055847,
      "learning_rate": 1.536853263115127e-05,
      "loss": 0.373,
      "step": 7603
    },
    {
      "epoch": 1.3192227619708536,
      "grad_norm": 1.167948842048645,
      "learning_rate": 1.536190484509273e-05,
      "loss": 0.3993,
      "step": 7604
    },
    {
      "epoch": 1.3193962526023595,
      "grad_norm": 1.0073474645614624,
      "learning_rate": 1.5355277597290537e-05,
      "loss": 0.4735,
      "step": 7605
    },
    {
      "epoch": 1.3195697432338653,
      "grad_norm": 1.5348100662231445,
      "learning_rate": 1.5348650888513798e-05,
      "loss": 0.5296,
      "step": 7606
    },
    {
      "epoch": 1.3197432338653714,
      "grad_norm": 0.7548238039016724,
      "learning_rate": 1.5342024719531536e-05,
      "loss": 0.557,
      "step": 7607
    },
    {
      "epoch": 1.3199167244968772,
      "grad_norm": 0.7464498281478882,
      "learning_rate": 1.533539909111273e-05,
      "loss": 0.557,
      "step": 7608
    },
    {
      "epoch": 1.320090215128383,
      "grad_norm": 0.7466408610343933,
      "learning_rate": 1.5328774004026304e-05,
      "loss": 0.5782,
      "step": 7609
    },
    {
      "epoch": 1.320263705759889,
      "grad_norm": 0.8689438700675964,
      "learning_rate": 1.5322149459041097e-05,
      "loss": 0.4982,
      "step": 7610
    },
    {
      "epoch": 1.3204371963913948,
      "grad_norm": 0.6298958659172058,
      "learning_rate": 1.531552545692589e-05,
      "loss": 0.5846,
      "step": 7611
    },
    {
      "epoch": 1.3206106870229006,
      "grad_norm": 0.8127974271774292,
      "learning_rate": 1.5308901998449415e-05,
      "loss": 0.4543,
      "step": 7612
    },
    {
      "epoch": 1.3207841776544067,
      "grad_norm": 0.6954392194747925,
      "learning_rate": 1.5302279084380328e-05,
      "loss": 0.6154,
      "step": 7613
    },
    {
      "epoch": 1.3209576682859125,
      "grad_norm": 0.8068144917488098,
      "learning_rate": 1.5295656715487226e-05,
      "loss": 0.4207,
      "step": 7614
    },
    {
      "epoch": 1.3211311589174184,
      "grad_norm": 0.8445618748664856,
      "learning_rate": 1.528903489253865e-05,
      "loss": 0.4735,
      "step": 7615
    },
    {
      "epoch": 1.3213046495489245,
      "grad_norm": 0.7115435600280762,
      "learning_rate": 1.5282413616303063e-05,
      "loss": 0.4727,
      "step": 7616
    },
    {
      "epoch": 1.3214781401804303,
      "grad_norm": 0.7735037207603455,
      "learning_rate": 1.5275792887548866e-05,
      "loss": 0.4672,
      "step": 7617
    },
    {
      "epoch": 1.3216516308119362,
      "grad_norm": 2.0262577533721924,
      "learning_rate": 1.526917270704441e-05,
      "loss": 0.4323,
      "step": 7618
    },
    {
      "epoch": 1.321825121443442,
      "grad_norm": 0.8620812892913818,
      "learning_rate": 1.5262553075557985e-05,
      "loss": 0.5571,
      "step": 7619
    },
    {
      "epoch": 1.3219986120749478,
      "grad_norm": 0.9684853553771973,
      "learning_rate": 1.5255933993857785e-05,
      "loss": 0.5067,
      "step": 7620
    },
    {
      "epoch": 1.322172102706454,
      "grad_norm": 0.9358473420143127,
      "learning_rate": 1.5249315462711974e-05,
      "loss": 0.411,
      "step": 7621
    },
    {
      "epoch": 1.3223455933379598,
      "grad_norm": 0.888228178024292,
      "learning_rate": 1.5242697482888649e-05,
      "loss": 0.4415,
      "step": 7622
    },
    {
      "epoch": 1.3225190839694656,
      "grad_norm": 1.0634689331054688,
      "learning_rate": 1.5236080055155812e-05,
      "loss": 0.5203,
      "step": 7623
    },
    {
      "epoch": 1.3226925746009717,
      "grad_norm": 3.2980542182922363,
      "learning_rate": 1.5229463180281441e-05,
      "loss": 0.4736,
      "step": 7624
    },
    {
      "epoch": 1.3228660652324775,
      "grad_norm": 0.9480955600738525,
      "learning_rate": 1.522284685903343e-05,
      "loss": 0.5311,
      "step": 7625
    },
    {
      "epoch": 1.3230395558639834,
      "grad_norm": 0.8151806592941284,
      "learning_rate": 1.5216231092179604e-05,
      "loss": 0.4984,
      "step": 7626
    },
    {
      "epoch": 1.3232130464954892,
      "grad_norm": 0.7603073120117188,
      "learning_rate": 1.5209615880487728e-05,
      "loss": 0.4894,
      "step": 7627
    },
    {
      "epoch": 1.323386537126995,
      "grad_norm": 0.7880406379699707,
      "learning_rate": 1.5203001224725525e-05,
      "loss": 0.4729,
      "step": 7628
    },
    {
      "epoch": 1.3235600277585011,
      "grad_norm": 0.7434356212615967,
      "learning_rate": 1.5196387125660607e-05,
      "loss": 0.444,
      "step": 7629
    },
    {
      "epoch": 1.323733518390007,
      "grad_norm": 1.1501247882843018,
      "learning_rate": 1.5189773584060563e-05,
      "loss": 0.4009,
      "step": 7630
    },
    {
      "epoch": 1.3239070090215128,
      "grad_norm": 0.5699131488800049,
      "learning_rate": 1.51831606006929e-05,
      "loss": 0.5969,
      "step": 7631
    },
    {
      "epoch": 1.3240804996530187,
      "grad_norm": 0.7818866968154907,
      "learning_rate": 1.517654817632507e-05,
      "loss": 0.5359,
      "step": 7632
    },
    {
      "epoch": 1.3242539902845247,
      "grad_norm": 0.7145308256149292,
      "learning_rate": 1.5169936311724434e-05,
      "loss": 0.4215,
      "step": 7633
    },
    {
      "epoch": 1.3244274809160306,
      "grad_norm": 0.8266650438308716,
      "learning_rate": 1.5163325007658319e-05,
      "loss": 0.616,
      "step": 7634
    },
    {
      "epoch": 1.3246009715475364,
      "grad_norm": 0.9614689946174622,
      "learning_rate": 1.5156714264893974e-05,
      "loss": 0.3827,
      "step": 7635
    },
    {
      "epoch": 1.3247744621790423,
      "grad_norm": 0.6380361914634705,
      "learning_rate": 1.5150104084198587e-05,
      "loss": 0.5,
      "step": 7636
    },
    {
      "epoch": 1.3249479528105481,
      "grad_norm": 0.7799776196479797,
      "learning_rate": 1.5143494466339266e-05,
      "loss": 0.5543,
      "step": 7637
    },
    {
      "epoch": 1.3251214434420542,
      "grad_norm": 0.8001444935798645,
      "learning_rate": 1.5136885412083073e-05,
      "loss": 0.3903,
      "step": 7638
    },
    {
      "epoch": 1.32529493407356,
      "grad_norm": 0.6268643140792847,
      "learning_rate": 1.5130276922196993e-05,
      "loss": 0.5321,
      "step": 7639
    },
    {
      "epoch": 1.3254684247050659,
      "grad_norm": 0.5907706618309021,
      "learning_rate": 1.5123668997447948e-05,
      "loss": 0.4918,
      "step": 7640
    },
    {
      "epoch": 1.325641915336572,
      "grad_norm": 1.132335901260376,
      "learning_rate": 1.5117061638602811e-05,
      "loss": 0.4606,
      "step": 7641
    },
    {
      "epoch": 1.3258154059680778,
      "grad_norm": 0.7444654107093811,
      "learning_rate": 1.5110454846428348e-05,
      "loss": 0.5414,
      "step": 7642
    },
    {
      "epoch": 1.3259888965995836,
      "grad_norm": 0.735350489616394,
      "learning_rate": 1.51038486216913e-05,
      "loss": 0.3879,
      "step": 7643
    },
    {
      "epoch": 1.3261623872310895,
      "grad_norm": 0.6652734279632568,
      "learning_rate": 1.5097242965158322e-05,
      "loss": 0.486,
      "step": 7644
    },
    {
      "epoch": 1.3263358778625953,
      "grad_norm": 0.5584680438041687,
      "learning_rate": 1.5090637877596022e-05,
      "loss": 0.5537,
      "step": 7645
    },
    {
      "epoch": 1.3265093684941014,
      "grad_norm": 0.6846016645431519,
      "learning_rate": 1.5084033359770907e-05,
      "loss": 0.4653,
      "step": 7646
    },
    {
      "epoch": 1.3266828591256072,
      "grad_norm": 0.9131779670715332,
      "learning_rate": 1.507742941244945e-05,
      "loss": 0.5026,
      "step": 7647
    },
    {
      "epoch": 1.326856349757113,
      "grad_norm": 0.5463833212852478,
      "learning_rate": 1.5070826036398052e-05,
      "loss": 0.6051,
      "step": 7648
    },
    {
      "epoch": 1.3270298403886192,
      "grad_norm": 1.3209789991378784,
      "learning_rate": 1.5064223232383028e-05,
      "loss": 0.4663,
      "step": 7649
    },
    {
      "epoch": 1.327203331020125,
      "grad_norm": 0.8686030507087708,
      "learning_rate": 1.505762100117065e-05,
      "loss": 0.4188,
      "step": 7650
    },
    {
      "epoch": 1.3273768216516308,
      "grad_norm": 0.7723866105079651,
      "learning_rate": 1.5051019343527123e-05,
      "loss": 0.4111,
      "step": 7651
    },
    {
      "epoch": 1.3275503122831367,
      "grad_norm": 0.7040600180625916,
      "learning_rate": 1.5044418260218559e-05,
      "loss": 0.4075,
      "step": 7652
    },
    {
      "epoch": 1.3277238029146425,
      "grad_norm": 0.5787320137023926,
      "learning_rate": 1.503781775201103e-05,
      "loss": 0.5212,
      "step": 7653
    },
    {
      "epoch": 1.3278972935461484,
      "grad_norm": 0.7481715679168701,
      "learning_rate": 1.503121781967054e-05,
      "loss": 0.4248,
      "step": 7654
    },
    {
      "epoch": 1.3280707841776545,
      "grad_norm": 0.8455163240432739,
      "learning_rate": 1.5024618463963006e-05,
      "loss": 0.4517,
      "step": 7655
    },
    {
      "epoch": 1.3282442748091603,
      "grad_norm": 0.8468508124351501,
      "learning_rate": 1.5018019685654295e-05,
      "loss": 0.5693,
      "step": 7656
    },
    {
      "epoch": 1.3284177654406661,
      "grad_norm": 0.8724422454833984,
      "learning_rate": 1.5011421485510204e-05,
      "loss": 0.4233,
      "step": 7657
    },
    {
      "epoch": 1.3285912560721722,
      "grad_norm": 0.8734956979751587,
      "learning_rate": 1.5004823864296472e-05,
      "loss": 0.4835,
      "step": 7658
    },
    {
      "epoch": 1.328764746703678,
      "grad_norm": 1.4164583683013916,
      "learning_rate": 1.4998226822778743e-05,
      "loss": 0.419,
      "step": 7659
    },
    {
      "epoch": 1.328938237335184,
      "grad_norm": 0.7992679476737976,
      "learning_rate": 1.4991630361722619e-05,
      "loss": 0.5253,
      "step": 7660
    },
    {
      "epoch": 1.3291117279666897,
      "grad_norm": 0.8235474824905396,
      "learning_rate": 1.498503448189363e-05,
      "loss": 0.4349,
      "step": 7661
    },
    {
      "epoch": 1.3292852185981956,
      "grad_norm": 0.9846152663230896,
      "learning_rate": 1.4978439184057233e-05,
      "loss": 0.4272,
      "step": 7662
    },
    {
      "epoch": 1.3294587092297017,
      "grad_norm": 0.7669355273246765,
      "learning_rate": 1.497184446897882e-05,
      "loss": 0.3984,
      "step": 7663
    },
    {
      "epoch": 1.3296321998612075,
      "grad_norm": 0.9644906520843506,
      "learning_rate": 1.4965250337423718e-05,
      "loss": 0.342,
      "step": 7664
    },
    {
      "epoch": 1.3298056904927134,
      "grad_norm": 0.8945387601852417,
      "learning_rate": 1.4958656790157176e-05,
      "loss": 0.4828,
      "step": 7665
    },
    {
      "epoch": 1.3299791811242194,
      "grad_norm": 0.7948451042175293,
      "learning_rate": 1.4952063827944385e-05,
      "loss": 0.4005,
      "step": 7666
    },
    {
      "epoch": 1.3301526717557253,
      "grad_norm": 0.598304271697998,
      "learning_rate": 1.4945471451550481e-05,
      "loss": 0.5308,
      "step": 7667
    },
    {
      "epoch": 1.3303261623872311,
      "grad_norm": 0.7321178317070007,
      "learning_rate": 1.4938879661740495e-05,
      "loss": 0.4996,
      "step": 7668
    },
    {
      "epoch": 1.330499653018737,
      "grad_norm": 0.7548737525939941,
      "learning_rate": 1.4932288459279423e-05,
      "loss": 0.3862,
      "step": 7669
    },
    {
      "epoch": 1.3306731436502428,
      "grad_norm": 0.8476508855819702,
      "learning_rate": 1.4925697844932185e-05,
      "loss": 0.5082,
      "step": 7670
    },
    {
      "epoch": 1.3308466342817487,
      "grad_norm": 0.5477104783058167,
      "learning_rate": 1.491910781946362e-05,
      "loss": 0.693,
      "step": 7671
    },
    {
      "epoch": 1.3310201249132547,
      "grad_norm": 0.7386735677719116,
      "learning_rate": 1.4912518383638512e-05,
      "loss": 0.5115,
      "step": 7672
    },
    {
      "epoch": 1.3311936155447606,
      "grad_norm": 0.7570234537124634,
      "learning_rate": 1.4905929538221574e-05,
      "loss": 0.4691,
      "step": 7673
    },
    {
      "epoch": 1.3313671061762664,
      "grad_norm": 0.8122114539146423,
      "learning_rate": 1.4899341283977457e-05,
      "loss": 0.3833,
      "step": 7674
    },
    {
      "epoch": 1.3315405968077725,
      "grad_norm": 2.640561103820801,
      "learning_rate": 1.489275362167072e-05,
      "loss": 0.3793,
      "step": 7675
    },
    {
      "epoch": 1.3317140874392783,
      "grad_norm": 1.0989961624145508,
      "learning_rate": 1.4886166552065873e-05,
      "loss": 0.434,
      "step": 7676
    },
    {
      "epoch": 1.3318875780707842,
      "grad_norm": 0.6833113431930542,
      "learning_rate": 1.4879580075927367e-05,
      "loss": 0.5037,
      "step": 7677
    },
    {
      "epoch": 1.33206106870229,
      "grad_norm": 0.8466157913208008,
      "learning_rate": 1.4872994194019553e-05,
      "loss": 0.519,
      "step": 7678
    },
    {
      "epoch": 1.3322345593337959,
      "grad_norm": 0.6697059273719788,
      "learning_rate": 1.4866408907106734e-05,
      "loss": 0.5159,
      "step": 7679
    },
    {
      "epoch": 1.332408049965302,
      "grad_norm": 0.6055539846420288,
      "learning_rate": 1.4859824215953154e-05,
      "loss": 0.5037,
      "step": 7680
    },
    {
      "epoch": 1.3325815405968078,
      "grad_norm": 0.713620126247406,
      "learning_rate": 1.4853240121322951e-05,
      "loss": 0.4952,
      "step": 7681
    },
    {
      "epoch": 1.3327550312283136,
      "grad_norm": 0.7720019817352295,
      "learning_rate": 1.4846656623980234e-05,
      "loss": 0.6401,
      "step": 7682
    },
    {
      "epoch": 1.3329285218598197,
      "grad_norm": 0.607806384563446,
      "learning_rate": 1.4840073724689021e-05,
      "loss": 0.6267,
      "step": 7683
    },
    {
      "epoch": 1.3331020124913255,
      "grad_norm": 1.852420687675476,
      "learning_rate": 1.4833491424213268e-05,
      "loss": 0.4603,
      "step": 7684
    },
    {
      "epoch": 1.3332755031228314,
      "grad_norm": 0.6995673179626465,
      "learning_rate": 1.482690972331685e-05,
      "loss": 0.4906,
      "step": 7685
    },
    {
      "epoch": 1.3334489937543372,
      "grad_norm": 0.805355966091156,
      "learning_rate": 1.4820328622763584e-05,
      "loss": 0.4603,
      "step": 7686
    },
    {
      "epoch": 1.333622484385843,
      "grad_norm": 0.6964371204376221,
      "learning_rate": 1.4813748123317223e-05,
      "loss": 0.5219,
      "step": 7687
    },
    {
      "epoch": 1.3337959750173491,
      "grad_norm": 0.960204005241394,
      "learning_rate": 1.4807168225741433e-05,
      "loss": 0.3848,
      "step": 7688
    },
    {
      "epoch": 1.333969465648855,
      "grad_norm": 0.9914819598197937,
      "learning_rate": 1.4800588930799822e-05,
      "loss": 0.4637,
      "step": 7689
    },
    {
      "epoch": 1.3341429562803608,
      "grad_norm": 0.7618035078048706,
      "learning_rate": 1.4794010239255925e-05,
      "loss": 0.5688,
      "step": 7690
    },
    {
      "epoch": 1.3343164469118667,
      "grad_norm": 0.8735632300376892,
      "learning_rate": 1.4787432151873202e-05,
      "loss": 0.4575,
      "step": 7691
    },
    {
      "epoch": 1.3344899375433728,
      "grad_norm": 1.0803961753845215,
      "learning_rate": 1.4780854669415053e-05,
      "loss": 0.4144,
      "step": 7692
    },
    {
      "epoch": 1.3346634281748786,
      "grad_norm": 0.7738522291183472,
      "learning_rate": 1.4774277792644812e-05,
      "loss": 0.5168,
      "step": 7693
    },
    {
      "epoch": 1.3348369188063844,
      "grad_norm": 1.2571601867675781,
      "learning_rate": 1.4767701522325708e-05,
      "loss": 0.5652,
      "step": 7694
    },
    {
      "epoch": 1.3350104094378903,
      "grad_norm": 0.5889428853988647,
      "learning_rate": 1.4761125859220942e-05,
      "loss": 0.5564,
      "step": 7695
    },
    {
      "epoch": 1.3351839000693961,
      "grad_norm": 0.9888232946395874,
      "learning_rate": 1.4754550804093633e-05,
      "loss": 0.4918,
      "step": 7696
    },
    {
      "epoch": 1.3353573907009022,
      "grad_norm": 0.6778308749198914,
      "learning_rate": 1.4747976357706806e-05,
      "loss": 0.4269,
      "step": 7697
    },
    {
      "epoch": 1.335530881332408,
      "grad_norm": 0.6476753950119019,
      "learning_rate": 1.4741402520823442e-05,
      "loss": 0.5807,
      "step": 7698
    },
    {
      "epoch": 1.335704371963914,
      "grad_norm": 0.7573915719985962,
      "learning_rate": 1.4734829294206455e-05,
      "loss": 0.4604,
      "step": 7699
    },
    {
      "epoch": 1.33587786259542,
      "grad_norm": 0.7617419362068176,
      "learning_rate": 1.4728256678618652e-05,
      "loss": 0.603,
      "step": 7700
    },
    {
      "epoch": 1.3360513532269258,
      "grad_norm": 0.8916187286376953,
      "learning_rate": 1.4721684674822805e-05,
      "loss": 0.3832,
      "step": 7701
    },
    {
      "epoch": 1.3362248438584317,
      "grad_norm": 0.652928352355957,
      "learning_rate": 1.47151132835816e-05,
      "loss": 0.5734,
      "step": 7702
    },
    {
      "epoch": 1.3363983344899375,
      "grad_norm": 0.5686636567115784,
      "learning_rate": 1.4708542505657668e-05,
      "loss": 0.6082,
      "step": 7703
    },
    {
      "epoch": 1.3365718251214433,
      "grad_norm": 0.6646944284439087,
      "learning_rate": 1.4701972341813533e-05,
      "loss": 0.5999,
      "step": 7704
    },
    {
      "epoch": 1.3367453157529494,
      "grad_norm": 0.8325403332710266,
      "learning_rate": 1.4695402792811684e-05,
      "loss": 0.4495,
      "step": 7705
    },
    {
      "epoch": 1.3369188063844553,
      "grad_norm": 0.7930448055267334,
      "learning_rate": 1.4688833859414529e-05,
      "loss": 0.4022,
      "step": 7706
    },
    {
      "epoch": 1.337092297015961,
      "grad_norm": 0.7062762379646301,
      "learning_rate": 1.4682265542384384e-05,
      "loss": 0.4838,
      "step": 7707
    },
    {
      "epoch": 1.3372657876474672,
      "grad_norm": 0.6182875037193298,
      "learning_rate": 1.467569784248352e-05,
      "loss": 0.548,
      "step": 7708
    },
    {
      "epoch": 1.337439278278973,
      "grad_norm": 1.4004443883895874,
      "learning_rate": 1.466913076047413e-05,
      "loss": 0.4725,
      "step": 7709
    },
    {
      "epoch": 1.3376127689104789,
      "grad_norm": 0.8085479140281677,
      "learning_rate": 1.4662564297118325e-05,
      "loss": 0.4496,
      "step": 7710
    },
    {
      "epoch": 1.3377862595419847,
      "grad_norm": 0.6966636180877686,
      "learning_rate": 1.465599845317815e-05,
      "loss": 0.6305,
      "step": 7711
    },
    {
      "epoch": 1.3379597501734906,
      "grad_norm": 0.5736681222915649,
      "learning_rate": 1.4649433229415588e-05,
      "loss": 0.5409,
      "step": 7712
    },
    {
      "epoch": 1.3381332408049964,
      "grad_norm": 0.6678444147109985,
      "learning_rate": 1.4642868626592529e-05,
      "loss": 0.5149,
      "step": 7713
    },
    {
      "epoch": 1.3383067314365025,
      "grad_norm": 0.7446213960647583,
      "learning_rate": 1.4636304645470807e-05,
      "loss": 0.584,
      "step": 7714
    },
    {
      "epoch": 1.3384802220680083,
      "grad_norm": 0.7427680492401123,
      "learning_rate": 1.462974128681218e-05,
      "loss": 0.4988,
      "step": 7715
    },
    {
      "epoch": 1.3386537126995142,
      "grad_norm": 0.7200520634651184,
      "learning_rate": 1.4623178551378346e-05,
      "loss": 0.5967,
      "step": 7716
    },
    {
      "epoch": 1.3388272033310202,
      "grad_norm": 0.9318338632583618,
      "learning_rate": 1.4616616439930895e-05,
      "loss": 0.4989,
      "step": 7717
    },
    {
      "epoch": 1.339000693962526,
      "grad_norm": 0.6261063814163208,
      "learning_rate": 1.4610054953231379e-05,
      "loss": 0.507,
      "step": 7718
    },
    {
      "epoch": 1.339174184594032,
      "grad_norm": 0.6261553168296814,
      "learning_rate": 1.4603494092041275e-05,
      "loss": 0.6486,
      "step": 7719
    },
    {
      "epoch": 1.3393476752255378,
      "grad_norm": 0.7913298010826111,
      "learning_rate": 1.4596933857121963e-05,
      "loss": 0.52,
      "step": 7720
    },
    {
      "epoch": 1.3395211658570436,
      "grad_norm": 0.7656115293502808,
      "learning_rate": 1.4590374249234768e-05,
      "loss": 0.5051,
      "step": 7721
    },
    {
      "epoch": 1.3396946564885497,
      "grad_norm": 0.7011547684669495,
      "learning_rate": 1.4583815269140957e-05,
      "loss": 0.5421,
      "step": 7722
    },
    {
      "epoch": 1.3398681471200555,
      "grad_norm": 0.7271060943603516,
      "learning_rate": 1.4577256917601688e-05,
      "loss": 0.5173,
      "step": 7723
    },
    {
      "epoch": 1.3400416377515614,
      "grad_norm": 0.8287790417671204,
      "learning_rate": 1.4570699195378071e-05,
      "loss": 0.4644,
      "step": 7724
    },
    {
      "epoch": 1.3402151283830674,
      "grad_norm": 0.8536555767059326,
      "learning_rate": 1.4564142103231148e-05,
      "loss": 0.5801,
      "step": 7725
    },
    {
      "epoch": 1.3403886190145733,
      "grad_norm": 0.6832239627838135,
      "learning_rate": 1.4557585641921859e-05,
      "loss": 0.5763,
      "step": 7726
    },
    {
      "epoch": 1.3405621096460791,
      "grad_norm": 0.7031279802322388,
      "learning_rate": 1.4551029812211095e-05,
      "loss": 0.4196,
      "step": 7727
    },
    {
      "epoch": 1.340735600277585,
      "grad_norm": 0.6169347763061523,
      "learning_rate": 1.4544474614859683e-05,
      "loss": 0.4151,
      "step": 7728
    },
    {
      "epoch": 1.3409090909090908,
      "grad_norm": 0.7564440369606018,
      "learning_rate": 1.4537920050628338e-05,
      "loss": 0.3553,
      "step": 7729
    },
    {
      "epoch": 1.3410825815405967,
      "grad_norm": 1.0365413427352905,
      "learning_rate": 1.4531366120277736e-05,
      "loss": 0.4156,
      "step": 7730
    },
    {
      "epoch": 1.3412560721721027,
      "grad_norm": 0.6548910140991211,
      "learning_rate": 1.4524812824568471e-05,
      "loss": 0.631,
      "step": 7731
    },
    {
      "epoch": 1.3414295628036086,
      "grad_norm": 0.7457895874977112,
      "learning_rate": 1.4518260164261058e-05,
      "loss": 0.418,
      "step": 7732
    },
    {
      "epoch": 1.3416030534351144,
      "grad_norm": 0.6473345756530762,
      "learning_rate": 1.4511708140115942e-05,
      "loss": 0.5637,
      "step": 7733
    },
    {
      "epoch": 1.3417765440666205,
      "grad_norm": 1.1007072925567627,
      "learning_rate": 1.4505156752893488e-05,
      "loss": 0.5497,
      "step": 7734
    },
    {
      "epoch": 1.3419500346981263,
      "grad_norm": 1.1076034307479858,
      "learning_rate": 1.4498606003353998e-05,
      "loss": 0.4563,
      "step": 7735
    },
    {
      "epoch": 1.3421235253296322,
      "grad_norm": 0.6503068208694458,
      "learning_rate": 1.4492055892257688e-05,
      "loss": 0.5977,
      "step": 7736
    },
    {
      "epoch": 1.342297015961138,
      "grad_norm": 0.9386075735092163,
      "learning_rate": 1.4485506420364715e-05,
      "loss": 0.468,
      "step": 7737
    },
    {
      "epoch": 1.3424705065926439,
      "grad_norm": 0.7115505933761597,
      "learning_rate": 1.4478957588435148e-05,
      "loss": 0.5226,
      "step": 7738
    },
    {
      "epoch": 1.34264399722415,
      "grad_norm": 0.7394730448722839,
      "learning_rate": 1.4472409397228979e-05,
      "loss": 0.4031,
      "step": 7739
    },
    {
      "epoch": 1.3428174878556558,
      "grad_norm": 0.7652788758277893,
      "learning_rate": 1.4465861847506142e-05,
      "loss": 0.4783,
      "step": 7740
    },
    {
      "epoch": 1.3429909784871616,
      "grad_norm": 0.8783412575721741,
      "learning_rate": 1.4459314940026495e-05,
      "loss": 0.5023,
      "step": 7741
    },
    {
      "epoch": 1.3431644691186677,
      "grad_norm": 0.7272326350212097,
      "learning_rate": 1.4452768675549798e-05,
      "loss": 0.3974,
      "step": 7742
    },
    {
      "epoch": 1.3433379597501736,
      "grad_norm": 1.0041476488113403,
      "learning_rate": 1.4446223054835758e-05,
      "loss": 0.3462,
      "step": 7743
    },
    {
      "epoch": 1.3435114503816794,
      "grad_norm": 0.655902624130249,
      "learning_rate": 1.4439678078644004e-05,
      "loss": 0.5696,
      "step": 7744
    },
    {
      "epoch": 1.3436849410131853,
      "grad_norm": 0.8266571760177612,
      "learning_rate": 1.4433133747734097e-05,
      "loss": 0.5107,
      "step": 7745
    },
    {
      "epoch": 1.343858431644691,
      "grad_norm": 0.7135486602783203,
      "learning_rate": 1.4426590062865497e-05,
      "loss": 0.4677,
      "step": 7746
    },
    {
      "epoch": 1.3440319222761972,
      "grad_norm": 0.6802731156349182,
      "learning_rate": 1.4420047024797614e-05,
      "loss": 0.6051,
      "step": 7747
    },
    {
      "epoch": 1.344205412907703,
      "grad_norm": 1.187752366065979,
      "learning_rate": 1.4413504634289785e-05,
      "loss": 0.4467,
      "step": 7748
    },
    {
      "epoch": 1.3443789035392089,
      "grad_norm": 0.5917049050331116,
      "learning_rate": 1.4406962892101243e-05,
      "loss": 0.4995,
      "step": 7749
    },
    {
      "epoch": 1.3445523941707147,
      "grad_norm": 0.6279724836349487,
      "learning_rate": 1.4400421798991178e-05,
      "loss": 0.4962,
      "step": 7750
    },
    {
      "epoch": 1.3447258848022208,
      "grad_norm": 0.6500610113143921,
      "learning_rate": 1.4393881355718694e-05,
      "loss": 0.4487,
      "step": 7751
    },
    {
      "epoch": 1.3448993754337266,
      "grad_norm": 0.7289589047431946,
      "learning_rate": 1.4387341563042801e-05,
      "loss": 0.5084,
      "step": 7752
    },
    {
      "epoch": 1.3450728660652325,
      "grad_norm": 0.6606741547584534,
      "learning_rate": 1.4380802421722461e-05,
      "loss": 0.4774,
      "step": 7753
    },
    {
      "epoch": 1.3452463566967383,
      "grad_norm": 0.6247408986091614,
      "learning_rate": 1.4374263932516557e-05,
      "loss": 0.501,
      "step": 7754
    },
    {
      "epoch": 1.3454198473282442,
      "grad_norm": 0.8095901012420654,
      "learning_rate": 1.436772609618387e-05,
      "loss": 0.5699,
      "step": 7755
    },
    {
      "epoch": 1.3455933379597502,
      "grad_norm": 0.7537714242935181,
      "learning_rate": 1.4361188913483132e-05,
      "loss": 0.525,
      "step": 7756
    },
    {
      "epoch": 1.345766828591256,
      "grad_norm": 0.7632372975349426,
      "learning_rate": 1.4354652385172995e-05,
      "loss": 0.5256,
      "step": 7757
    },
    {
      "epoch": 1.345940319222762,
      "grad_norm": 0.6155702471733093,
      "learning_rate": 1.4348116512012024e-05,
      "loss": 0.5352,
      "step": 7758
    },
    {
      "epoch": 1.346113809854268,
      "grad_norm": 0.7520788311958313,
      "learning_rate": 1.4341581294758722e-05,
      "loss": 0.5274,
      "step": 7759
    },
    {
      "epoch": 1.3462873004857738,
      "grad_norm": 1.4066622257232666,
      "learning_rate": 1.4335046734171499e-05,
      "loss": 0.48,
      "step": 7760
    },
    {
      "epoch": 1.3464607911172797,
      "grad_norm": 0.667074978351593,
      "learning_rate": 1.4328512831008708e-05,
      "loss": 0.4421,
      "step": 7761
    },
    {
      "epoch": 1.3466342817487855,
      "grad_norm": 0.6374412178993225,
      "learning_rate": 1.4321979586028607e-05,
      "loss": 0.4266,
      "step": 7762
    },
    {
      "epoch": 1.3468077723802914,
      "grad_norm": 0.7195127606391907,
      "learning_rate": 1.431544699998939e-05,
      "loss": 0.5153,
      "step": 7763
    },
    {
      "epoch": 1.3469812630117974,
      "grad_norm": 0.8725728392601013,
      "learning_rate": 1.4308915073649182e-05,
      "loss": 0.3697,
      "step": 7764
    },
    {
      "epoch": 1.3471547536433033,
      "grad_norm": 0.8734724521636963,
      "learning_rate": 1.4302383807766003e-05,
      "loss": 0.3995,
      "step": 7765
    },
    {
      "epoch": 1.3473282442748091,
      "grad_norm": 0.7318059206008911,
      "learning_rate": 1.4295853203097823e-05,
      "loss": 0.4504,
      "step": 7766
    },
    {
      "epoch": 1.3475017349063152,
      "grad_norm": 0.759671151638031,
      "learning_rate": 1.4289323260402533e-05,
      "loss": 0.4802,
      "step": 7767
    },
    {
      "epoch": 1.347675225537821,
      "grad_norm": 0.8502726554870605,
      "learning_rate": 1.4282793980437923e-05,
      "loss": 0.5551,
      "step": 7768
    },
    {
      "epoch": 1.3478487161693269,
      "grad_norm": 0.7244731783866882,
      "learning_rate": 1.4276265363961735e-05,
      "loss": 0.4489,
      "step": 7769
    },
    {
      "epoch": 1.3480222068008327,
      "grad_norm": 0.6123337149620056,
      "learning_rate": 1.4269737411731627e-05,
      "loss": 0.5626,
      "step": 7770
    },
    {
      "epoch": 1.3481956974323386,
      "grad_norm": 0.876191258430481,
      "learning_rate": 1.4263210124505164e-05,
      "loss": 0.4436,
      "step": 7771
    },
    {
      "epoch": 1.3483691880638444,
      "grad_norm": 0.8973270058631897,
      "learning_rate": 1.425668350303985e-05,
      "loss": 0.4598,
      "step": 7772
    },
    {
      "epoch": 1.3485426786953505,
      "grad_norm": 0.6726403832435608,
      "learning_rate": 1.425015754809311e-05,
      "loss": 0.6379,
      "step": 7773
    },
    {
      "epoch": 1.3487161693268563,
      "grad_norm": 0.7155255675315857,
      "learning_rate": 1.4243632260422292e-05,
      "loss": 0.5647,
      "step": 7774
    },
    {
      "epoch": 1.3488896599583622,
      "grad_norm": 0.7165016531944275,
      "learning_rate": 1.4237107640784648e-05,
      "loss": 0.4246,
      "step": 7775
    },
    {
      "epoch": 1.3490631505898683,
      "grad_norm": 0.770491898059845,
      "learning_rate": 1.4230583689937381e-05,
      "loss": 0.5459,
      "step": 7776
    },
    {
      "epoch": 1.349236641221374,
      "grad_norm": 0.8517716526985168,
      "learning_rate": 1.4224060408637605e-05,
      "loss": 0.4296,
      "step": 7777
    },
    {
      "epoch": 1.34941013185288,
      "grad_norm": 0.6960669159889221,
      "learning_rate": 1.4217537797642343e-05,
      "loss": 0.4739,
      "step": 7778
    },
    {
      "epoch": 1.3495836224843858,
      "grad_norm": 0.9173792004585266,
      "learning_rate": 1.4211015857708555e-05,
      "loss": 0.4238,
      "step": 7779
    },
    {
      "epoch": 1.3497571131158916,
      "grad_norm": 0.6874178051948547,
      "learning_rate": 1.4204494589593127e-05,
      "loss": 0.4902,
      "step": 7780
    },
    {
      "epoch": 1.3499306037473977,
      "grad_norm": 0.5907986164093018,
      "learning_rate": 1.4197973994052855e-05,
      "loss": 0.574,
      "step": 7781
    },
    {
      "epoch": 1.3501040943789036,
      "grad_norm": 0.6873245239257812,
      "learning_rate": 1.4191454071844457e-05,
      "loss": 0.6541,
      "step": 7782
    },
    {
      "epoch": 1.3502775850104094,
      "grad_norm": 0.6468162536621094,
      "learning_rate": 1.4184934823724588e-05,
      "loss": 0.5618,
      "step": 7783
    },
    {
      "epoch": 1.3504510756419155,
      "grad_norm": 0.8056586384773254,
      "learning_rate": 1.4178416250449804e-05,
      "loss": 0.449,
      "step": 7784
    },
    {
      "epoch": 1.3506245662734213,
      "grad_norm": 0.9317519664764404,
      "learning_rate": 1.41718983527766e-05,
      "loss": 0.4087,
      "step": 7785
    },
    {
      "epoch": 1.3507980569049272,
      "grad_norm": 0.8486621975898743,
      "learning_rate": 1.4165381131461388e-05,
      "loss": 0.4053,
      "step": 7786
    },
    {
      "epoch": 1.350971547536433,
      "grad_norm": 0.7225809693336487,
      "learning_rate": 1.4158864587260488e-05,
      "loss": 0.4386,
      "step": 7787
    },
    {
      "epoch": 1.3511450381679388,
      "grad_norm": 1.2607461214065552,
      "learning_rate": 1.4152348720930156e-05,
      "loss": 0.5332,
      "step": 7788
    },
    {
      "epoch": 1.3513185287994447,
      "grad_norm": 0.5642040967941284,
      "learning_rate": 1.414583353322657e-05,
      "loss": 0.5754,
      "step": 7789
    },
    {
      "epoch": 1.3514920194309508,
      "grad_norm": 1.3561378717422485,
      "learning_rate": 1.4139319024905836e-05,
      "loss": 0.6034,
      "step": 7790
    },
    {
      "epoch": 1.3516655100624566,
      "grad_norm": 0.5607908368110657,
      "learning_rate": 1.4132805196723944e-05,
      "loss": 0.6348,
      "step": 7791
    },
    {
      "epoch": 1.3518390006939625,
      "grad_norm": 1.1165635585784912,
      "learning_rate": 1.412629204943685e-05,
      "loss": 0.4183,
      "step": 7792
    },
    {
      "epoch": 1.3520124913254685,
      "grad_norm": 0.7449413537979126,
      "learning_rate": 1.4119779583800414e-05,
      "loss": 0.4342,
      "step": 7793
    },
    {
      "epoch": 1.3521859819569744,
      "grad_norm": 0.7623336911201477,
      "learning_rate": 1.4113267800570402e-05,
      "loss": 0.489,
      "step": 7794
    },
    {
      "epoch": 1.3523594725884802,
      "grad_norm": 0.8422659635543823,
      "learning_rate": 1.4106756700502522e-05,
      "loss": 0.4011,
      "step": 7795
    },
    {
      "epoch": 1.352532963219986,
      "grad_norm": 0.778367817401886,
      "learning_rate": 1.41002462843524e-05,
      "loss": 0.4458,
      "step": 7796
    },
    {
      "epoch": 1.352706453851492,
      "grad_norm": 0.5561218857765198,
      "learning_rate": 1.4093736552875563e-05,
      "loss": 0.5618,
      "step": 7797
    },
    {
      "epoch": 1.352879944482998,
      "grad_norm": 0.8073115944862366,
      "learning_rate": 1.4087227506827482e-05,
      "loss": 0.4069,
      "step": 7798
    },
    {
      "epoch": 1.3530534351145038,
      "grad_norm": 0.7060088515281677,
      "learning_rate": 1.4080719146963548e-05,
      "loss": 0.478,
      "step": 7799
    },
    {
      "epoch": 1.3532269257460097,
      "grad_norm": 0.6544910669326782,
      "learning_rate": 1.4074211474039046e-05,
      "loss": 0.5046,
      "step": 7800
    },
    {
      "epoch": 1.3534004163775157,
      "grad_norm": 0.7442007660865784,
      "learning_rate": 1.4067704488809204e-05,
      "loss": 0.5033,
      "step": 7801
    },
    {
      "epoch": 1.3535739070090216,
      "grad_norm": 0.929850161075592,
      "learning_rate": 1.406119819202917e-05,
      "loss": 0.4822,
      "step": 7802
    },
    {
      "epoch": 1.3537473976405274,
      "grad_norm": 0.6624704003334045,
      "learning_rate": 1.4054692584454017e-05,
      "loss": 0.5205,
      "step": 7803
    },
    {
      "epoch": 1.3539208882720333,
      "grad_norm": 1.0278087854385376,
      "learning_rate": 1.4048187666838707e-05,
      "loss": 0.4707,
      "step": 7804
    },
    {
      "epoch": 1.3540943789035391,
      "grad_norm": 0.5705901980400085,
      "learning_rate": 1.4041683439938152e-05,
      "loss": 0.4578,
      "step": 7805
    },
    {
      "epoch": 1.3542678695350452,
      "grad_norm": 0.7489475011825562,
      "learning_rate": 1.4035179904507184e-05,
      "loss": 0.4791,
      "step": 7806
    },
    {
      "epoch": 1.354441360166551,
      "grad_norm": 0.8546077609062195,
      "learning_rate": 1.4028677061300535e-05,
      "loss": 0.5017,
      "step": 7807
    },
    {
      "epoch": 1.3546148507980569,
      "grad_norm": 0.6993005871772766,
      "learning_rate": 1.4022174911072868e-05,
      "loss": 0.5601,
      "step": 7808
    },
    {
      "epoch": 1.3547883414295627,
      "grad_norm": 1.622098445892334,
      "learning_rate": 1.4015673454578776e-05,
      "loss": 0.5212,
      "step": 7809
    },
    {
      "epoch": 1.3549618320610688,
      "grad_norm": 0.8998701572418213,
      "learning_rate": 1.4009172692572743e-05,
      "loss": 0.5009,
      "step": 7810
    },
    {
      "epoch": 1.3551353226925746,
      "grad_norm": 0.8079712390899658,
      "learning_rate": 1.4002672625809201e-05,
      "loss": 0.4517,
      "step": 7811
    },
    {
      "epoch": 1.3553088133240805,
      "grad_norm": 0.8419730067253113,
      "learning_rate": 1.39961732550425e-05,
      "loss": 0.5302,
      "step": 7812
    },
    {
      "epoch": 1.3554823039555863,
      "grad_norm": 0.980035126209259,
      "learning_rate": 1.3989674581026878e-05,
      "loss": 0.4274,
      "step": 7813
    },
    {
      "epoch": 1.3556557945870922,
      "grad_norm": 0.8754704594612122,
      "learning_rate": 1.3983176604516526e-05,
      "loss": 0.5271,
      "step": 7814
    },
    {
      "epoch": 1.3558292852185982,
      "grad_norm": 0.7613076567649841,
      "learning_rate": 1.3976679326265542e-05,
      "loss": 0.5587,
      "step": 7815
    },
    {
      "epoch": 1.356002775850104,
      "grad_norm": 0.7278582453727722,
      "learning_rate": 1.3970182747027944e-05,
      "loss": 0.4357,
      "step": 7816
    },
    {
      "epoch": 1.35617626648161,
      "grad_norm": 0.7281306385993958,
      "learning_rate": 1.3963686867557658e-05,
      "loss": 0.4153,
      "step": 7817
    },
    {
      "epoch": 1.356349757113116,
      "grad_norm": 0.8076234459877014,
      "learning_rate": 1.3957191688608544e-05,
      "loss": 0.4236,
      "step": 7818
    },
    {
      "epoch": 1.3565232477446219,
      "grad_norm": 0.7760826349258423,
      "learning_rate": 1.3950697210934387e-05,
      "loss": 0.4036,
      "step": 7819
    },
    {
      "epoch": 1.3566967383761277,
      "grad_norm": 1.1662112474441528,
      "learning_rate": 1.3944203435288857e-05,
      "loss": 0.4236,
      "step": 7820
    },
    {
      "epoch": 1.3568702290076335,
      "grad_norm": 0.9857275485992432,
      "learning_rate": 1.3937710362425574e-05,
      "loss": 0.4192,
      "step": 7821
    },
    {
      "epoch": 1.3570437196391394,
      "grad_norm": 0.9380022287368774,
      "learning_rate": 1.3931217993098076e-05,
      "loss": 0.5424,
      "step": 7822
    },
    {
      "epoch": 1.3572172102706455,
      "grad_norm": 0.9351032972335815,
      "learning_rate": 1.3924726328059794e-05,
      "loss": 0.3819,
      "step": 7823
    },
    {
      "epoch": 1.3573907009021513,
      "grad_norm": 0.7847961187362671,
      "learning_rate": 1.3918235368064102e-05,
      "loss": 0.4368,
      "step": 7824
    },
    {
      "epoch": 1.3575641915336571,
      "grad_norm": 0.6209611296653748,
      "learning_rate": 1.3911745113864288e-05,
      "loss": 0.6069,
      "step": 7825
    },
    {
      "epoch": 1.3577376821651632,
      "grad_norm": 1.0030800104141235,
      "learning_rate": 1.3905255566213542e-05,
      "loss": 0.417,
      "step": 7826
    },
    {
      "epoch": 1.357911172796669,
      "grad_norm": 0.8271725177764893,
      "learning_rate": 1.3898766725864988e-05,
      "loss": 0.3507,
      "step": 7827
    },
    {
      "epoch": 1.358084663428175,
      "grad_norm": 0.965467631816864,
      "learning_rate": 1.3892278593571669e-05,
      "loss": 0.4448,
      "step": 7828
    },
    {
      "epoch": 1.3582581540596808,
      "grad_norm": 0.6716170907020569,
      "learning_rate": 1.3885791170086535e-05,
      "loss": 0.5402,
      "step": 7829
    },
    {
      "epoch": 1.3584316446911866,
      "grad_norm": 0.8228726983070374,
      "learning_rate": 1.3879304456162457e-05,
      "loss": 0.5143,
      "step": 7830
    },
    {
      "epoch": 1.3586051353226924,
      "grad_norm": 0.9074976444244385,
      "learning_rate": 1.3872818452552227e-05,
      "loss": 0.4987,
      "step": 7831
    },
    {
      "epoch": 1.3587786259541985,
      "grad_norm": 0.6851913332939148,
      "learning_rate": 1.3866333160008562e-05,
      "loss": 0.436,
      "step": 7832
    },
    {
      "epoch": 1.3589521165857044,
      "grad_norm": 0.8769930601119995,
      "learning_rate": 1.3859848579284076e-05,
      "loss": 0.3749,
      "step": 7833
    },
    {
      "epoch": 1.3591256072172102,
      "grad_norm": 1.0876624584197998,
      "learning_rate": 1.3853364711131324e-05,
      "loss": 0.4473,
      "step": 7834
    },
    {
      "epoch": 1.3592990978487163,
      "grad_norm": 0.9927643537521362,
      "learning_rate": 1.3846881556302757e-05,
      "loss": 0.465,
      "step": 7835
    },
    {
      "epoch": 1.3594725884802221,
      "grad_norm": 0.7676509618759155,
      "learning_rate": 1.3840399115550748e-05,
      "loss": 0.491,
      "step": 7836
    },
    {
      "epoch": 1.359646079111728,
      "grad_norm": 0.8637543320655823,
      "learning_rate": 1.3833917389627603e-05,
      "loss": 0.4051,
      "step": 7837
    },
    {
      "epoch": 1.3598195697432338,
      "grad_norm": 0.7441905736923218,
      "learning_rate": 1.3827436379285537e-05,
      "loss": 0.3986,
      "step": 7838
    },
    {
      "epoch": 1.3599930603747397,
      "grad_norm": 0.8019585013389587,
      "learning_rate": 1.3820956085276661e-05,
      "loss": 0.3824,
      "step": 7839
    },
    {
      "epoch": 1.3601665510062457,
      "grad_norm": 1.0437533855438232,
      "learning_rate": 1.3814476508353036e-05,
      "loss": 0.4912,
      "step": 7840
    },
    {
      "epoch": 1.3603400416377516,
      "grad_norm": 0.6700484156608582,
      "learning_rate": 1.3807997649266625e-05,
      "loss": 0.5547,
      "step": 7841
    },
    {
      "epoch": 1.3605135322692574,
      "grad_norm": 0.743519127368927,
      "learning_rate": 1.3801519508769295e-05,
      "loss": 0.4938,
      "step": 7842
    },
    {
      "epoch": 1.3606870229007635,
      "grad_norm": 0.8325339555740356,
      "learning_rate": 1.3795042087612847e-05,
      "loss": 0.4314,
      "step": 7843
    },
    {
      "epoch": 1.3608605135322693,
      "grad_norm": 0.788335382938385,
      "learning_rate": 1.3788565386548996e-05,
      "loss": 0.4155,
      "step": 7844
    },
    {
      "epoch": 1.3610340041637752,
      "grad_norm": 3.8572866916656494,
      "learning_rate": 1.3782089406329377e-05,
      "loss": 0.4501,
      "step": 7845
    },
    {
      "epoch": 1.361207494795281,
      "grad_norm": 0.6721601486206055,
      "learning_rate": 1.3775614147705521e-05,
      "loss": 0.527,
      "step": 7846
    },
    {
      "epoch": 1.3613809854267869,
      "grad_norm": 0.9929586052894592,
      "learning_rate": 1.3769139611428895e-05,
      "loss": 0.3642,
      "step": 7847
    },
    {
      "epoch": 1.3615544760582927,
      "grad_norm": 0.8101338148117065,
      "learning_rate": 1.3762665798250887e-05,
      "loss": 0.3476,
      "step": 7848
    },
    {
      "epoch": 1.3617279666897988,
      "grad_norm": 0.8143279552459717,
      "learning_rate": 1.375619270892277e-05,
      "loss": 0.4586,
      "step": 7849
    },
    {
      "epoch": 1.3619014573213046,
      "grad_norm": 0.7993375658988953,
      "learning_rate": 1.3749720344195768e-05,
      "loss": 0.4391,
      "step": 7850
    },
    {
      "epoch": 1.3620749479528105,
      "grad_norm": 0.7568285465240479,
      "learning_rate": 1.3743248704821008e-05,
      "loss": 0.4366,
      "step": 7851
    },
    {
      "epoch": 1.3622484385843165,
      "grad_norm": 0.934909999370575,
      "learning_rate": 1.373677779154952e-05,
      "loss": 0.4734,
      "step": 7852
    },
    {
      "epoch": 1.3624219292158224,
      "grad_norm": 0.7335373759269714,
      "learning_rate": 1.3730307605132268e-05,
      "loss": 0.4347,
      "step": 7853
    },
    {
      "epoch": 1.3625954198473282,
      "grad_norm": 0.8993807435035706,
      "learning_rate": 1.3723838146320128e-05,
      "loss": 0.4152,
      "step": 7854
    },
    {
      "epoch": 1.362768910478834,
      "grad_norm": 0.7619749903678894,
      "learning_rate": 1.3717369415863884e-05,
      "loss": 0.3834,
      "step": 7855
    },
    {
      "epoch": 1.36294240111034,
      "grad_norm": 0.71795654296875,
      "learning_rate": 1.3710901414514235e-05,
      "loss": 0.4917,
      "step": 7856
    },
    {
      "epoch": 1.363115891741846,
      "grad_norm": 0.8929448127746582,
      "learning_rate": 1.370443414302181e-05,
      "loss": 0.5934,
      "step": 7857
    },
    {
      "epoch": 1.3632893823733518,
      "grad_norm": 0.7523466348648071,
      "learning_rate": 1.3697967602137135e-05,
      "loss": 0.6012,
      "step": 7858
    },
    {
      "epoch": 1.3634628730048577,
      "grad_norm": 0.6525294184684753,
      "learning_rate": 1.3691501792610662e-05,
      "loss": 0.6115,
      "step": 7859
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.7265623211860657,
      "learning_rate": 1.368503671519276e-05,
      "loss": 0.4817,
      "step": 7860
    },
    {
      "epoch": 1.3638098542678696,
      "grad_norm": 0.6999167203903198,
      "learning_rate": 1.3678572370633708e-05,
      "loss": 0.6019,
      "step": 7861
    },
    {
      "epoch": 1.3639833448993754,
      "grad_norm": 0.7755110263824463,
      "learning_rate": 1.3672108759683694e-05,
      "loss": 0.3886,
      "step": 7862
    },
    {
      "epoch": 1.3641568355308813,
      "grad_norm": 0.9417155385017395,
      "learning_rate": 1.366564588309283e-05,
      "loss": 0.4704,
      "step": 7863
    },
    {
      "epoch": 1.3643303261623871,
      "grad_norm": 0.7990795373916626,
      "learning_rate": 1.3659183741611154e-05,
      "loss": 0.4494,
      "step": 7864
    },
    {
      "epoch": 1.3645038167938932,
      "grad_norm": 1.172098994255066,
      "learning_rate": 1.3652722335988579e-05,
      "loss": 0.4503,
      "step": 7865
    },
    {
      "epoch": 1.364677307425399,
      "grad_norm": 0.8514444231987,
      "learning_rate": 1.3646261666974976e-05,
      "loss": 0.5867,
      "step": 7866
    },
    {
      "epoch": 1.364850798056905,
      "grad_norm": 0.704356849193573,
      "learning_rate": 1.3639801735320122e-05,
      "loss": 0.6355,
      "step": 7867
    },
    {
      "epoch": 1.3650242886884107,
      "grad_norm": 0.6933794617652893,
      "learning_rate": 1.3633342541773673e-05,
      "loss": 0.5706,
      "step": 7868
    },
    {
      "epoch": 1.3651977793199168,
      "grad_norm": 0.7712671756744385,
      "learning_rate": 1.3626884087085246e-05,
      "loss": 0.402,
      "step": 7869
    },
    {
      "epoch": 1.3653712699514227,
      "grad_norm": 0.6848683953285217,
      "learning_rate": 1.3620426372004353e-05,
      "loss": 0.5499,
      "step": 7870
    },
    {
      "epoch": 1.3655447605829285,
      "grad_norm": 0.7330373525619507,
      "learning_rate": 1.3613969397280405e-05,
      "loss": 0.5287,
      "step": 7871
    },
    {
      "epoch": 1.3657182512144344,
      "grad_norm": 1.203633189201355,
      "learning_rate": 1.360751316366275e-05,
      "loss": 0.5663,
      "step": 7872
    },
    {
      "epoch": 1.3658917418459402,
      "grad_norm": 0.7536115646362305,
      "learning_rate": 1.3601057671900639e-05,
      "loss": 0.5173,
      "step": 7873
    },
    {
      "epoch": 1.3660652324774463,
      "grad_norm": 0.7513604164123535,
      "learning_rate": 1.3594602922743252e-05,
      "loss": 0.3983,
      "step": 7874
    },
    {
      "epoch": 1.3662387231089521,
      "grad_norm": 0.7157559394836426,
      "learning_rate": 1.3588148916939651e-05,
      "loss": 0.4396,
      "step": 7875
    },
    {
      "epoch": 1.366412213740458,
      "grad_norm": 0.6541614532470703,
      "learning_rate": 1.358169565523884e-05,
      "loss": 0.4288,
      "step": 7876
    },
    {
      "epoch": 1.366585704371964,
      "grad_norm": 1.4620234966278076,
      "learning_rate": 1.3575243138389733e-05,
      "loss": 0.517,
      "step": 7877
    },
    {
      "epoch": 1.3667591950034699,
      "grad_norm": 0.8827812671661377,
      "learning_rate": 1.356879136714114e-05,
      "loss": 0.3667,
      "step": 7878
    },
    {
      "epoch": 1.3669326856349757,
      "grad_norm": 0.8430798053741455,
      "learning_rate": 1.3562340342241802e-05,
      "loss": 0.3443,
      "step": 7879
    },
    {
      "epoch": 1.3671061762664816,
      "grad_norm": 0.7463581562042236,
      "learning_rate": 1.3555890064440374e-05,
      "loss": 0.3945,
      "step": 7880
    },
    {
      "epoch": 1.3672796668979874,
      "grad_norm": 0.8372496366500854,
      "learning_rate": 1.3549440534485407e-05,
      "loss": 0.3915,
      "step": 7881
    },
    {
      "epoch": 1.3674531575294935,
      "grad_norm": 0.6953713893890381,
      "learning_rate": 1.3542991753125387e-05,
      "loss": 0.452,
      "step": 7882
    },
    {
      "epoch": 1.3676266481609993,
      "grad_norm": 0.8361397385597229,
      "learning_rate": 1.3536543721108698e-05,
      "loss": 0.5,
      "step": 7883
    },
    {
      "epoch": 1.3678001387925052,
      "grad_norm": 0.558370053768158,
      "learning_rate": 1.3530096439183637e-05,
      "loss": 0.5347,
      "step": 7884
    },
    {
      "epoch": 1.3679736294240112,
      "grad_norm": 0.7438952326774597,
      "learning_rate": 1.3523649908098423e-05,
      "loss": 0.3577,
      "step": 7885
    },
    {
      "epoch": 1.368147120055517,
      "grad_norm": 0.6980698108673096,
      "learning_rate": 1.3517204128601193e-05,
      "loss": 0.4368,
      "step": 7886
    },
    {
      "epoch": 1.368320610687023,
      "grad_norm": 0.9308164715766907,
      "learning_rate": 1.351075910143997e-05,
      "loss": 0.4532,
      "step": 7887
    },
    {
      "epoch": 1.3684941013185288,
      "grad_norm": 0.8904773592948914,
      "learning_rate": 1.3504314827362715e-05,
      "loss": 0.4336,
      "step": 7888
    },
    {
      "epoch": 1.3686675919500346,
      "grad_norm": 0.9000870585441589,
      "learning_rate": 1.3497871307117291e-05,
      "loss": 0.4104,
      "step": 7889
    },
    {
      "epoch": 1.3688410825815405,
      "grad_norm": 0.7551040053367615,
      "learning_rate": 1.3491428541451487e-05,
      "loss": 0.5902,
      "step": 7890
    },
    {
      "epoch": 1.3690145732130465,
      "grad_norm": 1.0482600927352905,
      "learning_rate": 1.3484986531112977e-05,
      "loss": 0.6021,
      "step": 7891
    },
    {
      "epoch": 1.3691880638445524,
      "grad_norm": 0.6290242671966553,
      "learning_rate": 1.3478545276849373e-05,
      "loss": 0.5288,
      "step": 7892
    },
    {
      "epoch": 1.3693615544760582,
      "grad_norm": 0.7598239183425903,
      "learning_rate": 1.3472104779408195e-05,
      "loss": 0.535,
      "step": 7893
    },
    {
      "epoch": 1.3695350451075643,
      "grad_norm": 0.7155095338821411,
      "learning_rate": 1.3465665039536857e-05,
      "loss": 0.5695,
      "step": 7894
    },
    {
      "epoch": 1.3697085357390701,
      "grad_norm": 1.1650316715240479,
      "learning_rate": 1.3459226057982706e-05,
      "loss": 0.3976,
      "step": 7895
    },
    {
      "epoch": 1.369882026370576,
      "grad_norm": 0.7217416167259216,
      "learning_rate": 1.3452787835492998e-05,
      "loss": 0.4963,
      "step": 7896
    },
    {
      "epoch": 1.3700555170020818,
      "grad_norm": 0.7166260480880737,
      "learning_rate": 1.3446350372814884e-05,
      "loss": 0.433,
      "step": 7897
    },
    {
      "epoch": 1.3702290076335877,
      "grad_norm": 0.7416203618049622,
      "learning_rate": 1.3439913670695445e-05,
      "loss": 0.4707,
      "step": 7898
    },
    {
      "epoch": 1.3704024982650937,
      "grad_norm": 0.6370003819465637,
      "learning_rate": 1.3433477729881679e-05,
      "loss": 0.5367,
      "step": 7899
    },
    {
      "epoch": 1.3705759888965996,
      "grad_norm": 0.7328839302062988,
      "learning_rate": 1.3427042551120461e-05,
      "loss": 0.4386,
      "step": 7900
    },
    {
      "epoch": 1.3707494795281054,
      "grad_norm": 1.436739444732666,
      "learning_rate": 1.342060813515862e-05,
      "loss": 0.452,
      "step": 7901
    },
    {
      "epoch": 1.3709229701596115,
      "grad_norm": 0.6831396222114563,
      "learning_rate": 1.3414174482742865e-05,
      "loss": 0.5085,
      "step": 7902
    },
    {
      "epoch": 1.3710964607911174,
      "grad_norm": 0.6140052676200867,
      "learning_rate": 1.3407741594619844e-05,
      "loss": 0.5034,
      "step": 7903
    },
    {
      "epoch": 1.3712699514226232,
      "grad_norm": 1.0730401277542114,
      "learning_rate": 1.3401309471536092e-05,
      "loss": 0.3562,
      "step": 7904
    },
    {
      "epoch": 1.371443442054129,
      "grad_norm": 0.6170622110366821,
      "learning_rate": 1.3394878114238059e-05,
      "loss": 0.4937,
      "step": 7905
    },
    {
      "epoch": 1.371616932685635,
      "grad_norm": 0.8561860918998718,
      "learning_rate": 1.3388447523472122e-05,
      "loss": 0.4633,
      "step": 7906
    },
    {
      "epoch": 1.3717904233171407,
      "grad_norm": 0.9009454250335693,
      "learning_rate": 1.3382017699984551e-05,
      "loss": 0.5392,
      "step": 7907
    },
    {
      "epoch": 1.3719639139486468,
      "grad_norm": 1.0146112442016602,
      "learning_rate": 1.337558864452154e-05,
      "loss": 0.467,
      "step": 7908
    },
    {
      "epoch": 1.3721374045801527,
      "grad_norm": 0.7898365259170532,
      "learning_rate": 1.3369160357829185e-05,
      "loss": 0.5266,
      "step": 7909
    },
    {
      "epoch": 1.3723108952116585,
      "grad_norm": 1.2049388885498047,
      "learning_rate": 1.3362732840653494e-05,
      "loss": 0.3904,
      "step": 7910
    },
    {
      "epoch": 1.3724843858431646,
      "grad_norm": 0.6061626076698303,
      "learning_rate": 1.3356306093740392e-05,
      "loss": 0.465,
      "step": 7911
    },
    {
      "epoch": 1.3726578764746704,
      "grad_norm": 0.7990296483039856,
      "learning_rate": 1.3349880117835716e-05,
      "loss": 0.494,
      "step": 7912
    },
    {
      "epoch": 1.3728313671061763,
      "grad_norm": 0.7539100646972656,
      "learning_rate": 1.3343454913685195e-05,
      "loss": 0.3859,
      "step": 7913
    },
    {
      "epoch": 1.373004857737682,
      "grad_norm": 0.7247158288955688,
      "learning_rate": 1.3337030482034485e-05,
      "loss": 0.4126,
      "step": 7914
    },
    {
      "epoch": 1.373178348369188,
      "grad_norm": 0.939651608467102,
      "learning_rate": 1.3330606823629161e-05,
      "loss": 0.3953,
      "step": 7915
    },
    {
      "epoch": 1.373351839000694,
      "grad_norm": 0.6934583783149719,
      "learning_rate": 1.332418393921468e-05,
      "loss": 0.4112,
      "step": 7916
    },
    {
      "epoch": 1.3735253296321999,
      "grad_norm": 0.6739574670791626,
      "learning_rate": 1.3317761829536428e-05,
      "loss": 0.5436,
      "step": 7917
    },
    {
      "epoch": 1.3736988202637057,
      "grad_norm": 1.6639131307601929,
      "learning_rate": 1.3311340495339704e-05,
      "loss": 0.5269,
      "step": 7918
    },
    {
      "epoch": 1.3738723108952118,
      "grad_norm": 0.8540711402893066,
      "learning_rate": 1.3304919937369718e-05,
      "loss": 0.4225,
      "step": 7919
    },
    {
      "epoch": 1.3740458015267176,
      "grad_norm": 0.7971619367599487,
      "learning_rate": 1.3298500156371565e-05,
      "loss": 0.3648,
      "step": 7920
    },
    {
      "epoch": 1.3742192921582235,
      "grad_norm": 0.9696105122566223,
      "learning_rate": 1.329208115309028e-05,
      "loss": 0.5089,
      "step": 7921
    },
    {
      "epoch": 1.3743927827897293,
      "grad_norm": 1.033799409866333,
      "learning_rate": 1.32856629282708e-05,
      "loss": 0.5649,
      "step": 7922
    },
    {
      "epoch": 1.3745662734212352,
      "grad_norm": 0.7602754831314087,
      "learning_rate": 1.3279245482657953e-05,
      "loss": 0.5388,
      "step": 7923
    },
    {
      "epoch": 1.3747397640527412,
      "grad_norm": 0.8027068972587585,
      "learning_rate": 1.3272828816996498e-05,
      "loss": 0.3746,
      "step": 7924
    },
    {
      "epoch": 1.374913254684247,
      "grad_norm": 0.7596278786659241,
      "learning_rate": 1.3266412932031108e-05,
      "loss": 0.4908,
      "step": 7925
    },
    {
      "epoch": 1.375086745315753,
      "grad_norm": 1.0552160739898682,
      "learning_rate": 1.3259997828506333e-05,
      "loss": 0.4192,
      "step": 7926
    },
    {
      "epoch": 1.3752602359472588,
      "grad_norm": 0.7753838300704956,
      "learning_rate": 1.3253583507166666e-05,
      "loss": 0.4374,
      "step": 7927
    },
    {
      "epoch": 1.3754337265787648,
      "grad_norm": 0.9890931248664856,
      "learning_rate": 1.3247169968756494e-05,
      "loss": 0.5284,
      "step": 7928
    },
    {
      "epoch": 1.3756072172102707,
      "grad_norm": 1.1649266481399536,
      "learning_rate": 1.3240757214020117e-05,
      "loss": 0.4145,
      "step": 7929
    },
    {
      "epoch": 1.3757807078417765,
      "grad_norm": 0.7576749324798584,
      "learning_rate": 1.3234345243701743e-05,
      "loss": 0.6349,
      "step": 7930
    },
    {
      "epoch": 1.3759541984732824,
      "grad_norm": 0.7057588696479797,
      "learning_rate": 1.3227934058545481e-05,
      "loss": 0.5475,
      "step": 7931
    },
    {
      "epoch": 1.3761276891047882,
      "grad_norm": 1.1025787591934204,
      "learning_rate": 1.322152365929537e-05,
      "loss": 0.3412,
      "step": 7932
    },
    {
      "epoch": 1.3763011797362943,
      "grad_norm": 0.7550604343414307,
      "learning_rate": 1.321511404669533e-05,
      "loss": 0.4082,
      "step": 7933
    },
    {
      "epoch": 1.3764746703678001,
      "grad_norm": 0.5956621170043945,
      "learning_rate": 1.320870522148921e-05,
      "loss": 0.6117,
      "step": 7934
    },
    {
      "epoch": 1.376648160999306,
      "grad_norm": 0.9815237522125244,
      "learning_rate": 1.3202297184420774e-05,
      "loss": 0.3808,
      "step": 7935
    },
    {
      "epoch": 1.376821651630812,
      "grad_norm": 0.6306245923042297,
      "learning_rate": 1.3195889936233662e-05,
      "loss": 0.4378,
      "step": 7936
    },
    {
      "epoch": 1.376995142262318,
      "grad_norm": 0.6237532496452332,
      "learning_rate": 1.318948347767145e-05,
      "loss": 0.521,
      "step": 7937
    },
    {
      "epoch": 1.3771686328938237,
      "grad_norm": 0.72686368227005,
      "learning_rate": 1.3183077809477625e-05,
      "loss": 0.4746,
      "step": 7938
    },
    {
      "epoch": 1.3773421235253296,
      "grad_norm": 0.8104560971260071,
      "learning_rate": 1.3176672932395556e-05,
      "loss": 0.4727,
      "step": 7939
    },
    {
      "epoch": 1.3775156141568354,
      "grad_norm": 0.7111600041389465,
      "learning_rate": 1.3170268847168541e-05,
      "loss": 0.4973,
      "step": 7940
    },
    {
      "epoch": 1.3776891047883415,
      "grad_norm": 0.9007184505462646,
      "learning_rate": 1.3163865554539794e-05,
      "loss": 0.3983,
      "step": 7941
    },
    {
      "epoch": 1.3778625954198473,
      "grad_norm": 0.7875963449478149,
      "learning_rate": 1.315746305525241e-05,
      "loss": 0.5021,
      "step": 7942
    },
    {
      "epoch": 1.3780360860513532,
      "grad_norm": 1.3286610841751099,
      "learning_rate": 1.3151061350049408e-05,
      "loss": 0.4929,
      "step": 7943
    },
    {
      "epoch": 1.3782095766828593,
      "grad_norm": 1.5478802919387817,
      "learning_rate": 1.3144660439673727e-05,
      "loss": 0.4792,
      "step": 7944
    },
    {
      "epoch": 1.378383067314365,
      "grad_norm": 0.7513118386268616,
      "learning_rate": 1.3138260324868179e-05,
      "loss": 0.5482,
      "step": 7945
    },
    {
      "epoch": 1.378556557945871,
      "grad_norm": 0.686283528804779,
      "learning_rate": 1.3131861006375519e-05,
      "loss": 0.4866,
      "step": 7946
    },
    {
      "epoch": 1.3787300485773768,
      "grad_norm": 0.8788372278213501,
      "learning_rate": 1.3125462484938391e-05,
      "loss": 0.3917,
      "step": 7947
    },
    {
      "epoch": 1.3789035392088826,
      "grad_norm": 0.772336483001709,
      "learning_rate": 1.311906476129936e-05,
      "loss": 0.5665,
      "step": 7948
    },
    {
      "epoch": 1.3790770298403885,
      "grad_norm": 0.776218831539154,
      "learning_rate": 1.3112667836200873e-05,
      "loss": 0.5111,
      "step": 7949
    },
    {
      "epoch": 1.3792505204718946,
      "grad_norm": 0.7619222402572632,
      "learning_rate": 1.3106271710385312e-05,
      "loss": 0.5242,
      "step": 7950
    },
    {
      "epoch": 1.3794240111034004,
      "grad_norm": 0.7595698833465576,
      "learning_rate": 1.3099876384594951e-05,
      "loss": 0.5463,
      "step": 7951
    },
    {
      "epoch": 1.3795975017349063,
      "grad_norm": 0.759110689163208,
      "learning_rate": 1.3093481859571981e-05,
      "loss": 0.4614,
      "step": 7952
    },
    {
      "epoch": 1.3797709923664123,
      "grad_norm": 1.032598614692688,
      "learning_rate": 1.3087088136058481e-05,
      "loss": 0.4301,
      "step": 7953
    },
    {
      "epoch": 1.3799444829979182,
      "grad_norm": 0.6489108204841614,
      "learning_rate": 1.3080695214796464e-05,
      "loss": 0.5071,
      "step": 7954
    },
    {
      "epoch": 1.380117973629424,
      "grad_norm": 0.8093173503875732,
      "learning_rate": 1.3074303096527823e-05,
      "loss": 0.3673,
      "step": 7955
    },
    {
      "epoch": 1.3802914642609299,
      "grad_norm": 1.596784234046936,
      "learning_rate": 1.3067911781994384e-05,
      "loss": 0.3483,
      "step": 7956
    },
    {
      "epoch": 1.3804649548924357,
      "grad_norm": 0.7278069853782654,
      "learning_rate": 1.306152127193786e-05,
      "loss": 0.5435,
      "step": 7957
    },
    {
      "epoch": 1.3806384455239418,
      "grad_norm": 0.5162386298179626,
      "learning_rate": 1.3055131567099872e-05,
      "loss": 0.5865,
      "step": 7958
    },
    {
      "epoch": 1.3808119361554476,
      "grad_norm": 0.8442279696464539,
      "learning_rate": 1.3048742668221959e-05,
      "loss": 0.5,
      "step": 7959
    },
    {
      "epoch": 1.3809854267869535,
      "grad_norm": 0.6834975481033325,
      "learning_rate": 1.3042354576045559e-05,
      "loss": 0.4366,
      "step": 7960
    },
    {
      "epoch": 1.3811589174184595,
      "grad_norm": 1.4897080659866333,
      "learning_rate": 1.3035967291312029e-05,
      "loss": 0.4402,
      "step": 7961
    },
    {
      "epoch": 1.3813324080499654,
      "grad_norm": 1.2184573411941528,
      "learning_rate": 1.3029580814762596e-05,
      "loss": 0.5431,
      "step": 7962
    },
    {
      "epoch": 1.3815058986814712,
      "grad_norm": 1.0185627937316895,
      "learning_rate": 1.3023195147138434e-05,
      "loss": 0.4384,
      "step": 7963
    },
    {
      "epoch": 1.381679389312977,
      "grad_norm": 0.6986583471298218,
      "learning_rate": 1.3016810289180615e-05,
      "loss": 0.5607,
      "step": 7964
    },
    {
      "epoch": 1.381852879944483,
      "grad_norm": 0.6124531626701355,
      "learning_rate": 1.3010426241630088e-05,
      "loss": 0.4775,
      "step": 7965
    },
    {
      "epoch": 1.3820263705759888,
      "grad_norm": 1.0028398036956787,
      "learning_rate": 1.3004043005227742e-05,
      "loss": 0.4349,
      "step": 7966
    },
    {
      "epoch": 1.3821998612074948,
      "grad_norm": 0.9703394770622253,
      "learning_rate": 1.2997660580714364e-05,
      "loss": 0.3924,
      "step": 7967
    },
    {
      "epoch": 1.3823733518390007,
      "grad_norm": 1.0974111557006836,
      "learning_rate": 1.2991278968830625e-05,
      "loss": 0.528,
      "step": 7968
    },
    {
      "epoch": 1.3825468424705065,
      "grad_norm": 1.0687956809997559,
      "learning_rate": 1.298489817031713e-05,
      "loss": 0.3678,
      "step": 7969
    },
    {
      "epoch": 1.3827203331020126,
      "grad_norm": 0.835202157497406,
      "learning_rate": 1.2978518185914388e-05,
      "loss": 0.3934,
      "step": 7970
    },
    {
      "epoch": 1.3828938237335184,
      "grad_norm": 0.6531467437744141,
      "learning_rate": 1.2972139016362782e-05,
      "loss": 0.4285,
      "step": 7971
    },
    {
      "epoch": 1.3830673143650243,
      "grad_norm": 0.8614683747291565,
      "learning_rate": 1.296576066240263e-05,
      "loss": 0.4725,
      "step": 7972
    },
    {
      "epoch": 1.3832408049965301,
      "grad_norm": 0.8402180075645447,
      "learning_rate": 1.2959383124774152e-05,
      "loss": 0.4806,
      "step": 7973
    },
    {
      "epoch": 1.383414295628036,
      "grad_norm": 0.6812103986740112,
      "learning_rate": 1.2953006404217474e-05,
      "loss": 0.3644,
      "step": 7974
    },
    {
      "epoch": 1.383587786259542,
      "grad_norm": 0.8534111976623535,
      "learning_rate": 1.2946630501472607e-05,
      "loss": 0.365,
      "step": 7975
    },
    {
      "epoch": 1.3837612768910479,
      "grad_norm": 0.6894952058792114,
      "learning_rate": 1.2940255417279486e-05,
      "loss": 0.3879,
      "step": 7976
    },
    {
      "epoch": 1.3839347675225537,
      "grad_norm": 0.7686868906021118,
      "learning_rate": 1.2933881152377956e-05,
      "loss": 0.4373,
      "step": 7977
    },
    {
      "epoch": 1.3841082581540598,
      "grad_norm": 0.5888801217079163,
      "learning_rate": 1.2927507707507751e-05,
      "loss": 0.412,
      "step": 7978
    },
    {
      "epoch": 1.3842817487855656,
      "grad_norm": 0.8832677006721497,
      "learning_rate": 1.2921135083408513e-05,
      "loss": 0.4,
      "step": 7979
    },
    {
      "epoch": 1.3844552394170715,
      "grad_norm": 0.8359270095825195,
      "learning_rate": 1.2914763280819804e-05,
      "loss": 0.5376,
      "step": 7980
    },
    {
      "epoch": 1.3846287300485773,
      "grad_norm": 0.8548023104667664,
      "learning_rate": 1.2908392300481067e-05,
      "loss": 0.4343,
      "step": 7981
    },
    {
      "epoch": 1.3848022206800832,
      "grad_norm": 0.7461701035499573,
      "learning_rate": 1.2902022143131668e-05,
      "loss": 0.3782,
      "step": 7982
    },
    {
      "epoch": 1.3849757113115893,
      "grad_norm": 1.1255135536193848,
      "learning_rate": 1.2895652809510875e-05,
      "loss": 0.4425,
      "step": 7983
    },
    {
      "epoch": 1.385149201943095,
      "grad_norm": 0.6272336840629578,
      "learning_rate": 1.2889284300357847e-05,
      "loss": 0.5726,
      "step": 7984
    },
    {
      "epoch": 1.385322692574601,
      "grad_norm": 0.8208845853805542,
      "learning_rate": 1.2882916616411658e-05,
      "loss": 0.4878,
      "step": 7985
    },
    {
      "epoch": 1.3854961832061068,
      "grad_norm": 0.6497074365615845,
      "learning_rate": 1.28765497584113e-05,
      "loss": 0.5112,
      "step": 7986
    },
    {
      "epoch": 1.3856696738376129,
      "grad_norm": 0.8497748970985413,
      "learning_rate": 1.2870183727095633e-05,
      "loss": 0.5624,
      "step": 7987
    },
    {
      "epoch": 1.3858431644691187,
      "grad_norm": 0.7245793342590332,
      "learning_rate": 1.2863818523203452e-05,
      "loss": 0.4678,
      "step": 7988
    },
    {
      "epoch": 1.3860166551006246,
      "grad_norm": 0.7548524737358093,
      "learning_rate": 1.285745414747345e-05,
      "loss": 0.4174,
      "step": 7989
    },
    {
      "epoch": 1.3861901457321304,
      "grad_norm": 1.084741473197937,
      "learning_rate": 1.2851090600644223e-05,
      "loss": 0.3726,
      "step": 7990
    },
    {
      "epoch": 1.3863636363636362,
      "grad_norm": 0.7011428475379944,
      "learning_rate": 1.284472788345425e-05,
      "loss": 0.4595,
      "step": 7991
    },
    {
      "epoch": 1.3865371269951423,
      "grad_norm": 0.8405042290687561,
      "learning_rate": 1.2838365996641949e-05,
      "loss": 0.5168,
      "step": 7992
    },
    {
      "epoch": 1.3867106176266482,
      "grad_norm": 0.7136561870574951,
      "learning_rate": 1.2832004940945626e-05,
      "loss": 0.477,
      "step": 7993
    },
    {
      "epoch": 1.386884108258154,
      "grad_norm": 0.668769896030426,
      "learning_rate": 1.2825644717103472e-05,
      "loss": 0.5544,
      "step": 7994
    },
    {
      "epoch": 1.38705759888966,
      "grad_norm": 2.888563394546509,
      "learning_rate": 1.281928532585361e-05,
      "loss": 0.4764,
      "step": 7995
    },
    {
      "epoch": 1.387231089521166,
      "grad_norm": 0.8526228666305542,
      "learning_rate": 1.281292676793406e-05,
      "loss": 0.4258,
      "step": 7996
    },
    {
      "epoch": 1.3874045801526718,
      "grad_norm": 0.8274763822555542,
      "learning_rate": 1.2806569044082729e-05,
      "loss": 0.504,
      "step": 7997
    },
    {
      "epoch": 1.3875780707841776,
      "grad_norm": 0.8843476176261902,
      "learning_rate": 1.2800212155037437e-05,
      "loss": 0.4702,
      "step": 7998
    },
    {
      "epoch": 1.3877515614156835,
      "grad_norm": 1.0373793840408325,
      "learning_rate": 1.2793856101535922e-05,
      "loss": 0.5594,
      "step": 7999
    },
    {
      "epoch": 1.3879250520471895,
      "grad_norm": 1.3321309089660645,
      "learning_rate": 1.27875008843158e-05,
      "loss": 0.5873,
      "step": 8000
    },
    {
      "epoch": 1.3880985426786954,
      "grad_norm": 0.6675841808319092,
      "learning_rate": 1.2781146504114604e-05,
      "loss": 0.4432,
      "step": 8001
    },
    {
      "epoch": 1.3882720333102012,
      "grad_norm": 0.6967312693595886,
      "learning_rate": 1.2774792961669764e-05,
      "loss": 0.5146,
      "step": 8002
    },
    {
      "epoch": 1.3884455239417073,
      "grad_norm": 0.7670105695724487,
      "learning_rate": 1.2768440257718626e-05,
      "loss": 0.4835,
      "step": 8003
    },
    {
      "epoch": 1.3886190145732131,
      "grad_norm": 0.7322806715965271,
      "learning_rate": 1.2762088392998417e-05,
      "loss": 0.4613,
      "step": 8004
    },
    {
      "epoch": 1.388792505204719,
      "grad_norm": 0.8436444997787476,
      "learning_rate": 1.275573736824629e-05,
      "loss": 0.4445,
      "step": 8005
    },
    {
      "epoch": 1.3889659958362248,
      "grad_norm": 2.392430067062378,
      "learning_rate": 1.2749387184199283e-05,
      "loss": 0.3987,
      "step": 8006
    },
    {
      "epoch": 1.3891394864677307,
      "grad_norm": 0.6554961204528809,
      "learning_rate": 1.2743037841594334e-05,
      "loss": 0.6082,
      "step": 8007
    },
    {
      "epoch": 1.3893129770992365,
      "grad_norm": 1.2686378955841064,
      "learning_rate": 1.2736689341168304e-05,
      "loss": 0.4048,
      "step": 8008
    },
    {
      "epoch": 1.3894864677307426,
      "grad_norm": 1.6640623807907104,
      "learning_rate": 1.2730341683657945e-05,
      "loss": 0.4293,
      "step": 8009
    },
    {
      "epoch": 1.3896599583622484,
      "grad_norm": 0.7305620908737183,
      "learning_rate": 1.2723994869799898e-05,
      "loss": 0.4161,
      "step": 8010
    },
    {
      "epoch": 1.3898334489937543,
      "grad_norm": 0.7039326429367065,
      "learning_rate": 1.2717648900330723e-05,
      "loss": 0.3781,
      "step": 8011
    },
    {
      "epoch": 1.3900069396252603,
      "grad_norm": 0.8124909400939941,
      "learning_rate": 1.2711303775986888e-05,
      "loss": 0.4572,
      "step": 8012
    },
    {
      "epoch": 1.3901804302567662,
      "grad_norm": 0.8309138417243958,
      "learning_rate": 1.2704959497504734e-05,
      "loss": 0.5268,
      "step": 8013
    },
    {
      "epoch": 1.390353920888272,
      "grad_norm": 0.7094196677207947,
      "learning_rate": 1.2698616065620528e-05,
      "loss": 0.4626,
      "step": 8014
    },
    {
      "epoch": 1.3905274115197779,
      "grad_norm": 0.8486608862876892,
      "learning_rate": 1.269227348107045e-05,
      "loss": 0.4437,
      "step": 8015
    },
    {
      "epoch": 1.3907009021512837,
      "grad_norm": 0.6205337047576904,
      "learning_rate": 1.2685931744590536e-05,
      "loss": 0.4705,
      "step": 8016
    },
    {
      "epoch": 1.3908743927827898,
      "grad_norm": 0.936288058757782,
      "learning_rate": 1.2679590856916769e-05,
      "loss": 0.4498,
      "step": 8017
    },
    {
      "epoch": 1.3910478834142956,
      "grad_norm": 0.8357410430908203,
      "learning_rate": 1.267325081878501e-05,
      "loss": 0.4055,
      "step": 8018
    },
    {
      "epoch": 1.3912213740458015,
      "grad_norm": 1.7587887048721313,
      "learning_rate": 1.266691163093104e-05,
      "loss": 0.4345,
      "step": 8019
    },
    {
      "epoch": 1.3913948646773076,
      "grad_norm": 0.9359058737754822,
      "learning_rate": 1.2660573294090512e-05,
      "loss": 0.41,
      "step": 8020
    },
    {
      "epoch": 1.3915683553088134,
      "grad_norm": 0.8059520721435547,
      "learning_rate": 1.2654235808999007e-05,
      "loss": 0.4606,
      "step": 8021
    },
    {
      "epoch": 1.3917418459403192,
      "grad_norm": 0.7084975242614746,
      "learning_rate": 1.2647899176392003e-05,
      "loss": 0.4529,
      "step": 8022
    },
    {
      "epoch": 1.391915336571825,
      "grad_norm": 0.7561855912208557,
      "learning_rate": 1.2641563397004859e-05,
      "loss": 0.5632,
      "step": 8023
    },
    {
      "epoch": 1.392088827203331,
      "grad_norm": 0.7674224376678467,
      "learning_rate": 1.2635228471572853e-05,
      "loss": 0.4802,
      "step": 8024
    },
    {
      "epoch": 1.3922623178348368,
      "grad_norm": 0.6567704081535339,
      "learning_rate": 1.2628894400831175e-05,
      "loss": 0.5457,
      "step": 8025
    },
    {
      "epoch": 1.3924358084663429,
      "grad_norm": 0.653133749961853,
      "learning_rate": 1.2622561185514886e-05,
      "loss": 0.61,
      "step": 8026
    },
    {
      "epoch": 1.3926092990978487,
      "grad_norm": 1.3273907899856567,
      "learning_rate": 1.2616228826358965e-05,
      "loss": 0.3828,
      "step": 8027
    },
    {
      "epoch": 1.3927827897293545,
      "grad_norm": 0.6018370389938354,
      "learning_rate": 1.2609897324098297e-05,
      "loss": 0.4529,
      "step": 8028
    },
    {
      "epoch": 1.3929562803608606,
      "grad_norm": 1.0048967599868774,
      "learning_rate": 1.2603566679467654e-05,
      "loss": 0.4092,
      "step": 8029
    },
    {
      "epoch": 1.3931297709923665,
      "grad_norm": 0.7821022272109985,
      "learning_rate": 1.2597236893201712e-05,
      "loss": 0.4672,
      "step": 8030
    },
    {
      "epoch": 1.3933032616238723,
      "grad_norm": 0.8553732633590698,
      "learning_rate": 1.259090796603506e-05,
      "loss": 0.4114,
      "step": 8031
    },
    {
      "epoch": 1.3934767522553781,
      "grad_norm": 1.0681912899017334,
      "learning_rate": 1.2584579898702175e-05,
      "loss": 0.462,
      "step": 8032
    },
    {
      "epoch": 1.393650242886884,
      "grad_norm": 0.9106276631355286,
      "learning_rate": 1.2578252691937425e-05,
      "loss": 0.4395,
      "step": 8033
    },
    {
      "epoch": 1.39382373351839,
      "grad_norm": 0.7456284761428833,
      "learning_rate": 1.25719263464751e-05,
      "loss": 0.4952,
      "step": 8034
    },
    {
      "epoch": 1.393997224149896,
      "grad_norm": 0.5790156126022339,
      "learning_rate": 1.2565600863049387e-05,
      "loss": 0.5056,
      "step": 8035
    },
    {
      "epoch": 1.3941707147814018,
      "grad_norm": 0.7689415216445923,
      "learning_rate": 1.2559276242394347e-05,
      "loss": 0.486,
      "step": 8036
    },
    {
      "epoch": 1.3943442054129078,
      "grad_norm": 1.6128205060958862,
      "learning_rate": 1.255295248524397e-05,
      "loss": 0.3565,
      "step": 8037
    },
    {
      "epoch": 1.3945176960444137,
      "grad_norm": 0.8087097406387329,
      "learning_rate": 1.254662959233214e-05,
      "loss": 0.4202,
      "step": 8038
    },
    {
      "epoch": 1.3946911866759195,
      "grad_norm": 0.8839709758758545,
      "learning_rate": 1.2540307564392627e-05,
      "loss": 0.4677,
      "step": 8039
    },
    {
      "epoch": 1.3948646773074254,
      "grad_norm": 0.7029903531074524,
      "learning_rate": 1.2533986402159113e-05,
      "loss": 0.59,
      "step": 8040
    },
    {
      "epoch": 1.3950381679389312,
      "grad_norm": 0.5902888178825378,
      "learning_rate": 1.2527666106365182e-05,
      "loss": 0.6255,
      "step": 8041
    },
    {
      "epoch": 1.3952116585704373,
      "grad_norm": 0.8269442915916443,
      "learning_rate": 1.25213466777443e-05,
      "loss": 0.4564,
      "step": 8042
    },
    {
      "epoch": 1.3953851492019431,
      "grad_norm": 0.6043368577957153,
      "learning_rate": 1.2515028117029848e-05,
      "loss": 0.4556,
      "step": 8043
    },
    {
      "epoch": 1.395558639833449,
      "grad_norm": 2.77972674369812,
      "learning_rate": 1.2508710424955117e-05,
      "loss": 0.5302,
      "step": 8044
    },
    {
      "epoch": 1.3957321304649548,
      "grad_norm": 0.7138830423355103,
      "learning_rate": 1.2502393602253258e-05,
      "loss": 0.5239,
      "step": 8045
    },
    {
      "epoch": 1.3959056210964609,
      "grad_norm": 0.6649674773216248,
      "learning_rate": 1.249607764965736e-05,
      "loss": 0.6154,
      "step": 8046
    },
    {
      "epoch": 1.3960791117279667,
      "grad_norm": 2.444833517074585,
      "learning_rate": 1.2489762567900395e-05,
      "loss": 0.4222,
      "step": 8047
    },
    {
      "epoch": 1.3962526023594726,
      "grad_norm": 1.1484051942825317,
      "learning_rate": 1.2483448357715242e-05,
      "loss": 0.3544,
      "step": 8048
    },
    {
      "epoch": 1.3964260929909784,
      "grad_norm": 1.0488779544830322,
      "learning_rate": 1.2477135019834659e-05,
      "loss": 0.51,
      "step": 8049
    },
    {
      "epoch": 1.3965995836224843,
      "grad_norm": 1.3062036037445068,
      "learning_rate": 1.2470822554991321e-05,
      "loss": 0.3418,
      "step": 8050
    },
    {
      "epoch": 1.3967730742539903,
      "grad_norm": 0.8650899529457092,
      "learning_rate": 1.2464510963917803e-05,
      "loss": 0.4396,
      "step": 8051
    },
    {
      "epoch": 1.3969465648854962,
      "grad_norm": 0.8141747713088989,
      "learning_rate": 1.2458200247346569e-05,
      "loss": 0.4513,
      "step": 8052
    },
    {
      "epoch": 1.397120055517002,
      "grad_norm": 0.8654581904411316,
      "learning_rate": 1.2451890406009981e-05,
      "loss": 0.3401,
      "step": 8053
    },
    {
      "epoch": 1.397293546148508,
      "grad_norm": 0.8306574821472168,
      "learning_rate": 1.2445581440640312e-05,
      "loss": 0.4193,
      "step": 8054
    },
    {
      "epoch": 1.397467036780014,
      "grad_norm": 1.0073142051696777,
      "learning_rate": 1.2439273351969712e-05,
      "loss": 0.5129,
      "step": 8055
    },
    {
      "epoch": 1.3976405274115198,
      "grad_norm": 0.6483321189880371,
      "learning_rate": 1.243296614073025e-05,
      "loss": 0.6597,
      "step": 8056
    },
    {
      "epoch": 1.3978140180430256,
      "grad_norm": 0.6348764896392822,
      "learning_rate": 1.2426659807653896e-05,
      "loss": 0.4689,
      "step": 8057
    },
    {
      "epoch": 1.3979875086745315,
      "grad_norm": 0.7256723642349243,
      "learning_rate": 1.2420354353472483e-05,
      "loss": 0.6156,
      "step": 8058
    },
    {
      "epoch": 1.3981609993060375,
      "grad_norm": 0.8737062215805054,
      "learning_rate": 1.2414049778917782e-05,
      "loss": 0.4009,
      "step": 8059
    },
    {
      "epoch": 1.3983344899375434,
      "grad_norm": 0.6082401275634766,
      "learning_rate": 1.2407746084721444e-05,
      "loss": 0.4656,
      "step": 8060
    },
    {
      "epoch": 1.3985079805690492,
      "grad_norm": 0.8343032598495483,
      "learning_rate": 1.2401443271615028e-05,
      "loss": 0.4202,
      "step": 8061
    },
    {
      "epoch": 1.3986814712005553,
      "grad_norm": 4.274116039276123,
      "learning_rate": 1.2395141340329966e-05,
      "loss": 0.4369,
      "step": 8062
    },
    {
      "epoch": 1.3988549618320612,
      "grad_norm": 0.8232946991920471,
      "learning_rate": 1.2388840291597611e-05,
      "loss": 0.3423,
      "step": 8063
    },
    {
      "epoch": 1.399028452463567,
      "grad_norm": 0.7979413270950317,
      "learning_rate": 1.2382540126149218e-05,
      "loss": 0.3979,
      "step": 8064
    },
    {
      "epoch": 1.3992019430950728,
      "grad_norm": 1.2604224681854248,
      "learning_rate": 1.2376240844715912e-05,
      "loss": 0.4266,
      "step": 8065
    },
    {
      "epoch": 1.3993754337265787,
      "grad_norm": 1.2358559370040894,
      "learning_rate": 1.2369942448028738e-05,
      "loss": 0.5542,
      "step": 8066
    },
    {
      "epoch": 1.3995489243580845,
      "grad_norm": 0.8104420900344849,
      "learning_rate": 1.2363644936818645e-05,
      "loss": 0.4565,
      "step": 8067
    },
    {
      "epoch": 1.3997224149895906,
      "grad_norm": 0.9057925343513489,
      "learning_rate": 1.2357348311816444e-05,
      "loss": 0.4686,
      "step": 8068
    },
    {
      "epoch": 1.3998959056210964,
      "grad_norm": 0.8612111806869507,
      "learning_rate": 1.2351052573752878e-05,
      "loss": 0.4067,
      "step": 8069
    },
    {
      "epoch": 1.4000693962526023,
      "grad_norm": 0.8033817410469055,
      "learning_rate": 1.2344757723358583e-05,
      "loss": 0.519,
      "step": 8070
    },
    {
      "epoch": 1.4002428868841084,
      "grad_norm": 0.6826604008674622,
      "learning_rate": 1.2338463761364063e-05,
      "loss": 0.5049,
      "step": 8071
    },
    {
      "epoch": 1.4004163775156142,
      "grad_norm": 0.7676969766616821,
      "learning_rate": 1.2332170688499753e-05,
      "loss": 0.5406,
      "step": 8072
    },
    {
      "epoch": 1.40058986814712,
      "grad_norm": 0.7458460330963135,
      "learning_rate": 1.2325878505495971e-05,
      "loss": 0.5112,
      "step": 8073
    },
    {
      "epoch": 1.400763358778626,
      "grad_norm": 0.8685405254364014,
      "learning_rate": 1.2319587213082931e-05,
      "loss": 0.5762,
      "step": 8074
    },
    {
      "epoch": 1.4009368494101317,
      "grad_norm": 0.9314450621604919,
      "learning_rate": 1.2313296811990747e-05,
      "loss": 0.4232,
      "step": 8075
    },
    {
      "epoch": 1.4011103400416378,
      "grad_norm": 2.373419761657715,
      "learning_rate": 1.230700730294942e-05,
      "loss": 0.5314,
      "step": 8076
    },
    {
      "epoch": 1.4012838306731437,
      "grad_norm": 0.9044128656387329,
      "learning_rate": 1.2300718686688863e-05,
      "loss": 0.4974,
      "step": 8077
    },
    {
      "epoch": 1.4014573213046495,
      "grad_norm": 0.7750056982040405,
      "learning_rate": 1.229443096393887e-05,
      "loss": 0.535,
      "step": 8078
    },
    {
      "epoch": 1.4016308119361556,
      "grad_norm": 0.7632639408111572,
      "learning_rate": 1.2288144135429146e-05,
      "loss": 0.4065,
      "step": 8079
    },
    {
      "epoch": 1.4018043025676614,
      "grad_norm": 0.8691059350967407,
      "learning_rate": 1.2281858201889283e-05,
      "loss": 0.505,
      "step": 8080
    },
    {
      "epoch": 1.4019777931991673,
      "grad_norm": 0.6620309352874756,
      "learning_rate": 1.2275573164048765e-05,
      "loss": 0.4806,
      "step": 8081
    },
    {
      "epoch": 1.4021512838306731,
      "grad_norm": 0.7421913743019104,
      "learning_rate": 1.226928902263698e-05,
      "loss": 0.5088,
      "step": 8082
    },
    {
      "epoch": 1.402324774462179,
      "grad_norm": 0.8207002282142639,
      "learning_rate": 1.2263005778383222e-05,
      "loss": 0.4769,
      "step": 8083
    },
    {
      "epoch": 1.4024982650936848,
      "grad_norm": 0.8392714262008667,
      "learning_rate": 1.2256723432016648e-05,
      "loss": 0.5535,
      "step": 8084
    },
    {
      "epoch": 1.4026717557251909,
      "grad_norm": 0.8502095341682434,
      "learning_rate": 1.2250441984266343e-05,
      "loss": 0.5287,
      "step": 8085
    },
    {
      "epoch": 1.4028452463566967,
      "grad_norm": 0.666627824306488,
      "learning_rate": 1.2244161435861286e-05,
      "loss": 0.5157,
      "step": 8086
    },
    {
      "epoch": 1.4030187369882026,
      "grad_norm": 0.8406854271888733,
      "learning_rate": 1.2237881787530324e-05,
      "loss": 0.4941,
      "step": 8087
    },
    {
      "epoch": 1.4031922276197086,
      "grad_norm": 0.7217394113540649,
      "learning_rate": 1.2231603040002225e-05,
      "loss": 0.4692,
      "step": 8088
    },
    {
      "epoch": 1.4033657182512145,
      "grad_norm": 6.5075297355651855,
      "learning_rate": 1.2225325194005642e-05,
      "loss": 0.4331,
      "step": 8089
    },
    {
      "epoch": 1.4035392088827203,
      "grad_norm": 0.8759097456932068,
      "learning_rate": 1.2219048250269141e-05,
      "loss": 0.4757,
      "step": 8090
    },
    {
      "epoch": 1.4037126995142262,
      "grad_norm": 0.802541196346283,
      "learning_rate": 1.2212772209521146e-05,
      "loss": 0.5006,
      "step": 8091
    },
    {
      "epoch": 1.403886190145732,
      "grad_norm": 0.6343971490859985,
      "learning_rate": 1.2206497072490014e-05,
      "loss": 0.4535,
      "step": 8092
    },
    {
      "epoch": 1.404059680777238,
      "grad_norm": 1.4083843231201172,
      "learning_rate": 1.2200222839903983e-05,
      "loss": 0.4753,
      "step": 8093
    },
    {
      "epoch": 1.404233171408744,
      "grad_norm": 1.0108699798583984,
      "learning_rate": 1.2193949512491172e-05,
      "loss": 0.4808,
      "step": 8094
    },
    {
      "epoch": 1.4044066620402498,
      "grad_norm": 0.9857318997383118,
      "learning_rate": 1.218767709097962e-05,
      "loss": 0.3835,
      "step": 8095
    },
    {
      "epoch": 1.4045801526717558,
      "grad_norm": 0.6723509430885315,
      "learning_rate": 1.2181405576097247e-05,
      "loss": 0.4135,
      "step": 8096
    },
    {
      "epoch": 1.4047536433032617,
      "grad_norm": 0.8481411337852478,
      "learning_rate": 1.2175134968571863e-05,
      "loss": 0.491,
      "step": 8097
    },
    {
      "epoch": 1.4049271339347675,
      "grad_norm": 1.2493733167648315,
      "learning_rate": 1.2168865269131182e-05,
      "loss": 0.4707,
      "step": 8098
    },
    {
      "epoch": 1.4051006245662734,
      "grad_norm": 0.6661719679832458,
      "learning_rate": 1.2162596478502815e-05,
      "loss": 0.5647,
      "step": 8099
    },
    {
      "epoch": 1.4052741151977792,
      "grad_norm": 0.7146643996238708,
      "learning_rate": 1.2156328597414258e-05,
      "loss": 0.5984,
      "step": 8100
    },
    {
      "epoch": 1.4054476058292853,
      "grad_norm": 0.996522068977356,
      "learning_rate": 1.2150061626592908e-05,
      "loss": 0.5195,
      "step": 8101
    },
    {
      "epoch": 1.4056210964607911,
      "grad_norm": 1.018540382385254,
      "learning_rate": 1.2143795566766054e-05,
      "loss": 0.5179,
      "step": 8102
    },
    {
      "epoch": 1.405794587092297,
      "grad_norm": 0.928839921951294,
      "learning_rate": 1.2137530418660877e-05,
      "loss": 0.3743,
      "step": 8103
    },
    {
      "epoch": 1.4059680777238028,
      "grad_norm": 0.7133684754371643,
      "learning_rate": 1.2131266183004455e-05,
      "loss": 0.5493,
      "step": 8104
    },
    {
      "epoch": 1.406141568355309,
      "grad_norm": 0.6732192635536194,
      "learning_rate": 1.212500286052376e-05,
      "loss": 0.408,
      "step": 8105
    },
    {
      "epoch": 1.4063150589868147,
      "grad_norm": 0.6635958552360535,
      "learning_rate": 1.2118740451945668e-05,
      "loss": 0.5432,
      "step": 8106
    },
    {
      "epoch": 1.4064885496183206,
      "grad_norm": 0.9566813111305237,
      "learning_rate": 1.2112478957996922e-05,
      "loss": 0.5143,
      "step": 8107
    },
    {
      "epoch": 1.4066620402498264,
      "grad_norm": 0.7447433471679688,
      "learning_rate": 1.2106218379404187e-05,
      "loss": 0.4912,
      "step": 8108
    },
    {
      "epoch": 1.4068355308813323,
      "grad_norm": 0.7607671618461609,
      "learning_rate": 1.2099958716894008e-05,
      "loss": 0.4187,
      "step": 8109
    },
    {
      "epoch": 1.4070090215128384,
      "grad_norm": 0.8422075510025024,
      "learning_rate": 1.2093699971192825e-05,
      "loss": 0.4257,
      "step": 8110
    },
    {
      "epoch": 1.4071825121443442,
      "grad_norm": 0.897028386592865,
      "learning_rate": 1.2087442143026968e-05,
      "loss": 0.6093,
      "step": 8111
    },
    {
      "epoch": 1.40735600277585,
      "grad_norm": 0.6197953224182129,
      "learning_rate": 1.208118523312268e-05,
      "loss": 0.4557,
      "step": 8112
    },
    {
      "epoch": 1.4075294934073561,
      "grad_norm": 1.1465601921081543,
      "learning_rate": 1.2074929242206066e-05,
      "loss": 0.3566,
      "step": 8113
    },
    {
      "epoch": 1.407702984038862,
      "grad_norm": 0.7143644094467163,
      "learning_rate": 1.2068674171003146e-05,
      "loss": 0.4926,
      "step": 8114
    },
    {
      "epoch": 1.4078764746703678,
      "grad_norm": 0.6614172458648682,
      "learning_rate": 1.2062420020239838e-05,
      "loss": 0.5656,
      "step": 8115
    },
    {
      "epoch": 1.4080499653018737,
      "grad_norm": 1.2392299175262451,
      "learning_rate": 1.205616679064193e-05,
      "loss": 0.5135,
      "step": 8116
    },
    {
      "epoch": 1.4082234559333795,
      "grad_norm": 0.8379489183425903,
      "learning_rate": 1.204991448293512e-05,
      "loss": 0.3663,
      "step": 8117
    },
    {
      "epoch": 1.4083969465648856,
      "grad_norm": 0.696752667427063,
      "learning_rate": 1.2043663097844999e-05,
      "loss": 0.3864,
      "step": 8118
    },
    {
      "epoch": 1.4085704371963914,
      "grad_norm": 0.7961304783821106,
      "learning_rate": 1.2037412636097056e-05,
      "loss": 0.5007,
      "step": 8119
    },
    {
      "epoch": 1.4087439278278973,
      "grad_norm": 0.6261388659477234,
      "learning_rate": 1.2031163098416644e-05,
      "loss": 0.5752,
      "step": 8120
    },
    {
      "epoch": 1.4089174184594033,
      "grad_norm": 0.8483225107192993,
      "learning_rate": 1.2024914485529042e-05,
      "loss": 0.3861,
      "step": 8121
    },
    {
      "epoch": 1.4090909090909092,
      "grad_norm": 0.7405794858932495,
      "learning_rate": 1.2018666798159408e-05,
      "loss": 0.3491,
      "step": 8122
    },
    {
      "epoch": 1.409264399722415,
      "grad_norm": 0.8894197344779968,
      "learning_rate": 1.2012420037032797e-05,
      "loss": 0.4347,
      "step": 8123
    },
    {
      "epoch": 1.4094378903539209,
      "grad_norm": 0.7734290361404419,
      "learning_rate": 1.2006174202874141e-05,
      "loss": 0.3259,
      "step": 8124
    },
    {
      "epoch": 1.4096113809854267,
      "grad_norm": 0.8192522525787354,
      "learning_rate": 1.1999929296408288e-05,
      "loss": 0.4576,
      "step": 8125
    },
    {
      "epoch": 1.4097848716169326,
      "grad_norm": 0.8636189699172974,
      "learning_rate": 1.1993685318359956e-05,
      "loss": 0.5552,
      "step": 8126
    },
    {
      "epoch": 1.4099583622484386,
      "grad_norm": 0.7944595217704773,
      "learning_rate": 1.1987442269453779e-05,
      "loss": 0.5653,
      "step": 8127
    },
    {
      "epoch": 1.4101318528799445,
      "grad_norm": 0.7815333008766174,
      "learning_rate": 1.1981200150414262e-05,
      "loss": 0.5242,
      "step": 8128
    },
    {
      "epoch": 1.4103053435114503,
      "grad_norm": 0.762512743473053,
      "learning_rate": 1.1974958961965806e-05,
      "loss": 0.5216,
      "step": 8129
    },
    {
      "epoch": 1.4104788341429564,
      "grad_norm": 0.6927229166030884,
      "learning_rate": 1.1968718704832716e-05,
      "loss": 0.5153,
      "step": 8130
    },
    {
      "epoch": 1.4106523247744622,
      "grad_norm": 0.5905316472053528,
      "learning_rate": 1.1962479379739179e-05,
      "loss": 0.5563,
      "step": 8131
    },
    {
      "epoch": 1.410825815405968,
      "grad_norm": 0.772675096988678,
      "learning_rate": 1.195624098740928e-05,
      "loss": 0.5208,
      "step": 8132
    },
    {
      "epoch": 1.410999306037474,
      "grad_norm": 0.6897779107093811,
      "learning_rate": 1.1950003528566981e-05,
      "loss": 0.4327,
      "step": 8133
    },
    {
      "epoch": 1.4111727966689798,
      "grad_norm": 1.1361875534057617,
      "learning_rate": 1.1943767003936152e-05,
      "loss": 0.4803,
      "step": 8134
    },
    {
      "epoch": 1.4113462873004858,
      "grad_norm": 0.6257690191268921,
      "learning_rate": 1.193753141424056e-05,
      "loss": 0.5568,
      "step": 8135
    },
    {
      "epoch": 1.4115197779319917,
      "grad_norm": 0.8639114499092102,
      "learning_rate": 1.1931296760203831e-05,
      "loss": 0.5244,
      "step": 8136
    },
    {
      "epoch": 1.4116932685634975,
      "grad_norm": 0.7131174802780151,
      "learning_rate": 1.1925063042549519e-05,
      "loss": 0.4972,
      "step": 8137
    },
    {
      "epoch": 1.4118667591950036,
      "grad_norm": 0.8591305613517761,
      "learning_rate": 1.1918830262001055e-05,
      "loss": 0.4943,
      "step": 8138
    },
    {
      "epoch": 1.4120402498265094,
      "grad_norm": 0.687491238117218,
      "learning_rate": 1.1912598419281747e-05,
      "loss": 0.4995,
      "step": 8139
    },
    {
      "epoch": 1.4122137404580153,
      "grad_norm": 0.7206785082817078,
      "learning_rate": 1.1906367515114816e-05,
      "loss": 0.5199,
      "step": 8140
    },
    {
      "epoch": 1.4123872310895211,
      "grad_norm": 1.3990131616592407,
      "learning_rate": 1.1900137550223376e-05,
      "loss": 0.3428,
      "step": 8141
    },
    {
      "epoch": 1.412560721721027,
      "grad_norm": 0.663081169128418,
      "learning_rate": 1.1893908525330401e-05,
      "loss": 0.3266,
      "step": 8142
    },
    {
      "epoch": 1.4127342123525328,
      "grad_norm": 1.1218020915985107,
      "learning_rate": 1.1887680441158791e-05,
      "loss": 0.4956,
      "step": 8143
    },
    {
      "epoch": 1.412907702984039,
      "grad_norm": 0.7569759488105774,
      "learning_rate": 1.1881453298431323e-05,
      "loss": 0.3933,
      "step": 8144
    },
    {
      "epoch": 1.4130811936155447,
      "grad_norm": 0.8735617399215698,
      "learning_rate": 1.1875227097870652e-05,
      "loss": 0.4637,
      "step": 8145
    },
    {
      "epoch": 1.4132546842470506,
      "grad_norm": 0.9186040163040161,
      "learning_rate": 1.1869001840199345e-05,
      "loss": 0.3828,
      "step": 8146
    },
    {
      "epoch": 1.4134281748785567,
      "grad_norm": 0.6188430786132812,
      "learning_rate": 1.1862777526139848e-05,
      "loss": 0.431,
      "step": 8147
    },
    {
      "epoch": 1.4136016655100625,
      "grad_norm": 0.6487597227096558,
      "learning_rate": 1.1856554156414503e-05,
      "loss": 0.4985,
      "step": 8148
    },
    {
      "epoch": 1.4137751561415683,
      "grad_norm": 0.7717430591583252,
      "learning_rate": 1.1850331731745541e-05,
      "loss": 0.3992,
      "step": 8149
    },
    {
      "epoch": 1.4139486467730742,
      "grad_norm": 0.8516520857810974,
      "learning_rate": 1.1844110252855072e-05,
      "loss": 0.4482,
      "step": 8150
    },
    {
      "epoch": 1.41412213740458,
      "grad_norm": 0.9156954884529114,
      "learning_rate": 1.1837889720465117e-05,
      "loss": 0.5394,
      "step": 8151
    },
    {
      "epoch": 1.414295628036086,
      "grad_norm": 1.0913176536560059,
      "learning_rate": 1.1831670135297564e-05,
      "loss": 0.3937,
      "step": 8152
    },
    {
      "epoch": 1.414469118667592,
      "grad_norm": 1.863344430923462,
      "learning_rate": 1.1825451498074209e-05,
      "loss": 0.431,
      "step": 8153
    },
    {
      "epoch": 1.4146426092990978,
      "grad_norm": 0.9412972927093506,
      "learning_rate": 1.1819233809516746e-05,
      "loss": 0.3961,
      "step": 8154
    },
    {
      "epoch": 1.4148160999306039,
      "grad_norm": 0.7449471950531006,
      "learning_rate": 1.181301707034672e-05,
      "loss": 0.5623,
      "step": 8155
    },
    {
      "epoch": 1.4149895905621097,
      "grad_norm": 1.3430240154266357,
      "learning_rate": 1.1806801281285604e-05,
      "loss": 0.4568,
      "step": 8156
    },
    {
      "epoch": 1.4151630811936156,
      "grad_norm": 2.3827271461486816,
      "learning_rate": 1.1800586443054754e-05,
      "loss": 0.447,
      "step": 8157
    },
    {
      "epoch": 1.4153365718251214,
      "grad_norm": 1.2478195428848267,
      "learning_rate": 1.1794372556375392e-05,
      "loss": 0.4565,
      "step": 8158
    },
    {
      "epoch": 1.4155100624566272,
      "grad_norm": 0.7910694479942322,
      "learning_rate": 1.178815962196866e-05,
      "loss": 0.4298,
      "step": 8159
    },
    {
      "epoch": 1.4156835530881333,
      "grad_norm": 0.7412468194961548,
      "learning_rate": 1.178194764055557e-05,
      "loss": 0.5117,
      "step": 8160
    },
    {
      "epoch": 1.4158570437196392,
      "grad_norm": 0.6640676856040955,
      "learning_rate": 1.1775736612857038e-05,
      "loss": 0.4578,
      "step": 8161
    },
    {
      "epoch": 1.416030534351145,
      "grad_norm": 0.8816553950309753,
      "learning_rate": 1.176952653959385e-05,
      "loss": 0.5081,
      "step": 8162
    },
    {
      "epoch": 1.4162040249826509,
      "grad_norm": 0.6530722975730896,
      "learning_rate": 1.17633174214867e-05,
      "loss": 0.604,
      "step": 8163
    },
    {
      "epoch": 1.416377515614157,
      "grad_norm": 1.0684694051742554,
      "learning_rate": 1.1757109259256163e-05,
      "loss": 0.4286,
      "step": 8164
    },
    {
      "epoch": 1.4165510062456628,
      "grad_norm": 0.7710397243499756,
      "learning_rate": 1.17509020536227e-05,
      "loss": 0.5076,
      "step": 8165
    },
    {
      "epoch": 1.4167244968771686,
      "grad_norm": 1.031315565109253,
      "learning_rate": 1.174469580530666e-05,
      "loss": 0.514,
      "step": 8166
    },
    {
      "epoch": 1.4168979875086745,
      "grad_norm": 0.7741178274154663,
      "learning_rate": 1.1738490515028303e-05,
      "loss": 0.4115,
      "step": 8167
    },
    {
      "epoch": 1.4170714781401803,
      "grad_norm": 0.9059246778488159,
      "learning_rate": 1.1732286183507738e-05,
      "loss": 0.4238,
      "step": 8168
    },
    {
      "epoch": 1.4172449687716864,
      "grad_norm": 0.7780495285987854,
      "learning_rate": 1.1726082811464994e-05,
      "loss": 0.5532,
      "step": 8169
    },
    {
      "epoch": 1.4174184594031922,
      "grad_norm": 0.7769747376441956,
      "learning_rate": 1.1719880399619987e-05,
      "loss": 0.4147,
      "step": 8170
    },
    {
      "epoch": 1.417591950034698,
      "grad_norm": 0.7542340159416199,
      "learning_rate": 1.171367894869251e-05,
      "loss": 0.517,
      "step": 8171
    },
    {
      "epoch": 1.4177654406662041,
      "grad_norm": 0.9824731349945068,
      "learning_rate": 1.1707478459402236e-05,
      "loss": 0.3795,
      "step": 8172
    },
    {
      "epoch": 1.41793893129771,
      "grad_norm": 1.423109531402588,
      "learning_rate": 1.170127893246876e-05,
      "loss": 0.4789,
      "step": 8173
    },
    {
      "epoch": 1.4181124219292158,
      "grad_norm": 0.9541971683502197,
      "learning_rate": 1.1695080368611526e-05,
      "loss": 0.4475,
      "step": 8174
    },
    {
      "epoch": 1.4182859125607217,
      "grad_norm": 0.9733325839042664,
      "learning_rate": 1.1688882768549895e-05,
      "loss": 0.5002,
      "step": 8175
    },
    {
      "epoch": 1.4184594031922275,
      "grad_norm": 0.6772567629814148,
      "learning_rate": 1.1682686133003105e-05,
      "loss": 0.4119,
      "step": 8176
    },
    {
      "epoch": 1.4186328938237336,
      "grad_norm": 0.6227893829345703,
      "learning_rate": 1.1676490462690282e-05,
      "loss": 0.4996,
      "step": 8177
    },
    {
      "epoch": 1.4188063844552394,
      "grad_norm": 0.6760939955711365,
      "learning_rate": 1.167029575833044e-05,
      "loss": 0.4785,
      "step": 8178
    },
    {
      "epoch": 1.4189798750867453,
      "grad_norm": 0.7768404483795166,
      "learning_rate": 1.1664102020642472e-05,
      "loss": 0.3849,
      "step": 8179
    },
    {
      "epoch": 1.4191533657182513,
      "grad_norm": 0.6399132609367371,
      "learning_rate": 1.1657909250345194e-05,
      "loss": 0.5537,
      "step": 8180
    },
    {
      "epoch": 1.4193268563497572,
      "grad_norm": 1.3256938457489014,
      "learning_rate": 1.1651717448157252e-05,
      "loss": 0.5266,
      "step": 8181
    },
    {
      "epoch": 1.419500346981263,
      "grad_norm": 1.1580417156219482,
      "learning_rate": 1.1645526614797235e-05,
      "loss": 0.3728,
      "step": 8182
    },
    {
      "epoch": 1.4196738376127689,
      "grad_norm": 2.0219199657440186,
      "learning_rate": 1.1639336750983593e-05,
      "loss": 0.4918,
      "step": 8183
    },
    {
      "epoch": 1.4198473282442747,
      "grad_norm": 5.228805065155029,
      "learning_rate": 1.1633147857434658e-05,
      "loss": 0.4169,
      "step": 8184
    },
    {
      "epoch": 1.4200208188757806,
      "grad_norm": 0.7683346271514893,
      "learning_rate": 1.162695993486866e-05,
      "loss": 0.392,
      "step": 8185
    },
    {
      "epoch": 1.4201943095072866,
      "grad_norm": 0.7288727760314941,
      "learning_rate": 1.1620772984003724e-05,
      "loss": 0.4781,
      "step": 8186
    },
    {
      "epoch": 1.4203678001387925,
      "grad_norm": 0.7506224513053894,
      "learning_rate": 1.1614587005557847e-05,
      "loss": 0.3904,
      "step": 8187
    },
    {
      "epoch": 1.4205412907702983,
      "grad_norm": 0.6392959356307983,
      "learning_rate": 1.1608402000248908e-05,
      "loss": 0.5076,
      "step": 8188
    },
    {
      "epoch": 1.4207147814018044,
      "grad_norm": 0.8153432607650757,
      "learning_rate": 1.160221796879471e-05,
      "loss": 0.4668,
      "step": 8189
    },
    {
      "epoch": 1.4208882720333103,
      "grad_norm": 0.6652383804321289,
      "learning_rate": 1.1596034911912896e-05,
      "loss": 0.5255,
      "step": 8190
    },
    {
      "epoch": 1.421061762664816,
      "grad_norm": 0.6237618327140808,
      "learning_rate": 1.1589852830321024e-05,
      "loss": 0.5607,
      "step": 8191
    },
    {
      "epoch": 1.421235253296322,
      "grad_norm": 0.9686141014099121,
      "learning_rate": 1.1583671724736526e-05,
      "loss": 0.5532,
      "step": 8192
    },
    {
      "epoch": 1.4214087439278278,
      "grad_norm": 2.3930442333221436,
      "learning_rate": 1.1577491595876744e-05,
      "loss": 0.4435,
      "step": 8193
    },
    {
      "epoch": 1.4215822345593339,
      "grad_norm": 0.7133120894432068,
      "learning_rate": 1.157131244445886e-05,
      "loss": 0.4365,
      "step": 8194
    },
    {
      "epoch": 1.4217557251908397,
      "grad_norm": 0.7545367479324341,
      "learning_rate": 1.1565134271199999e-05,
      "loss": 0.5017,
      "step": 8195
    },
    {
      "epoch": 1.4219292158223455,
      "grad_norm": 0.6678562164306641,
      "learning_rate": 1.1558957076817135e-05,
      "loss": 0.6106,
      "step": 8196
    },
    {
      "epoch": 1.4221027064538516,
      "grad_norm": 0.6045330762863159,
      "learning_rate": 1.1552780862027136e-05,
      "loss": 0.4875,
      "step": 8197
    },
    {
      "epoch": 1.4222761970853575,
      "grad_norm": 0.8746786117553711,
      "learning_rate": 1.1546605627546752e-05,
      "loss": 0.3557,
      "step": 8198
    },
    {
      "epoch": 1.4224496877168633,
      "grad_norm": 0.7303048372268677,
      "learning_rate": 1.154043137409265e-05,
      "loss": 0.4138,
      "step": 8199
    },
    {
      "epoch": 1.4226231783483692,
      "grad_norm": 1.039693832397461,
      "learning_rate": 1.1534258102381332e-05,
      "loss": 0.5621,
      "step": 8200
    },
    {
      "epoch": 1.422796668979875,
      "grad_norm": 0.7501853108406067,
      "learning_rate": 1.1528085813129233e-05,
      "loss": 0.5427,
      "step": 8201
    },
    {
      "epoch": 1.4229701596113808,
      "grad_norm": 1.3978708982467651,
      "learning_rate": 1.1521914507052646e-05,
      "loss": 0.4719,
      "step": 8202
    },
    {
      "epoch": 1.423143650242887,
      "grad_norm": 0.7245519757270813,
      "learning_rate": 1.151574418486776e-05,
      "loss": 0.3704,
      "step": 8203
    },
    {
      "epoch": 1.4233171408743928,
      "grad_norm": 1.0005288124084473,
      "learning_rate": 1.1509574847290641e-05,
      "loss": 0.3488,
      "step": 8204
    },
    {
      "epoch": 1.4234906315058986,
      "grad_norm": 1.342585563659668,
      "learning_rate": 1.150340649503726e-05,
      "loss": 0.5226,
      "step": 8205
    },
    {
      "epoch": 1.4236641221374047,
      "grad_norm": 1.026921272277832,
      "learning_rate": 1.1497239128823456e-05,
      "loss": 0.3259,
      "step": 8206
    },
    {
      "epoch": 1.4238376127689105,
      "grad_norm": 0.8890343308448792,
      "learning_rate": 1.149107274936496e-05,
      "loss": 0.4583,
      "step": 8207
    },
    {
      "epoch": 1.4240111034004164,
      "grad_norm": 1.0013446807861328,
      "learning_rate": 1.1484907357377378e-05,
      "loss": 0.3878,
      "step": 8208
    },
    {
      "epoch": 1.4241845940319222,
      "grad_norm": 0.8156121969223022,
      "learning_rate": 1.1478742953576225e-05,
      "loss": 0.4158,
      "step": 8209
    },
    {
      "epoch": 1.424358084663428,
      "grad_norm": 0.8877331018447876,
      "learning_rate": 1.1472579538676883e-05,
      "loss": 0.3945,
      "step": 8210
    },
    {
      "epoch": 1.4245315752949341,
      "grad_norm": 0.7141983509063721,
      "learning_rate": 1.1466417113394624e-05,
      "loss": 0.5415,
      "step": 8211
    },
    {
      "epoch": 1.42470506592644,
      "grad_norm": 0.8154292702674866,
      "learning_rate": 1.1460255678444598e-05,
      "loss": 0.5631,
      "step": 8212
    },
    {
      "epoch": 1.4248785565579458,
      "grad_norm": 0.8014858365058899,
      "learning_rate": 1.1454095234541848e-05,
      "loss": 0.4602,
      "step": 8213
    },
    {
      "epoch": 1.4250520471894519,
      "grad_norm": 0.803516149520874,
      "learning_rate": 1.144793578240131e-05,
      "loss": 0.4619,
      "step": 8214
    },
    {
      "epoch": 1.4252255378209577,
      "grad_norm": 0.6020379066467285,
      "learning_rate": 1.1441777322737791e-05,
      "loss": 0.5419,
      "step": 8215
    },
    {
      "epoch": 1.4253990284524636,
      "grad_norm": 0.6872659921646118,
      "learning_rate": 1.1435619856265982e-05,
      "loss": 0.5328,
      "step": 8216
    },
    {
      "epoch": 1.4255725190839694,
      "grad_norm": 0.6719970703125,
      "learning_rate": 1.1429463383700465e-05,
      "loss": 0.529,
      "step": 8217
    },
    {
      "epoch": 1.4257460097154753,
      "grad_norm": 0.7338549494743347,
      "learning_rate": 1.1423307905755714e-05,
      "loss": 0.4821,
      "step": 8218
    },
    {
      "epoch": 1.4259195003469813,
      "grad_norm": 1.0120731592178345,
      "learning_rate": 1.1417153423146074e-05,
      "loss": 0.4324,
      "step": 8219
    },
    {
      "epoch": 1.4260929909784872,
      "grad_norm": 0.7835046648979187,
      "learning_rate": 1.1410999936585782e-05,
      "loss": 0.5813,
      "step": 8220
    },
    {
      "epoch": 1.426266481609993,
      "grad_norm": 0.609506368637085,
      "learning_rate": 1.1404847446788948e-05,
      "loss": 0.4217,
      "step": 8221
    },
    {
      "epoch": 1.4264399722414989,
      "grad_norm": 0.7258155345916748,
      "learning_rate": 1.1398695954469598e-05,
      "loss": 0.4923,
      "step": 8222
    },
    {
      "epoch": 1.426613462873005,
      "grad_norm": 1.3275835514068604,
      "learning_rate": 1.1392545460341588e-05,
      "loss": 0.4771,
      "step": 8223
    },
    {
      "epoch": 1.4267869535045108,
      "grad_norm": 0.7754116654396057,
      "learning_rate": 1.1386395965118715e-05,
      "loss": 0.4517,
      "step": 8224
    },
    {
      "epoch": 1.4269604441360166,
      "grad_norm": 0.7081223130226135,
      "learning_rate": 1.1380247469514628e-05,
      "loss": 0.5043,
      "step": 8225
    },
    {
      "epoch": 1.4271339347675225,
      "grad_norm": 1.3863879442214966,
      "learning_rate": 1.1374099974242867e-05,
      "loss": 0.5066,
      "step": 8226
    },
    {
      "epoch": 1.4273074253990283,
      "grad_norm": 0.8166904449462891,
      "learning_rate": 1.1367953480016848e-05,
      "loss": 0.3375,
      "step": 8227
    },
    {
      "epoch": 1.4274809160305344,
      "grad_norm": 1.2641361951828003,
      "learning_rate": 1.13618079875499e-05,
      "loss": 0.377,
      "step": 8228
    },
    {
      "epoch": 1.4276544066620402,
      "grad_norm": 0.9420011639595032,
      "learning_rate": 1.1355663497555186e-05,
      "loss": 0.3428,
      "step": 8229
    },
    {
      "epoch": 1.427827897293546,
      "grad_norm": 1.1718153953552246,
      "learning_rate": 1.1349520010745802e-05,
      "loss": 0.3717,
      "step": 8230
    },
    {
      "epoch": 1.4280013879250522,
      "grad_norm": 1.6591047048568726,
      "learning_rate": 1.1343377527834703e-05,
      "loss": 0.5181,
      "step": 8231
    },
    {
      "epoch": 1.428174878556558,
      "grad_norm": 1.0140023231506348,
      "learning_rate": 1.1337236049534726e-05,
      "loss": 0.3816,
      "step": 8232
    },
    {
      "epoch": 1.4283483691880638,
      "grad_norm": 1.085556149482727,
      "learning_rate": 1.133109557655859e-05,
      "loss": 0.4036,
      "step": 8233
    },
    {
      "epoch": 1.4285218598195697,
      "grad_norm": 1.0881836414337158,
      "learning_rate": 1.1324956109618927e-05,
      "loss": 0.4258,
      "step": 8234
    },
    {
      "epoch": 1.4286953504510755,
      "grad_norm": 0.7150824069976807,
      "learning_rate": 1.1318817649428213e-05,
      "loss": 0.4537,
      "step": 8235
    },
    {
      "epoch": 1.4288688410825816,
      "grad_norm": 0.6293311715126038,
      "learning_rate": 1.1312680196698817e-05,
      "loss": 0.6064,
      "step": 8236
    },
    {
      "epoch": 1.4290423317140875,
      "grad_norm": 0.8089287877082825,
      "learning_rate": 1.1306543752143018e-05,
      "loss": 0.4554,
      "step": 8237
    },
    {
      "epoch": 1.4292158223455933,
      "grad_norm": 1.0295014381408691,
      "learning_rate": 1.1300408316472944e-05,
      "loss": 0.3726,
      "step": 8238
    },
    {
      "epoch": 1.4293893129770994,
      "grad_norm": 2.3466145992279053,
      "learning_rate": 1.1294273890400625e-05,
      "loss": 0.3907,
      "step": 8239
    },
    {
      "epoch": 1.4295628036086052,
      "grad_norm": 0.6298247575759888,
      "learning_rate": 1.1288140474637953e-05,
      "loss": 0.5355,
      "step": 8240
    },
    {
      "epoch": 1.429736294240111,
      "grad_norm": 0.9479881525039673,
      "learning_rate": 1.1282008069896749e-05,
      "loss": 0.4679,
      "step": 8241
    },
    {
      "epoch": 1.429909784871617,
      "grad_norm": 0.9374826550483704,
      "learning_rate": 1.127587667688865e-05,
      "loss": 0.489,
      "step": 8242
    },
    {
      "epoch": 1.4300832755031228,
      "grad_norm": 0.7527461647987366,
      "learning_rate": 1.1269746296325236e-05,
      "loss": 0.4746,
      "step": 8243
    },
    {
      "epoch": 1.4302567661346286,
      "grad_norm": 0.6833534240722656,
      "learning_rate": 1.126361692891794e-05,
      "loss": 0.5043,
      "step": 8244
    },
    {
      "epoch": 1.4304302567661347,
      "grad_norm": 0.6030701994895935,
      "learning_rate": 1.1257488575378077e-05,
      "loss": 0.4475,
      "step": 8245
    },
    {
      "epoch": 1.4306037473976405,
      "grad_norm": 0.8309581875801086,
      "learning_rate": 1.1251361236416845e-05,
      "loss": 0.4567,
      "step": 8246
    },
    {
      "epoch": 1.4307772380291464,
      "grad_norm": 0.63411545753479,
      "learning_rate": 1.124523491274534e-05,
      "loss": 0.5254,
      "step": 8247
    },
    {
      "epoch": 1.4309507286606524,
      "grad_norm": 0.8961930274963379,
      "learning_rate": 1.1239109605074527e-05,
      "loss": 0.3545,
      "step": 8248
    },
    {
      "epoch": 1.4311242192921583,
      "grad_norm": 0.7756201028823853,
      "learning_rate": 1.1232985314115255e-05,
      "loss": 0.4612,
      "step": 8249
    },
    {
      "epoch": 1.4312977099236641,
      "grad_norm": 1.0690734386444092,
      "learning_rate": 1.1226862040578244e-05,
      "loss": 0.3627,
      "step": 8250
    },
    {
      "epoch": 1.43147120055517,
      "grad_norm": 0.6681637167930603,
      "learning_rate": 1.1220739785174129e-05,
      "loss": 0.5686,
      "step": 8251
    },
    {
      "epoch": 1.4316446911866758,
      "grad_norm": 1.3204238414764404,
      "learning_rate": 1.1214618548613379e-05,
      "loss": 0.4663,
      "step": 8252
    },
    {
      "epoch": 1.4318181818181819,
      "grad_norm": 0.7250412702560425,
      "learning_rate": 1.120849833160639e-05,
      "loss": 0.5106,
      "step": 8253
    },
    {
      "epoch": 1.4319916724496877,
      "grad_norm": 0.6920293569564819,
      "learning_rate": 1.1202379134863412e-05,
      "loss": 0.5026,
      "step": 8254
    },
    {
      "epoch": 1.4321651630811936,
      "grad_norm": 0.9779270887374878,
      "learning_rate": 1.1196260959094588e-05,
      "loss": 0.4744,
      "step": 8255
    },
    {
      "epoch": 1.4323386537126996,
      "grad_norm": 0.7436539530754089,
      "learning_rate": 1.1190143805009934e-05,
      "loss": 0.4041,
      "step": 8256
    },
    {
      "epoch": 1.4325121443442055,
      "grad_norm": 0.9834410548210144,
      "learning_rate": 1.118402767331936e-05,
      "loss": 0.5546,
      "step": 8257
    },
    {
      "epoch": 1.4326856349757113,
      "grad_norm": 0.6436958909034729,
      "learning_rate": 1.117791256473265e-05,
      "loss": 0.4659,
      "step": 8258
    },
    {
      "epoch": 1.4328591256072172,
      "grad_norm": 1.0978981256484985,
      "learning_rate": 1.1171798479959466e-05,
      "loss": 0.4434,
      "step": 8259
    },
    {
      "epoch": 1.433032616238723,
      "grad_norm": 0.8727421164512634,
      "learning_rate": 1.1165685419709353e-05,
      "loss": 0.5299,
      "step": 8260
    },
    {
      "epoch": 1.433206106870229,
      "grad_norm": 0.7201048135757446,
      "learning_rate": 1.1159573384691736e-05,
      "loss": 0.4921,
      "step": 8261
    },
    {
      "epoch": 1.433379597501735,
      "grad_norm": 1.7645478248596191,
      "learning_rate": 1.1153462375615934e-05,
      "loss": 0.4193,
      "step": 8262
    },
    {
      "epoch": 1.4335530881332408,
      "grad_norm": 0.8043879866600037,
      "learning_rate": 1.1147352393191135e-05,
      "loss": 0.4173,
      "step": 8263
    },
    {
      "epoch": 1.4337265787647466,
      "grad_norm": 0.6321791410446167,
      "learning_rate": 1.1141243438126403e-05,
      "loss": 0.5363,
      "step": 8264
    },
    {
      "epoch": 1.4339000693962527,
      "grad_norm": 1.3232882022857666,
      "learning_rate": 1.1135135511130685e-05,
      "loss": 0.3771,
      "step": 8265
    },
    {
      "epoch": 1.4340735600277585,
      "grad_norm": 1.0101134777069092,
      "learning_rate": 1.1129028612912832e-05,
      "loss": 0.3775,
      "step": 8266
    },
    {
      "epoch": 1.4342470506592644,
      "grad_norm": 0.903439462184906,
      "learning_rate": 1.1122922744181541e-05,
      "loss": 0.3851,
      "step": 8267
    },
    {
      "epoch": 1.4344205412907702,
      "grad_norm": 1.0646275281906128,
      "learning_rate": 1.1116817905645411e-05,
      "loss": 0.4194,
      "step": 8268
    },
    {
      "epoch": 1.434594031922276,
      "grad_norm": 0.6139266490936279,
      "learning_rate": 1.1110714098012907e-05,
      "loss": 0.5493,
      "step": 8269
    },
    {
      "epoch": 1.4347675225537821,
      "grad_norm": 0.7320806980133057,
      "learning_rate": 1.1104611321992404e-05,
      "loss": 0.3738,
      "step": 8270
    },
    {
      "epoch": 1.434941013185288,
      "grad_norm": 2.575880527496338,
      "learning_rate": 1.1098509578292109e-05,
      "loss": 0.4389,
      "step": 8271
    },
    {
      "epoch": 1.4351145038167938,
      "grad_norm": 0.7725823521614075,
      "learning_rate": 1.1092408867620155e-05,
      "loss": 0.5675,
      "step": 8272
    },
    {
      "epoch": 1.4352879944483,
      "grad_norm": 1.3586474657058716,
      "learning_rate": 1.1086309190684532e-05,
      "loss": 0.3756,
      "step": 8273
    },
    {
      "epoch": 1.4354614850798058,
      "grad_norm": 0.8224762678146362,
      "learning_rate": 1.1080210548193113e-05,
      "loss": 0.3847,
      "step": 8274
    },
    {
      "epoch": 1.4356349757113116,
      "grad_norm": 1.038537621498108,
      "learning_rate": 1.107411294085365e-05,
      "loss": 0.4352,
      "step": 8275
    },
    {
      "epoch": 1.4358084663428174,
      "grad_norm": 0.828065812587738,
      "learning_rate": 1.1068016369373784e-05,
      "loss": 0.4152,
      "step": 8276
    },
    {
      "epoch": 1.4359819569743233,
      "grad_norm": 0.7691086530685425,
      "learning_rate": 1.1061920834461026e-05,
      "loss": 0.463,
      "step": 8277
    },
    {
      "epoch": 1.4361554476058294,
      "grad_norm": 0.9992594718933105,
      "learning_rate": 1.1055826336822775e-05,
      "loss": 0.3401,
      "step": 8278
    },
    {
      "epoch": 1.4363289382373352,
      "grad_norm": 0.7811552286148071,
      "learning_rate": 1.1049732877166286e-05,
      "loss": 0.3811,
      "step": 8279
    },
    {
      "epoch": 1.436502428868841,
      "grad_norm": 0.9243087768554688,
      "learning_rate": 1.1043640456198745e-05,
      "loss": 0.4268,
      "step": 8280
    },
    {
      "epoch": 1.436675919500347,
      "grad_norm": 0.9331871867179871,
      "learning_rate": 1.103754907462715e-05,
      "loss": 0.4979,
      "step": 8281
    },
    {
      "epoch": 1.436849410131853,
      "grad_norm": 0.6397832632064819,
      "learning_rate": 1.1031458733158434e-05,
      "loss": 0.421,
      "step": 8282
    },
    {
      "epoch": 1.4370229007633588,
      "grad_norm": 0.9624772071838379,
      "learning_rate": 1.1025369432499385e-05,
      "loss": 0.4974,
      "step": 8283
    },
    {
      "epoch": 1.4371963913948647,
      "grad_norm": 1.1527369022369385,
      "learning_rate": 1.101928117335666e-05,
      "loss": 0.4113,
      "step": 8284
    },
    {
      "epoch": 1.4373698820263705,
      "grad_norm": 0.6153997778892517,
      "learning_rate": 1.101319395643683e-05,
      "loss": 0.5398,
      "step": 8285
    },
    {
      "epoch": 1.4375433726578764,
      "grad_norm": 0.5979682803153992,
      "learning_rate": 1.100710778244631e-05,
      "loss": 0.568,
      "step": 8286
    },
    {
      "epoch": 1.4377168632893824,
      "grad_norm": 0.8310407996177673,
      "learning_rate": 1.1001022652091413e-05,
      "loss": 0.488,
      "step": 8287
    },
    {
      "epoch": 1.4378903539208883,
      "grad_norm": 0.6937515735626221,
      "learning_rate": 1.0994938566078315e-05,
      "loss": 0.4478,
      "step": 8288
    },
    {
      "epoch": 1.438063844552394,
      "grad_norm": 1.24624764919281,
      "learning_rate": 1.0988855525113096e-05,
      "loss": 0.4032,
      "step": 8289
    },
    {
      "epoch": 1.4382373351839002,
      "grad_norm": 1.329052448272705,
      "learning_rate": 1.0982773529901696e-05,
      "loss": 0.4254,
      "step": 8290
    },
    {
      "epoch": 1.438410825815406,
      "grad_norm": 0.8210911750793457,
      "learning_rate": 1.0976692581149933e-05,
      "loss": 0.4188,
      "step": 8291
    },
    {
      "epoch": 1.4385843164469119,
      "grad_norm": 0.8221972584724426,
      "learning_rate": 1.0970612679563501e-05,
      "loss": 0.4941,
      "step": 8292
    },
    {
      "epoch": 1.4387578070784177,
      "grad_norm": 0.7580887079238892,
      "learning_rate": 1.0964533825848008e-05,
      "loss": 0.386,
      "step": 8293
    },
    {
      "epoch": 1.4389312977099236,
      "grad_norm": 0.8661448359489441,
      "learning_rate": 1.0958456020708875e-05,
      "loss": 0.3766,
      "step": 8294
    },
    {
      "epoch": 1.4391047883414296,
      "grad_norm": 1.0701017379760742,
      "learning_rate": 1.0952379264851464e-05,
      "loss": 0.4092,
      "step": 8295
    },
    {
      "epoch": 1.4392782789729355,
      "grad_norm": 0.7596119046211243,
      "learning_rate": 1.0946303558980981e-05,
      "loss": 0.5391,
      "step": 8296
    },
    {
      "epoch": 1.4394517696044413,
      "grad_norm": 0.7729180455207825,
      "learning_rate": 1.094022890380252e-05,
      "loss": 0.4447,
      "step": 8297
    },
    {
      "epoch": 1.4396252602359474,
      "grad_norm": 2.3617026805877686,
      "learning_rate": 1.0934155300021048e-05,
      "loss": 0.4341,
      "step": 8298
    },
    {
      "epoch": 1.4397987508674532,
      "grad_norm": 1.3820737600326538,
      "learning_rate": 1.0928082748341429e-05,
      "loss": 0.3708,
      "step": 8299
    },
    {
      "epoch": 1.439972241498959,
      "grad_norm": 0.9820736646652222,
      "learning_rate": 1.0922011249468362e-05,
      "loss": 0.4109,
      "step": 8300
    },
    {
      "epoch": 1.440145732130465,
      "grad_norm": 0.7744532227516174,
      "learning_rate": 1.0915940804106477e-05,
      "loss": 0.5073,
      "step": 8301
    },
    {
      "epoch": 1.4403192227619708,
      "grad_norm": 0.7360996603965759,
      "learning_rate": 1.0909871412960245e-05,
      "loss": 0.6124,
      "step": 8302
    },
    {
      "epoch": 1.4404927133934766,
      "grad_norm": 0.8232945799827576,
      "learning_rate": 1.090380307673403e-05,
      "loss": 0.5385,
      "step": 8303
    },
    {
      "epoch": 1.4406662040249827,
      "grad_norm": 0.6191974878311157,
      "learning_rate": 1.0897735796132056e-05,
      "loss": 0.6143,
      "step": 8304
    },
    {
      "epoch": 1.4408396946564885,
      "grad_norm": 1.2847371101379395,
      "learning_rate": 1.0891669571858458e-05,
      "loss": 0.4058,
      "step": 8305
    },
    {
      "epoch": 1.4410131852879944,
      "grad_norm": 0.8297262191772461,
      "learning_rate": 1.0885604404617221e-05,
      "loss": 0.3998,
      "step": 8306
    },
    {
      "epoch": 1.4411866759195004,
      "grad_norm": 0.9757704138755798,
      "learning_rate": 1.0879540295112212e-05,
      "loss": 0.3265,
      "step": 8307
    },
    {
      "epoch": 1.4413601665510063,
      "grad_norm": 0.7426693439483643,
      "learning_rate": 1.087347724404717e-05,
      "loss": 0.4854,
      "step": 8308
    },
    {
      "epoch": 1.4415336571825121,
      "grad_norm": 1.0658361911773682,
      "learning_rate": 1.086741525212574e-05,
      "loss": 0.4758,
      "step": 8309
    },
    {
      "epoch": 1.441707147814018,
      "grad_norm": 0.6992179751396179,
      "learning_rate": 1.086135432005141e-05,
      "loss": 0.4962,
      "step": 8310
    },
    {
      "epoch": 1.4418806384455238,
      "grad_norm": 0.7144151926040649,
      "learning_rate": 1.0855294448527548e-05,
      "loss": 0.4519,
      "step": 8311
    },
    {
      "epoch": 1.44205412907703,
      "grad_norm": 0.6744188070297241,
      "learning_rate": 1.0849235638257442e-05,
      "loss": 0.405,
      "step": 8312
    },
    {
      "epoch": 1.4422276197085357,
      "grad_norm": 0.6769935488700867,
      "learning_rate": 1.0843177889944187e-05,
      "loss": 0.5189,
      "step": 8313
    },
    {
      "epoch": 1.4424011103400416,
      "grad_norm": 0.5383853912353516,
      "learning_rate": 1.0837121204290812e-05,
      "loss": 0.6127,
      "step": 8314
    },
    {
      "epoch": 1.4425746009715477,
      "grad_norm": 0.7602445483207703,
      "learning_rate": 1.08310655820002e-05,
      "loss": 0.6104,
      "step": 8315
    },
    {
      "epoch": 1.4427480916030535,
      "grad_norm": 1.0592349767684937,
      "learning_rate": 1.0825011023775111e-05,
      "loss": 0.5861,
      "step": 8316
    },
    {
      "epoch": 1.4429215822345594,
      "grad_norm": 0.5987264513969421,
      "learning_rate": 1.0818957530318177e-05,
      "loss": 0.5736,
      "step": 8317
    },
    {
      "epoch": 1.4430950728660652,
      "grad_norm": 0.6851949691772461,
      "learning_rate": 1.0812905102331927e-05,
      "loss": 0.4696,
      "step": 8318
    },
    {
      "epoch": 1.443268563497571,
      "grad_norm": 0.5215808153152466,
      "learning_rate": 1.0806853740518743e-05,
      "loss": 0.5856,
      "step": 8319
    },
    {
      "epoch": 1.4434420541290771,
      "grad_norm": 5.401055335998535,
      "learning_rate": 1.0800803445580896e-05,
      "loss": 0.4019,
      "step": 8320
    },
    {
      "epoch": 1.443615544760583,
      "grad_norm": 0.598903238773346,
      "learning_rate": 1.079475421822052e-05,
      "loss": 0.4912,
      "step": 8321
    },
    {
      "epoch": 1.4437890353920888,
      "grad_norm": 1.0459188222885132,
      "learning_rate": 1.078870605913966e-05,
      "loss": 0.3865,
      "step": 8322
    },
    {
      "epoch": 1.4439625260235947,
      "grad_norm": 0.783470630645752,
      "learning_rate": 1.0782658969040179e-05,
      "loss": 0.3664,
      "step": 8323
    },
    {
      "epoch": 1.4441360166551007,
      "grad_norm": 0.9712731242179871,
      "learning_rate": 1.0776612948623874e-05,
      "loss": 0.3561,
      "step": 8324
    },
    {
      "epoch": 1.4443095072866066,
      "grad_norm": 1.0471014976501465,
      "learning_rate": 1.0770567998592383e-05,
      "loss": 0.3859,
      "step": 8325
    },
    {
      "epoch": 1.4444829979181124,
      "grad_norm": 1.1087979078292847,
      "learning_rate": 1.0764524119647228e-05,
      "loss": 0.3643,
      "step": 8326
    },
    {
      "epoch": 1.4446564885496183,
      "grad_norm": 0.8103334307670593,
      "learning_rate": 1.0758481312489804e-05,
      "loss": 0.5364,
      "step": 8327
    },
    {
      "epoch": 1.444829979181124,
      "grad_norm": 1.0769774913787842,
      "learning_rate": 1.0752439577821398e-05,
      "loss": 0.5149,
      "step": 8328
    },
    {
      "epoch": 1.4450034698126302,
      "grad_norm": 0.6957473754882812,
      "learning_rate": 1.0746398916343158e-05,
      "loss": 0.3832,
      "step": 8329
    },
    {
      "epoch": 1.445176960444136,
      "grad_norm": 1.0740876197814941,
      "learning_rate": 1.0740359328756105e-05,
      "loss": 0.3268,
      "step": 8330
    },
    {
      "epoch": 1.4453504510756419,
      "grad_norm": 1.0540927648544312,
      "learning_rate": 1.0734320815761143e-05,
      "loss": 0.3591,
      "step": 8331
    },
    {
      "epoch": 1.445523941707148,
      "grad_norm": 0.8353700637817383,
      "learning_rate": 1.0728283378059036e-05,
      "loss": 0.5391,
      "step": 8332
    },
    {
      "epoch": 1.4456974323386538,
      "grad_norm": 1.2079349756240845,
      "learning_rate": 1.0722247016350456e-05,
      "loss": 0.5454,
      "step": 8333
    },
    {
      "epoch": 1.4458709229701596,
      "grad_norm": 0.9697757959365845,
      "learning_rate": 1.0716211731335922e-05,
      "loss": 0.4517,
      "step": 8334
    },
    {
      "epoch": 1.4460444136016655,
      "grad_norm": 0.704738199710846,
      "learning_rate": 1.0710177523715833e-05,
      "loss": 0.5906,
      "step": 8335
    },
    {
      "epoch": 1.4462179042331713,
      "grad_norm": 0.8852899074554443,
      "learning_rate": 1.0704144394190458e-05,
      "loss": 0.4235,
      "step": 8336
    },
    {
      "epoch": 1.4463913948646774,
      "grad_norm": 0.9528529047966003,
      "learning_rate": 1.0698112343459967e-05,
      "loss": 0.4167,
      "step": 8337
    },
    {
      "epoch": 1.4465648854961832,
      "grad_norm": 1.396693468093872,
      "learning_rate": 1.0692081372224378e-05,
      "loss": 0.3618,
      "step": 8338
    },
    {
      "epoch": 1.446738376127689,
      "grad_norm": 0.8811051845550537,
      "learning_rate": 1.0686051481183593e-05,
      "loss": 0.5481,
      "step": 8339
    },
    {
      "epoch": 1.4469118667591951,
      "grad_norm": 0.7508476376533508,
      "learning_rate": 1.0680022671037376e-05,
      "loss": 0.4497,
      "step": 8340
    },
    {
      "epoch": 1.447085357390701,
      "grad_norm": 1.0145187377929688,
      "learning_rate": 1.0673994942485404e-05,
      "loss": 0.5071,
      "step": 8341
    },
    {
      "epoch": 1.4472588480222068,
      "grad_norm": 0.7528433799743652,
      "learning_rate": 1.0667968296227169e-05,
      "loss": 0.4768,
      "step": 8342
    },
    {
      "epoch": 1.4474323386537127,
      "grad_norm": 1.0493375062942505,
      "learning_rate": 1.0661942732962096e-05,
      "loss": 0.3898,
      "step": 8343
    },
    {
      "epoch": 1.4476058292852185,
      "grad_norm": 0.8907434940338135,
      "learning_rate": 1.0655918253389452e-05,
      "loss": 0.4526,
      "step": 8344
    },
    {
      "epoch": 1.4477793199167244,
      "grad_norm": 1.0421587228775024,
      "learning_rate": 1.0649894858208381e-05,
      "loss": 0.4113,
      "step": 8345
    },
    {
      "epoch": 1.4479528105482304,
      "grad_norm": 0.9041519165039062,
      "learning_rate": 1.06438725481179e-05,
      "loss": 0.3333,
      "step": 8346
    },
    {
      "epoch": 1.4481263011797363,
      "grad_norm": 0.6785004734992981,
      "learning_rate": 1.0637851323816918e-05,
      "loss": 0.4388,
      "step": 8347
    },
    {
      "epoch": 1.4482997918112421,
      "grad_norm": 0.6762832403182983,
      "learning_rate": 1.06318311860042e-05,
      "loss": 0.4838,
      "step": 8348
    },
    {
      "epoch": 1.4484732824427482,
      "grad_norm": 0.6441954970359802,
      "learning_rate": 1.0625812135378388e-05,
      "loss": 0.59,
      "step": 8349
    },
    {
      "epoch": 1.448646773074254,
      "grad_norm": 0.781283438205719,
      "learning_rate": 1.0619794172637995e-05,
      "loss": 0.4589,
      "step": 8350
    },
    {
      "epoch": 1.44882026370576,
      "grad_norm": 1.0100747346878052,
      "learning_rate": 1.0613777298481431e-05,
      "loss": 0.3864,
      "step": 8351
    },
    {
      "epoch": 1.4489937543372657,
      "grad_norm": 0.6544135808944702,
      "learning_rate": 1.0607761513606935e-05,
      "loss": 0.5074,
      "step": 8352
    },
    {
      "epoch": 1.4491672449687716,
      "grad_norm": 1.044804573059082,
      "learning_rate": 1.0601746818712666e-05,
      "loss": 0.5249,
      "step": 8353
    },
    {
      "epoch": 1.4493407356002777,
      "grad_norm": 0.859351396560669,
      "learning_rate": 1.0595733214496633e-05,
      "loss": 0.5231,
      "step": 8354
    },
    {
      "epoch": 1.4495142262317835,
      "grad_norm": 0.7557289600372314,
      "learning_rate": 1.0589720701656714e-05,
      "loss": 0.4941,
      "step": 8355
    },
    {
      "epoch": 1.4496877168632893,
      "grad_norm": 0.5786538124084473,
      "learning_rate": 1.0583709280890668e-05,
      "loss": 0.5795,
      "step": 8356
    },
    {
      "epoch": 1.4498612074947954,
      "grad_norm": 1.1952705383300781,
      "learning_rate": 1.057769895289614e-05,
      "loss": 0.3909,
      "step": 8357
    },
    {
      "epoch": 1.4500346981263013,
      "grad_norm": 0.8314424157142639,
      "learning_rate": 1.0571689718370629e-05,
      "loss": 0.4208,
      "step": 8358
    },
    {
      "epoch": 1.450208188757807,
      "grad_norm": 0.8545066118240356,
      "learning_rate": 1.05656815780115e-05,
      "loss": 0.5289,
      "step": 8359
    },
    {
      "epoch": 1.450381679389313,
      "grad_norm": 0.5665410161018372,
      "learning_rate": 1.0559674532516033e-05,
      "loss": 0.5892,
      "step": 8360
    },
    {
      "epoch": 1.4505551700208188,
      "grad_norm": 0.7000844478607178,
      "learning_rate": 1.0553668582581324e-05,
      "loss": 0.4115,
      "step": 8361
    },
    {
      "epoch": 1.4507286606523246,
      "grad_norm": 0.8924859166145325,
      "learning_rate": 1.0547663728904392e-05,
      "loss": 0.3994,
      "step": 8362
    },
    {
      "epoch": 1.4509021512838307,
      "grad_norm": 0.938998281955719,
      "learning_rate": 1.0541659972182088e-05,
      "loss": 0.5281,
      "step": 8363
    },
    {
      "epoch": 1.4510756419153366,
      "grad_norm": 0.7375452518463135,
      "learning_rate": 1.0535657313111183e-05,
      "loss": 0.4556,
      "step": 8364
    },
    {
      "epoch": 1.4512491325468424,
      "grad_norm": 0.7703595757484436,
      "learning_rate": 1.0529655752388254e-05,
      "loss": 0.545,
      "step": 8365
    },
    {
      "epoch": 1.4514226231783485,
      "grad_norm": 0.7838109731674194,
      "learning_rate": 1.0523655290709825e-05,
      "loss": 0.5122,
      "step": 8366
    },
    {
      "epoch": 1.4515961138098543,
      "grad_norm": 0.7033941745758057,
      "learning_rate": 1.051765592877224e-05,
      "loss": 0.4962,
      "step": 8367
    },
    {
      "epoch": 1.4517696044413602,
      "grad_norm": 0.6633094549179077,
      "learning_rate": 1.0511657667271731e-05,
      "loss": 0.5167,
      "step": 8368
    },
    {
      "epoch": 1.451943095072866,
      "grad_norm": 0.8117217421531677,
      "learning_rate": 1.0505660506904398e-05,
      "loss": 0.554,
      "step": 8369
    },
    {
      "epoch": 1.4521165857043719,
      "grad_norm": 1.3190199136734009,
      "learning_rate": 1.0499664448366245e-05,
      "loss": 0.3636,
      "step": 8370
    },
    {
      "epoch": 1.452290076335878,
      "grad_norm": 0.8043152093887329,
      "learning_rate": 1.0493669492353082e-05,
      "loss": 0.3423,
      "step": 8371
    },
    {
      "epoch": 1.4524635669673838,
      "grad_norm": 1.0273507833480835,
      "learning_rate": 1.0487675639560664e-05,
      "loss": 0.317,
      "step": 8372
    },
    {
      "epoch": 1.4526370575988896,
      "grad_norm": 0.7748756408691406,
      "learning_rate": 1.048168289068457e-05,
      "loss": 0.4421,
      "step": 8373
    },
    {
      "epoch": 1.4528105482303957,
      "grad_norm": 1.1627253293991089,
      "learning_rate": 1.0475691246420267e-05,
      "loss": 0.5059,
      "step": 8374
    },
    {
      "epoch": 1.4529840388619015,
      "grad_norm": 0.6672224998474121,
      "learning_rate": 1.0469700707463089e-05,
      "loss": 0.5135,
      "step": 8375
    },
    {
      "epoch": 1.4531575294934074,
      "grad_norm": 0.732526957988739,
      "learning_rate": 1.0463711274508253e-05,
      "loss": 0.5013,
      "step": 8376
    },
    {
      "epoch": 1.4533310201249132,
      "grad_norm": 0.9877192378044128,
      "learning_rate": 1.0457722948250837e-05,
      "loss": 0.5612,
      "step": 8377
    },
    {
      "epoch": 1.453504510756419,
      "grad_norm": 0.9189408421516418,
      "learning_rate": 1.045173572938579e-05,
      "loss": 0.3949,
      "step": 8378
    },
    {
      "epoch": 1.4536780013879251,
      "grad_norm": 0.7369690537452698,
      "learning_rate": 1.0445749618607932e-05,
      "loss": 0.5655,
      "step": 8379
    },
    {
      "epoch": 1.453851492019431,
      "grad_norm": 1.3301423788070679,
      "learning_rate": 1.0439764616611972e-05,
      "loss": 0.3816,
      "step": 8380
    },
    {
      "epoch": 1.4540249826509368,
      "grad_norm": 0.7214325070381165,
      "learning_rate": 1.0433780724092466e-05,
      "loss": 0.3775,
      "step": 8381
    },
    {
      "epoch": 1.4541984732824427,
      "grad_norm": 1.0076757669448853,
      "learning_rate": 1.0427797941743854e-05,
      "loss": 0.5013,
      "step": 8382
    },
    {
      "epoch": 1.4543719639139487,
      "grad_norm": 0.5548766851425171,
      "learning_rate": 1.0421816270260447e-05,
      "loss": 0.6019,
      "step": 8383
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.7440147995948792,
      "learning_rate": 1.041583571033641e-05,
      "loss": 0.603,
      "step": 8384
    },
    {
      "epoch": 1.4547189451769604,
      "grad_norm": 0.7528059482574463,
      "learning_rate": 1.0409856262665818e-05,
      "loss": 0.4189,
      "step": 8385
    },
    {
      "epoch": 1.4548924358084663,
      "grad_norm": 0.8353288769721985,
      "learning_rate": 1.0403877927942582e-05,
      "loss": 0.5216,
      "step": 8386
    },
    {
      "epoch": 1.4550659264399721,
      "grad_norm": 0.7426737546920776,
      "learning_rate": 1.0397900706860493e-05,
      "loss": 0.4143,
      "step": 8387
    },
    {
      "epoch": 1.4552394170714782,
      "grad_norm": 0.7790663242340088,
      "learning_rate": 1.0391924600113211e-05,
      "loss": 0.4835,
      "step": 8388
    },
    {
      "epoch": 1.455412907702984,
      "grad_norm": 0.8680485486984253,
      "learning_rate": 1.0385949608394283e-05,
      "loss": 0.5001,
      "step": 8389
    },
    {
      "epoch": 1.4555863983344899,
      "grad_norm": 0.7560135126113892,
      "learning_rate": 1.0379975732397096e-05,
      "loss": 0.4662,
      "step": 8390
    },
    {
      "epoch": 1.455759888965996,
      "grad_norm": 0.6263082027435303,
      "learning_rate": 1.0374002972814941e-05,
      "loss": 0.5312,
      "step": 8391
    },
    {
      "epoch": 1.4559333795975018,
      "grad_norm": 1.1819356679916382,
      "learning_rate": 1.0368031330340948e-05,
      "loss": 0.3561,
      "step": 8392
    },
    {
      "epoch": 1.4561068702290076,
      "grad_norm": 0.944335401058197,
      "learning_rate": 1.0362060805668162e-05,
      "loss": 0.5069,
      "step": 8393
    },
    {
      "epoch": 1.4562803608605135,
      "grad_norm": 0.9007725715637207,
      "learning_rate": 1.0356091399489431e-05,
      "loss": 0.4227,
      "step": 8394
    },
    {
      "epoch": 1.4564538514920193,
      "grad_norm": 0.6720866560935974,
      "learning_rate": 1.0350123112497541e-05,
      "loss": 0.4968,
      "step": 8395
    },
    {
      "epoch": 1.4566273421235254,
      "grad_norm": 0.7603023648262024,
      "learning_rate": 1.0344155945385106e-05,
      "loss": 0.3953,
      "step": 8396
    },
    {
      "epoch": 1.4568008327550312,
      "grad_norm": 0.8988627791404724,
      "learning_rate": 1.0338189898844626e-05,
      "loss": 0.4646,
      "step": 8397
    },
    {
      "epoch": 1.456974323386537,
      "grad_norm": 0.9700048565864563,
      "learning_rate": 1.0332224973568458e-05,
      "loss": 0.3648,
      "step": 8398
    },
    {
      "epoch": 1.4571478140180432,
      "grad_norm": 0.9551883339881897,
      "learning_rate": 1.0326261170248863e-05,
      "loss": 0.4601,
      "step": 8399
    },
    {
      "epoch": 1.457321304649549,
      "grad_norm": 0.7829773426055908,
      "learning_rate": 1.0320298489577913e-05,
      "loss": 0.4734,
      "step": 8400
    },
    {
      "epoch": 1.4574947952810549,
      "grad_norm": 0.7974841594696045,
      "learning_rate": 1.0314336932247615e-05,
      "loss": 0.3765,
      "step": 8401
    },
    {
      "epoch": 1.4576682859125607,
      "grad_norm": 0.8945306539535522,
      "learning_rate": 1.03083764989498e-05,
      "loss": 0.5234,
      "step": 8402
    },
    {
      "epoch": 1.4578417765440665,
      "grad_norm": 0.5802792906761169,
      "learning_rate": 1.0302417190376184e-05,
      "loss": 0.4392,
      "step": 8403
    },
    {
      "epoch": 1.4580152671755724,
      "grad_norm": 1.2883274555206299,
      "learning_rate": 1.0296459007218345e-05,
      "loss": 0.4371,
      "step": 8404
    },
    {
      "epoch": 1.4581887578070785,
      "grad_norm": 0.8092826008796692,
      "learning_rate": 1.0290501950167755e-05,
      "loss": 0.5762,
      "step": 8405
    },
    {
      "epoch": 1.4583622484385843,
      "grad_norm": 0.9017000198364258,
      "learning_rate": 1.0284546019915727e-05,
      "loss": 0.4286,
      "step": 8406
    },
    {
      "epoch": 1.4585357390700902,
      "grad_norm": 0.9883742332458496,
      "learning_rate": 1.0278591217153444e-05,
      "loss": 0.3618,
      "step": 8407
    },
    {
      "epoch": 1.4587092297015962,
      "grad_norm": 0.8880689144134521,
      "learning_rate": 1.0272637542571988e-05,
      "loss": 0.3259,
      "step": 8408
    },
    {
      "epoch": 1.458882720333102,
      "grad_norm": 0.6883126497268677,
      "learning_rate": 1.026668499686228e-05,
      "loss": 0.5259,
      "step": 8409
    },
    {
      "epoch": 1.459056210964608,
      "grad_norm": 0.6304786205291748,
      "learning_rate": 1.026073358071512e-05,
      "loss": 0.5785,
      "step": 8410
    },
    {
      "epoch": 1.4592297015961138,
      "grad_norm": 0.9625796675682068,
      "learning_rate": 1.0254783294821166e-05,
      "loss": 0.442,
      "step": 8411
    },
    {
      "epoch": 1.4594031922276196,
      "grad_norm": 0.7708973288536072,
      "learning_rate": 1.0248834139870985e-05,
      "loss": 0.5096,
      "step": 8412
    },
    {
      "epoch": 1.4595766828591257,
      "grad_norm": 0.6290324926376343,
      "learning_rate": 1.0242886116554947e-05,
      "loss": 0.5823,
      "step": 8413
    },
    {
      "epoch": 1.4597501734906315,
      "grad_norm": 0.7336987257003784,
      "learning_rate": 1.0236939225563351e-05,
      "loss": 0.6217,
      "step": 8414
    },
    {
      "epoch": 1.4599236641221374,
      "grad_norm": 0.8475745320320129,
      "learning_rate": 1.0230993467586336e-05,
      "loss": 0.3936,
      "step": 8415
    },
    {
      "epoch": 1.4600971547536434,
      "grad_norm": 0.76064532995224,
      "learning_rate": 1.0225048843313914e-05,
      "loss": 0.4185,
      "step": 8416
    },
    {
      "epoch": 1.4602706453851493,
      "grad_norm": 0.7234097719192505,
      "learning_rate": 1.0219105353435951e-05,
      "loss": 0.4302,
      "step": 8417
    },
    {
      "epoch": 1.4604441360166551,
      "grad_norm": 2.1421239376068115,
      "learning_rate": 1.021316299864223e-05,
      "loss": 0.3957,
      "step": 8418
    },
    {
      "epoch": 1.460617626648161,
      "grad_norm": 0.7063815593719482,
      "learning_rate": 1.0207221779622327e-05,
      "loss": 0.4518,
      "step": 8419
    },
    {
      "epoch": 1.4607911172796668,
      "grad_norm": 0.7108697891235352,
      "learning_rate": 1.0201281697065757e-05,
      "loss": 0.4972,
      "step": 8420
    },
    {
      "epoch": 1.4609646079111727,
      "grad_norm": 0.9781704545021057,
      "learning_rate": 1.0195342751661856e-05,
      "loss": 0.3401,
      "step": 8421
    },
    {
      "epoch": 1.4611380985426787,
      "grad_norm": 0.6981971859931946,
      "learning_rate": 1.0189404944099867e-05,
      "loss": 0.4618,
      "step": 8422
    },
    {
      "epoch": 1.4613115891741846,
      "grad_norm": 0.8739231824874878,
      "learning_rate": 1.0183468275068853e-05,
      "loss": 0.3482,
      "step": 8423
    },
    {
      "epoch": 1.4614850798056904,
      "grad_norm": 1.831498146057129,
      "learning_rate": 1.017753274525779e-05,
      "loss": 0.4212,
      "step": 8424
    },
    {
      "epoch": 1.4616585704371965,
      "grad_norm": 0.7566888928413391,
      "learning_rate": 1.01715983553555e-05,
      "loss": 0.4188,
      "step": 8425
    },
    {
      "epoch": 1.4618320610687023,
      "grad_norm": 2.4486005306243896,
      "learning_rate": 1.016566510605067e-05,
      "loss": 0.3857,
      "step": 8426
    },
    {
      "epoch": 1.4620055517002082,
      "grad_norm": 0.7806375622749329,
      "learning_rate": 1.0159732998031857e-05,
      "loss": 0.3994,
      "step": 8427
    },
    {
      "epoch": 1.462179042331714,
      "grad_norm": 0.7226378321647644,
      "learning_rate": 1.0153802031987504e-05,
      "loss": 0.5568,
      "step": 8428
    },
    {
      "epoch": 1.4623525329632199,
      "grad_norm": 0.7658690810203552,
      "learning_rate": 1.01478722086059e-05,
      "loss": 0.4004,
      "step": 8429
    },
    {
      "epoch": 1.462526023594726,
      "grad_norm": 0.7795974612236023,
      "learning_rate": 1.0141943528575205e-05,
      "loss": 0.4243,
      "step": 8430
    },
    {
      "epoch": 1.4626995142262318,
      "grad_norm": 0.6728429198265076,
      "learning_rate": 1.0136015992583449e-05,
      "loss": 0.4589,
      "step": 8431
    },
    {
      "epoch": 1.4628730048577376,
      "grad_norm": 0.896666944026947,
      "learning_rate": 1.0130089601318525e-05,
      "loss": 0.4868,
      "step": 8432
    },
    {
      "epoch": 1.4630464954892437,
      "grad_norm": 0.9360397458076477,
      "learning_rate": 1.0124164355468208e-05,
      "loss": 0.41,
      "step": 8433
    },
    {
      "epoch": 1.4632199861207495,
      "grad_norm": 0.8924311399459839,
      "learning_rate": 1.0118240255720128e-05,
      "loss": 0.4838,
      "step": 8434
    },
    {
      "epoch": 1.4633934767522554,
      "grad_norm": 0.8122832775115967,
      "learning_rate": 1.011231730276178e-05,
      "loss": 0.4663,
      "step": 8435
    },
    {
      "epoch": 1.4635669673837612,
      "grad_norm": 0.73357093334198,
      "learning_rate": 1.0106395497280524e-05,
      "loss": 0.5358,
      "step": 8436
    },
    {
      "epoch": 1.463740458015267,
      "grad_norm": 0.5891599059104919,
      "learning_rate": 1.0100474839963605e-05,
      "loss": 0.4646,
      "step": 8437
    },
    {
      "epoch": 1.4639139486467732,
      "grad_norm": 0.6430240273475647,
      "learning_rate": 1.0094555331498118e-05,
      "loss": 0.4349,
      "step": 8438
    },
    {
      "epoch": 1.464087439278279,
      "grad_norm": 1.0415815114974976,
      "learning_rate": 1.0088636972571027e-05,
      "loss": 0.4403,
      "step": 8439
    },
    {
      "epoch": 1.4642609299097848,
      "grad_norm": 0.7313327789306641,
      "learning_rate": 1.0082719763869153e-05,
      "loss": 0.51,
      "step": 8440
    },
    {
      "epoch": 1.4644344205412907,
      "grad_norm": 0.6586468815803528,
      "learning_rate": 1.0076803706079224e-05,
      "loss": 0.3967,
      "step": 8441
    },
    {
      "epoch": 1.4646079111727968,
      "grad_norm": 0.6369693279266357,
      "learning_rate": 1.0070888799887772e-05,
      "loss": 0.5192,
      "step": 8442
    },
    {
      "epoch": 1.4647814018043026,
      "grad_norm": 0.8158454895019531,
      "learning_rate": 1.0064975045981254e-05,
      "loss": 0.5299,
      "step": 8443
    },
    {
      "epoch": 1.4649548924358085,
      "grad_norm": 0.5639089345932007,
      "learning_rate": 1.0059062445045957e-05,
      "loss": 0.5651,
      "step": 8444
    },
    {
      "epoch": 1.4651283830673143,
      "grad_norm": 0.6220554709434509,
      "learning_rate": 1.0053150997768046e-05,
      "loss": 0.4736,
      "step": 8445
    },
    {
      "epoch": 1.4653018736988201,
      "grad_norm": 1.0816049575805664,
      "learning_rate": 1.0047240704833544e-05,
      "loss": 0.405,
      "step": 8446
    },
    {
      "epoch": 1.4654753643303262,
      "grad_norm": 1.5779757499694824,
      "learning_rate": 1.0041331566928365e-05,
      "loss": 0.5348,
      "step": 8447
    },
    {
      "epoch": 1.465648854961832,
      "grad_norm": 0.7839391827583313,
      "learning_rate": 1.0035423584738262e-05,
      "loss": 0.4214,
      "step": 8448
    },
    {
      "epoch": 1.465822345593338,
      "grad_norm": 0.9254438281059265,
      "learning_rate": 1.0029516758948863e-05,
      "loss": 0.5283,
      "step": 8449
    },
    {
      "epoch": 1.465995836224844,
      "grad_norm": 0.7670827507972717,
      "learning_rate": 1.0023611090245653e-05,
      "loss": 0.53,
      "step": 8450
    },
    {
      "epoch": 1.4661693268563498,
      "grad_norm": 0.9193515181541443,
      "learning_rate": 1.0017706579314016e-05,
      "loss": 0.5215,
      "step": 8451
    },
    {
      "epoch": 1.4663428174878557,
      "grad_norm": 0.755422055721283,
      "learning_rate": 1.0011803226839148e-05,
      "loss": 0.4807,
      "step": 8452
    },
    {
      "epoch": 1.4665163081193615,
      "grad_norm": 0.7575535774230957,
      "learning_rate": 1.000590103350616e-05,
      "loss": 0.4962,
      "step": 8453
    },
    {
      "epoch": 1.4666897987508674,
      "grad_norm": 0.8128533959388733,
      "learning_rate": 1.0000000000000006e-05,
      "loss": 0.5104,
      "step": 8454
    },
    {
      "epoch": 1.4668632893823734,
      "grad_norm": 0.6965898275375366,
      "learning_rate": 9.994100127005492e-06,
      "loss": 0.5612,
      "step": 8455
    },
    {
      "epoch": 1.4670367800138793,
      "grad_norm": 0.7350049614906311,
      "learning_rate": 9.988201415207327e-06,
      "loss": 0.4757,
      "step": 8456
    },
    {
      "epoch": 1.4672102706453851,
      "grad_norm": 0.7166589498519897,
      "learning_rate": 9.982303865290055e-06,
      "loss": 0.3915,
      "step": 8457
    },
    {
      "epoch": 1.4673837612768912,
      "grad_norm": 0.5868585705757141,
      "learning_rate": 9.976407477938092e-06,
      "loss": 0.538,
      "step": 8458
    },
    {
      "epoch": 1.467557251908397,
      "grad_norm": 0.6436400413513184,
      "learning_rate": 9.970512253835713e-06,
      "loss": 0.427,
      "step": 8459
    },
    {
      "epoch": 1.4677307425399029,
      "grad_norm": 0.7440900206565857,
      "learning_rate": 9.96461819366709e-06,
      "loss": 0.3804,
      "step": 8460
    },
    {
      "epoch": 1.4679042331714087,
      "grad_norm": 1.111206293106079,
      "learning_rate": 9.958725298116204e-06,
      "loss": 0.3649,
      "step": 8461
    },
    {
      "epoch": 1.4680777238029146,
      "grad_norm": 0.9442750215530396,
      "learning_rate": 9.952833567866954e-06,
      "loss": 0.4076,
      "step": 8462
    },
    {
      "epoch": 1.4682512144344204,
      "grad_norm": 1.032470703125,
      "learning_rate": 9.946943003603067e-06,
      "loss": 0.4896,
      "step": 8463
    },
    {
      "epoch": 1.4684247050659265,
      "grad_norm": 0.8886162042617798,
      "learning_rate": 9.941053606008176e-06,
      "loss": 0.314,
      "step": 8464
    },
    {
      "epoch": 1.4685981956974323,
      "grad_norm": 0.9300706386566162,
      "learning_rate": 9.935165375765718e-06,
      "loss": 0.3347,
      "step": 8465
    },
    {
      "epoch": 1.4687716863289382,
      "grad_norm": 0.73177570104599,
      "learning_rate": 9.929278313559054e-06,
      "loss": 0.4554,
      "step": 8466
    },
    {
      "epoch": 1.4689451769604442,
      "grad_norm": 0.6601850986480713,
      "learning_rate": 9.923392420071376e-06,
      "loss": 0.5767,
      "step": 8467
    },
    {
      "epoch": 1.46911866759195,
      "grad_norm": 1.0624281167984009,
      "learning_rate": 9.917507695985752e-06,
      "loss": 0.4116,
      "step": 8468
    },
    {
      "epoch": 1.469292158223456,
      "grad_norm": 1.227677345275879,
      "learning_rate": 9.911624141985096e-06,
      "loss": 0.323,
      "step": 8469
    },
    {
      "epoch": 1.4694656488549618,
      "grad_norm": 0.7922270894050598,
      "learning_rate": 9.905741758752234e-06,
      "loss": 0.371,
      "step": 8470
    },
    {
      "epoch": 1.4696391394864676,
      "grad_norm": 0.6427243947982788,
      "learning_rate": 9.899860546969785e-06,
      "loss": 0.3859,
      "step": 8471
    },
    {
      "epoch": 1.4698126301179737,
      "grad_norm": 0.7310827970504761,
      "learning_rate": 9.893980507320295e-06,
      "loss": 0.4584,
      "step": 8472
    },
    {
      "epoch": 1.4699861207494795,
      "grad_norm": 0.6126799583435059,
      "learning_rate": 9.888101640486146e-06,
      "loss": 0.4736,
      "step": 8473
    },
    {
      "epoch": 1.4701596113809854,
      "grad_norm": 0.5600416660308838,
      "learning_rate": 9.882223947149583e-06,
      "loss": 0.6259,
      "step": 8474
    },
    {
      "epoch": 1.4703331020124915,
      "grad_norm": 0.8366650938987732,
      "learning_rate": 9.876347427992712e-06,
      "loss": 0.5151,
      "step": 8475
    },
    {
      "epoch": 1.4705065926439973,
      "grad_norm": 0.9180997014045715,
      "learning_rate": 9.870472083697526e-06,
      "loss": 0.4323,
      "step": 8476
    },
    {
      "epoch": 1.4706800832755031,
      "grad_norm": 0.7303565740585327,
      "learning_rate": 9.864597914945859e-06,
      "loss": 0.4814,
      "step": 8477
    },
    {
      "epoch": 1.470853573907009,
      "grad_norm": 1.0823837518692017,
      "learning_rate": 9.858724922419413e-06,
      "loss": 0.4498,
      "step": 8478
    },
    {
      "epoch": 1.4710270645385148,
      "grad_norm": 0.7902705073356628,
      "learning_rate": 9.852853106799752e-06,
      "loss": 0.4369,
      "step": 8479
    },
    {
      "epoch": 1.4712005551700207,
      "grad_norm": 1.0457099676132202,
      "learning_rate": 9.846982468768316e-06,
      "loss": 0.4037,
      "step": 8480
    },
    {
      "epoch": 1.4713740458015268,
      "grad_norm": 0.9382363557815552,
      "learning_rate": 9.841113009006395e-06,
      "loss": 0.5115,
      "step": 8481
    },
    {
      "epoch": 1.4715475364330326,
      "grad_norm": 0.6638898849487305,
      "learning_rate": 9.83524472819515e-06,
      "loss": 0.5068,
      "step": 8482
    },
    {
      "epoch": 1.4717210270645384,
      "grad_norm": 0.670421838760376,
      "learning_rate": 9.829377627015595e-06,
      "loss": 0.5934,
      "step": 8483
    },
    {
      "epoch": 1.4718945176960445,
      "grad_norm": 0.8659542798995972,
      "learning_rate": 9.823511706148612e-06,
      "loss": 0.5544,
      "step": 8484
    },
    {
      "epoch": 1.4720680083275504,
      "grad_norm": 0.6149436831474304,
      "learning_rate": 9.81764696627496e-06,
      "loss": 0.5089,
      "step": 8485
    },
    {
      "epoch": 1.4722414989590562,
      "grad_norm": 0.7267339825630188,
      "learning_rate": 9.811783408075244e-06,
      "loss": 0.3921,
      "step": 8486
    },
    {
      "epoch": 1.472414989590562,
      "grad_norm": 0.8999056816101074,
      "learning_rate": 9.805921032229935e-06,
      "loss": 0.4633,
      "step": 8487
    },
    {
      "epoch": 1.472588480222068,
      "grad_norm": 0.96303790807724,
      "learning_rate": 9.800059839419358e-06,
      "loss": 0.4055,
      "step": 8488
    },
    {
      "epoch": 1.472761970853574,
      "grad_norm": 0.785336434841156,
      "learning_rate": 9.794199830323741e-06,
      "loss": 0.4161,
      "step": 8489
    },
    {
      "epoch": 1.4729354614850798,
      "grad_norm": 4.219667911529541,
      "learning_rate": 9.788341005623107e-06,
      "loss": 0.4529,
      "step": 8490
    },
    {
      "epoch": 1.4731089521165857,
      "grad_norm": 0.7926281094551086,
      "learning_rate": 9.782483365997409e-06,
      "loss": 0.3579,
      "step": 8491
    },
    {
      "epoch": 1.4732824427480917,
      "grad_norm": 0.8731950521469116,
      "learning_rate": 9.776626912126413e-06,
      "loss": 0.3644,
      "step": 8492
    },
    {
      "epoch": 1.4734559333795976,
      "grad_norm": 0.8293495774269104,
      "learning_rate": 9.770771644689792e-06,
      "loss": 0.5417,
      "step": 8493
    },
    {
      "epoch": 1.4736294240111034,
      "grad_norm": 0.7870762944221497,
      "learning_rate": 9.764917564367025e-06,
      "loss": 0.3804,
      "step": 8494
    },
    {
      "epoch": 1.4738029146426093,
      "grad_norm": 0.8626502752304077,
      "learning_rate": 9.75906467183751e-06,
      "loss": 0.5247,
      "step": 8495
    },
    {
      "epoch": 1.473976405274115,
      "grad_norm": 0.7849081754684448,
      "learning_rate": 9.753212967780472e-06,
      "loss": 0.5177,
      "step": 8496
    },
    {
      "epoch": 1.4741498959056212,
      "grad_norm": 1.1207414865493774,
      "learning_rate": 9.747362452875009e-06,
      "loss": 0.4625,
      "step": 8497
    },
    {
      "epoch": 1.474323386537127,
      "grad_norm": 0.8799973130226135,
      "learning_rate": 9.741513127800072e-06,
      "loss": 0.4646,
      "step": 8498
    },
    {
      "epoch": 1.4744968771686329,
      "grad_norm": 0.6825425028800964,
      "learning_rate": 9.735664993234499e-06,
      "loss": 0.5953,
      "step": 8499
    },
    {
      "epoch": 1.4746703678001387,
      "grad_norm": 0.7112365961074829,
      "learning_rate": 9.729818049856963e-06,
      "loss": 0.3888,
      "step": 8500
    },
    {
      "epoch": 1.4748438584316448,
      "grad_norm": 1.3058369159698486,
      "learning_rate": 9.72397229834601e-06,
      "loss": 0.4067,
      "step": 8501
    },
    {
      "epoch": 1.4750173490631506,
      "grad_norm": 0.6349422931671143,
      "learning_rate": 9.718127739380043e-06,
      "loss": 0.5435,
      "step": 8502
    },
    {
      "epoch": 1.4751908396946565,
      "grad_norm": 0.765207052230835,
      "learning_rate": 9.712284373637329e-06,
      "loss": 0.3588,
      "step": 8503
    },
    {
      "epoch": 1.4753643303261623,
      "grad_norm": 0.6961727738380432,
      "learning_rate": 9.706442201796007e-06,
      "loss": 0.4391,
      "step": 8504
    },
    {
      "epoch": 1.4755378209576682,
      "grad_norm": 0.7369248867034912,
      "learning_rate": 9.700601224534061e-06,
      "loss": 0.5142,
      "step": 8505
    },
    {
      "epoch": 1.4757113115891742,
      "grad_norm": 0.7529562711715698,
      "learning_rate": 9.694761442529345e-06,
      "loss": 0.3741,
      "step": 8506
    },
    {
      "epoch": 1.47588480222068,
      "grad_norm": 0.8767305612564087,
      "learning_rate": 9.688922856459563e-06,
      "loss": 0.3862,
      "step": 8507
    },
    {
      "epoch": 1.476058292852186,
      "grad_norm": 1.0340005159378052,
      "learning_rate": 9.683085467002306e-06,
      "loss": 0.4139,
      "step": 8508
    },
    {
      "epoch": 1.476231783483692,
      "grad_norm": 0.6615685820579529,
      "learning_rate": 9.677249274835003e-06,
      "loss": 0.4683,
      "step": 8509
    },
    {
      "epoch": 1.4764052741151978,
      "grad_norm": 0.6835951209068298,
      "learning_rate": 9.67141428063495e-06,
      "loss": 0.4952,
      "step": 8510
    },
    {
      "epoch": 1.4765787647467037,
      "grad_norm": 0.7067520618438721,
      "learning_rate": 9.665580485079297e-06,
      "loss": 0.4879,
      "step": 8511
    },
    {
      "epoch": 1.4767522553782095,
      "grad_norm": 0.8401545882225037,
      "learning_rate": 9.659747888845087e-06,
      "loss": 0.4802,
      "step": 8512
    },
    {
      "epoch": 1.4769257460097154,
      "grad_norm": 0.8900430202484131,
      "learning_rate": 9.653916492609168e-06,
      "loss": 0.4176,
      "step": 8513
    },
    {
      "epoch": 1.4770992366412214,
      "grad_norm": 0.8312796354293823,
      "learning_rate": 9.648086297048302e-06,
      "loss": 0.3748,
      "step": 8514
    },
    {
      "epoch": 1.4772727272727273,
      "grad_norm": 0.7063842415809631,
      "learning_rate": 9.642257302839085e-06,
      "loss": 0.4597,
      "step": 8515
    },
    {
      "epoch": 1.4774462179042331,
      "grad_norm": 1.0829441547393799,
      "learning_rate": 9.636429510657974e-06,
      "loss": 0.5219,
      "step": 8516
    },
    {
      "epoch": 1.4776197085357392,
      "grad_norm": 0.7372325658798218,
      "learning_rate": 9.63060292118129e-06,
      "loss": 0.4382,
      "step": 8517
    },
    {
      "epoch": 1.477793199167245,
      "grad_norm": 0.7412036061286926,
      "learning_rate": 9.624777535085233e-06,
      "loss": 0.386,
      "step": 8518
    },
    {
      "epoch": 1.477966689798751,
      "grad_norm": 0.8647828102111816,
      "learning_rate": 9.61895335304582e-06,
      "loss": 0.4579,
      "step": 8519
    },
    {
      "epoch": 1.4781401804302567,
      "grad_norm": 1.84585702419281,
      "learning_rate": 9.61313037573897e-06,
      "loss": 0.453,
      "step": 8520
    },
    {
      "epoch": 1.4783136710617626,
      "grad_norm": 0.5941470265388489,
      "learning_rate": 9.607308603840437e-06,
      "loss": 0.4928,
      "step": 8521
    },
    {
      "epoch": 1.4784871616932684,
      "grad_norm": 0.8481560349464417,
      "learning_rate": 9.601488038025869e-06,
      "loss": 0.4034,
      "step": 8522
    },
    {
      "epoch": 1.4786606523247745,
      "grad_norm": 0.9535527229309082,
      "learning_rate": 9.59566867897071e-06,
      "loss": 0.3519,
      "step": 8523
    },
    {
      "epoch": 1.4788341429562804,
      "grad_norm": 0.7511053085327148,
      "learning_rate": 9.589850527350337e-06,
      "loss": 0.4575,
      "step": 8524
    },
    {
      "epoch": 1.4790076335877862,
      "grad_norm": 1.0763275623321533,
      "learning_rate": 9.584033583839938e-06,
      "loss": 0.491,
      "step": 8525
    },
    {
      "epoch": 1.4791811242192923,
      "grad_norm": 1.197890281677246,
      "learning_rate": 9.578217849114579e-06,
      "loss": 0.3983,
      "step": 8526
    },
    {
      "epoch": 1.479354614850798,
      "grad_norm": 0.7234625816345215,
      "learning_rate": 9.572403323849175e-06,
      "loss": 0.3718,
      "step": 8527
    },
    {
      "epoch": 1.479528105482304,
      "grad_norm": 0.7402774691581726,
      "learning_rate": 9.566590008718524e-06,
      "loss": 0.5287,
      "step": 8528
    },
    {
      "epoch": 1.4797015961138098,
      "grad_norm": 0.7728129625320435,
      "learning_rate": 9.560777904397258e-06,
      "loss": 0.5115,
      "step": 8529
    },
    {
      "epoch": 1.4798750867453156,
      "grad_norm": 0.915511965751648,
      "learning_rate": 9.554967011559874e-06,
      "loss": 0.4138,
      "step": 8530
    },
    {
      "epoch": 1.4800485773768217,
      "grad_norm": 1.1457018852233887,
      "learning_rate": 9.549157330880753e-06,
      "loss": 0.3912,
      "step": 8531
    },
    {
      "epoch": 1.4802220680083276,
      "grad_norm": 0.864634096622467,
      "learning_rate": 9.54334886303409e-06,
      "loss": 0.4006,
      "step": 8532
    },
    {
      "epoch": 1.4803955586398334,
      "grad_norm": 0.7768258452415466,
      "learning_rate": 9.537541608693982e-06,
      "loss": 0.5658,
      "step": 8533
    },
    {
      "epoch": 1.4805690492713395,
      "grad_norm": 0.7090941667556763,
      "learning_rate": 9.53173556853435e-06,
      "loss": 0.4668,
      "step": 8534
    },
    {
      "epoch": 1.4807425399028453,
      "grad_norm": 1.6957459449768066,
      "learning_rate": 9.52593074322902e-06,
      "loss": 0.5034,
      "step": 8535
    },
    {
      "epoch": 1.4809160305343512,
      "grad_norm": 0.8803064227104187,
      "learning_rate": 9.520127133451619e-06,
      "loss": 0.4086,
      "step": 8536
    },
    {
      "epoch": 1.481089521165857,
      "grad_norm": 0.7819961905479431,
      "learning_rate": 9.514324739875684e-06,
      "loss": 0.4628,
      "step": 8537
    },
    {
      "epoch": 1.4812630117973629,
      "grad_norm": 0.6098188757896423,
      "learning_rate": 9.508523563174578e-06,
      "loss": 0.5854,
      "step": 8538
    },
    {
      "epoch": 1.4814365024288687,
      "grad_norm": 0.7409568428993225,
      "learning_rate": 9.50272360402154e-06,
      "loss": 0.389,
      "step": 8539
    },
    {
      "epoch": 1.4816099930603748,
      "grad_norm": 0.8341870307922363,
      "learning_rate": 9.496924863089652e-06,
      "loss": 0.394,
      "step": 8540
    },
    {
      "epoch": 1.4817834836918806,
      "grad_norm": 0.759617269039154,
      "learning_rate": 9.491127341051884e-06,
      "loss": 0.4193,
      "step": 8541
    },
    {
      "epoch": 1.4819569743233865,
      "grad_norm": 1.3312665224075317,
      "learning_rate": 9.485331038581021e-06,
      "loss": 0.4683,
      "step": 8542
    },
    {
      "epoch": 1.4821304649548925,
      "grad_norm": 0.7491121888160706,
      "learning_rate": 9.479535956349751e-06,
      "loss": 0.4744,
      "step": 8543
    },
    {
      "epoch": 1.4823039555863984,
      "grad_norm": 0.7190536856651306,
      "learning_rate": 9.473742095030588e-06,
      "loss": 0.431,
      "step": 8544
    },
    {
      "epoch": 1.4824774462179042,
      "grad_norm": 1.0622676610946655,
      "learning_rate": 9.467949455295925e-06,
      "loss": 0.4563,
      "step": 8545
    },
    {
      "epoch": 1.48265093684941,
      "grad_norm": 0.8268857598304749,
      "learning_rate": 9.46215803781799e-06,
      "loss": 0.4945,
      "step": 8546
    },
    {
      "epoch": 1.482824427480916,
      "grad_norm": 0.7199467420578003,
      "learning_rate": 9.456367843268903e-06,
      "loss": 0.4816,
      "step": 8547
    },
    {
      "epoch": 1.482997918112422,
      "grad_norm": 0.6562536954879761,
      "learning_rate": 9.450578872320613e-06,
      "loss": 0.5212,
      "step": 8548
    },
    {
      "epoch": 1.4831714087439278,
      "grad_norm": 1.1030664443969727,
      "learning_rate": 9.444791125644937e-06,
      "loss": 0.362,
      "step": 8549
    },
    {
      "epoch": 1.4833448993754337,
      "grad_norm": 0.7876076698303223,
      "learning_rate": 9.439004603913542e-06,
      "loss": 0.4969,
      "step": 8550
    },
    {
      "epoch": 1.4835183900069397,
      "grad_norm": 1.0008448362350464,
      "learning_rate": 9.433219307797977e-06,
      "loss": 0.3895,
      "step": 8551
    },
    {
      "epoch": 1.4836918806384456,
      "grad_norm": 0.7976511120796204,
      "learning_rate": 9.427435237969624e-06,
      "loss": 0.5105,
      "step": 8552
    },
    {
      "epoch": 1.4838653712699514,
      "grad_norm": 0.8462479114532471,
      "learning_rate": 9.421652395099732e-06,
      "loss": 0.462,
      "step": 8553
    },
    {
      "epoch": 1.4840388619014573,
      "grad_norm": 0.7487668991088867,
      "learning_rate": 9.415870779859405e-06,
      "loss": 0.4177,
      "step": 8554
    },
    {
      "epoch": 1.4842123525329631,
      "grad_norm": 0.5780131220817566,
      "learning_rate": 9.410090392919598e-06,
      "loss": 0.5214,
      "step": 8555
    },
    {
      "epoch": 1.4843858431644692,
      "grad_norm": 3.076767683029175,
      "learning_rate": 9.404311234951148e-06,
      "loss": 0.4138,
      "step": 8556
    },
    {
      "epoch": 1.484559333795975,
      "grad_norm": 0.7333472967147827,
      "learning_rate": 9.39853330662473e-06,
      "loss": 0.4052,
      "step": 8557
    },
    {
      "epoch": 1.484732824427481,
      "grad_norm": 0.7892191410064697,
      "learning_rate": 9.392756608610871e-06,
      "loss": 0.5435,
      "step": 8558
    },
    {
      "epoch": 1.4849063150589867,
      "grad_norm": 0.7384252548217773,
      "learning_rate": 9.386981141579961e-06,
      "loss": 0.3242,
      "step": 8559
    },
    {
      "epoch": 1.4850798056904928,
      "grad_norm": 0.9419588446617126,
      "learning_rate": 9.381206906202268e-06,
      "loss": 0.3159,
      "step": 8560
    },
    {
      "epoch": 1.4852532963219987,
      "grad_norm": 0.7519877552986145,
      "learning_rate": 9.375433903147877e-06,
      "loss": 0.5704,
      "step": 8561
    },
    {
      "epoch": 1.4854267869535045,
      "grad_norm": 0.8963461518287659,
      "learning_rate": 9.369662133086768e-06,
      "loss": 0.3737,
      "step": 8562
    },
    {
      "epoch": 1.4856002775850103,
      "grad_norm": 1.7488105297088623,
      "learning_rate": 9.363891596688745e-06,
      "loss": 0.3864,
      "step": 8563
    },
    {
      "epoch": 1.4857737682165162,
      "grad_norm": 0.8128525018692017,
      "learning_rate": 9.358122294623514e-06,
      "loss": 0.4784,
      "step": 8564
    },
    {
      "epoch": 1.4859472588480223,
      "grad_norm": 0.6044059991836548,
      "learning_rate": 9.352354227560575e-06,
      "loss": 0.5121,
      "step": 8565
    },
    {
      "epoch": 1.486120749479528,
      "grad_norm": 2.582504987716675,
      "learning_rate": 9.34658739616934e-06,
      "loss": 0.5166,
      "step": 8566
    },
    {
      "epoch": 1.486294240111034,
      "grad_norm": 0.9345373511314392,
      "learning_rate": 9.340821801119053e-06,
      "loss": 0.4183,
      "step": 8567
    },
    {
      "epoch": 1.48646773074254,
      "grad_norm": 3.323610305786133,
      "learning_rate": 9.335057443078817e-06,
      "loss": 0.5702,
      "step": 8568
    },
    {
      "epoch": 1.4866412213740459,
      "grad_norm": 0.6034752726554871,
      "learning_rate": 9.329294322717584e-06,
      "loss": 0.4974,
      "step": 8569
    },
    {
      "epoch": 1.4868147120055517,
      "grad_norm": 0.6210371851921082,
      "learning_rate": 9.323532440704196e-06,
      "loss": 0.3853,
      "step": 8570
    },
    {
      "epoch": 1.4869882026370576,
      "grad_norm": 1.0118930339813232,
      "learning_rate": 9.31777179770729e-06,
      "loss": 0.4394,
      "step": 8571
    },
    {
      "epoch": 1.4871616932685634,
      "grad_norm": 0.7408242225646973,
      "learning_rate": 9.312012394395423e-06,
      "loss": 0.4777,
      "step": 8572
    },
    {
      "epoch": 1.4873351839000695,
      "grad_norm": 0.700859785079956,
      "learning_rate": 9.30625423143697e-06,
      "loss": 0.5621,
      "step": 8573
    },
    {
      "epoch": 1.4875086745315753,
      "grad_norm": 0.7190759181976318,
      "learning_rate": 9.300497309500176e-06,
      "loss": 0.3806,
      "step": 8574
    },
    {
      "epoch": 1.4876821651630812,
      "grad_norm": 0.6803382039070129,
      "learning_rate": 9.29474162925313e-06,
      "loss": 0.4049,
      "step": 8575
    },
    {
      "epoch": 1.4878556557945872,
      "grad_norm": 0.657941460609436,
      "learning_rate": 9.288987191363799e-06,
      "loss": 0.4789,
      "step": 8576
    },
    {
      "epoch": 1.488029146426093,
      "grad_norm": 0.6799991130828857,
      "learning_rate": 9.283233996499984e-06,
      "loss": 0.6385,
      "step": 8577
    },
    {
      "epoch": 1.488202637057599,
      "grad_norm": 0.7778498530387878,
      "learning_rate": 9.277482045329344e-06,
      "loss": 0.4995,
      "step": 8578
    },
    {
      "epoch": 1.4883761276891048,
      "grad_norm": 1.108154296875,
      "learning_rate": 9.271731338519415e-06,
      "loss": 0.55,
      "step": 8579
    },
    {
      "epoch": 1.4885496183206106,
      "grad_norm": 0.638132631778717,
      "learning_rate": 9.265981876737566e-06,
      "loss": 0.4076,
      "step": 8580
    },
    {
      "epoch": 1.4887231089521165,
      "grad_norm": 0.942566990852356,
      "learning_rate": 9.260233660651025e-06,
      "loss": 0.4186,
      "step": 8581
    },
    {
      "epoch": 1.4888965995836225,
      "grad_norm": 0.8701571822166443,
      "learning_rate": 9.254486690926878e-06,
      "loss": 0.3516,
      "step": 8582
    },
    {
      "epoch": 1.4890700902151284,
      "grad_norm": 0.7635884881019592,
      "learning_rate": 9.248740968232084e-06,
      "loss": 0.3903,
      "step": 8583
    },
    {
      "epoch": 1.4892435808466342,
      "grad_norm": 0.742935836315155,
      "learning_rate": 9.242996493233414e-06,
      "loss": 0.4371,
      "step": 8584
    },
    {
      "epoch": 1.4894170714781403,
      "grad_norm": 0.7536942958831787,
      "learning_rate": 9.237253266597544e-06,
      "loss": 0.3566,
      "step": 8585
    },
    {
      "epoch": 1.4895905621096461,
      "grad_norm": 0.9712985157966614,
      "learning_rate": 9.23151128899097e-06,
      "loss": 0.4584,
      "step": 8586
    },
    {
      "epoch": 1.489764052741152,
      "grad_norm": 1.0012733936309814,
      "learning_rate": 9.225770561080062e-06,
      "loss": 0.4162,
      "step": 8587
    },
    {
      "epoch": 1.4899375433726578,
      "grad_norm": 0.856743335723877,
      "learning_rate": 9.220031083531026e-06,
      "loss": 0.491,
      "step": 8588
    },
    {
      "epoch": 1.4901110340041637,
      "grad_norm": 0.7103423476219177,
      "learning_rate": 9.214292857009961e-06,
      "loss": 0.5468,
      "step": 8589
    },
    {
      "epoch": 1.4902845246356697,
      "grad_norm": 0.6725903749465942,
      "learning_rate": 9.208555882182762e-06,
      "loss": 0.5104,
      "step": 8590
    },
    {
      "epoch": 1.4904580152671756,
      "grad_norm": 0.6370904445648193,
      "learning_rate": 9.202820159715234e-06,
      "loss": 0.5353,
      "step": 8591
    },
    {
      "epoch": 1.4906315058986814,
      "grad_norm": 0.7135043740272522,
      "learning_rate": 9.197085690273e-06,
      "loss": 0.4349,
      "step": 8592
    },
    {
      "epoch": 1.4908049965301875,
      "grad_norm": 0.8648028373718262,
      "learning_rate": 9.191352474521572e-06,
      "loss": 0.5393,
      "step": 8593
    },
    {
      "epoch": 1.4909784871616933,
      "grad_norm": 0.8293433785438538,
      "learning_rate": 9.185620513126275e-06,
      "loss": 0.423,
      "step": 8594
    },
    {
      "epoch": 1.4911519777931992,
      "grad_norm": 0.7110719084739685,
      "learning_rate": 9.179889806752322e-06,
      "loss": 0.4988,
      "step": 8595
    },
    {
      "epoch": 1.491325468424705,
      "grad_norm": 0.7662829160690308,
      "learning_rate": 9.174160356064765e-06,
      "loss": 0.4418,
      "step": 8596
    },
    {
      "epoch": 1.4914989590562109,
      "grad_norm": 0.8652961850166321,
      "learning_rate": 9.168432161728515e-06,
      "loss": 0.369,
      "step": 8597
    },
    {
      "epoch": 1.4916724496877167,
      "grad_norm": 0.8832250833511353,
      "learning_rate": 9.162705224408326e-06,
      "loss": 0.3846,
      "step": 8598
    },
    {
      "epoch": 1.4918459403192228,
      "grad_norm": 0.8304434418678284,
      "learning_rate": 9.15697954476883e-06,
      "loss": 0.4884,
      "step": 8599
    },
    {
      "epoch": 1.4920194309507286,
      "grad_norm": 0.7447237372398376,
      "learning_rate": 9.151255123474493e-06,
      "loss": 0.4503,
      "step": 8600
    },
    {
      "epoch": 1.4921929215822345,
      "grad_norm": 0.7981603741645813,
      "learning_rate": 9.14553196118964e-06,
      "loss": 0.4208,
      "step": 8601
    },
    {
      "epoch": 1.4923664122137406,
      "grad_norm": 0.7032742500305176,
      "learning_rate": 9.139810058578451e-06,
      "loss": 0.4658,
      "step": 8602
    },
    {
      "epoch": 1.4925399028452464,
      "grad_norm": 0.7949456572532654,
      "learning_rate": 9.134089416304951e-06,
      "loss": 0.4136,
      "step": 8603
    },
    {
      "epoch": 1.4927133934767522,
      "grad_norm": 0.7762060165405273,
      "learning_rate": 9.128370035033046e-06,
      "loss": 0.4445,
      "step": 8604
    },
    {
      "epoch": 1.492886884108258,
      "grad_norm": 0.7838228344917297,
      "learning_rate": 9.122651915426464e-06,
      "loss": 0.5227,
      "step": 8605
    },
    {
      "epoch": 1.493060374739764,
      "grad_norm": 0.682068943977356,
      "learning_rate": 9.116935058148801e-06,
      "loss": 0.4818,
      "step": 8606
    },
    {
      "epoch": 1.49323386537127,
      "grad_norm": 0.6018600463867188,
      "learning_rate": 9.111219463863501e-06,
      "loss": 0.5979,
      "step": 8607
    },
    {
      "epoch": 1.4934073560027759,
      "grad_norm": 0.7784172296524048,
      "learning_rate": 9.105505133233876e-06,
      "loss": 0.4379,
      "step": 8608
    },
    {
      "epoch": 1.4935808466342817,
      "grad_norm": 0.692943811416626,
      "learning_rate": 9.099792066923077e-06,
      "loss": 0.4481,
      "step": 8609
    },
    {
      "epoch": 1.4937543372657878,
      "grad_norm": 0.7734472751617432,
      "learning_rate": 9.094080265594108e-06,
      "loss": 0.5063,
      "step": 8610
    },
    {
      "epoch": 1.4939278278972936,
      "grad_norm": 1.248750925064087,
      "learning_rate": 9.088369729909823e-06,
      "loss": 0.5038,
      "step": 8611
    },
    {
      "epoch": 1.4941013185287995,
      "grad_norm": 0.6195592284202576,
      "learning_rate": 9.082660460532961e-06,
      "loss": 0.5079,
      "step": 8612
    },
    {
      "epoch": 1.4942748091603053,
      "grad_norm": 1.2637933492660522,
      "learning_rate": 9.07695245812606e-06,
      "loss": 0.3878,
      "step": 8613
    },
    {
      "epoch": 1.4944482997918112,
      "grad_norm": 0.6877127289772034,
      "learning_rate": 9.071245723351563e-06,
      "loss": 0.4902,
      "step": 8614
    },
    {
      "epoch": 1.4946217904233172,
      "grad_norm": 1.2176055908203125,
      "learning_rate": 9.065540256871733e-06,
      "loss": 0.5112,
      "step": 8615
    },
    {
      "epoch": 1.494795281054823,
      "grad_norm": 1.0032249689102173,
      "learning_rate": 9.059836059348696e-06,
      "loss": 0.4208,
      "step": 8616
    },
    {
      "epoch": 1.494968771686329,
      "grad_norm": 0.6759541034698486,
      "learning_rate": 9.054133131444427e-06,
      "loss": 0.5089,
      "step": 8617
    },
    {
      "epoch": 1.4951422623178348,
      "grad_norm": 0.9625713229179382,
      "learning_rate": 9.048431473820776e-06,
      "loss": 0.443,
      "step": 8618
    },
    {
      "epoch": 1.4953157529493408,
      "grad_norm": 0.6422067880630493,
      "learning_rate": 9.042731087139398e-06,
      "loss": 0.4221,
      "step": 8619
    },
    {
      "epoch": 1.4954892435808467,
      "grad_norm": 0.674849808216095,
      "learning_rate": 9.037031972061854e-06,
      "loss": 0.4745,
      "step": 8620
    },
    {
      "epoch": 1.4956627342123525,
      "grad_norm": 0.791410505771637,
      "learning_rate": 9.031334129249515e-06,
      "loss": 0.3588,
      "step": 8621
    },
    {
      "epoch": 1.4958362248438584,
      "grad_norm": 0.7717758417129517,
      "learning_rate": 9.02563755936365e-06,
      "loss": 0.4924,
      "step": 8622
    },
    {
      "epoch": 1.4960097154753642,
      "grad_norm": 0.800813615322113,
      "learning_rate": 9.019942263065316e-06,
      "loss": 0.559,
      "step": 8623
    },
    {
      "epoch": 1.4961832061068703,
      "grad_norm": 0.6531893014907837,
      "learning_rate": 9.014248241015484e-06,
      "loss": 0.583,
      "step": 8624
    },
    {
      "epoch": 1.4963566967383761,
      "grad_norm": 0.6513347029685974,
      "learning_rate": 9.008555493874944e-06,
      "loss": 0.5826,
      "step": 8625
    },
    {
      "epoch": 1.496530187369882,
      "grad_norm": 3.2361068725585938,
      "learning_rate": 9.00286402230434e-06,
      "loss": 0.3813,
      "step": 8626
    },
    {
      "epoch": 1.496703678001388,
      "grad_norm": 0.7543614506721497,
      "learning_rate": 8.99717382696419e-06,
      "loss": 0.3718,
      "step": 8627
    },
    {
      "epoch": 1.4968771686328939,
      "grad_norm": 1.0834203958511353,
      "learning_rate": 8.991484908514835e-06,
      "loss": 0.4679,
      "step": 8628
    },
    {
      "epoch": 1.4970506592643997,
      "grad_norm": 0.8064327836036682,
      "learning_rate": 8.985797267616485e-06,
      "loss": 0.51,
      "step": 8629
    },
    {
      "epoch": 1.4972241498959056,
      "grad_norm": 0.9959091544151306,
      "learning_rate": 8.980110904929189e-06,
      "loss": 0.3596,
      "step": 8630
    },
    {
      "epoch": 1.4973976405274114,
      "grad_norm": 0.8047711849212646,
      "learning_rate": 8.974425821112877e-06,
      "loss": 0.5043,
      "step": 8631
    },
    {
      "epoch": 1.4975711311589175,
      "grad_norm": 0.778866171836853,
      "learning_rate": 8.968742016827283e-06,
      "loss": 0.5012,
      "step": 8632
    },
    {
      "epoch": 1.4977446217904233,
      "grad_norm": 1.1322171688079834,
      "learning_rate": 8.963059492732039e-06,
      "loss": 0.3735,
      "step": 8633
    },
    {
      "epoch": 1.4979181124219292,
      "grad_norm": 0.9971731901168823,
      "learning_rate": 8.957378249486592e-06,
      "loss": 0.3782,
      "step": 8634
    },
    {
      "epoch": 1.4980916030534353,
      "grad_norm": 0.7050626277923584,
      "learning_rate": 8.951698287750283e-06,
      "loss": 0.6083,
      "step": 8635
    },
    {
      "epoch": 1.498265093684941,
      "grad_norm": 0.870461642742157,
      "learning_rate": 8.946019608182245e-06,
      "loss": 0.459,
      "step": 8636
    },
    {
      "epoch": 1.498438584316447,
      "grad_norm": 0.8623781204223633,
      "learning_rate": 8.940342211441522e-06,
      "loss": 0.3542,
      "step": 8637
    },
    {
      "epoch": 1.4986120749479528,
      "grad_norm": 1.6813430786132812,
      "learning_rate": 8.93466609818697e-06,
      "loss": 0.3652,
      "step": 8638
    },
    {
      "epoch": 1.4987855655794586,
      "grad_norm": 0.8649514317512512,
      "learning_rate": 8.928991269077311e-06,
      "loss": 0.4236,
      "step": 8639
    },
    {
      "epoch": 1.4989590562109645,
      "grad_norm": 0.8307508826255798,
      "learning_rate": 8.92331772477111e-06,
      "loss": 0.5244,
      "step": 8640
    },
    {
      "epoch": 1.4991325468424705,
      "grad_norm": 0.8284005522727966,
      "learning_rate": 8.917645465926807e-06,
      "loss": 0.3849,
      "step": 8641
    },
    {
      "epoch": 1.4993060374739764,
      "grad_norm": 1.0204592943191528,
      "learning_rate": 8.91197449320265e-06,
      "loss": 0.3115,
      "step": 8642
    },
    {
      "epoch": 1.4994795281054822,
      "grad_norm": 1.2717472314834595,
      "learning_rate": 8.906304807256775e-06,
      "loss": 0.3308,
      "step": 8643
    },
    {
      "epoch": 1.4996530187369883,
      "grad_norm": 1.0788602828979492,
      "learning_rate": 8.900636408747156e-06,
      "loss": 0.363,
      "step": 8644
    },
    {
      "epoch": 1.4998265093684942,
      "grad_norm": 0.6880956292152405,
      "learning_rate": 8.894969298331617e-06,
      "loss": 0.4625,
      "step": 8645
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6167536377906799,
      "learning_rate": 8.889303476667823e-06,
      "loss": 0.4945,
      "step": 8646
    },
    {
      "epoch": 1.5001734906315058,
      "grad_norm": 0.7947840094566345,
      "learning_rate": 8.883638944413313e-06,
      "loss": 0.4093,
      "step": 8647
    },
    {
      "epoch": 1.5003469812630117,
      "grad_norm": 0.6716247200965881,
      "learning_rate": 8.877975702225457e-06,
      "loss": 0.5023,
      "step": 8648
    },
    {
      "epoch": 1.5005204718945175,
      "grad_norm": 1.0151311159133911,
      "learning_rate": 8.872313750761482e-06,
      "loss": 0.4816,
      "step": 8649
    },
    {
      "epoch": 1.5006939625260236,
      "grad_norm": 0.9199911952018738,
      "learning_rate": 8.866653090678452e-06,
      "loss": 0.5542,
      "step": 8650
    },
    {
      "epoch": 1.5008674531575295,
      "grad_norm": 0.9588578343391418,
      "learning_rate": 8.860993722633312e-06,
      "loss": 0.4092,
      "step": 8651
    },
    {
      "epoch": 1.5010409437890355,
      "grad_norm": 1.4194256067276,
      "learning_rate": 8.855335647282833e-06,
      "loss": 0.3568,
      "step": 8652
    },
    {
      "epoch": 1.5012144344205414,
      "grad_norm": 0.8696090579032898,
      "learning_rate": 8.849678865283635e-06,
      "loss": 0.3328,
      "step": 8653
    },
    {
      "epoch": 1.5013879250520472,
      "grad_norm": 1.2177104949951172,
      "learning_rate": 8.844023377292198e-06,
      "loss": 0.5056,
      "step": 8654
    },
    {
      "epoch": 1.501561415683553,
      "grad_norm": 0.9737895727157593,
      "learning_rate": 8.838369183964841e-06,
      "loss": 0.3463,
      "step": 8655
    },
    {
      "epoch": 1.501734906315059,
      "grad_norm": 1.043319821357727,
      "learning_rate": 8.832716285957754e-06,
      "loss": 0.3523,
      "step": 8656
    },
    {
      "epoch": 1.5019083969465647,
      "grad_norm": 1.0447510480880737,
      "learning_rate": 8.827064683926954e-06,
      "loss": 0.3477,
      "step": 8657
    },
    {
      "epoch": 1.5020818875780708,
      "grad_norm": 1.3222448825836182,
      "learning_rate": 8.821414378528314e-06,
      "loss": 0.3904,
      "step": 8658
    },
    {
      "epoch": 1.5022553782095767,
      "grad_norm": 0.7870173454284668,
      "learning_rate": 8.815765370417557e-06,
      "loss": 0.4387,
      "step": 8659
    },
    {
      "epoch": 1.5024288688410827,
      "grad_norm": 0.757055938243866,
      "learning_rate": 8.810117660250275e-06,
      "loss": 0.5533,
      "step": 8660
    },
    {
      "epoch": 1.5026023594725886,
      "grad_norm": 0.9041140079498291,
      "learning_rate": 8.80447124868186e-06,
      "loss": 0.3798,
      "step": 8661
    },
    {
      "epoch": 1.5027758501040944,
      "grad_norm": 1.1670129299163818,
      "learning_rate": 8.798826136367613e-06,
      "loss": 0.3694,
      "step": 8662
    },
    {
      "epoch": 1.5029493407356003,
      "grad_norm": 0.834130048751831,
      "learning_rate": 8.793182323962635e-06,
      "loss": 0.3368,
      "step": 8663
    },
    {
      "epoch": 1.5031228313671061,
      "grad_norm": 0.6511291861534119,
      "learning_rate": 8.787539812121924e-06,
      "loss": 0.4883,
      "step": 8664
    },
    {
      "epoch": 1.503296321998612,
      "grad_norm": 0.7934287786483765,
      "learning_rate": 8.781898601500265e-06,
      "loss": 0.4287,
      "step": 8665
    },
    {
      "epoch": 1.5034698126301178,
      "grad_norm": 0.7822073698043823,
      "learning_rate": 8.776258692752355e-06,
      "loss": 0.4736,
      "step": 8666
    },
    {
      "epoch": 1.5036433032616239,
      "grad_norm": 0.7018622756004333,
      "learning_rate": 8.7706200865327e-06,
      "loss": 0.5229,
      "step": 8667
    },
    {
      "epoch": 1.5038167938931297,
      "grad_norm": 0.7407475709915161,
      "learning_rate": 8.76498278349567e-06,
      "loss": 0.4767,
      "step": 8668
    },
    {
      "epoch": 1.5039902845246358,
      "grad_norm": 0.7079533338546753,
      "learning_rate": 8.75934678429547e-06,
      "loss": 0.6644,
      "step": 8669
    },
    {
      "epoch": 1.5041637751561416,
      "grad_norm": 0.6935982704162598,
      "learning_rate": 8.753712089586184e-06,
      "loss": 0.5565,
      "step": 8670
    },
    {
      "epoch": 1.5043372657876475,
      "grad_norm": 0.6961166262626648,
      "learning_rate": 8.748078700021714e-06,
      "loss": 0.3792,
      "step": 8671
    },
    {
      "epoch": 1.5045107564191533,
      "grad_norm": 0.8431174755096436,
      "learning_rate": 8.74244661625582e-06,
      "loss": 0.459,
      "step": 8672
    },
    {
      "epoch": 1.5046842470506592,
      "grad_norm": 0.6829946637153625,
      "learning_rate": 8.736815838942114e-06,
      "loss": 0.4969,
      "step": 8673
    },
    {
      "epoch": 1.504857737682165,
      "grad_norm": 0.9456247687339783,
      "learning_rate": 8.731186368734049e-06,
      "loss": 0.349,
      "step": 8674
    },
    {
      "epoch": 1.505031228313671,
      "grad_norm": 0.8031119704246521,
      "learning_rate": 8.725558206284944e-06,
      "loss": 0.4606,
      "step": 8675
    },
    {
      "epoch": 1.505204718945177,
      "grad_norm": 0.6166511178016663,
      "learning_rate": 8.719931352247944e-06,
      "loss": 0.4719,
      "step": 8676
    },
    {
      "epoch": 1.505378209576683,
      "grad_norm": 1.3053230047225952,
      "learning_rate": 8.714305807276056e-06,
      "loss": 0.4058,
      "step": 8677
    },
    {
      "epoch": 1.5055517002081888,
      "grad_norm": 0.7026805877685547,
      "learning_rate": 8.708681572022122e-06,
      "loss": 0.4995,
      "step": 8678
    },
    {
      "epoch": 1.5057251908396947,
      "grad_norm": 0.8382639288902283,
      "learning_rate": 8.703058647138855e-06,
      "loss": 0.4978,
      "step": 8679
    },
    {
      "epoch": 1.5058986814712005,
      "grad_norm": 0.9075166583061218,
      "learning_rate": 8.697437033278797e-06,
      "loss": 0.3594,
      "step": 8680
    },
    {
      "epoch": 1.5060721721027064,
      "grad_norm": 1.3929052352905273,
      "learning_rate": 8.691816731094341e-06,
      "loss": 0.5269,
      "step": 8681
    },
    {
      "epoch": 1.5062456627342122,
      "grad_norm": 1.178870677947998,
      "learning_rate": 8.686197741237722e-06,
      "loss": 0.4796,
      "step": 8682
    },
    {
      "epoch": 1.5064191533657183,
      "grad_norm": 0.7761985063552856,
      "learning_rate": 8.680580064361053e-06,
      "loss": 0.3702,
      "step": 8683
    },
    {
      "epoch": 1.5065926439972241,
      "grad_norm": 0.7987775802612305,
      "learning_rate": 8.674963701116243e-06,
      "loss": 0.5496,
      "step": 8684
    },
    {
      "epoch": 1.5067661346287302,
      "grad_norm": 0.9020431637763977,
      "learning_rate": 8.669348652155101e-06,
      "loss": 0.4366,
      "step": 8685
    },
    {
      "epoch": 1.506939625260236,
      "grad_norm": 0.9262181520462036,
      "learning_rate": 8.663734918129247e-06,
      "loss": 0.4659,
      "step": 8686
    },
    {
      "epoch": 1.507113115891742,
      "grad_norm": 0.819610059261322,
      "learning_rate": 8.658122499690166e-06,
      "loss": 0.4401,
      "step": 8687
    },
    {
      "epoch": 1.5072866065232478,
      "grad_norm": 0.8673333525657654,
      "learning_rate": 8.652511397489181e-06,
      "loss": 0.3316,
      "step": 8688
    },
    {
      "epoch": 1.5074600971547536,
      "grad_norm": 0.950929582118988,
      "learning_rate": 8.646901612177484e-06,
      "loss": 0.4189,
      "step": 8689
    },
    {
      "epoch": 1.5076335877862594,
      "grad_norm": 0.7968010306358337,
      "learning_rate": 8.641293144406067e-06,
      "loss": 0.4417,
      "step": 8690
    },
    {
      "epoch": 1.5078070784177653,
      "grad_norm": 1.0768781900405884,
      "learning_rate": 8.635685994825824e-06,
      "loss": 0.3964,
      "step": 8691
    },
    {
      "epoch": 1.5079805690492714,
      "grad_norm": 0.8976142406463623,
      "learning_rate": 8.630080164087456e-06,
      "loss": 0.3446,
      "step": 8692
    },
    {
      "epoch": 1.5081540596807772,
      "grad_norm": 3.6228787899017334,
      "learning_rate": 8.624475652841549e-06,
      "loss": 0.3589,
      "step": 8693
    },
    {
      "epoch": 1.5083275503122833,
      "grad_norm": 0.6739755868911743,
      "learning_rate": 8.618872461738483e-06,
      "loss": 0.4677,
      "step": 8694
    },
    {
      "epoch": 1.5085010409437891,
      "grad_norm": 0.9304718971252441,
      "learning_rate": 8.613270591428537e-06,
      "loss": 0.4839,
      "step": 8695
    },
    {
      "epoch": 1.508674531575295,
      "grad_norm": 2.7039785385131836,
      "learning_rate": 8.607670042561807e-06,
      "loss": 0.5592,
      "step": 8696
    },
    {
      "epoch": 1.5088480222068008,
      "grad_norm": 0.783592700958252,
      "learning_rate": 8.602070815788241e-06,
      "loss": 0.4778,
      "step": 8697
    },
    {
      "epoch": 1.5090215128383067,
      "grad_norm": 0.6445301175117493,
      "learning_rate": 8.596472911757633e-06,
      "loss": 0.5728,
      "step": 8698
    },
    {
      "epoch": 1.5091950034698125,
      "grad_norm": 0.8287010788917542,
      "learning_rate": 8.59087633111964e-06,
      "loss": 0.4379,
      "step": 8699
    },
    {
      "epoch": 1.5093684941013186,
      "grad_norm": 0.8014488816261292,
      "learning_rate": 8.58528107452374e-06,
      "loss": 0.4491,
      "step": 8700
    },
    {
      "epoch": 1.5095419847328244,
      "grad_norm": 0.7545841932296753,
      "learning_rate": 8.579687142619267e-06,
      "loss": 0.4193,
      "step": 8701
    },
    {
      "epoch": 1.5097154753643305,
      "grad_norm": 0.8282356262207031,
      "learning_rate": 8.574094536055423e-06,
      "loss": 0.4635,
      "step": 8702
    },
    {
      "epoch": 1.5098889659958363,
      "grad_norm": 1.024894118309021,
      "learning_rate": 8.568503255481204e-06,
      "loss": 0.4114,
      "step": 8703
    },
    {
      "epoch": 1.5100624566273422,
      "grad_norm": 0.867028534412384,
      "learning_rate": 8.562913301545513e-06,
      "loss": 0.5336,
      "step": 8704
    },
    {
      "epoch": 1.510235947258848,
      "grad_norm": 0.6761842966079712,
      "learning_rate": 8.557324674897061e-06,
      "loss": 0.3893,
      "step": 8705
    },
    {
      "epoch": 1.5104094378903539,
      "grad_norm": 0.8481341004371643,
      "learning_rate": 8.55173737618441e-06,
      "loss": 0.4073,
      "step": 8706
    },
    {
      "epoch": 1.5105829285218597,
      "grad_norm": 0.7954474091529846,
      "learning_rate": 8.54615140605597e-06,
      "loss": 0.4332,
      "step": 8707
    },
    {
      "epoch": 1.5107564191533656,
      "grad_norm": 0.8756697773933411,
      "learning_rate": 8.540566765160016e-06,
      "loss": 0.4514,
      "step": 8708
    },
    {
      "epoch": 1.5109299097848716,
      "grad_norm": 2.885711669921875,
      "learning_rate": 8.53498345414464e-06,
      "loss": 0.5441,
      "step": 8709
    },
    {
      "epoch": 1.5111034004163775,
      "grad_norm": 0.6980843544006348,
      "learning_rate": 8.529401473657795e-06,
      "loss": 0.453,
      "step": 8710
    },
    {
      "epoch": 1.5112768910478835,
      "grad_norm": 1.0166019201278687,
      "learning_rate": 8.523820824347266e-06,
      "loss": 0.4214,
      "step": 8711
    },
    {
      "epoch": 1.5114503816793894,
      "grad_norm": 0.7295387387275696,
      "learning_rate": 8.518241506860719e-06,
      "loss": 0.3721,
      "step": 8712
    },
    {
      "epoch": 1.5116238723108952,
      "grad_norm": 0.9436020851135254,
      "learning_rate": 8.512663521845607e-06,
      "loss": 0.5176,
      "step": 8713
    },
    {
      "epoch": 1.511797362942401,
      "grad_norm": 0.7077387571334839,
      "learning_rate": 8.507086869949287e-06,
      "loss": 0.4146,
      "step": 8714
    },
    {
      "epoch": 1.511970853573907,
      "grad_norm": 1.8903543949127197,
      "learning_rate": 8.50151155181893e-06,
      "loss": 0.5031,
      "step": 8715
    },
    {
      "epoch": 1.5121443442054128,
      "grad_norm": 0.7740080952644348,
      "learning_rate": 8.495937568101551e-06,
      "loss": 0.5549,
      "step": 8716
    },
    {
      "epoch": 1.5123178348369188,
      "grad_norm": 0.7523136138916016,
      "learning_rate": 8.49036491944402e-06,
      "loss": 0.4861,
      "step": 8717
    },
    {
      "epoch": 1.5124913254684247,
      "grad_norm": 0.785311758518219,
      "learning_rate": 8.484793606493054e-06,
      "loss": 0.351,
      "step": 8718
    },
    {
      "epoch": 1.5126648160999308,
      "grad_norm": 1.0966126918792725,
      "learning_rate": 8.47922362989521e-06,
      "loss": 0.4734,
      "step": 8719
    },
    {
      "epoch": 1.5128383067314366,
      "grad_norm": 0.9584879875183105,
      "learning_rate": 8.473654990296887e-06,
      "loss": 0.3885,
      "step": 8720
    },
    {
      "epoch": 1.5130117973629424,
      "grad_norm": 1.2425463199615479,
      "learning_rate": 8.468087688344329e-06,
      "loss": 0.4983,
      "step": 8721
    },
    {
      "epoch": 1.5131852879944483,
      "grad_norm": 0.9456865191459656,
      "learning_rate": 8.462521724683637e-06,
      "loss": 0.3493,
      "step": 8722
    },
    {
      "epoch": 1.5133587786259541,
      "grad_norm": 1.2924953699111938,
      "learning_rate": 8.456957099960743e-06,
      "loss": 0.3552,
      "step": 8723
    },
    {
      "epoch": 1.51353226925746,
      "grad_norm": 0.617445707321167,
      "learning_rate": 8.451393814821427e-06,
      "loss": 0.571,
      "step": 8724
    },
    {
      "epoch": 1.5137057598889658,
      "grad_norm": 0.6202148199081421,
      "learning_rate": 8.445831869911317e-06,
      "loss": 0.5447,
      "step": 8725
    },
    {
      "epoch": 1.513879250520472,
      "grad_norm": 0.9100032448768616,
      "learning_rate": 8.440271265875875e-06,
      "loss": 0.3827,
      "step": 8726
    },
    {
      "epoch": 1.5140527411519777,
      "grad_norm": 1.0329697132110596,
      "learning_rate": 8.434712003360427e-06,
      "loss": 0.408,
      "step": 8727
    },
    {
      "epoch": 1.5142262317834838,
      "grad_norm": 0.7562122344970703,
      "learning_rate": 8.42915408301013e-06,
      "loss": 0.5271,
      "step": 8728
    },
    {
      "epoch": 1.5143997224149897,
      "grad_norm": 0.7683006525039673,
      "learning_rate": 8.423597505469983e-06,
      "loss": 0.4318,
      "step": 8729
    },
    {
      "epoch": 1.5145732130464955,
      "grad_norm": 1.2715370655059814,
      "learning_rate": 8.418042271384828e-06,
      "loss": 0.6125,
      "step": 8730
    },
    {
      "epoch": 1.5147467036780013,
      "grad_norm": 0.8350828289985657,
      "learning_rate": 8.412488381399378e-06,
      "loss": 0.4047,
      "step": 8731
    },
    {
      "epoch": 1.5149201943095072,
      "grad_norm": 0.763030469417572,
      "learning_rate": 8.406935836158138e-06,
      "loss": 0.4677,
      "step": 8732
    },
    {
      "epoch": 1.515093684941013,
      "grad_norm": 0.8802626132965088,
      "learning_rate": 8.401384636305509e-06,
      "loss": 0.3602,
      "step": 8733
    },
    {
      "epoch": 1.515267175572519,
      "grad_norm": 0.8387312293052673,
      "learning_rate": 8.39583478248571e-06,
      "loss": 0.413,
      "step": 8734
    },
    {
      "epoch": 1.515440666204025,
      "grad_norm": 0.7499859929084778,
      "learning_rate": 8.390286275342805e-06,
      "loss": 0.4001,
      "step": 8735
    },
    {
      "epoch": 1.515614156835531,
      "grad_norm": 0.7287644147872925,
      "learning_rate": 8.3847391155207e-06,
      "loss": 0.4792,
      "step": 8736
    },
    {
      "epoch": 1.5157876474670369,
      "grad_norm": 0.6044920086860657,
      "learning_rate": 8.379193303663162e-06,
      "loss": 0.6472,
      "step": 8737
    },
    {
      "epoch": 1.5159611380985427,
      "grad_norm": 0.676855742931366,
      "learning_rate": 8.373648840413781e-06,
      "loss": 0.5596,
      "step": 8738
    },
    {
      "epoch": 1.5161346287300486,
      "grad_norm": 0.6197137236595154,
      "learning_rate": 8.368105726416e-06,
      "loss": 0.496,
      "step": 8739
    },
    {
      "epoch": 1.5163081193615544,
      "grad_norm": 0.7422799468040466,
      "learning_rate": 8.362563962313095e-06,
      "loss": 0.5498,
      "step": 8740
    },
    {
      "epoch": 1.5164816099930603,
      "grad_norm": 0.7818398475646973,
      "learning_rate": 8.357023548748217e-06,
      "loss": 0.5175,
      "step": 8741
    },
    {
      "epoch": 1.5166551006245663,
      "grad_norm": 1.221691370010376,
      "learning_rate": 8.35148448636431e-06,
      "loss": 0.3844,
      "step": 8742
    },
    {
      "epoch": 1.5168285912560722,
      "grad_norm": 0.6863657236099243,
      "learning_rate": 8.345946775804209e-06,
      "loss": 0.5272,
      "step": 8743
    },
    {
      "epoch": 1.5170020818875782,
      "grad_norm": 0.6155330538749695,
      "learning_rate": 8.340410417710562e-06,
      "loss": 0.5604,
      "step": 8744
    },
    {
      "epoch": 1.517175572519084,
      "grad_norm": 0.8316579461097717,
      "learning_rate": 8.334875412725874e-06,
      "loss": 0.3724,
      "step": 8745
    },
    {
      "epoch": 1.51734906315059,
      "grad_norm": 0.7292289733886719,
      "learning_rate": 8.32934176149248e-06,
      "loss": 0.4319,
      "step": 8746
    },
    {
      "epoch": 1.5175225537820958,
      "grad_norm": 1.0372445583343506,
      "learning_rate": 8.323809464652579e-06,
      "loss": 0.3347,
      "step": 8747
    },
    {
      "epoch": 1.5176960444136016,
      "grad_norm": 1.7314422130584717,
      "learning_rate": 8.318278522848198e-06,
      "loss": 0.3685,
      "step": 8748
    },
    {
      "epoch": 1.5178695350451075,
      "grad_norm": 0.6570568084716797,
      "learning_rate": 8.312748936721196e-06,
      "loss": 0.4059,
      "step": 8749
    },
    {
      "epoch": 1.5180430256766133,
      "grad_norm": 0.5814393162727356,
      "learning_rate": 8.307220706913308e-06,
      "loss": 0.5522,
      "step": 8750
    },
    {
      "epoch": 1.5182165163081194,
      "grad_norm": 1.0725975036621094,
      "learning_rate": 8.30169383406608e-06,
      "loss": 0.4053,
      "step": 8751
    },
    {
      "epoch": 1.5183900069396252,
      "grad_norm": 0.7847479581832886,
      "learning_rate": 8.296168318820914e-06,
      "loss": 0.4827,
      "step": 8752
    },
    {
      "epoch": 1.5185634975711313,
      "grad_norm": 0.8318679928779602,
      "learning_rate": 8.290644161819046e-06,
      "loss": 0.4396,
      "step": 8753
    },
    {
      "epoch": 1.5187369882026371,
      "grad_norm": 0.5768228769302368,
      "learning_rate": 8.28512136370158e-06,
      "loss": 0.496,
      "step": 8754
    },
    {
      "epoch": 1.518910478834143,
      "grad_norm": 0.9376530647277832,
      "learning_rate": 8.279599925109415e-06,
      "loss": 0.5413,
      "step": 8755
    },
    {
      "epoch": 1.5190839694656488,
      "grad_norm": 0.6636176705360413,
      "learning_rate": 8.274079846683346e-06,
      "loss": 0.4565,
      "step": 8756
    },
    {
      "epoch": 1.5192574600971547,
      "grad_norm": 0.828438401222229,
      "learning_rate": 8.268561129063975e-06,
      "loss": 0.4106,
      "step": 8757
    },
    {
      "epoch": 1.5194309507286605,
      "grad_norm": 0.7038265466690063,
      "learning_rate": 8.263043772891752e-06,
      "loss": 0.379,
      "step": 8758
    },
    {
      "epoch": 1.5196044413601666,
      "grad_norm": 0.6734606623649597,
      "learning_rate": 8.257527778806968e-06,
      "loss": 0.5026,
      "step": 8759
    },
    {
      "epoch": 1.5197779319916724,
      "grad_norm": 0.6494773030281067,
      "learning_rate": 8.252013147449785e-06,
      "loss": 0.4474,
      "step": 8760
    },
    {
      "epoch": 1.5199514226231785,
      "grad_norm": 0.595797061920166,
      "learning_rate": 8.246499879460149e-06,
      "loss": 0.5162,
      "step": 8761
    },
    {
      "epoch": 1.5201249132546844,
      "grad_norm": 1.0151066780090332,
      "learning_rate": 8.240987975477903e-06,
      "loss": 0.4301,
      "step": 8762
    },
    {
      "epoch": 1.5202984038861902,
      "grad_norm": 0.668760359287262,
      "learning_rate": 8.2354774361427e-06,
      "loss": 0.4299,
      "step": 8763
    },
    {
      "epoch": 1.520471894517696,
      "grad_norm": 1.0509576797485352,
      "learning_rate": 8.229968262094064e-06,
      "loss": 0.4872,
      "step": 8764
    },
    {
      "epoch": 1.520645385149202,
      "grad_norm": 0.7104885578155518,
      "learning_rate": 8.224460453971307e-06,
      "loss": 0.5177,
      "step": 8765
    },
    {
      "epoch": 1.5208188757807077,
      "grad_norm": 0.9281945824623108,
      "learning_rate": 8.218954012413647e-06,
      "loss": 0.439,
      "step": 8766
    },
    {
      "epoch": 1.5209923664122136,
      "grad_norm": 0.9999685287475586,
      "learning_rate": 8.2134489380601e-06,
      "loss": 0.5006,
      "step": 8767
    },
    {
      "epoch": 1.5211658570437196,
      "grad_norm": 1.2470293045043945,
      "learning_rate": 8.207945231549539e-06,
      "loss": 0.5039,
      "step": 8768
    },
    {
      "epoch": 1.5213393476752255,
      "grad_norm": 0.6648445725440979,
      "learning_rate": 8.202442893520666e-06,
      "loss": 0.4855,
      "step": 8769
    },
    {
      "epoch": 1.5215128383067316,
      "grad_norm": 0.7445406913757324,
      "learning_rate": 8.19694192461205e-06,
      "loss": 0.5251,
      "step": 8770
    },
    {
      "epoch": 1.5216863289382374,
      "grad_norm": 0.9122549295425415,
      "learning_rate": 8.191442325462075e-06,
      "loss": 0.3891,
      "step": 8771
    },
    {
      "epoch": 1.5218598195697433,
      "grad_norm": 0.8672484755516052,
      "learning_rate": 8.185944096708982e-06,
      "loss": 0.4661,
      "step": 8772
    },
    {
      "epoch": 1.522033310201249,
      "grad_norm": 1.122320532798767,
      "learning_rate": 8.180447238990838e-06,
      "loss": 0.4647,
      "step": 8773
    },
    {
      "epoch": 1.522206800832755,
      "grad_norm": 0.8131636381149292,
      "learning_rate": 8.17495175294556e-06,
      "loss": 0.5242,
      "step": 8774
    },
    {
      "epoch": 1.5223802914642608,
      "grad_norm": 0.5619968771934509,
      "learning_rate": 8.169457639210916e-06,
      "loss": 0.5197,
      "step": 8775
    },
    {
      "epoch": 1.5225537820957669,
      "grad_norm": 1.2358834743499756,
      "learning_rate": 8.1639648984245e-06,
      "loss": 0.4531,
      "step": 8776
    },
    {
      "epoch": 1.5227272727272727,
      "grad_norm": 1.154571294784546,
      "learning_rate": 8.158473531223748e-06,
      "loss": 0.3551,
      "step": 8777
    },
    {
      "epoch": 1.5229007633587788,
      "grad_norm": 0.7980679869651794,
      "learning_rate": 8.152983538245933e-06,
      "loss": 0.4188,
      "step": 8778
    },
    {
      "epoch": 1.5230742539902846,
      "grad_norm": 0.8363901376724243,
      "learning_rate": 8.147494920128192e-06,
      "loss": 0.4165,
      "step": 8779
    },
    {
      "epoch": 1.5232477446217905,
      "grad_norm": 0.8565142750740051,
      "learning_rate": 8.142007677507475e-06,
      "loss": 0.4725,
      "step": 8780
    },
    {
      "epoch": 1.5234212352532963,
      "grad_norm": 0.6684045791625977,
      "learning_rate": 8.136521811020584e-06,
      "loss": 0.4329,
      "step": 8781
    },
    {
      "epoch": 1.5235947258848022,
      "grad_norm": 0.7264367341995239,
      "learning_rate": 8.131037321304154e-06,
      "loss": 0.4521,
      "step": 8782
    },
    {
      "epoch": 1.523768216516308,
      "grad_norm": 0.7267224788665771,
      "learning_rate": 8.125554208994688e-06,
      "loss": 0.3289,
      "step": 8783
    },
    {
      "epoch": 1.523941707147814,
      "grad_norm": 0.7770065069198608,
      "learning_rate": 8.120072474728476e-06,
      "loss": 0.5278,
      "step": 8784
    },
    {
      "epoch": 1.52411519777932,
      "grad_norm": 0.8099251985549927,
      "learning_rate": 8.114592119141704e-06,
      "loss": 0.4939,
      "step": 8785
    },
    {
      "epoch": 1.5242886884108258,
      "grad_norm": 0.9952824711799622,
      "learning_rate": 8.10911314287037e-06,
      "loss": 0.5514,
      "step": 8786
    },
    {
      "epoch": 1.5244621790423318,
      "grad_norm": 0.8321950435638428,
      "learning_rate": 8.103635546550305e-06,
      "loss": 0.4597,
      "step": 8787
    },
    {
      "epoch": 1.5246356696738377,
      "grad_norm": 0.8479383587837219,
      "learning_rate": 8.098159330817192e-06,
      "loss": 0.36,
      "step": 8788
    },
    {
      "epoch": 1.5248091603053435,
      "grad_norm": 0.7072434425354004,
      "learning_rate": 8.092684496306573e-06,
      "loss": 0.4403,
      "step": 8789
    },
    {
      "epoch": 1.5249826509368494,
      "grad_norm": 0.6384987235069275,
      "learning_rate": 8.087211043653777e-06,
      "loss": 0.6069,
      "step": 8790
    },
    {
      "epoch": 1.5251561415683552,
      "grad_norm": 0.8090570569038391,
      "learning_rate": 8.08173897349403e-06,
      "loss": 0.4767,
      "step": 8791
    },
    {
      "epoch": 1.525329632199861,
      "grad_norm": 0.794032871723175,
      "learning_rate": 8.076268286462352e-06,
      "loss": 0.4619,
      "step": 8792
    },
    {
      "epoch": 1.5255031228313671,
      "grad_norm": 0.839773952960968,
      "learning_rate": 8.070798983193651e-06,
      "loss": 0.3959,
      "step": 8793
    },
    {
      "epoch": 1.525676613462873,
      "grad_norm": 0.9397363662719727,
      "learning_rate": 8.065331064322612e-06,
      "loss": 0.4071,
      "step": 8794
    },
    {
      "epoch": 1.525850104094379,
      "grad_norm": 1.0764880180358887,
      "learning_rate": 8.059864530483816e-06,
      "loss": 0.4209,
      "step": 8795
    },
    {
      "epoch": 1.526023594725885,
      "grad_norm": 0.7488556504249573,
      "learning_rate": 8.054399382311657e-06,
      "loss": 0.6305,
      "step": 8796
    },
    {
      "epoch": 1.5261970853573907,
      "grad_norm": 0.7319439649581909,
      "learning_rate": 8.048935620440359e-06,
      "loss": 0.5018,
      "step": 8797
    },
    {
      "epoch": 1.5263705759888966,
      "grad_norm": 3.2551815509796143,
      "learning_rate": 8.043473245504017e-06,
      "loss": 0.5421,
      "step": 8798
    },
    {
      "epoch": 1.5265440666204024,
      "grad_norm": 0.7339943051338196,
      "learning_rate": 8.038012258136536e-06,
      "loss": 0.4573,
      "step": 8799
    },
    {
      "epoch": 1.5267175572519083,
      "grad_norm": 0.7562931776046753,
      "learning_rate": 8.032552658971672e-06,
      "loss": 0.4332,
      "step": 8800
    },
    {
      "epoch": 1.5268910478834143,
      "grad_norm": 1.1799920797348022,
      "learning_rate": 8.02709444864301e-06,
      "loss": 0.4946,
      "step": 8801
    },
    {
      "epoch": 1.5270645385149202,
      "grad_norm": 1.3458267450332642,
      "learning_rate": 8.021637627784e-06,
      "loss": 0.3823,
      "step": 8802
    },
    {
      "epoch": 1.5272380291464263,
      "grad_norm": 0.8133507966995239,
      "learning_rate": 8.01618219702789e-06,
      "loss": 0.4359,
      "step": 8803
    },
    {
      "epoch": 1.527411519777932,
      "grad_norm": 0.8063660860061646,
      "learning_rate": 8.010728157007805e-06,
      "loss": 0.4825,
      "step": 8804
    },
    {
      "epoch": 1.527585010409438,
      "grad_norm": 0.7743210196495056,
      "learning_rate": 8.005275508356689e-06,
      "loss": 0.5447,
      "step": 8805
    },
    {
      "epoch": 1.5277585010409438,
      "grad_norm": 0.9252443909645081,
      "learning_rate": 7.999824251707324e-06,
      "loss": 0.4867,
      "step": 8806
    },
    {
      "epoch": 1.5279319916724496,
      "grad_norm": 0.9197413921356201,
      "learning_rate": 7.994374387692335e-06,
      "loss": 0.3779,
      "step": 8807
    },
    {
      "epoch": 1.5281054823039555,
      "grad_norm": 0.9694995284080505,
      "learning_rate": 7.98892591694419e-06,
      "loss": 0.4221,
      "step": 8808
    },
    {
      "epoch": 1.5282789729354613,
      "grad_norm": 0.7210115194320679,
      "learning_rate": 7.98347884009519e-06,
      "loss": 0.5466,
      "step": 8809
    },
    {
      "epoch": 1.5284524635669674,
      "grad_norm": 0.7455983757972717,
      "learning_rate": 7.978033157777473e-06,
      "loss": 0.3805,
      "step": 8810
    },
    {
      "epoch": 1.5286259541984732,
      "grad_norm": 0.7473982572555542,
      "learning_rate": 7.97258887062301e-06,
      "loss": 0.449,
      "step": 8811
    },
    {
      "epoch": 1.5287994448299793,
      "grad_norm": 0.7718564867973328,
      "learning_rate": 7.967145979263637e-06,
      "loss": 0.4937,
      "step": 8812
    },
    {
      "epoch": 1.5289729354614852,
      "grad_norm": 0.6857679486274719,
      "learning_rate": 7.961704484330979e-06,
      "loss": 0.5232,
      "step": 8813
    },
    {
      "epoch": 1.529146426092991,
      "grad_norm": 0.7407565116882324,
      "learning_rate": 7.956264386456551e-06,
      "loss": 0.4028,
      "step": 8814
    },
    {
      "epoch": 1.5293199167244969,
      "grad_norm": 0.9997866749763489,
      "learning_rate": 7.950825686271674e-06,
      "loss": 0.4673,
      "step": 8815
    },
    {
      "epoch": 1.5294934073560027,
      "grad_norm": 0.706325352191925,
      "learning_rate": 7.945388384407518e-06,
      "loss": 0.4778,
      "step": 8816
    },
    {
      "epoch": 1.5296668979875085,
      "grad_norm": 0.9359931349754333,
      "learning_rate": 7.939952481495076e-06,
      "loss": 0.3251,
      "step": 8817
    },
    {
      "epoch": 1.5298403886190146,
      "grad_norm": 5.180793285369873,
      "learning_rate": 7.934517978165211e-06,
      "loss": 0.382,
      "step": 8818
    },
    {
      "epoch": 1.5300138792505205,
      "grad_norm": 0.7217835187911987,
      "learning_rate": 7.929084875048594e-06,
      "loss": 0.4458,
      "step": 8819
    },
    {
      "epoch": 1.5301873698820265,
      "grad_norm": 0.6770364046096802,
      "learning_rate": 7.92365317277574e-06,
      "loss": 0.5565,
      "step": 8820
    },
    {
      "epoch": 1.5303608605135324,
      "grad_norm": 0.9632218480110168,
      "learning_rate": 7.918222871977004e-06,
      "loss": 0.4833,
      "step": 8821
    },
    {
      "epoch": 1.5305343511450382,
      "grad_norm": 0.9420800805091858,
      "learning_rate": 7.912793973282584e-06,
      "loss": 0.4531,
      "step": 8822
    },
    {
      "epoch": 1.530707841776544,
      "grad_norm": 0.8090986609458923,
      "learning_rate": 7.90736647732251e-06,
      "loss": 0.4592,
      "step": 8823
    },
    {
      "epoch": 1.53088133240805,
      "grad_norm": 0.8491949439048767,
      "learning_rate": 7.90194038472665e-06,
      "loss": 0.3547,
      "step": 8824
    },
    {
      "epoch": 1.5310548230395558,
      "grad_norm": 0.8736350536346436,
      "learning_rate": 7.896515696124703e-06,
      "loss": 0.3782,
      "step": 8825
    },
    {
      "epoch": 1.5312283136710616,
      "grad_norm": 0.7773745656013489,
      "learning_rate": 7.891092412146204e-06,
      "loss": 0.4202,
      "step": 8826
    },
    {
      "epoch": 1.5314018043025677,
      "grad_norm": 0.7690825462341309,
      "learning_rate": 7.885670533420548e-06,
      "loss": 0.4293,
      "step": 8827
    },
    {
      "epoch": 1.5315752949340735,
      "grad_norm": 2.417379379272461,
      "learning_rate": 7.880250060576943e-06,
      "loss": 0.3743,
      "step": 8828
    },
    {
      "epoch": 1.5317487855655796,
      "grad_norm": 0.6922433376312256,
      "learning_rate": 7.874830994244443e-06,
      "loss": 0.5067,
      "step": 8829
    },
    {
      "epoch": 1.5319222761970854,
      "grad_norm": 0.7780335545539856,
      "learning_rate": 7.869413335051926e-06,
      "loss": 0.535,
      "step": 8830
    },
    {
      "epoch": 1.5320957668285913,
      "grad_norm": 0.9590572714805603,
      "learning_rate": 7.863997083628139e-06,
      "loss": 0.5477,
      "step": 8831
    },
    {
      "epoch": 1.5322692574600971,
      "grad_norm": 0.8635021448135376,
      "learning_rate": 7.85858224060162e-06,
      "loss": 0.4388,
      "step": 8832
    },
    {
      "epoch": 1.532442748091603,
      "grad_norm": 0.629139244556427,
      "learning_rate": 7.853168806600784e-06,
      "loss": 0.5092,
      "step": 8833
    },
    {
      "epoch": 1.5326162387231088,
      "grad_norm": 1.41582453250885,
      "learning_rate": 7.847756782253864e-06,
      "loss": 0.456,
      "step": 8834
    },
    {
      "epoch": 1.5327897293546149,
      "grad_norm": 1.408156156539917,
      "learning_rate": 7.842346168188927e-06,
      "loss": 0.4902,
      "step": 8835
    },
    {
      "epoch": 1.5329632199861207,
      "grad_norm": 0.7910613417625427,
      "learning_rate": 7.836936965033881e-06,
      "loss": 0.4017,
      "step": 8836
    },
    {
      "epoch": 1.5331367106176268,
      "grad_norm": 0.842945396900177,
      "learning_rate": 7.831529173416476e-06,
      "loss": 0.4017,
      "step": 8837
    },
    {
      "epoch": 1.5333102012491326,
      "grad_norm": 0.7382654547691345,
      "learning_rate": 7.826122793964293e-06,
      "loss": 0.3668,
      "step": 8838
    },
    {
      "epoch": 1.5334836918806385,
      "grad_norm": 0.6733341813087463,
      "learning_rate": 7.820717827304743e-06,
      "loss": 0.48,
      "step": 8839
    },
    {
      "epoch": 1.5336571825121443,
      "grad_norm": 1.4515665769577026,
      "learning_rate": 7.815314274065074e-06,
      "loss": 0.3545,
      "step": 8840
    },
    {
      "epoch": 1.5338306731436502,
      "grad_norm": 0.868407130241394,
      "learning_rate": 7.809912134872395e-06,
      "loss": 0.3721,
      "step": 8841
    },
    {
      "epoch": 1.534004163775156,
      "grad_norm": 0.7506738305091858,
      "learning_rate": 7.804511410353603e-06,
      "loss": 0.4023,
      "step": 8842
    },
    {
      "epoch": 1.534177654406662,
      "grad_norm": 0.7305961847305298,
      "learning_rate": 7.79911210113548e-06,
      "loss": 0.5459,
      "step": 8843
    },
    {
      "epoch": 1.534351145038168,
      "grad_norm": 0.6782276630401611,
      "learning_rate": 7.793714207844616e-06,
      "loss": 0.4285,
      "step": 8844
    },
    {
      "epoch": 1.5345246356696738,
      "grad_norm": 0.6881431341171265,
      "learning_rate": 7.788317731107431e-06,
      "loss": 0.5073,
      "step": 8845
    },
    {
      "epoch": 1.5346981263011799,
      "grad_norm": 0.6842974424362183,
      "learning_rate": 7.782922671550213e-06,
      "loss": 0.4827,
      "step": 8846
    },
    {
      "epoch": 1.5348716169326857,
      "grad_norm": 0.9670971632003784,
      "learning_rate": 7.777529029799053e-06,
      "loss": 0.5076,
      "step": 8847
    },
    {
      "epoch": 1.5350451075641915,
      "grad_norm": 0.9345824718475342,
      "learning_rate": 7.772136806479891e-06,
      "loss": 0.5463,
      "step": 8848
    },
    {
      "epoch": 1.5352185981956974,
      "grad_norm": 0.8835152983665466,
      "learning_rate": 7.766746002218495e-06,
      "loss": 0.4781,
      "step": 8849
    },
    {
      "epoch": 1.5353920888272032,
      "grad_norm": 0.7977238297462463,
      "learning_rate": 7.761356617640485e-06,
      "loss": 0.3837,
      "step": 8850
    },
    {
      "epoch": 1.535565579458709,
      "grad_norm": 0.5723086595535278,
      "learning_rate": 7.755968653371302e-06,
      "loss": 0.4725,
      "step": 8851
    },
    {
      "epoch": 1.5357390700902152,
      "grad_norm": 0.766066312789917,
      "learning_rate": 7.750582110036225e-06,
      "loss": 0.3985,
      "step": 8852
    },
    {
      "epoch": 1.535912560721721,
      "grad_norm": 0.8079668879508972,
      "learning_rate": 7.745196988260357e-06,
      "loss": 0.5059,
      "step": 8853
    },
    {
      "epoch": 1.536086051353227,
      "grad_norm": 1.4065872430801392,
      "learning_rate": 7.739813288668677e-06,
      "loss": 0.3893,
      "step": 8854
    },
    {
      "epoch": 1.536259541984733,
      "grad_norm": 0.8298816680908203,
      "learning_rate": 7.734431011885932e-06,
      "loss": 0.5157,
      "step": 8855
    },
    {
      "epoch": 1.5364330326162388,
      "grad_norm": 0.7164503931999207,
      "learning_rate": 7.72905015853677e-06,
      "loss": 0.4236,
      "step": 8856
    },
    {
      "epoch": 1.5366065232477446,
      "grad_norm": 0.7510810494422913,
      "learning_rate": 7.723670729245636e-06,
      "loss": 0.4879,
      "step": 8857
    },
    {
      "epoch": 1.5367800138792505,
      "grad_norm": 1.4038009643554688,
      "learning_rate": 7.718292724636815e-06,
      "loss": 0.4797,
      "step": 8858
    },
    {
      "epoch": 1.5369535045107563,
      "grad_norm": 0.6696471571922302,
      "learning_rate": 7.712916145334432e-06,
      "loss": 0.6,
      "step": 8859
    },
    {
      "epoch": 1.5371269951422624,
      "grad_norm": 1.284012794494629,
      "learning_rate": 7.70754099196246e-06,
      "loss": 0.5154,
      "step": 8860
    },
    {
      "epoch": 1.5373004857737682,
      "grad_norm": 1.2538352012634277,
      "learning_rate": 7.702167265144664e-06,
      "loss": 0.4362,
      "step": 8861
    },
    {
      "epoch": 1.5374739764052743,
      "grad_norm": 0.7628593444824219,
      "learning_rate": 7.696794965504695e-06,
      "loss": 0.4001,
      "step": 8862
    },
    {
      "epoch": 1.5376474670367801,
      "grad_norm": 1.3229341506958008,
      "learning_rate": 7.691424093666007e-06,
      "loss": 0.5374,
      "step": 8863
    },
    {
      "epoch": 1.537820957668286,
      "grad_norm": 0.7463794350624084,
      "learning_rate": 7.686054650251893e-06,
      "loss": 0.5352,
      "step": 8864
    },
    {
      "epoch": 1.5379944482997918,
      "grad_norm": 0.8578668236732483,
      "learning_rate": 7.68068663588548e-06,
      "loss": 0.3442,
      "step": 8865
    },
    {
      "epoch": 1.5381679389312977,
      "grad_norm": 0.9100556373596191,
      "learning_rate": 7.675320051189746e-06,
      "loss": 0.3804,
      "step": 8866
    },
    {
      "epoch": 1.5383414295628035,
      "grad_norm": 0.9146038889884949,
      "learning_rate": 7.66995489678748e-06,
      "loss": 0.4378,
      "step": 8867
    },
    {
      "epoch": 1.5385149201943094,
      "grad_norm": 0.9123438000679016,
      "learning_rate": 7.664591173301315e-06,
      "loss": 0.4573,
      "step": 8868
    },
    {
      "epoch": 1.5386884108258154,
      "grad_norm": 0.6044626832008362,
      "learning_rate": 7.65922888135371e-06,
      "loss": 0.5498,
      "step": 8869
    },
    {
      "epoch": 1.5388619014573213,
      "grad_norm": 0.7664366960525513,
      "learning_rate": 7.65386802156698e-06,
      "loss": 0.3813,
      "step": 8870
    },
    {
      "epoch": 1.5390353920888273,
      "grad_norm": 0.5493521690368652,
      "learning_rate": 7.648508594563256e-06,
      "loss": 0.4487,
      "step": 8871
    },
    {
      "epoch": 1.5392088827203332,
      "grad_norm": 0.8492365479469299,
      "learning_rate": 7.64315060096449e-06,
      "loss": 0.4982,
      "step": 8872
    },
    {
      "epoch": 1.539382373351839,
      "grad_norm": 0.7712075114250183,
      "learning_rate": 7.637794041392512e-06,
      "loss": 0.4443,
      "step": 8873
    },
    {
      "epoch": 1.5395558639833449,
      "grad_norm": 1.18946373462677,
      "learning_rate": 7.632438916468928e-06,
      "loss": 0.3827,
      "step": 8874
    },
    {
      "epoch": 1.5397293546148507,
      "grad_norm": 0.9963278770446777,
      "learning_rate": 7.627085226815223e-06,
      "loss": 0.3531,
      "step": 8875
    },
    {
      "epoch": 1.5399028452463566,
      "grad_norm": 0.8705530166625977,
      "learning_rate": 7.621732973052696e-06,
      "loss": 0.4442,
      "step": 8876
    },
    {
      "epoch": 1.5400763358778626,
      "grad_norm": 0.7073082327842712,
      "learning_rate": 7.616382155802482e-06,
      "loss": 0.4034,
      "step": 8877
    },
    {
      "epoch": 1.5402498265093685,
      "grad_norm": 0.617682695388794,
      "learning_rate": 7.611032775685541e-06,
      "loss": 0.3966,
      "step": 8878
    },
    {
      "epoch": 1.5404233171408745,
      "grad_norm": 0.7195889353752136,
      "learning_rate": 7.605684833322686e-06,
      "loss": 0.5928,
      "step": 8879
    },
    {
      "epoch": 1.5405968077723804,
      "grad_norm": 0.8288052082061768,
      "learning_rate": 7.600338329334554e-06,
      "loss": 0.4183,
      "step": 8880
    },
    {
      "epoch": 1.5407702984038862,
      "grad_norm": 0.8420101404190063,
      "learning_rate": 7.5949932643416036e-06,
      "loss": 0.4926,
      "step": 8881
    },
    {
      "epoch": 1.540943789035392,
      "grad_norm": 0.6889872550964355,
      "learning_rate": 7.5896496389641336e-06,
      "loss": 0.597,
      "step": 8882
    },
    {
      "epoch": 1.541117279666898,
      "grad_norm": 0.6035048961639404,
      "learning_rate": 7.584307453822297e-06,
      "loss": 0.4974,
      "step": 8883
    },
    {
      "epoch": 1.5412907702984038,
      "grad_norm": 0.8604782223701477,
      "learning_rate": 7.5789667095360355e-06,
      "loss": 0.3951,
      "step": 8884
    },
    {
      "epoch": 1.5414642609299096,
      "grad_norm": 0.8894672393798828,
      "learning_rate": 7.5736274067251655e-06,
      "loss": 0.5442,
      "step": 8885
    },
    {
      "epoch": 1.5416377515614157,
      "grad_norm": 0.9153072834014893,
      "learning_rate": 7.568289546009316e-06,
      "loss": 0.4735,
      "step": 8886
    },
    {
      "epoch": 1.5418112421929215,
      "grad_norm": 1.0435465574264526,
      "learning_rate": 7.562953128007951e-06,
      "loss": 0.4873,
      "step": 8887
    },
    {
      "epoch": 1.5419847328244276,
      "grad_norm": 0.8539049029350281,
      "learning_rate": 7.557618153340358e-06,
      "loss": 0.4045,
      "step": 8888
    },
    {
      "epoch": 1.5421582234559335,
      "grad_norm": 1.1509201526641846,
      "learning_rate": 7.5522846226256855e-06,
      "loss": 0.4175,
      "step": 8889
    },
    {
      "epoch": 1.5423317140874393,
      "grad_norm": 0.8957439064979553,
      "learning_rate": 7.546952536482888e-06,
      "loss": 0.4139,
      "step": 8890
    },
    {
      "epoch": 1.5425052047189451,
      "grad_norm": 0.7355961799621582,
      "learning_rate": 7.541621895530757e-06,
      "loss": 0.4144,
      "step": 8891
    },
    {
      "epoch": 1.542678695350451,
      "grad_norm": 1.029009461402893,
      "learning_rate": 7.536292700387924e-06,
      "loss": 0.5231,
      "step": 8892
    },
    {
      "epoch": 1.5428521859819568,
      "grad_norm": 0.7891607284545898,
      "learning_rate": 7.530964951672836e-06,
      "loss": 0.448,
      "step": 8893
    },
    {
      "epoch": 1.543025676613463,
      "grad_norm": 0.8846710324287415,
      "learning_rate": 7.5256386500038055e-06,
      "loss": 0.4792,
      "step": 8894
    },
    {
      "epoch": 1.5431991672449688,
      "grad_norm": 0.9755485653877258,
      "learning_rate": 7.520313795998943e-06,
      "loss": 0.4188,
      "step": 8895
    },
    {
      "epoch": 1.5433726578764748,
      "grad_norm": 0.7196856737136841,
      "learning_rate": 7.5149903902762066e-06,
      "loss": 0.3968,
      "step": 8896
    },
    {
      "epoch": 1.5435461485079807,
      "grad_norm": 0.8937692642211914,
      "learning_rate": 7.5096684334533765e-06,
      "loss": 0.3875,
      "step": 8897
    },
    {
      "epoch": 1.5437196391394865,
      "grad_norm": 0.7358959317207336,
      "learning_rate": 7.504347926148086e-06,
      "loss": 0.4679,
      "step": 8898
    },
    {
      "epoch": 1.5438931297709924,
      "grad_norm": 0.9264071583747864,
      "learning_rate": 7.499028868977778e-06,
      "loss": 0.5386,
      "step": 8899
    },
    {
      "epoch": 1.5440666204024982,
      "grad_norm": 0.8067649006843567,
      "learning_rate": 7.4937112625597375e-06,
      "loss": 0.4251,
      "step": 8900
    },
    {
      "epoch": 1.544240111034004,
      "grad_norm": 0.9063412547111511,
      "learning_rate": 7.48839510751107e-06,
      "loss": 0.4148,
      "step": 8901
    },
    {
      "epoch": 1.5444136016655101,
      "grad_norm": 0.7851597666740417,
      "learning_rate": 7.483080404448744e-06,
      "loss": 0.4479,
      "step": 8902
    },
    {
      "epoch": 1.544587092297016,
      "grad_norm": 1.094315528869629,
      "learning_rate": 7.477767153989508e-06,
      "loss": 0.3554,
      "step": 8903
    },
    {
      "epoch": 1.5447605829285218,
      "grad_norm": 0.879088819026947,
      "learning_rate": 7.472455356749992e-06,
      "loss": 0.5227,
      "step": 8904
    },
    {
      "epoch": 1.5449340735600279,
      "grad_norm": 0.6144444346427917,
      "learning_rate": 7.4671450133466285e-06,
      "loss": 0.5631,
      "step": 8905
    },
    {
      "epoch": 1.5451075641915337,
      "grad_norm": 0.6775422692298889,
      "learning_rate": 7.461836124395692e-06,
      "loss": 0.4895,
      "step": 8906
    },
    {
      "epoch": 1.5452810548230396,
      "grad_norm": 0.8490018248558044,
      "learning_rate": 7.456528690513274e-06,
      "loss": 0.4784,
      "step": 8907
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.8370487093925476,
      "learning_rate": 7.451222712315325e-06,
      "loss": 0.4385,
      "step": 8908
    },
    {
      "epoch": 1.5456280360860513,
      "grad_norm": 0.7490215301513672,
      "learning_rate": 7.445918190417603e-06,
      "loss": 0.3976,
      "step": 8909
    },
    {
      "epoch": 1.545801526717557,
      "grad_norm": 0.7567493319511414,
      "learning_rate": 7.440615125435702e-06,
      "loss": 0.4846,
      "step": 8910
    },
    {
      "epoch": 1.5459750173490632,
      "grad_norm": 1.395071268081665,
      "learning_rate": 7.435313517985043e-06,
      "loss": 0.3939,
      "step": 8911
    },
    {
      "epoch": 1.546148507980569,
      "grad_norm": 1.2977927923202515,
      "learning_rate": 7.430013368680908e-06,
      "loss": 0.4623,
      "step": 8912
    },
    {
      "epoch": 1.546321998612075,
      "grad_norm": 0.7360656261444092,
      "learning_rate": 7.424714678138351e-06,
      "loss": 0.4561,
      "step": 8913
    },
    {
      "epoch": 1.546495489243581,
      "grad_norm": 0.6926393508911133,
      "learning_rate": 7.419417446972319e-06,
      "loss": 0.4658,
      "step": 8914
    },
    {
      "epoch": 1.5466689798750868,
      "grad_norm": 0.8517128229141235,
      "learning_rate": 7.41412167579755e-06,
      "loss": 0.5496,
      "step": 8915
    },
    {
      "epoch": 1.5468424705065926,
      "grad_norm": 2.9771463871002197,
      "learning_rate": 7.408827365228625e-06,
      "loss": 0.5194,
      "step": 8916
    },
    {
      "epoch": 1.5470159611380985,
      "grad_norm": 0.8298816084861755,
      "learning_rate": 7.403534515879951e-06,
      "loss": 0.4966,
      "step": 8917
    },
    {
      "epoch": 1.5471894517696043,
      "grad_norm": 0.7964062690734863,
      "learning_rate": 7.3982431283657805e-06,
      "loss": 0.4346,
      "step": 8918
    },
    {
      "epoch": 1.5473629424011104,
      "grad_norm": 0.8984449505805969,
      "learning_rate": 7.392953203300177e-06,
      "loss": 0.4348,
      "step": 8919
    },
    {
      "epoch": 1.5475364330326162,
      "grad_norm": 0.8992365598678589,
      "learning_rate": 7.38766474129704e-06,
      "loss": 0.4463,
      "step": 8920
    },
    {
      "epoch": 1.5477099236641223,
      "grad_norm": 0.724549412727356,
      "learning_rate": 7.3823777429701125e-06,
      "loss": 0.4171,
      "step": 8921
    },
    {
      "epoch": 1.5478834142956281,
      "grad_norm": 0.8949272632598877,
      "learning_rate": 7.37709220893295e-06,
      "loss": 0.5023,
      "step": 8922
    },
    {
      "epoch": 1.548056904927134,
      "grad_norm": 0.8588585257530212,
      "learning_rate": 7.371808139798944e-06,
      "loss": 0.3899,
      "step": 8923
    },
    {
      "epoch": 1.5482303955586398,
      "grad_norm": 0.6372013092041016,
      "learning_rate": 7.3665255361813125e-06,
      "loss": 0.4033,
      "step": 8924
    },
    {
      "epoch": 1.5484038861901457,
      "grad_norm": 0.9637125730514526,
      "learning_rate": 7.361244398693128e-06,
      "loss": 0.363,
      "step": 8925
    },
    {
      "epoch": 1.5485773768216515,
      "grad_norm": 0.5929570198059082,
      "learning_rate": 7.355964727947242e-06,
      "loss": 0.3713,
      "step": 8926
    },
    {
      "epoch": 1.5487508674531574,
      "grad_norm": 0.9674267768859863,
      "learning_rate": 7.350686524556392e-06,
      "loss": 0.4985,
      "step": 8927
    },
    {
      "epoch": 1.5489243580846634,
      "grad_norm": 0.7561858296394348,
      "learning_rate": 7.3454097891331085e-06,
      "loss": 0.527,
      "step": 8928
    },
    {
      "epoch": 1.5490978487161693,
      "grad_norm": 0.7522987127304077,
      "learning_rate": 7.340134522289766e-06,
      "loss": 0.3781,
      "step": 8929
    },
    {
      "epoch": 1.5492713393476754,
      "grad_norm": 0.8440223336219788,
      "learning_rate": 7.334860724638555e-06,
      "loss": 0.3956,
      "step": 8930
    },
    {
      "epoch": 1.5494448299791812,
      "grad_norm": 0.9758820533752441,
      "learning_rate": 7.329588396791529e-06,
      "loss": 0.3809,
      "step": 8931
    },
    {
      "epoch": 1.549618320610687,
      "grad_norm": 0.6462194323539734,
      "learning_rate": 7.3243175393605215e-06,
      "loss": 0.4856,
      "step": 8932
    },
    {
      "epoch": 1.549791811242193,
      "grad_norm": 0.7615544199943542,
      "learning_rate": 7.31904815295724e-06,
      "loss": 0.5265,
      "step": 8933
    },
    {
      "epoch": 1.5499653018736987,
      "grad_norm": 0.737786054611206,
      "learning_rate": 7.313780238193195e-06,
      "loss": 0.5164,
      "step": 8934
    },
    {
      "epoch": 1.5501387925052046,
      "grad_norm": 1.0981014966964722,
      "learning_rate": 7.308513795679737e-06,
      "loss": 0.485,
      "step": 8935
    },
    {
      "epoch": 1.5503122831367107,
      "grad_norm": 1.1777385473251343,
      "learning_rate": 7.303248826028036e-06,
      "loss": 0.4087,
      "step": 8936
    },
    {
      "epoch": 1.5504857737682165,
      "grad_norm": 0.7753485441207886,
      "learning_rate": 7.297985329849106e-06,
      "loss": 0.4117,
      "step": 8937
    },
    {
      "epoch": 1.5506592643997226,
      "grad_norm": 1.2526369094848633,
      "learning_rate": 7.292723307753784e-06,
      "loss": 0.476,
      "step": 8938
    },
    {
      "epoch": 1.5508327550312284,
      "grad_norm": 0.814134955406189,
      "learning_rate": 7.287462760352726e-06,
      "loss": 0.5787,
      "step": 8939
    },
    {
      "epoch": 1.5510062456627343,
      "grad_norm": 0.7823880910873413,
      "learning_rate": 7.282203688256422e-06,
      "loss": 0.5027,
      "step": 8940
    },
    {
      "epoch": 1.55117973629424,
      "grad_norm": 0.5917524099349976,
      "learning_rate": 7.276946092075205e-06,
      "loss": 0.4935,
      "step": 8941
    },
    {
      "epoch": 1.551353226925746,
      "grad_norm": 1.1361775398254395,
      "learning_rate": 7.27168997241922e-06,
      "loss": 0.4821,
      "step": 8942
    },
    {
      "epoch": 1.5515267175572518,
      "grad_norm": 0.9524984955787659,
      "learning_rate": 7.266435329898447e-06,
      "loss": 0.4102,
      "step": 8943
    },
    {
      "epoch": 1.5517002081887576,
      "grad_norm": 0.6795886158943176,
      "learning_rate": 7.261182165122689e-06,
      "loss": 0.4523,
      "step": 8944
    },
    {
      "epoch": 1.5518736988202637,
      "grad_norm": 0.6872532963752747,
      "learning_rate": 7.255930478701578e-06,
      "loss": 0.5601,
      "step": 8945
    },
    {
      "epoch": 1.5520471894517696,
      "grad_norm": 1.181846261024475,
      "learning_rate": 7.250680271244593e-06,
      "loss": 0.4717,
      "step": 8946
    },
    {
      "epoch": 1.5522206800832756,
      "grad_norm": 0.7594887614250183,
      "learning_rate": 7.245431543361016e-06,
      "loss": 0.4582,
      "step": 8947
    },
    {
      "epoch": 1.5523941707147815,
      "grad_norm": 1.0892949104309082,
      "learning_rate": 7.240184295659971e-06,
      "loss": 0.3826,
      "step": 8948
    },
    {
      "epoch": 1.5525676613462873,
      "grad_norm": 0.7497468590736389,
      "learning_rate": 7.234938528750402e-06,
      "loss": 0.4995,
      "step": 8949
    },
    {
      "epoch": 1.5527411519777932,
      "grad_norm": 0.7543135285377502,
      "learning_rate": 7.229694243241097e-06,
      "loss": 0.4598,
      "step": 8950
    },
    {
      "epoch": 1.552914642609299,
      "grad_norm": 0.8440332412719727,
      "learning_rate": 7.224451439740654e-06,
      "loss": 0.5125,
      "step": 8951
    },
    {
      "epoch": 1.5530881332408049,
      "grad_norm": 0.9373429417610168,
      "learning_rate": 7.219210118857509e-06,
      "loss": 0.5015,
      "step": 8952
    },
    {
      "epoch": 1.553261623872311,
      "grad_norm": 0.8574715256690979,
      "learning_rate": 7.213970281199913e-06,
      "loss": 0.3992,
      "step": 8953
    },
    {
      "epoch": 1.5534351145038168,
      "grad_norm": 1.7464864253997803,
      "learning_rate": 7.208731927375982e-06,
      "loss": 0.344,
      "step": 8954
    },
    {
      "epoch": 1.5536086051353228,
      "grad_norm": 0.7708041071891785,
      "learning_rate": 7.203495057993599e-06,
      "loss": 0.3959,
      "step": 8955
    },
    {
      "epoch": 1.5537820957668287,
      "grad_norm": 0.8572695851325989,
      "learning_rate": 7.198259673660535e-06,
      "loss": 0.4666,
      "step": 8956
    },
    {
      "epoch": 1.5539555863983345,
      "grad_norm": 0.9208317399024963,
      "learning_rate": 7.19302577498435e-06,
      "loss": 0.3711,
      "step": 8957
    },
    {
      "epoch": 1.5541290770298404,
      "grad_norm": 0.8072161078453064,
      "learning_rate": 7.187793362572451e-06,
      "loss": 0.4554,
      "step": 8958
    },
    {
      "epoch": 1.5543025676613462,
      "grad_norm": 0.6844647526741028,
      "learning_rate": 7.1825624370320505e-06,
      "loss": 0.3895,
      "step": 8959
    },
    {
      "epoch": 1.554476058292852,
      "grad_norm": 0.5709481835365295,
      "learning_rate": 7.17733299897023e-06,
      "loss": 0.6091,
      "step": 8960
    },
    {
      "epoch": 1.5546495489243581,
      "grad_norm": 0.6680960059165955,
      "learning_rate": 7.172105048993842e-06,
      "loss": 0.5458,
      "step": 8961
    },
    {
      "epoch": 1.554823039555864,
      "grad_norm": 0.6700970530509949,
      "learning_rate": 7.166878587709618e-06,
      "loss": 0.4829,
      "step": 8962
    },
    {
      "epoch": 1.5549965301873698,
      "grad_norm": 0.7285874485969543,
      "learning_rate": 7.1616536157240865e-06,
      "loss": 0.4807,
      "step": 8963
    },
    {
      "epoch": 1.555170020818876,
      "grad_norm": 0.7929092645645142,
      "learning_rate": 7.156430133643613e-06,
      "loss": 0.4189,
      "step": 8964
    },
    {
      "epoch": 1.5553435114503817,
      "grad_norm": 1.0222623348236084,
      "learning_rate": 7.151208142074382e-06,
      "loss": 0.5461,
      "step": 8965
    },
    {
      "epoch": 1.5555170020818876,
      "grad_norm": 0.7129715085029602,
      "learning_rate": 7.145987641622423e-06,
      "loss": 0.6161,
      "step": 8966
    },
    {
      "epoch": 1.5556904927133934,
      "grad_norm": 0.897832453250885,
      "learning_rate": 7.1407686328935774e-06,
      "loss": 0.4286,
      "step": 8967
    },
    {
      "epoch": 1.5558639833448993,
      "grad_norm": 0.5735976696014404,
      "learning_rate": 7.1355511164935085e-06,
      "loss": 0.4882,
      "step": 8968
    },
    {
      "epoch": 1.5560374739764051,
      "grad_norm": 0.7005171775817871,
      "learning_rate": 7.130335093027731e-06,
      "loss": 0.4387,
      "step": 8969
    },
    {
      "epoch": 1.5562109646079112,
      "grad_norm": 0.9279868602752686,
      "learning_rate": 7.125120563101562e-06,
      "loss": 0.42,
      "step": 8970
    },
    {
      "epoch": 1.556384455239417,
      "grad_norm": 0.9232456088066101,
      "learning_rate": 7.1199075273201515e-06,
      "loss": 0.5034,
      "step": 8971
    },
    {
      "epoch": 1.556557945870923,
      "grad_norm": 0.8271253705024719,
      "learning_rate": 7.114695986288476e-06,
      "loss": 0.5229,
      "step": 8972
    },
    {
      "epoch": 1.556731436502429,
      "grad_norm": 0.8400318026542664,
      "learning_rate": 7.109485940611358e-06,
      "loss": 0.41,
      "step": 8973
    },
    {
      "epoch": 1.5569049271339348,
      "grad_norm": 0.868891179561615,
      "learning_rate": 7.104277390893404e-06,
      "loss": 0.5298,
      "step": 8974
    },
    {
      "epoch": 1.5570784177654406,
      "grad_norm": 0.691235363483429,
      "learning_rate": 7.099070337739094e-06,
      "loss": 0.5015,
      "step": 8975
    },
    {
      "epoch": 1.5572519083969465,
      "grad_norm": 0.6555090546607971,
      "learning_rate": 7.0938647817527014e-06,
      "loss": 0.5098,
      "step": 8976
    },
    {
      "epoch": 1.5574253990284523,
      "grad_norm": 0.8201124668121338,
      "learning_rate": 7.088660723538339e-06,
      "loss": 0.3494,
      "step": 8977
    },
    {
      "epoch": 1.5575988896599584,
      "grad_norm": 1.1224524974822998,
      "learning_rate": 7.083458163699939e-06,
      "loss": 0.3679,
      "step": 8978
    },
    {
      "epoch": 1.5577723802914643,
      "grad_norm": 0.7555831074714661,
      "learning_rate": 7.0782571028412774e-06,
      "loss": 0.3779,
      "step": 8979
    },
    {
      "epoch": 1.5579458709229703,
      "grad_norm": 0.7846662402153015,
      "learning_rate": 7.073057541565933e-06,
      "loss": 0.4219,
      "step": 8980
    },
    {
      "epoch": 1.5581193615544762,
      "grad_norm": 0.8801530003547668,
      "learning_rate": 7.067859480477326e-06,
      "loss": 0.4624,
      "step": 8981
    },
    {
      "epoch": 1.558292852185982,
      "grad_norm": 0.9899958968162537,
      "learning_rate": 7.062662920178689e-06,
      "loss": 0.4891,
      "step": 8982
    },
    {
      "epoch": 1.5584663428174879,
      "grad_norm": 1.2896184921264648,
      "learning_rate": 7.057467861273106e-06,
      "loss": 0.4803,
      "step": 8983
    },
    {
      "epoch": 1.5586398334489937,
      "grad_norm": 0.8861548900604248,
      "learning_rate": 7.052274304363449e-06,
      "loss": 0.4966,
      "step": 8984
    },
    {
      "epoch": 1.5588133240804996,
      "grad_norm": 0.8386437892913818,
      "learning_rate": 7.0470822500524504e-06,
      "loss": 0.3698,
      "step": 8985
    },
    {
      "epoch": 1.5589868147120054,
      "grad_norm": 0.657148003578186,
      "learning_rate": 7.041891698942649e-06,
      "loss": 0.4625,
      "step": 8986
    },
    {
      "epoch": 1.5591603053435115,
      "grad_norm": 0.688218891620636,
      "learning_rate": 7.036702651636416e-06,
      "loss": 0.554,
      "step": 8987
    },
    {
      "epoch": 1.5593337959750173,
      "grad_norm": 0.76467365026474,
      "learning_rate": 7.03151510873594e-06,
      "loss": 0.4149,
      "step": 8988
    },
    {
      "epoch": 1.5595072866065234,
      "grad_norm": 0.6468123197555542,
      "learning_rate": 7.0263290708432515e-06,
      "loss": 0.5258,
      "step": 8989
    },
    {
      "epoch": 1.5596807772380292,
      "grad_norm": 0.8571133017539978,
      "learning_rate": 7.021144538560194e-06,
      "loss": 0.554,
      "step": 8990
    },
    {
      "epoch": 1.559854267869535,
      "grad_norm": 0.6320785880088806,
      "learning_rate": 7.015961512488434e-06,
      "loss": 0.4454,
      "step": 8991
    },
    {
      "epoch": 1.560027758501041,
      "grad_norm": 0.7255579233169556,
      "learning_rate": 7.010779993229471e-06,
      "loss": 0.3877,
      "step": 8992
    },
    {
      "epoch": 1.5602012491325468,
      "grad_norm": 0.8437013626098633,
      "learning_rate": 7.005599981384618e-06,
      "loss": 0.459,
      "step": 8993
    },
    {
      "epoch": 1.5603747397640526,
      "grad_norm": 0.7917951345443726,
      "learning_rate": 7.000421477555038e-06,
      "loss": 0.5112,
      "step": 8994
    },
    {
      "epoch": 1.5605482303955587,
      "grad_norm": 0.7351046800613403,
      "learning_rate": 6.9952444823416894e-06,
      "loss": 0.5686,
      "step": 8995
    },
    {
      "epoch": 1.5607217210270645,
      "grad_norm": 0.6726951003074646,
      "learning_rate": 6.9900689963453734e-06,
      "loss": 0.6199,
      "step": 8996
    },
    {
      "epoch": 1.5608952116585706,
      "grad_norm": 0.7336606383323669,
      "learning_rate": 6.9848950201667045e-06,
      "loss": 0.478,
      "step": 8997
    },
    {
      "epoch": 1.5610687022900764,
      "grad_norm": 0.9243895411491394,
      "learning_rate": 6.9797225544061385e-06,
      "loss": 0.4578,
      "step": 8998
    },
    {
      "epoch": 1.5612421929215823,
      "grad_norm": 0.7041562795639038,
      "learning_rate": 6.974551599663944e-06,
      "loss": 0.4706,
      "step": 8999
    },
    {
      "epoch": 1.5614156835530881,
      "grad_norm": 0.7184816598892212,
      "learning_rate": 6.969382156540212e-06,
      "loss": 0.5055,
      "step": 9000
    },
    {
      "epoch": 1.561589174184594,
      "grad_norm": 0.6358767151832581,
      "learning_rate": 6.96421422563486e-06,
      "loss": 0.3832,
      "step": 9001
    },
    {
      "epoch": 1.5617626648160998,
      "grad_norm": 0.9298535585403442,
      "learning_rate": 6.9590478075476475e-06,
      "loss": 0.3671,
      "step": 9002
    },
    {
      "epoch": 1.5619361554476057,
      "grad_norm": 0.8680828213691711,
      "learning_rate": 6.953882902878122e-06,
      "loss": 0.4718,
      "step": 9003
    },
    {
      "epoch": 1.5621096460791117,
      "grad_norm": 0.783250093460083,
      "learning_rate": 6.9487195122256925e-06,
      "loss": 0.3511,
      "step": 9004
    },
    {
      "epoch": 1.5622831367106176,
      "grad_norm": 0.7922610640525818,
      "learning_rate": 6.943557636189571e-06,
      "loss": 0.4148,
      "step": 9005
    },
    {
      "epoch": 1.5624566273421236,
      "grad_norm": 0.6474125385284424,
      "learning_rate": 6.9383972753688e-06,
      "loss": 0.54,
      "step": 9006
    },
    {
      "epoch": 1.5626301179736295,
      "grad_norm": 0.6304345726966858,
      "learning_rate": 6.933238430362239e-06,
      "loss": 0.5142,
      "step": 9007
    },
    {
      "epoch": 1.5628036086051353,
      "grad_norm": 0.7479380369186401,
      "learning_rate": 6.928081101768589e-06,
      "loss": 0.4434,
      "step": 9008
    },
    {
      "epoch": 1.5629770992366412,
      "grad_norm": 0.6971745491027832,
      "learning_rate": 6.922925290186362e-06,
      "loss": 0.387,
      "step": 9009
    },
    {
      "epoch": 1.563150589868147,
      "grad_norm": 0.7164735794067383,
      "learning_rate": 6.9177709962138905e-06,
      "loss": 0.3875,
      "step": 9010
    },
    {
      "epoch": 1.5633240804996529,
      "grad_norm": 0.6971276998519897,
      "learning_rate": 6.912618220449332e-06,
      "loss": 0.4354,
      "step": 9011
    },
    {
      "epoch": 1.563497571131159,
      "grad_norm": 1.1500219106674194,
      "learning_rate": 6.907466963490692e-06,
      "loss": 0.4811,
      "step": 9012
    },
    {
      "epoch": 1.5636710617626648,
      "grad_norm": 0.7223638892173767,
      "learning_rate": 6.9023172259357555e-06,
      "loss": 0.4069,
      "step": 9013
    },
    {
      "epoch": 1.5638445523941709,
      "grad_norm": 0.9820303320884705,
      "learning_rate": 6.897169008382172e-06,
      "loss": 0.3965,
      "step": 9014
    },
    {
      "epoch": 1.5640180430256767,
      "grad_norm": 0.6734370589256287,
      "learning_rate": 6.892022311427393e-06,
      "loss": 0.552,
      "step": 9015
    },
    {
      "epoch": 1.5641915336571826,
      "grad_norm": 0.7944628596305847,
      "learning_rate": 6.88687713566869e-06,
      "loss": 0.4819,
      "step": 9016
    },
    {
      "epoch": 1.5643650242886884,
      "grad_norm": 0.9793569445610046,
      "learning_rate": 6.881733481703186e-06,
      "loss": 0.4395,
      "step": 9017
    },
    {
      "epoch": 1.5645385149201942,
      "grad_norm": 0.8730150461196899,
      "learning_rate": 6.876591350127795e-06,
      "loss": 0.4686,
      "step": 9018
    },
    {
      "epoch": 1.5647120055517,
      "grad_norm": 0.9695137739181519,
      "learning_rate": 6.87145074153927e-06,
      "loss": 0.4552,
      "step": 9019
    },
    {
      "epoch": 1.5648854961832062,
      "grad_norm": 0.7851256132125854,
      "learning_rate": 6.866311656534177e-06,
      "loss": 0.4071,
      "step": 9020
    },
    {
      "epoch": 1.565058986814712,
      "grad_norm": 0.7857309579849243,
      "learning_rate": 6.861174095708934e-06,
      "loss": 0.3802,
      "step": 9021
    },
    {
      "epoch": 1.5652324774462179,
      "grad_norm": 0.8684077858924866,
      "learning_rate": 6.856038059659731e-06,
      "loss": 0.3909,
      "step": 9022
    },
    {
      "epoch": 1.565405968077724,
      "grad_norm": 0.9386990666389465,
      "learning_rate": 6.850903548982637e-06,
      "loss": 0.4572,
      "step": 9023
    },
    {
      "epoch": 1.5655794587092298,
      "grad_norm": 0.9225912690162659,
      "learning_rate": 6.8457705642734994e-06,
      "loss": 0.4894,
      "step": 9024
    },
    {
      "epoch": 1.5657529493407356,
      "grad_norm": 1.114875316619873,
      "learning_rate": 6.840639106128031e-06,
      "loss": 0.3694,
      "step": 9025
    },
    {
      "epoch": 1.5659264399722415,
      "grad_norm": 1.1497195959091187,
      "learning_rate": 6.835509175141713e-06,
      "loss": 0.5614,
      "step": 9026
    },
    {
      "epoch": 1.5660999306037473,
      "grad_norm": 0.636439323425293,
      "learning_rate": 6.830380771909901e-06,
      "loss": 0.5386,
      "step": 9027
    },
    {
      "epoch": 1.5662734212352531,
      "grad_norm": 0.8100028038024902,
      "learning_rate": 6.825253897027746e-06,
      "loss": 0.4321,
      "step": 9028
    },
    {
      "epoch": 1.5664469118667592,
      "grad_norm": 0.8574302792549133,
      "learning_rate": 6.82012855109023e-06,
      "loss": 0.4034,
      "step": 9029
    },
    {
      "epoch": 1.566620402498265,
      "grad_norm": 0.8456642627716064,
      "learning_rate": 6.815004734692146e-06,
      "loss": 0.4025,
      "step": 9030
    },
    {
      "epoch": 1.5667938931297711,
      "grad_norm": 1.075279951095581,
      "learning_rate": 6.8098824484281375e-06,
      "loss": 0.4342,
      "step": 9031
    },
    {
      "epoch": 1.566967383761277,
      "grad_norm": 0.8899574279785156,
      "learning_rate": 6.804761692892627e-06,
      "loss": 0.4513,
      "step": 9032
    },
    {
      "epoch": 1.5671408743927828,
      "grad_norm": 0.7178186774253845,
      "learning_rate": 6.799642468679908e-06,
      "loss": 0.5368,
      "step": 9033
    },
    {
      "epoch": 1.5673143650242887,
      "grad_norm": 0.8426361680030823,
      "learning_rate": 6.794524776384059e-06,
      "loss": 0.3991,
      "step": 9034
    },
    {
      "epoch": 1.5674878556557945,
      "grad_norm": 1.0742685794830322,
      "learning_rate": 6.7894086165989985e-06,
      "loss": 0.4621,
      "step": 9035
    },
    {
      "epoch": 1.5676613462873004,
      "grad_norm": 0.7841826677322388,
      "learning_rate": 6.784293989918454e-06,
      "loss": 0.3975,
      "step": 9036
    },
    {
      "epoch": 1.5678348369188064,
      "grad_norm": 0.732846736907959,
      "learning_rate": 6.779180896935997e-06,
      "loss": 0.5145,
      "step": 9037
    },
    {
      "epoch": 1.5680083275503123,
      "grad_norm": 0.9075619578361511,
      "learning_rate": 6.774069338245002e-06,
      "loss": 0.4452,
      "step": 9038
    },
    {
      "epoch": 1.5681818181818183,
      "grad_norm": 0.7694520950317383,
      "learning_rate": 6.7689593144386745e-06,
      "loss": 0.5419,
      "step": 9039
    },
    {
      "epoch": 1.5683553088133242,
      "grad_norm": 0.8426071405410767,
      "learning_rate": 6.763850826110025e-06,
      "loss": 0.3242,
      "step": 9040
    },
    {
      "epoch": 1.56852879944483,
      "grad_norm": 0.9523208141326904,
      "learning_rate": 6.758743873851921e-06,
      "loss": 0.327,
      "step": 9041
    },
    {
      "epoch": 1.5687022900763359,
      "grad_norm": 0.9337752461433411,
      "learning_rate": 6.753638458257017e-06,
      "loss": 0.4467,
      "step": 9042
    },
    {
      "epoch": 1.5688757807078417,
      "grad_norm": 0.6437515616416931,
      "learning_rate": 6.748534579917807e-06,
      "loss": 0.5056,
      "step": 9043
    },
    {
      "epoch": 1.5690492713393476,
      "grad_norm": 0.7449395656585693,
      "learning_rate": 6.743432239426599e-06,
      "loss": 0.4554,
      "step": 9044
    },
    {
      "epoch": 1.5692227619708534,
      "grad_norm": 0.8015483617782593,
      "learning_rate": 6.7383314373755184e-06,
      "loss": 0.3591,
      "step": 9045
    },
    {
      "epoch": 1.5693962526023595,
      "grad_norm": 0.6457865834236145,
      "learning_rate": 6.733232174356537e-06,
      "loss": 0.494,
      "step": 9046
    },
    {
      "epoch": 1.5695697432338653,
      "grad_norm": 0.7131983637809753,
      "learning_rate": 6.728134450961419e-06,
      "loss": 0.5073,
      "step": 9047
    },
    {
      "epoch": 1.5697432338653714,
      "grad_norm": 0.7374870181083679,
      "learning_rate": 6.723038267781763e-06,
      "loss": 0.53,
      "step": 9048
    },
    {
      "epoch": 1.5699167244968772,
      "grad_norm": 0.7074459195137024,
      "learning_rate": 6.71794362540898e-06,
      "loss": 0.5764,
      "step": 9049
    },
    {
      "epoch": 1.570090215128383,
      "grad_norm": 0.799924910068512,
      "learning_rate": 6.712850524434329e-06,
      "loss": 0.5123,
      "step": 9050
    },
    {
      "epoch": 1.570263705759889,
      "grad_norm": 0.7831860780715942,
      "learning_rate": 6.707758965448843e-06,
      "loss": 0.4191,
      "step": 9051
    },
    {
      "epoch": 1.5704371963913948,
      "grad_norm": 0.810056209564209,
      "learning_rate": 6.7026689490434275e-06,
      "loss": 0.3857,
      "step": 9052
    },
    {
      "epoch": 1.5706106870229006,
      "grad_norm": 0.9071383476257324,
      "learning_rate": 6.6975804758087645e-06,
      "loss": 0.3643,
      "step": 9053
    },
    {
      "epoch": 1.5707841776544067,
      "grad_norm": 0.8107170462608337,
      "learning_rate": 6.692493546335404e-06,
      "loss": 0.3964,
      "step": 9054
    },
    {
      "epoch": 1.5709576682859125,
      "grad_norm": 1.23489511013031,
      "learning_rate": 6.687408161213657e-06,
      "loss": 0.4023,
      "step": 9055
    },
    {
      "epoch": 1.5711311589174186,
      "grad_norm": 0.7870251536369324,
      "learning_rate": 6.682324321033715e-06,
      "loss": 0.4589,
      "step": 9056
    },
    {
      "epoch": 1.5713046495489245,
      "grad_norm": 0.6991247534751892,
      "learning_rate": 6.677242026385553e-06,
      "loss": 0.4662,
      "step": 9057
    },
    {
      "epoch": 1.5714781401804303,
      "grad_norm": 0.8230313062667847,
      "learning_rate": 6.672161277858977e-06,
      "loss": 0.3594,
      "step": 9058
    },
    {
      "epoch": 1.5716516308119362,
      "grad_norm": 0.966713547706604,
      "learning_rate": 6.667082076043609e-06,
      "loss": 0.419,
      "step": 9059
    },
    {
      "epoch": 1.571825121443442,
      "grad_norm": 0.7259449362754822,
      "learning_rate": 6.662004421528909e-06,
      "loss": 0.5421,
      "step": 9060
    },
    {
      "epoch": 1.5719986120749478,
      "grad_norm": 1.3742544651031494,
      "learning_rate": 6.656928314904136e-06,
      "loss": 0.4172,
      "step": 9061
    },
    {
      "epoch": 1.5721721027064537,
      "grad_norm": 0.7404026985168457,
      "learning_rate": 6.651853756758382e-06,
      "loss": 0.5085,
      "step": 9062
    },
    {
      "epoch": 1.5723455933379598,
      "grad_norm": 0.9047279953956604,
      "learning_rate": 6.646780747680552e-06,
      "loss": 0.4746,
      "step": 9063
    },
    {
      "epoch": 1.5725190839694656,
      "grad_norm": 1.2601579427719116,
      "learning_rate": 6.641709288259368e-06,
      "loss": 0.5427,
      "step": 9064
    },
    {
      "epoch": 1.5726925746009717,
      "grad_norm": 0.6506098508834839,
      "learning_rate": 6.636639379083396e-06,
      "loss": 0.5492,
      "step": 9065
    },
    {
      "epoch": 1.5728660652324775,
      "grad_norm": 1.5154086351394653,
      "learning_rate": 6.6315710207409925e-06,
      "loss": 0.3955,
      "step": 9066
    },
    {
      "epoch": 1.5730395558639834,
      "grad_norm": 1.506515383720398,
      "learning_rate": 6.6265042138203505e-06,
      "loss": 0.3898,
      "step": 9067
    },
    {
      "epoch": 1.5732130464954892,
      "grad_norm": 0.9609709978103638,
      "learning_rate": 6.621438958909472e-06,
      "loss": 0.4889,
      "step": 9068
    },
    {
      "epoch": 1.573386537126995,
      "grad_norm": 0.9973898530006409,
      "learning_rate": 6.616375256596197e-06,
      "loss": 0.329,
      "step": 9069
    },
    {
      "epoch": 1.573560027758501,
      "grad_norm": 0.8524582386016846,
      "learning_rate": 6.6113131074681694e-06,
      "loss": 0.489,
      "step": 9070
    },
    {
      "epoch": 1.573733518390007,
      "grad_norm": 0.7121954560279846,
      "learning_rate": 6.606252512112856e-06,
      "loss": 0.3952,
      "step": 9071
    },
    {
      "epoch": 1.5739070090215128,
      "grad_norm": 0.7621756196022034,
      "learning_rate": 6.6011934711175395e-06,
      "loss": 0.5177,
      "step": 9072
    },
    {
      "epoch": 1.5740804996530189,
      "grad_norm": 0.810859739780426,
      "learning_rate": 6.596135985069347e-06,
      "loss": 0.4792,
      "step": 9073
    },
    {
      "epoch": 1.5742539902845247,
      "grad_norm": 0.7965677380561829,
      "learning_rate": 6.591080054555177e-06,
      "loss": 0.4496,
      "step": 9074
    },
    {
      "epoch": 1.5744274809160306,
      "grad_norm": 0.7577996253967285,
      "learning_rate": 6.586025680161799e-06,
      "loss": 0.5212,
      "step": 9075
    },
    {
      "epoch": 1.5746009715475364,
      "grad_norm": 0.7506447434425354,
      "learning_rate": 6.580972862475769e-06,
      "loss": 0.4076,
      "step": 9076
    },
    {
      "epoch": 1.5747744621790423,
      "grad_norm": 1.142518401145935,
      "learning_rate": 6.575921602083477e-06,
      "loss": 0.5144,
      "step": 9077
    },
    {
      "epoch": 1.5749479528105481,
      "grad_norm": 0.9225130677223206,
      "learning_rate": 6.570871899571119e-06,
      "loss": 0.4324,
      "step": 9078
    },
    {
      "epoch": 1.5751214434420542,
      "grad_norm": 0.6513336896896362,
      "learning_rate": 6.565823755524732e-06,
      "loss": 0.4908,
      "step": 9079
    },
    {
      "epoch": 1.57529493407356,
      "grad_norm": 0.613537073135376,
      "learning_rate": 6.56077717053015e-06,
      "loss": 0.5026,
      "step": 9080
    },
    {
      "epoch": 1.5754684247050659,
      "grad_norm": 0.7019462585449219,
      "learning_rate": 6.555732145173037e-06,
      "loss": 0.5442,
      "step": 9081
    },
    {
      "epoch": 1.575641915336572,
      "grad_norm": 0.9109891057014465,
      "learning_rate": 6.550688680038871e-06,
      "loss": 0.3792,
      "step": 9082
    },
    {
      "epoch": 1.5758154059680778,
      "grad_norm": 0.7040951251983643,
      "learning_rate": 6.545646775712964e-06,
      "loss": 0.5659,
      "step": 9083
    },
    {
      "epoch": 1.5759888965995836,
      "grad_norm": 0.7814924120903015,
      "learning_rate": 6.5406064327804165e-06,
      "loss": 0.4725,
      "step": 9084
    },
    {
      "epoch": 1.5761623872310895,
      "grad_norm": 1.5051835775375366,
      "learning_rate": 6.53556765182618e-06,
      "loss": 0.4648,
      "step": 9085
    },
    {
      "epoch": 1.5763358778625953,
      "grad_norm": 0.7355524897575378,
      "learning_rate": 6.5305304334350075e-06,
      "loss": 0.5163,
      "step": 9086
    },
    {
      "epoch": 1.5765093684941012,
      "grad_norm": 0.8153585195541382,
      "learning_rate": 6.525494778191473e-06,
      "loss": 0.418,
      "step": 9087
    },
    {
      "epoch": 1.5766828591256072,
      "grad_norm": 0.714627742767334,
      "learning_rate": 6.520460686679964e-06,
      "loss": 0.3392,
      "step": 9088
    },
    {
      "epoch": 1.576856349757113,
      "grad_norm": 0.6588917374610901,
      "learning_rate": 6.515428159484707e-06,
      "loss": 0.4119,
      "step": 9089
    },
    {
      "epoch": 1.5770298403886192,
      "grad_norm": 0.7315980792045593,
      "learning_rate": 6.510397197189724e-06,
      "loss": 0.3981,
      "step": 9090
    },
    {
      "epoch": 1.577203331020125,
      "grad_norm": 0.8116651177406311,
      "learning_rate": 6.505367800378856e-06,
      "loss": 0.3419,
      "step": 9091
    },
    {
      "epoch": 1.5773768216516308,
      "grad_norm": 0.7682046294212341,
      "learning_rate": 6.500339969635794e-06,
      "loss": 0.4075,
      "step": 9092
    },
    {
      "epoch": 1.5775503122831367,
      "grad_norm": 0.7009580731391907,
      "learning_rate": 6.495313705543997e-06,
      "loss": 0.3578,
      "step": 9093
    },
    {
      "epoch": 1.5777238029146425,
      "grad_norm": 0.8438929915428162,
      "learning_rate": 6.490289008686786e-06,
      "loss": 0.4176,
      "step": 9094
    },
    {
      "epoch": 1.5778972935461484,
      "grad_norm": 0.7142999768257141,
      "learning_rate": 6.485265879647269e-06,
      "loss": 0.5607,
      "step": 9095
    },
    {
      "epoch": 1.5780707841776545,
      "grad_norm": 0.7572633624076843,
      "learning_rate": 6.480244319008411e-06,
      "loss": 0.5898,
      "step": 9096
    },
    {
      "epoch": 1.5782442748091603,
      "grad_norm": 0.48166990280151367,
      "learning_rate": 6.475224327352938e-06,
      "loss": 0.5698,
      "step": 9097
    },
    {
      "epoch": 1.5784177654406664,
      "grad_norm": 0.8895741105079651,
      "learning_rate": 6.470205905263449e-06,
      "loss": 0.5183,
      "step": 9098
    },
    {
      "epoch": 1.5785912560721722,
      "grad_norm": 0.8576816916465759,
      "learning_rate": 6.4651890533223294e-06,
      "loss": 0.4637,
      "step": 9099
    },
    {
      "epoch": 1.578764746703678,
      "grad_norm": 0.9381229281425476,
      "learning_rate": 6.460173772111791e-06,
      "loss": 0.4706,
      "step": 9100
    },
    {
      "epoch": 1.578938237335184,
      "grad_norm": 0.8738219738006592,
      "learning_rate": 6.455160062213857e-06,
      "loss": 0.3947,
      "step": 9101
    },
    {
      "epoch": 1.5791117279666897,
      "grad_norm": 0.873576819896698,
      "learning_rate": 6.450147924210395e-06,
      "loss": 0.4266,
      "step": 9102
    },
    {
      "epoch": 1.5792852185981956,
      "grad_norm": 1.0661113262176514,
      "learning_rate": 6.44513735868304e-06,
      "loss": 0.4674,
      "step": 9103
    },
    {
      "epoch": 1.5794587092297014,
      "grad_norm": 0.8318861126899719,
      "learning_rate": 6.440128366213297e-06,
      "loss": 0.5085,
      "step": 9104
    },
    {
      "epoch": 1.5796321998612075,
      "grad_norm": 1.5399738550186157,
      "learning_rate": 6.435120947382456e-06,
      "loss": 0.4467,
      "step": 9105
    },
    {
      "epoch": 1.5798056904927134,
      "grad_norm": 0.8369775414466858,
      "learning_rate": 6.430115102771637e-06,
      "loss": 0.4502,
      "step": 9106
    },
    {
      "epoch": 1.5799791811242194,
      "grad_norm": 1.0189540386199951,
      "learning_rate": 6.425110832961765e-06,
      "loss": 0.3452,
      "step": 9107
    },
    {
      "epoch": 1.5801526717557253,
      "grad_norm": 1.0672706365585327,
      "learning_rate": 6.420108138533607e-06,
      "loss": 0.3255,
      "step": 9108
    },
    {
      "epoch": 1.5803261623872311,
      "grad_norm": 0.7228703498840332,
      "learning_rate": 6.415107020067721e-06,
      "loss": 0.3575,
      "step": 9109
    },
    {
      "epoch": 1.580499653018737,
      "grad_norm": 0.7414281368255615,
      "learning_rate": 6.410107478144496e-06,
      "loss": 0.3776,
      "step": 9110
    },
    {
      "epoch": 1.5806731436502428,
      "grad_norm": 1.7083401679992676,
      "learning_rate": 6.405109513344126e-06,
      "loss": 0.4216,
      "step": 9111
    },
    {
      "epoch": 1.5808466342817487,
      "grad_norm": 0.7078840732574463,
      "learning_rate": 6.400113126246645e-06,
      "loss": 0.5028,
      "step": 9112
    },
    {
      "epoch": 1.5810201249132547,
      "grad_norm": 1.244852900505066,
      "learning_rate": 6.395118317431883e-06,
      "loss": 0.3494,
      "step": 9113
    },
    {
      "epoch": 1.5811936155447606,
      "grad_norm": 1.4125138521194458,
      "learning_rate": 6.390125087479493e-06,
      "loss": 0.3657,
      "step": 9114
    },
    {
      "epoch": 1.5813671061762666,
      "grad_norm": 0.834786593914032,
      "learning_rate": 6.385133436968946e-06,
      "loss": 0.4061,
      "step": 9115
    },
    {
      "epoch": 1.5815405968077725,
      "grad_norm": 1.4661144018173218,
      "learning_rate": 6.380143366479521e-06,
      "loss": 0.4791,
      "step": 9116
    },
    {
      "epoch": 1.5817140874392783,
      "grad_norm": 0.7873803377151489,
      "learning_rate": 6.375154876590335e-06,
      "loss": 0.5126,
      "step": 9117
    },
    {
      "epoch": 1.5818875780707842,
      "grad_norm": 0.863389790058136,
      "learning_rate": 6.370167967880303e-06,
      "loss": 0.3486,
      "step": 9118
    },
    {
      "epoch": 1.58206106870229,
      "grad_norm": 0.9513432383537292,
      "learning_rate": 6.365182640928158e-06,
      "loss": 0.3654,
      "step": 9119
    },
    {
      "epoch": 1.5822345593337959,
      "grad_norm": 0.6658217906951904,
      "learning_rate": 6.360198896312451e-06,
      "loss": 0.4583,
      "step": 9120
    },
    {
      "epoch": 1.5824080499653017,
      "grad_norm": 0.8609620332717896,
      "learning_rate": 6.355216734611567e-06,
      "loss": 0.4985,
      "step": 9121
    },
    {
      "epoch": 1.5825815405968078,
      "grad_norm": 0.9660351872444153,
      "learning_rate": 6.350236156403666e-06,
      "loss": 0.4331,
      "step": 9122
    },
    {
      "epoch": 1.5827550312283136,
      "grad_norm": 0.7391103506088257,
      "learning_rate": 6.345257162266773e-06,
      "loss": 0.5355,
      "step": 9123
    },
    {
      "epoch": 1.5829285218598197,
      "grad_norm": 1.0405844449996948,
      "learning_rate": 6.3402797527786904e-06,
      "loss": 0.4579,
      "step": 9124
    },
    {
      "epoch": 1.5831020124913255,
      "grad_norm": 0.7017146348953247,
      "learning_rate": 6.335303928517071e-06,
      "loss": 0.3473,
      "step": 9125
    },
    {
      "epoch": 1.5832755031228314,
      "grad_norm": 0.8399113416671753,
      "learning_rate": 6.330329690059342e-06,
      "loss": 0.4696,
      "step": 9126
    },
    {
      "epoch": 1.5834489937543372,
      "grad_norm": 0.8594346046447754,
      "learning_rate": 6.3253570379827864e-06,
      "loss": 0.351,
      "step": 9127
    },
    {
      "epoch": 1.583622484385843,
      "grad_norm": 1.07136070728302,
      "learning_rate": 6.32038597286448e-06,
      "loss": 0.4559,
      "step": 9128
    },
    {
      "epoch": 1.583795975017349,
      "grad_norm": 0.8304911851882935,
      "learning_rate": 6.315416495281323e-06,
      "loss": 0.4612,
      "step": 9129
    },
    {
      "epoch": 1.583969465648855,
      "grad_norm": 0.9867230653762817,
      "learning_rate": 6.31044860581002e-06,
      "loss": 0.4049,
      "step": 9130
    },
    {
      "epoch": 1.5841429562803608,
      "grad_norm": 0.960951030254364,
      "learning_rate": 6.305482305027122e-06,
      "loss": 0.4299,
      "step": 9131
    },
    {
      "epoch": 1.584316446911867,
      "grad_norm": 0.8653309345245361,
      "learning_rate": 6.300517593508944e-06,
      "loss": 0.46,
      "step": 9132
    },
    {
      "epoch": 1.5844899375433728,
      "grad_norm": 0.7783135175704956,
      "learning_rate": 6.295554471831671e-06,
      "loss": 0.4358,
      "step": 9133
    },
    {
      "epoch": 1.5846634281748786,
      "grad_norm": 1.077301263809204,
      "learning_rate": 6.290592940571269e-06,
      "loss": 0.3718,
      "step": 9134
    },
    {
      "epoch": 1.5848369188063844,
      "grad_norm": 1.4728044271469116,
      "learning_rate": 6.2856330003035324e-06,
      "loss": 0.4374,
      "step": 9135
    },
    {
      "epoch": 1.5850104094378903,
      "grad_norm": 0.8031530976295471,
      "learning_rate": 6.280674651604059e-06,
      "loss": 0.6,
      "step": 9136
    },
    {
      "epoch": 1.5851839000693961,
      "grad_norm": 0.8494324088096619,
      "learning_rate": 6.275717895048285e-06,
      "loss": 0.5836,
      "step": 9137
    },
    {
      "epoch": 1.5853573907009022,
      "grad_norm": 0.7558203935623169,
      "learning_rate": 6.270762731211442e-06,
      "loss": 0.46,
      "step": 9138
    },
    {
      "epoch": 1.585530881332408,
      "grad_norm": 0.5446569323539734,
      "learning_rate": 6.2658091606685745e-06,
      "loss": 0.5767,
      "step": 9139
    },
    {
      "epoch": 1.585704371963914,
      "grad_norm": 0.7430805563926697,
      "learning_rate": 6.260857183994564e-06,
      "loss": 0.5164,
      "step": 9140
    },
    {
      "epoch": 1.58587786259542,
      "grad_norm": 1.3108949661254883,
      "learning_rate": 6.2559068017640865e-06,
      "loss": 0.3945,
      "step": 9141
    },
    {
      "epoch": 1.5860513532269258,
      "grad_norm": 0.6558985114097595,
      "learning_rate": 6.25095801455164e-06,
      "loss": 0.4688,
      "step": 9142
    },
    {
      "epoch": 1.5862248438584317,
      "grad_norm": 0.7553314566612244,
      "learning_rate": 6.246010822931532e-06,
      "loss": 0.4707,
      "step": 9143
    },
    {
      "epoch": 1.5863983344899375,
      "grad_norm": 0.7920790910720825,
      "learning_rate": 6.241065227477905e-06,
      "loss": 0.4611,
      "step": 9144
    },
    {
      "epoch": 1.5865718251214433,
      "grad_norm": 1.4178071022033691,
      "learning_rate": 6.2361212287646774e-06,
      "loss": 0.3625,
      "step": 9145
    },
    {
      "epoch": 1.5867453157529492,
      "grad_norm": 0.9829283356666565,
      "learning_rate": 6.231178827365627e-06,
      "loss": 0.4968,
      "step": 9146
    },
    {
      "epoch": 1.5869188063844553,
      "grad_norm": 0.7507148385047913,
      "learning_rate": 6.2262380238543185e-06,
      "loss": 0.4103,
      "step": 9147
    },
    {
      "epoch": 1.587092297015961,
      "grad_norm": 0.830384373664856,
      "learning_rate": 6.221298818804136e-06,
      "loss": 0.3536,
      "step": 9148
    },
    {
      "epoch": 1.5872657876474672,
      "grad_norm": 0.6961411237716675,
      "learning_rate": 6.216361212788276e-06,
      "loss": 0.5941,
      "step": 9149
    },
    {
      "epoch": 1.587439278278973,
      "grad_norm": 0.6326931715011597,
      "learning_rate": 6.211425206379769e-06,
      "loss": 0.4996,
      "step": 9150
    },
    {
      "epoch": 1.5876127689104789,
      "grad_norm": 0.8804559707641602,
      "learning_rate": 6.206490800151421e-06,
      "loss": 0.4768,
      "step": 9151
    },
    {
      "epoch": 1.5877862595419847,
      "grad_norm": 0.803973376750946,
      "learning_rate": 6.201557994675895e-06,
      "loss": 0.4097,
      "step": 9152
    },
    {
      "epoch": 1.5879597501734906,
      "grad_norm": 0.6925197839736938,
      "learning_rate": 6.196626790525635e-06,
      "loss": 0.4462,
      "step": 9153
    },
    {
      "epoch": 1.5881332408049964,
      "grad_norm": 0.6980529427528381,
      "learning_rate": 6.191697188272933e-06,
      "loss": 0.3606,
      "step": 9154
    },
    {
      "epoch": 1.5883067314365025,
      "grad_norm": 0.8955292701721191,
      "learning_rate": 6.186769188489852e-06,
      "loss": 0.5315,
      "step": 9155
    },
    {
      "epoch": 1.5884802220680083,
      "grad_norm": 0.6994858980178833,
      "learning_rate": 6.181842791748307e-06,
      "loss": 0.5776,
      "step": 9156
    },
    {
      "epoch": 1.5886537126995144,
      "grad_norm": 0.7811257243156433,
      "learning_rate": 6.1769179986200065e-06,
      "loss": 0.4552,
      "step": 9157
    },
    {
      "epoch": 1.5888272033310202,
      "grad_norm": 0.7694811224937439,
      "learning_rate": 6.17199480967648e-06,
      "loss": 0.4657,
      "step": 9158
    },
    {
      "epoch": 1.589000693962526,
      "grad_norm": 0.8114852905273438,
      "learning_rate": 6.167073225489062e-06,
      "loss": 0.3735,
      "step": 9159
    },
    {
      "epoch": 1.589174184594032,
      "grad_norm": 0.7839207053184509,
      "learning_rate": 6.162153246628921e-06,
      "loss": 0.3959,
      "step": 9160
    },
    {
      "epoch": 1.5893476752255378,
      "grad_norm": 0.5622174739837646,
      "learning_rate": 6.157234873667021e-06,
      "loss": 0.5399,
      "step": 9161
    },
    {
      "epoch": 1.5895211658570436,
      "grad_norm": 0.7764237523078918,
      "learning_rate": 6.152318107174144e-06,
      "loss": 0.4785,
      "step": 9162
    },
    {
      "epoch": 1.5896946564885495,
      "grad_norm": 0.7497486472129822,
      "learning_rate": 6.1474029477208864e-06,
      "loss": 0.5824,
      "step": 9163
    },
    {
      "epoch": 1.5898681471200555,
      "grad_norm": 1.0032864809036255,
      "learning_rate": 6.142489395877651e-06,
      "loss": 0.3083,
      "step": 9164
    },
    {
      "epoch": 1.5900416377515614,
      "grad_norm": 0.8805416226387024,
      "learning_rate": 6.137577452214676e-06,
      "loss": 0.4022,
      "step": 9165
    },
    {
      "epoch": 1.5902151283830674,
      "grad_norm": 0.8548498749732971,
      "learning_rate": 6.132667117301989e-06,
      "loss": 0.4919,
      "step": 9166
    },
    {
      "epoch": 1.5903886190145733,
      "grad_norm": 1.2921521663665771,
      "learning_rate": 6.127758391709442e-06,
      "loss": 0.3438,
      "step": 9167
    },
    {
      "epoch": 1.5905621096460791,
      "grad_norm": 0.9614955186843872,
      "learning_rate": 6.122851276006692e-06,
      "loss": 0.3766,
      "step": 9168
    },
    {
      "epoch": 1.590735600277585,
      "grad_norm": 1.2059763669967651,
      "learning_rate": 6.117945770763228e-06,
      "loss": 0.4922,
      "step": 9169
    },
    {
      "epoch": 1.5909090909090908,
      "grad_norm": 0.8131858706474304,
      "learning_rate": 6.113041876548333e-06,
      "loss": 0.4535,
      "step": 9170
    },
    {
      "epoch": 1.5910825815405967,
      "grad_norm": 0.736976683139801,
      "learning_rate": 6.1081395939311086e-06,
      "loss": 0.5316,
      "step": 9171
    },
    {
      "epoch": 1.5912560721721027,
      "grad_norm": 0.8057016134262085,
      "learning_rate": 6.103238923480468e-06,
      "loss": 0.4275,
      "step": 9172
    },
    {
      "epoch": 1.5914295628036086,
      "grad_norm": 0.9016966819763184,
      "learning_rate": 6.098339865765153e-06,
      "loss": 0.4287,
      "step": 9173
    },
    {
      "epoch": 1.5916030534351147,
      "grad_norm": 0.9611093401908875,
      "learning_rate": 6.093442421353683e-06,
      "loss": 0.5229,
      "step": 9174
    },
    {
      "epoch": 1.5917765440666205,
      "grad_norm": 0.6343430280685425,
      "learning_rate": 6.088546590814432e-06,
      "loss": 0.5574,
      "step": 9175
    },
    {
      "epoch": 1.5919500346981263,
      "grad_norm": 1.0098336935043335,
      "learning_rate": 6.083652374715561e-06,
      "loss": 0.608,
      "step": 9176
    },
    {
      "epoch": 1.5921235253296322,
      "grad_norm": 0.6942441463470459,
      "learning_rate": 6.078759773625045e-06,
      "loss": 0.4966,
      "step": 9177
    },
    {
      "epoch": 1.592297015961138,
      "grad_norm": 0.7739158272743225,
      "learning_rate": 6.073868788110673e-06,
      "loss": 0.4349,
      "step": 9178
    },
    {
      "epoch": 1.5924705065926439,
      "grad_norm": 0.6788617968559265,
      "learning_rate": 6.068979418740067e-06,
      "loss": 0.4912,
      "step": 9179
    },
    {
      "epoch": 1.5926439972241497,
      "grad_norm": 0.70098876953125,
      "learning_rate": 6.064091666080621e-06,
      "loss": 0.4116,
      "step": 9180
    },
    {
      "epoch": 1.5928174878556558,
      "grad_norm": 0.8632174730300903,
      "learning_rate": 6.059205530699581e-06,
      "loss": 0.4497,
      "step": 9181
    },
    {
      "epoch": 1.5929909784871616,
      "grad_norm": 0.9082633256912231,
      "learning_rate": 6.054321013163978e-06,
      "loss": 0.3707,
      "step": 9182
    },
    {
      "epoch": 1.5931644691186677,
      "grad_norm": 0.8024830222129822,
      "learning_rate": 6.049438114040682e-06,
      "loss": 0.4788,
      "step": 9183
    },
    {
      "epoch": 1.5933379597501736,
      "grad_norm": 0.7636101245880127,
      "learning_rate": 6.044556833896338e-06,
      "loss": 0.4121,
      "step": 9184
    },
    {
      "epoch": 1.5935114503816794,
      "grad_norm": 0.7734701633453369,
      "learning_rate": 6.03967717329744e-06,
      "loss": 0.3143,
      "step": 9185
    },
    {
      "epoch": 1.5936849410131853,
      "grad_norm": 0.7777693867683411,
      "learning_rate": 6.034799132810274e-06,
      "loss": 0.4668,
      "step": 9186
    },
    {
      "epoch": 1.593858431644691,
      "grad_norm": 0.7053585052490234,
      "learning_rate": 6.029922713000935e-06,
      "loss": 0.5001,
      "step": 9187
    },
    {
      "epoch": 1.594031922276197,
      "grad_norm": 0.8492000102996826,
      "learning_rate": 6.025047914435349e-06,
      "loss": 0.4404,
      "step": 9188
    },
    {
      "epoch": 1.594205412907703,
      "grad_norm": 0.7375931739807129,
      "learning_rate": 6.020174737679236e-06,
      "loss": 0.3701,
      "step": 9189
    },
    {
      "epoch": 1.5943789035392089,
      "grad_norm": 0.7666688561439514,
      "learning_rate": 6.015303183298135e-06,
      "loss": 0.3666,
      "step": 9190
    },
    {
      "epoch": 1.594552394170715,
      "grad_norm": 0.9085474610328674,
      "learning_rate": 6.01043325185739e-06,
      "loss": 0.4756,
      "step": 9191
    },
    {
      "epoch": 1.5947258848022208,
      "grad_norm": 0.7535820007324219,
      "learning_rate": 6.005564943922179e-06,
      "loss": 0.5159,
      "step": 9192
    },
    {
      "epoch": 1.5948993754337266,
      "grad_norm": 0.8561382293701172,
      "learning_rate": 6.000698260057452e-06,
      "loss": 0.3917,
      "step": 9193
    },
    {
      "epoch": 1.5950728660652325,
      "grad_norm": 0.6852325201034546,
      "learning_rate": 5.995833200828007e-06,
      "loss": 0.5533,
      "step": 9194
    },
    {
      "epoch": 1.5952463566967383,
      "grad_norm": 0.9255613088607788,
      "learning_rate": 5.990969766798433e-06,
      "loss": 0.4999,
      "step": 9195
    },
    {
      "epoch": 1.5954198473282442,
      "grad_norm": 0.8838304877281189,
      "learning_rate": 5.9861079585331535e-06,
      "loss": 0.3906,
      "step": 9196
    },
    {
      "epoch": 1.5955933379597502,
      "grad_norm": 0.8277800679206848,
      "learning_rate": 5.981247776596364e-06,
      "loss": 0.4397,
      "step": 9197
    },
    {
      "epoch": 1.595766828591256,
      "grad_norm": 0.7583249807357788,
      "learning_rate": 5.97638922155211e-06,
      "loss": 0.5585,
      "step": 9198
    },
    {
      "epoch": 1.595940319222762,
      "grad_norm": 0.7804730534553528,
      "learning_rate": 5.971532293964226e-06,
      "loss": 0.4011,
      "step": 9199
    },
    {
      "epoch": 1.596113809854268,
      "grad_norm": 0.7163670063018799,
      "learning_rate": 5.9666769943963675e-06,
      "loss": 0.4969,
      "step": 9200
    },
    {
      "epoch": 1.5962873004857738,
      "grad_norm": 0.7205650210380554,
      "learning_rate": 5.96182332341199e-06,
      "loss": 0.4364,
      "step": 9201
    },
    {
      "epoch": 1.5964607911172797,
      "grad_norm": 0.7018092274665833,
      "learning_rate": 5.956971281574386e-06,
      "loss": 0.5375,
      "step": 9202
    },
    {
      "epoch": 1.5966342817487855,
      "grad_norm": 0.6795221567153931,
      "learning_rate": 5.952120869446614e-06,
      "loss": 0.5612,
      "step": 9203
    },
    {
      "epoch": 1.5968077723802914,
      "grad_norm": 0.771239697933197,
      "learning_rate": 5.947272087591591e-06,
      "loss": 0.5967,
      "step": 9204
    },
    {
      "epoch": 1.5969812630117972,
      "grad_norm": 0.7565046548843384,
      "learning_rate": 5.942424936572017e-06,
      "loss": 0.4929,
      "step": 9205
    },
    {
      "epoch": 1.5971547536433033,
      "grad_norm": 0.8558209538459778,
      "learning_rate": 5.9375794169504095e-06,
      "loss": 0.494,
      "step": 9206
    },
    {
      "epoch": 1.5973282442748091,
      "grad_norm": 0.6613254547119141,
      "learning_rate": 5.932735529289091e-06,
      "loss": 0.4871,
      "step": 9207
    },
    {
      "epoch": 1.5975017349063152,
      "grad_norm": 0.8535557985305786,
      "learning_rate": 5.927893274150214e-06,
      "loss": 0.4371,
      "step": 9208
    },
    {
      "epoch": 1.597675225537821,
      "grad_norm": 0.6975634694099426,
      "learning_rate": 5.9230526520957154e-06,
      "loss": 0.4816,
      "step": 9209
    },
    {
      "epoch": 1.5978487161693269,
      "grad_norm": 0.8697851896286011,
      "learning_rate": 5.918213663687362e-06,
      "loss": 0.4951,
      "step": 9210
    },
    {
      "epoch": 1.5980222068008327,
      "grad_norm": 0.7481193542480469,
      "learning_rate": 5.913376309486716e-06,
      "loss": 0.5833,
      "step": 9211
    },
    {
      "epoch": 1.5981956974323386,
      "grad_norm": 0.6353183388710022,
      "learning_rate": 5.908540590055168e-06,
      "loss": 0.4703,
      "step": 9212
    },
    {
      "epoch": 1.5983691880638444,
      "grad_norm": 0.9413015246391296,
      "learning_rate": 5.903706505953906e-06,
      "loss": 0.3647,
      "step": 9213
    },
    {
      "epoch": 1.5985426786953505,
      "grad_norm": 0.8919071555137634,
      "learning_rate": 5.898874057743926e-06,
      "loss": 0.4026,
      "step": 9214
    },
    {
      "epoch": 1.5987161693268563,
      "grad_norm": 1.060607671737671,
      "learning_rate": 5.894043245986045e-06,
      "loss": 0.5665,
      "step": 9215
    },
    {
      "epoch": 1.5988896599583624,
      "grad_norm": 0.8731929063796997,
      "learning_rate": 5.889214071240876e-06,
      "loss": 0.4706,
      "step": 9216
    },
    {
      "epoch": 1.5990631505898683,
      "grad_norm": 0.7572540640830994,
      "learning_rate": 5.884386534068864e-06,
      "loss": 0.4324,
      "step": 9217
    },
    {
      "epoch": 1.599236641221374,
      "grad_norm": 0.6952676773071289,
      "learning_rate": 5.879560635030242e-06,
      "loss": 0.4614,
      "step": 9218
    },
    {
      "epoch": 1.59941013185288,
      "grad_norm": 0.8549386858940125,
      "learning_rate": 5.87473637468506e-06,
      "loss": 0.4067,
      "step": 9219
    },
    {
      "epoch": 1.5995836224843858,
      "grad_norm": 0.8678174018859863,
      "learning_rate": 5.869913753593175e-06,
      "loss": 0.4244,
      "step": 9220
    },
    {
      "epoch": 1.5997571131158916,
      "grad_norm": 0.8675947785377502,
      "learning_rate": 5.865092772314276e-06,
      "loss": 0.3524,
      "step": 9221
    },
    {
      "epoch": 1.5999306037473975,
      "grad_norm": 1.4709620475769043,
      "learning_rate": 5.860273431407821e-06,
      "loss": 0.4775,
      "step": 9222
    },
    {
      "epoch": 1.6001040943789036,
      "grad_norm": 0.8937550187110901,
      "learning_rate": 5.855455731433115e-06,
      "loss": 0.3348,
      "step": 9223
    },
    {
      "epoch": 1.6002775850104094,
      "grad_norm": 0.6757932901382446,
      "learning_rate": 5.8506396729492455e-06,
      "loss": 0.4922,
      "step": 9224
    },
    {
      "epoch": 1.6004510756419155,
      "grad_norm": 0.9056023359298706,
      "learning_rate": 5.845825256515145e-06,
      "loss": 0.5139,
      "step": 9225
    },
    {
      "epoch": 1.6006245662734213,
      "grad_norm": 1.0853079557418823,
      "learning_rate": 5.841012482689501e-06,
      "loss": 0.4449,
      "step": 9226
    },
    {
      "epoch": 1.6007980569049272,
      "grad_norm": 0.7655895948410034,
      "learning_rate": 5.836201352030863e-06,
      "loss": 0.5815,
      "step": 9227
    },
    {
      "epoch": 1.600971547536433,
      "grad_norm": 1.3326359987258911,
      "learning_rate": 5.831391865097564e-06,
      "loss": 0.4553,
      "step": 9228
    },
    {
      "epoch": 1.6011450381679388,
      "grad_norm": 0.7713728547096252,
      "learning_rate": 5.82658402244775e-06,
      "loss": 0.527,
      "step": 9229
    },
    {
      "epoch": 1.6013185287994447,
      "grad_norm": 10.568388938903809,
      "learning_rate": 5.821777824639365e-06,
      "loss": 0.5093,
      "step": 9230
    },
    {
      "epoch": 1.6014920194309508,
      "grad_norm": 0.7469318509101868,
      "learning_rate": 5.816973272230196e-06,
      "loss": 0.4376,
      "step": 9231
    },
    {
      "epoch": 1.6016655100624566,
      "grad_norm": 1.014448881149292,
      "learning_rate": 5.812170365777801e-06,
      "loss": 0.4911,
      "step": 9232
    },
    {
      "epoch": 1.6018390006939627,
      "grad_norm": 0.9525623917579651,
      "learning_rate": 5.807369105839569e-06,
      "loss": 0.3952,
      "step": 9233
    },
    {
      "epoch": 1.6020124913254685,
      "grad_norm": 1.4405517578125,
      "learning_rate": 5.802569492972687e-06,
      "loss": 0.5205,
      "step": 9234
    },
    {
      "epoch": 1.6021859819569744,
      "grad_norm": 0.782208263874054,
      "learning_rate": 5.797771527734155e-06,
      "loss": 0.3975,
      "step": 9235
    },
    {
      "epoch": 1.6023594725884802,
      "grad_norm": 0.7666157484054565,
      "learning_rate": 5.792975210680793e-06,
      "loss": 0.3898,
      "step": 9236
    },
    {
      "epoch": 1.602532963219986,
      "grad_norm": 0.8280892968177795,
      "learning_rate": 5.7881805423692105e-06,
      "loss": 0.3287,
      "step": 9237
    },
    {
      "epoch": 1.602706453851492,
      "grad_norm": 0.8459081053733826,
      "learning_rate": 5.783387523355839e-06,
      "loss": 0.5415,
      "step": 9238
    },
    {
      "epoch": 1.6028799444829978,
      "grad_norm": 0.8862437009811401,
      "learning_rate": 5.778596154196902e-06,
      "loss": 0.4484,
      "step": 9239
    },
    {
      "epoch": 1.6030534351145038,
      "grad_norm": 0.9152216911315918,
      "learning_rate": 5.773806435448459e-06,
      "loss": 0.4476,
      "step": 9240
    },
    {
      "epoch": 1.6032269257460097,
      "grad_norm": 1.9208900928497314,
      "learning_rate": 5.769018367666357e-06,
      "loss": 0.4103,
      "step": 9241
    },
    {
      "epoch": 1.6034004163775157,
      "grad_norm": 0.8232020735740662,
      "learning_rate": 5.764231951406256e-06,
      "loss": 0.3483,
      "step": 9242
    },
    {
      "epoch": 1.6035739070090216,
      "grad_norm": 1.0295816659927368,
      "learning_rate": 5.759447187223617e-06,
      "loss": 0.4777,
      "step": 9243
    },
    {
      "epoch": 1.6037473976405274,
      "grad_norm": 0.7637811899185181,
      "learning_rate": 5.75466407567374e-06,
      "loss": 0.411,
      "step": 9244
    },
    {
      "epoch": 1.6039208882720333,
      "grad_norm": 0.6895084381103516,
      "learning_rate": 5.749882617311682e-06,
      "loss": 0.5614,
      "step": 9245
    },
    {
      "epoch": 1.6040943789035391,
      "grad_norm": 0.8308559060096741,
      "learning_rate": 5.745102812692358e-06,
      "loss": 0.3559,
      "step": 9246
    },
    {
      "epoch": 1.604267869535045,
      "grad_norm": 1.082716464996338,
      "learning_rate": 5.740324662370462e-06,
      "loss": 0.345,
      "step": 9247
    },
    {
      "epoch": 1.604441360166551,
      "grad_norm": 1.375120759010315,
      "learning_rate": 5.735548166900506e-06,
      "loss": 0.3909,
      "step": 9248
    },
    {
      "epoch": 1.6046148507980569,
      "grad_norm": 0.8588771820068359,
      "learning_rate": 5.7307733268368e-06,
      "loss": 0.3923,
      "step": 9249
    },
    {
      "epoch": 1.604788341429563,
      "grad_norm": 0.931693434715271,
      "learning_rate": 5.72600014273349e-06,
      "loss": 0.3853,
      "step": 9250
    },
    {
      "epoch": 1.6049618320610688,
      "grad_norm": 1.8362689018249512,
      "learning_rate": 5.721228615144481e-06,
      "loss": 0.4402,
      "step": 9251
    },
    {
      "epoch": 1.6051353226925746,
      "grad_norm": 0.8088194131851196,
      "learning_rate": 5.716458744623536e-06,
      "loss": 0.4089,
      "step": 9252
    },
    {
      "epoch": 1.6053088133240805,
      "grad_norm": 0.77403724193573,
      "learning_rate": 5.711690531724192e-06,
      "loss": 0.3812,
      "step": 9253
    },
    {
      "epoch": 1.6054823039555863,
      "grad_norm": 0.6571146845817566,
      "learning_rate": 5.706923976999825e-06,
      "loss": 0.5425,
      "step": 9254
    },
    {
      "epoch": 1.6056557945870922,
      "grad_norm": 0.7870723605155945,
      "learning_rate": 5.702159081003571e-06,
      "loss": 0.374,
      "step": 9255
    },
    {
      "epoch": 1.6058292852185982,
      "grad_norm": 1.3615376949310303,
      "learning_rate": 5.697395844288423e-06,
      "loss": 0.3523,
      "step": 9256
    },
    {
      "epoch": 1.606002775850104,
      "grad_norm": 0.7027696371078491,
      "learning_rate": 5.692634267407151e-06,
      "loss": 0.4547,
      "step": 9257
    },
    {
      "epoch": 1.60617626648161,
      "grad_norm": 0.6760240197181702,
      "learning_rate": 5.687874350912346e-06,
      "loss": 0.5397,
      "step": 9258
    },
    {
      "epoch": 1.606349757113116,
      "grad_norm": 0.6329642534255981,
      "learning_rate": 5.683116095356391e-06,
      "loss": 0.4252,
      "step": 9259
    },
    {
      "epoch": 1.6065232477446219,
      "grad_norm": 0.8432753086090088,
      "learning_rate": 5.678359501291504e-06,
      "loss": 0.4285,
      "step": 9260
    },
    {
      "epoch": 1.6066967383761277,
      "grad_norm": 0.7299416661262512,
      "learning_rate": 5.673604569269684e-06,
      "loss": 0.4967,
      "step": 9261
    },
    {
      "epoch": 1.6068702290076335,
      "grad_norm": 0.7422299981117249,
      "learning_rate": 5.668851299842739e-06,
      "loss": 0.4343,
      "step": 9262
    },
    {
      "epoch": 1.6070437196391394,
      "grad_norm": 1.866463541984558,
      "learning_rate": 5.664099693562315e-06,
      "loss": 0.4298,
      "step": 9263
    },
    {
      "epoch": 1.6072172102706452,
      "grad_norm": 0.8768714666366577,
      "learning_rate": 5.659349750979814e-06,
      "loss": 0.3596,
      "step": 9264
    },
    {
      "epoch": 1.6073907009021513,
      "grad_norm": 0.7241788506507874,
      "learning_rate": 5.6546014726464906e-06,
      "loss": 0.4443,
      "step": 9265
    },
    {
      "epoch": 1.6075641915336571,
      "grad_norm": 0.7192736864089966,
      "learning_rate": 5.6498548591133725e-06,
      "loss": 0.3469,
      "step": 9266
    },
    {
      "epoch": 1.6077376821651632,
      "grad_norm": 0.7598190903663635,
      "learning_rate": 5.645109910931335e-06,
      "loss": 0.4667,
      "step": 9267
    },
    {
      "epoch": 1.607911172796669,
      "grad_norm": 0.7664093375205994,
      "learning_rate": 5.6403666286510065e-06,
      "loss": 0.5514,
      "step": 9268
    },
    {
      "epoch": 1.608084663428175,
      "grad_norm": 0.7578974366188049,
      "learning_rate": 5.635625012822869e-06,
      "loss": 0.3986,
      "step": 9269
    },
    {
      "epoch": 1.6082581540596808,
      "grad_norm": 0.7644089460372925,
      "learning_rate": 5.630885063997187e-06,
      "loss": 0.3786,
      "step": 9270
    },
    {
      "epoch": 1.6084316446911866,
      "grad_norm": 0.731870174407959,
      "learning_rate": 5.626146782724036e-06,
      "loss": 0.4619,
      "step": 9271
    },
    {
      "epoch": 1.6086051353226924,
      "grad_norm": 0.6918858289718628,
      "learning_rate": 5.621410169553292e-06,
      "loss": 0.5197,
      "step": 9272
    },
    {
      "epoch": 1.6087786259541985,
      "grad_norm": 1.2198553085327148,
      "learning_rate": 5.616675225034667e-06,
      "loss": 0.5334,
      "step": 9273
    },
    {
      "epoch": 1.6089521165857044,
      "grad_norm": 0.6842505931854248,
      "learning_rate": 5.6119419497176275e-06,
      "loss": 0.4431,
      "step": 9274
    },
    {
      "epoch": 1.6091256072172104,
      "grad_norm": 0.720130980014801,
      "learning_rate": 5.607210344151497e-06,
      "loss": 0.3934,
      "step": 9275
    },
    {
      "epoch": 1.6092990978487163,
      "grad_norm": 1.2143728733062744,
      "learning_rate": 5.6024804088853775e-06,
      "loss": 0.4303,
      "step": 9276
    },
    {
      "epoch": 1.6094725884802221,
      "grad_norm": 0.8116788268089294,
      "learning_rate": 5.597752144468185e-06,
      "loss": 0.4598,
      "step": 9277
    },
    {
      "epoch": 1.609646079111728,
      "grad_norm": 0.9019766449928284,
      "learning_rate": 5.59302555144863e-06,
      "loss": 0.4512,
      "step": 9278
    },
    {
      "epoch": 1.6098195697432338,
      "grad_norm": 0.8071489930152893,
      "learning_rate": 5.588300630375252e-06,
      "loss": 0.3835,
      "step": 9279
    },
    {
      "epoch": 1.6099930603747397,
      "grad_norm": 1.1320544481277466,
      "learning_rate": 5.5835773817963814e-06,
      "loss": 0.4166,
      "step": 9280
    },
    {
      "epoch": 1.6101665510062455,
      "grad_norm": 0.8258172869682312,
      "learning_rate": 5.57885580626015e-06,
      "loss": 0.4205,
      "step": 9281
    },
    {
      "epoch": 1.6103400416377516,
      "grad_norm": 0.739963948726654,
      "learning_rate": 5.574135904314504e-06,
      "loss": 0.4767,
      "step": 9282
    },
    {
      "epoch": 1.6105135322692574,
      "grad_norm": 0.7183013558387756,
      "learning_rate": 5.5694176765072005e-06,
      "loss": 0.5292,
      "step": 9283
    },
    {
      "epoch": 1.6106870229007635,
      "grad_norm": 1.0741134881973267,
      "learning_rate": 5.5647011233857915e-06,
      "loss": 0.4429,
      "step": 9284
    },
    {
      "epoch": 1.6108605135322693,
      "grad_norm": 0.8927547335624695,
      "learning_rate": 5.559986245497637e-06,
      "loss": 0.3763,
      "step": 9285
    },
    {
      "epoch": 1.6110340041637752,
      "grad_norm": 0.9037680625915527,
      "learning_rate": 5.555273043389906e-06,
      "loss": 0.4192,
      "step": 9286
    },
    {
      "epoch": 1.611207494795281,
      "grad_norm": 0.8425066471099854,
      "learning_rate": 5.55056151760956e-06,
      "loss": 0.4099,
      "step": 9287
    },
    {
      "epoch": 1.6113809854267869,
      "grad_norm": 0.7884397506713867,
      "learning_rate": 5.545851668703397e-06,
      "loss": 0.4019,
      "step": 9288
    },
    {
      "epoch": 1.6115544760582927,
      "grad_norm": 0.6788613200187683,
      "learning_rate": 5.541143497217989e-06,
      "loss": 0.4137,
      "step": 9289
    },
    {
      "epoch": 1.6117279666897988,
      "grad_norm": 0.6453644037246704,
      "learning_rate": 5.536437003699724e-06,
      "loss": 0.5056,
      "step": 9290
    },
    {
      "epoch": 1.6119014573213046,
      "grad_norm": 0.7041041254997253,
      "learning_rate": 5.531732188694794e-06,
      "loss": 0.5112,
      "step": 9291
    },
    {
      "epoch": 1.6120749479528107,
      "grad_norm": 0.8735839128494263,
      "learning_rate": 5.527029052749216e-06,
      "loss": 0.3748,
      "step": 9292
    },
    {
      "epoch": 1.6122484385843165,
      "grad_norm": 1.1022486686706543,
      "learning_rate": 5.522327596408766e-06,
      "loss": 0.3633,
      "step": 9293
    },
    {
      "epoch": 1.6124219292158224,
      "grad_norm": 0.8820821642875671,
      "learning_rate": 5.517627820219076e-06,
      "loss": 0.3546,
      "step": 9294
    },
    {
      "epoch": 1.6125954198473282,
      "grad_norm": 0.8135870695114136,
      "learning_rate": 5.512929724725544e-06,
      "loss": 0.5264,
      "step": 9295
    },
    {
      "epoch": 1.612768910478834,
      "grad_norm": 1.2711671590805054,
      "learning_rate": 5.508233310473412e-06,
      "loss": 0.4166,
      "step": 9296
    },
    {
      "epoch": 1.61294240111034,
      "grad_norm": 1.0111753940582275,
      "learning_rate": 5.503538578007679e-06,
      "loss": 0.474,
      "step": 9297
    },
    {
      "epoch": 1.6131158917418458,
      "grad_norm": 0.9204239249229431,
      "learning_rate": 5.498845527873193e-06,
      "loss": 0.6093,
      "step": 9298
    },
    {
      "epoch": 1.6132893823733518,
      "grad_norm": 0.8759139776229858,
      "learning_rate": 5.494154160614578e-06,
      "loss": 0.4791,
      "step": 9299
    },
    {
      "epoch": 1.6134628730048577,
      "grad_norm": 1.0911972522735596,
      "learning_rate": 5.489464476776276e-06,
      "loss": 0.3347,
      "step": 9300
    },
    {
      "epoch": 1.6136363636363638,
      "grad_norm": 0.7725734114646912,
      "learning_rate": 5.484776476902525e-06,
      "loss": 0.4252,
      "step": 9301
    },
    {
      "epoch": 1.6138098542678696,
      "grad_norm": 0.7725723385810852,
      "learning_rate": 5.480090161537388e-06,
      "loss": 0.4652,
      "step": 9302
    },
    {
      "epoch": 1.6139833448993754,
      "grad_norm": 0.825690507888794,
      "learning_rate": 5.475405531224696e-06,
      "loss": 0.3943,
      "step": 9303
    },
    {
      "epoch": 1.6141568355308813,
      "grad_norm": 0.8373029828071594,
      "learning_rate": 5.470722586508122e-06,
      "loss": 0.377,
      "step": 9304
    },
    {
      "epoch": 1.6143303261623871,
      "grad_norm": 1.112805962562561,
      "learning_rate": 5.466041327931122e-06,
      "loss": 0.3735,
      "step": 9305
    },
    {
      "epoch": 1.614503816793893,
      "grad_norm": 1.0779210329055786,
      "learning_rate": 5.4613617560369625e-06,
      "loss": 0.4561,
      "step": 9306
    },
    {
      "epoch": 1.614677307425399,
      "grad_norm": 0.8767843246459961,
      "learning_rate": 5.456683871368704e-06,
      "loss": 0.4266,
      "step": 9307
    },
    {
      "epoch": 1.614850798056905,
      "grad_norm": 0.8819618225097656,
      "learning_rate": 5.452007674469235e-06,
      "loss": 0.4556,
      "step": 9308
    },
    {
      "epoch": 1.615024288688411,
      "grad_norm": 0.8689743876457214,
      "learning_rate": 5.447333165881228e-06,
      "loss": 0.4421,
      "step": 9309
    },
    {
      "epoch": 1.6151977793199168,
      "grad_norm": 0.9669204354286194,
      "learning_rate": 5.442660346147157e-06,
      "loss": 0.5767,
      "step": 9310
    },
    {
      "epoch": 1.6153712699514227,
      "grad_norm": 0.6955989003181458,
      "learning_rate": 5.437989215809323e-06,
      "loss": 0.5322,
      "step": 9311
    },
    {
      "epoch": 1.6155447605829285,
      "grad_norm": 0.7663390040397644,
      "learning_rate": 5.433319775409807e-06,
      "loss": 0.527,
      "step": 9312
    },
    {
      "epoch": 1.6157182512144344,
      "grad_norm": 1.5488319396972656,
      "learning_rate": 5.428652025490506e-06,
      "loss": 0.5129,
      "step": 9313
    },
    {
      "epoch": 1.6158917418459402,
      "grad_norm": 1.026512622833252,
      "learning_rate": 5.4239859665931105e-06,
      "loss": 0.4865,
      "step": 9314
    },
    {
      "epoch": 1.6160652324774463,
      "grad_norm": 0.9221273064613342,
      "learning_rate": 5.41932159925914e-06,
      "loss": 0.4036,
      "step": 9315
    },
    {
      "epoch": 1.6162387231089521,
      "grad_norm": 1.0832653045654297,
      "learning_rate": 5.4146589240298745e-06,
      "loss": 0.4113,
      "step": 9316
    },
    {
      "epoch": 1.6164122137404582,
      "grad_norm": 1.9201949834823608,
      "learning_rate": 5.409997941446443e-06,
      "loss": 0.5138,
      "step": 9317
    },
    {
      "epoch": 1.616585704371964,
      "grad_norm": 1.0195437669754028,
      "learning_rate": 5.405338652049749e-06,
      "loss": 0.4425,
      "step": 9318
    },
    {
      "epoch": 1.6167591950034699,
      "grad_norm": 0.7928690910339355,
      "learning_rate": 5.400681056380515e-06,
      "loss": 0.4218,
      "step": 9319
    },
    {
      "epoch": 1.6169326856349757,
      "grad_norm": 0.8504543900489807,
      "learning_rate": 5.396025154979247e-06,
      "loss": 0.4165,
      "step": 9320
    },
    {
      "epoch": 1.6171061762664816,
      "grad_norm": 0.7713993191719055,
      "learning_rate": 5.39137094838629e-06,
      "loss": 0.5248,
      "step": 9321
    },
    {
      "epoch": 1.6172796668979874,
      "grad_norm": 0.5526061654090881,
      "learning_rate": 5.386718437141743e-06,
      "loss": 0.574,
      "step": 9322
    },
    {
      "epoch": 1.6174531575294933,
      "grad_norm": 0.8415552973747253,
      "learning_rate": 5.382067621785556e-06,
      "loss": 0.4799,
      "step": 9323
    },
    {
      "epoch": 1.6176266481609993,
      "grad_norm": 0.9677339792251587,
      "learning_rate": 5.37741850285745e-06,
      "loss": 0.3543,
      "step": 9324
    },
    {
      "epoch": 1.6178001387925052,
      "grad_norm": 0.7718323469161987,
      "learning_rate": 5.372771080896977e-06,
      "loss": 0.4779,
      "step": 9325
    },
    {
      "epoch": 1.6179736294240112,
      "grad_norm": 0.6921229362487793,
      "learning_rate": 5.368125356443452e-06,
      "loss": 0.3804,
      "step": 9326
    },
    {
      "epoch": 1.618147120055517,
      "grad_norm": 0.6649701595306396,
      "learning_rate": 5.3634813300360355e-06,
      "loss": 0.5355,
      "step": 9327
    },
    {
      "epoch": 1.618320610687023,
      "grad_norm": 0.6825523972511292,
      "learning_rate": 5.358839002213665e-06,
      "loss": 0.4819,
      "step": 9328
    },
    {
      "epoch": 1.6184941013185288,
      "grad_norm": 0.7959592938423157,
      "learning_rate": 5.354198373515087e-06,
      "loss": 0.4883,
      "step": 9329
    },
    {
      "epoch": 1.6186675919500346,
      "grad_norm": 0.6361215710639954,
      "learning_rate": 5.349559444478849e-06,
      "loss": 0.5021,
      "step": 9330
    },
    {
      "epoch": 1.6188410825815405,
      "grad_norm": 1.068855881690979,
      "learning_rate": 5.344922215643316e-06,
      "loss": 0.4326,
      "step": 9331
    },
    {
      "epoch": 1.6190145732130465,
      "grad_norm": 0.7132313251495361,
      "learning_rate": 5.3402866875466344e-06,
      "loss": 0.5419,
      "step": 9332
    },
    {
      "epoch": 1.6191880638445524,
      "grad_norm": 1.0609829425811768,
      "learning_rate": 5.335652860726765e-06,
      "loss": 0.4026,
      "step": 9333
    },
    {
      "epoch": 1.6193615544760585,
      "grad_norm": 0.8046808838844299,
      "learning_rate": 5.331020735721469e-06,
      "loss": 0.453,
      "step": 9334
    },
    {
      "epoch": 1.6195350451075643,
      "grad_norm": 1.76338791847229,
      "learning_rate": 5.326390313068303e-06,
      "loss": 0.3383,
      "step": 9335
    },
    {
      "epoch": 1.6197085357390701,
      "grad_norm": 0.7001862525939941,
      "learning_rate": 5.321761593304646e-06,
      "loss": 0.4744,
      "step": 9336
    },
    {
      "epoch": 1.619882026370576,
      "grad_norm": 0.6396026015281677,
      "learning_rate": 5.317134576967658e-06,
      "loss": 0.5968,
      "step": 9337
    },
    {
      "epoch": 1.6200555170020818,
      "grad_norm": 0.6747438907623291,
      "learning_rate": 5.312509264594312e-06,
      "loss": 0.4683,
      "step": 9338
    },
    {
      "epoch": 1.6202290076335877,
      "grad_norm": 1.0268335342407227,
      "learning_rate": 5.307885656721374e-06,
      "loss": 0.3291,
      "step": 9339
    },
    {
      "epoch": 1.6204024982650935,
      "grad_norm": 0.9441016316413879,
      "learning_rate": 5.303263753885433e-06,
      "loss": 0.4294,
      "step": 9340
    },
    {
      "epoch": 1.6205759888965996,
      "grad_norm": 0.9289868474006653,
      "learning_rate": 5.298643556622858e-06,
      "loss": 0.4991,
      "step": 9341
    },
    {
      "epoch": 1.6207494795281054,
      "grad_norm": 1.0585331916809082,
      "learning_rate": 5.294025065469827e-06,
      "loss": 0.5472,
      "step": 9342
    },
    {
      "epoch": 1.6209229701596115,
      "grad_norm": 0.7627704739570618,
      "learning_rate": 5.2894082809623185e-06,
      "loss": 0.5371,
      "step": 9343
    },
    {
      "epoch": 1.6210964607911174,
      "grad_norm": 0.6469327211380005,
      "learning_rate": 5.284793203636132e-06,
      "loss": 0.5803,
      "step": 9344
    },
    {
      "epoch": 1.6212699514226232,
      "grad_norm": 0.8668105006217957,
      "learning_rate": 5.280179834026828e-06,
      "loss": 0.4592,
      "step": 9345
    },
    {
      "epoch": 1.621443442054129,
      "grad_norm": 1.0167548656463623,
      "learning_rate": 5.2755681726698134e-06,
      "loss": 0.372,
      "step": 9346
    },
    {
      "epoch": 1.621616932685635,
      "grad_norm": 0.915577232837677,
      "learning_rate": 5.270958220100269e-06,
      "loss": 0.5203,
      "step": 9347
    },
    {
      "epoch": 1.6217904233171407,
      "grad_norm": 0.7445098161697388,
      "learning_rate": 5.2663499768531865e-06,
      "loss": 0.5134,
      "step": 9348
    },
    {
      "epoch": 1.6219639139486468,
      "grad_norm": 0.9516087174415588,
      "learning_rate": 5.26174344346335e-06,
      "loss": 0.5135,
      "step": 9349
    },
    {
      "epoch": 1.6221374045801527,
      "grad_norm": 0.8416820764541626,
      "learning_rate": 5.257138620465374e-06,
      "loss": 0.389,
      "step": 9350
    },
    {
      "epoch": 1.6223108952116587,
      "grad_norm": 0.6412305235862732,
      "learning_rate": 5.252535508393628e-06,
      "loss": 0.5183,
      "step": 9351
    },
    {
      "epoch": 1.6224843858431646,
      "grad_norm": 0.8845618963241577,
      "learning_rate": 5.247934107782324e-06,
      "loss": 0.4965,
      "step": 9352
    },
    {
      "epoch": 1.6226578764746704,
      "grad_norm": 1.1905583143234253,
      "learning_rate": 5.243334419165453e-06,
      "loss": 0.4443,
      "step": 9353
    },
    {
      "epoch": 1.6228313671061763,
      "grad_norm": 1.2265573740005493,
      "learning_rate": 5.238736443076828e-06,
      "loss": 0.5354,
      "step": 9354
    },
    {
      "epoch": 1.623004857737682,
      "grad_norm": 0.8113061189651489,
      "learning_rate": 5.234140180050029e-06,
      "loss": 0.3723,
      "step": 9355
    },
    {
      "epoch": 1.623178348369188,
      "grad_norm": 0.6684175729751587,
      "learning_rate": 5.2295456306184715e-06,
      "loss": 0.4166,
      "step": 9356
    },
    {
      "epoch": 1.6233518390006938,
      "grad_norm": 0.6276340484619141,
      "learning_rate": 5.2249527953153545e-06,
      "loss": 0.4747,
      "step": 9357
    },
    {
      "epoch": 1.6235253296321999,
      "grad_norm": 0.6321830153465271,
      "learning_rate": 5.220361674673677e-06,
      "loss": 0.4192,
      "step": 9358
    },
    {
      "epoch": 1.6236988202637057,
      "grad_norm": 0.7691701054573059,
      "learning_rate": 5.215772269226255e-06,
      "loss": 0.3533,
      "step": 9359
    },
    {
      "epoch": 1.6238723108952118,
      "grad_norm": 0.7306829690933228,
      "learning_rate": 5.211184579505688e-06,
      "loss": 0.3625,
      "step": 9360
    },
    {
      "epoch": 1.6240458015267176,
      "grad_norm": 0.8247324824333191,
      "learning_rate": 5.206598606044384e-06,
      "loss": 0.3588,
      "step": 9361
    },
    {
      "epoch": 1.6242192921582235,
      "grad_norm": 0.9606726169586182,
      "learning_rate": 5.2020143493745425e-06,
      "loss": 0.3992,
      "step": 9362
    },
    {
      "epoch": 1.6243927827897293,
      "grad_norm": 0.766910970211029,
      "learning_rate": 5.1974318100281905e-06,
      "loss": 0.5186,
      "step": 9363
    },
    {
      "epoch": 1.6245662734212352,
      "grad_norm": 0.795785665512085,
      "learning_rate": 5.1928509885371124e-06,
      "loss": 0.5034,
      "step": 9364
    },
    {
      "epoch": 1.624739764052741,
      "grad_norm": 0.7035911679267883,
      "learning_rate": 5.188271885432938e-06,
      "loss": 0.5208,
      "step": 9365
    },
    {
      "epoch": 1.624913254684247,
      "grad_norm": 0.8098400831222534,
      "learning_rate": 5.183694501247072e-06,
      "loss": 0.4284,
      "step": 9366
    },
    {
      "epoch": 1.625086745315753,
      "grad_norm": 0.9330823421478271,
      "learning_rate": 5.179118836510721e-06,
      "loss": 0.4572,
      "step": 9367
    },
    {
      "epoch": 1.625260235947259,
      "grad_norm": 0.8530337810516357,
      "learning_rate": 5.174544891754896e-06,
      "loss": 0.5416,
      "step": 9368
    },
    {
      "epoch": 1.6254337265787648,
      "grad_norm": 0.7967491149902344,
      "learning_rate": 5.169972667510414e-06,
      "loss": 0.4255,
      "step": 9369
    },
    {
      "epoch": 1.6256072172102707,
      "grad_norm": 0.7864426970481873,
      "learning_rate": 5.165402164307884e-06,
      "loss": 0.5245,
      "step": 9370
    },
    {
      "epoch": 1.6257807078417765,
      "grad_norm": 0.9544775485992432,
      "learning_rate": 5.160833382677721e-06,
      "loss": 0.4849,
      "step": 9371
    },
    {
      "epoch": 1.6259541984732824,
      "grad_norm": 0.9821032285690308,
      "learning_rate": 5.15626632315013e-06,
      "loss": 0.4447,
      "step": 9372
    },
    {
      "epoch": 1.6261276891047882,
      "grad_norm": 0.8680276870727539,
      "learning_rate": 5.151700986255137e-06,
      "loss": 0.4697,
      "step": 9373
    },
    {
      "epoch": 1.6263011797362943,
      "grad_norm": 0.8093870878219604,
      "learning_rate": 5.147137372522537e-06,
      "loss": 0.418,
      "step": 9374
    },
    {
      "epoch": 1.6264746703678001,
      "grad_norm": 1.3203651905059814,
      "learning_rate": 5.142575482481957e-06,
      "loss": 0.4877,
      "step": 9375
    },
    {
      "epoch": 1.6266481609993062,
      "grad_norm": 0.8356888294219971,
      "learning_rate": 5.138015316662803e-06,
      "loss": 0.4918,
      "step": 9376
    },
    {
      "epoch": 1.626821651630812,
      "grad_norm": 0.8403335809707642,
      "learning_rate": 5.1334568755942915e-06,
      "loss": 0.4158,
      "step": 9377
    },
    {
      "epoch": 1.626995142262318,
      "grad_norm": 0.6890471577644348,
      "learning_rate": 5.128900159805425e-06,
      "loss": 0.4746,
      "step": 9378
    },
    {
      "epoch": 1.6271686328938237,
      "grad_norm": 0.9023023247718811,
      "learning_rate": 5.124345169825031e-06,
      "loss": 0.4921,
      "step": 9379
    },
    {
      "epoch": 1.6273421235253296,
      "grad_norm": 0.777741551399231,
      "learning_rate": 5.119791906181713e-06,
      "loss": 0.4172,
      "step": 9380
    },
    {
      "epoch": 1.6275156141568354,
      "grad_norm": 0.7545634508132935,
      "learning_rate": 5.115240369403882e-06,
      "loss": 0.4095,
      "step": 9381
    },
    {
      "epoch": 1.6276891047883413,
      "grad_norm": 0.7467777132987976,
      "learning_rate": 5.110690560019744e-06,
      "loss": 0.4971,
      "step": 9382
    },
    {
      "epoch": 1.6278625954198473,
      "grad_norm": 0.7398180365562439,
      "learning_rate": 5.106142478557323e-06,
      "loss": 0.4148,
      "step": 9383
    },
    {
      "epoch": 1.6280360860513532,
      "grad_norm": 0.757480263710022,
      "learning_rate": 5.1015961255444235e-06,
      "loss": 0.3639,
      "step": 9384
    },
    {
      "epoch": 1.6282095766828593,
      "grad_norm": 0.8431456685066223,
      "learning_rate": 5.097051501508652e-06,
      "loss": 0.4777,
      "step": 9385
    },
    {
      "epoch": 1.628383067314365,
      "grad_norm": 0.9009338021278381,
      "learning_rate": 5.09250860697742e-06,
      "loss": 0.36,
      "step": 9386
    },
    {
      "epoch": 1.628556557945871,
      "grad_norm": 0.7599311470985413,
      "learning_rate": 5.087967442477928e-06,
      "loss": 0.3723,
      "step": 9387
    },
    {
      "epoch": 1.6287300485773768,
      "grad_norm": 0.7761737108230591,
      "learning_rate": 5.083428008537197e-06,
      "loss": 0.372,
      "step": 9388
    },
    {
      "epoch": 1.6289035392088826,
      "grad_norm": 0.7962995767593384,
      "learning_rate": 5.078890305682027e-06,
      "loss": 0.4002,
      "step": 9389
    },
    {
      "epoch": 1.6290770298403885,
      "grad_norm": 0.9658076167106628,
      "learning_rate": 5.074354334439022e-06,
      "loss": 0.5925,
      "step": 9390
    },
    {
      "epoch": 1.6292505204718946,
      "grad_norm": 0.8104629516601562,
      "learning_rate": 5.069820095334583e-06,
      "loss": 0.4004,
      "step": 9391
    },
    {
      "epoch": 1.6294240111034004,
      "grad_norm": 1.1162132024765015,
      "learning_rate": 5.065287588894933e-06,
      "loss": 0.4384,
      "step": 9392
    },
    {
      "epoch": 1.6295975017349065,
      "grad_norm": 1.3077670335769653,
      "learning_rate": 5.060756815646046e-06,
      "loss": 0.3755,
      "step": 9393
    },
    {
      "epoch": 1.6297709923664123,
      "grad_norm": 2.617023468017578,
      "learning_rate": 5.056227776113747e-06,
      "loss": 0.395,
      "step": 9394
    },
    {
      "epoch": 1.6299444829979182,
      "grad_norm": 0.7315855622291565,
      "learning_rate": 5.05170047082362e-06,
      "loss": 0.554,
      "step": 9395
    },
    {
      "epoch": 1.630117973629424,
      "grad_norm": 0.748149037361145,
      "learning_rate": 5.0471749003010835e-06,
      "loss": 0.3365,
      "step": 9396
    },
    {
      "epoch": 1.6302914642609299,
      "grad_norm": 0.771815836429596,
      "learning_rate": 5.0426510650713116e-06,
      "loss": 0.3942,
      "step": 9397
    },
    {
      "epoch": 1.6304649548924357,
      "grad_norm": 0.8909497857093811,
      "learning_rate": 5.038128965659317e-06,
      "loss": 0.3345,
      "step": 9398
    },
    {
      "epoch": 1.6306384455239415,
      "grad_norm": 0.802179753780365,
      "learning_rate": 5.033608602589892e-06,
      "loss": 0.4484,
      "step": 9399
    },
    {
      "epoch": 1.6308119361554476,
      "grad_norm": 0.7441188097000122,
      "learning_rate": 5.029089976387627e-06,
      "loss": 0.4084,
      "step": 9400
    },
    {
      "epoch": 1.6309854267869535,
      "grad_norm": 0.7388148903846741,
      "learning_rate": 5.0245730875769095e-06,
      "loss": 0.4401,
      "step": 9401
    },
    {
      "epoch": 1.6311589174184595,
      "grad_norm": 0.9957874417304993,
      "learning_rate": 5.020057936681939e-06,
      "loss": 0.417,
      "step": 9402
    },
    {
      "epoch": 1.6313324080499654,
      "grad_norm": 0.7852291464805603,
      "learning_rate": 5.0155445242267006e-06,
      "loss": 0.3761,
      "step": 9403
    },
    {
      "epoch": 1.6315058986814712,
      "grad_norm": 0.7357715964317322,
      "learning_rate": 5.011032850734983e-06,
      "loss": 0.399,
      "step": 9404
    },
    {
      "epoch": 1.631679389312977,
      "grad_norm": 0.8344091773033142,
      "learning_rate": 5.006522916730368e-06,
      "loss": 0.3878,
      "step": 9405
    },
    {
      "epoch": 1.631852879944483,
      "grad_norm": 0.6356186866760254,
      "learning_rate": 5.00201472273623e-06,
      "loss": 0.4965,
      "step": 9406
    },
    {
      "epoch": 1.6320263705759888,
      "grad_norm": 0.9192460179328918,
      "learning_rate": 4.9975082692757705e-06,
      "loss": 0.5393,
      "step": 9407
    },
    {
      "epoch": 1.6321998612074948,
      "grad_norm": 1.1186772584915161,
      "learning_rate": 4.993003556871954e-06,
      "loss": 0.3128,
      "step": 9408
    },
    {
      "epoch": 1.6323733518390007,
      "grad_norm": 0.6981306672096252,
      "learning_rate": 4.9885005860475626e-06,
      "loss": 0.5186,
      "step": 9409
    },
    {
      "epoch": 1.6325468424705067,
      "grad_norm": 1.1612515449523926,
      "learning_rate": 4.983999357325164e-06,
      "loss": 0.5011,
      "step": 9410
    },
    {
      "epoch": 1.6327203331020126,
      "grad_norm": 1.1904067993164062,
      "learning_rate": 4.9794998712271425e-06,
      "loss": 0.5631,
      "step": 9411
    },
    {
      "epoch": 1.6328938237335184,
      "grad_norm": 0.772208034992218,
      "learning_rate": 4.975002128275666e-06,
      "loss": 0.3475,
      "step": 9412
    },
    {
      "epoch": 1.6330673143650243,
      "grad_norm": 0.7802537083625793,
      "learning_rate": 4.970506128992696e-06,
      "loss": 0.5374,
      "step": 9413
    },
    {
      "epoch": 1.6332408049965301,
      "grad_norm": 1.3811105489730835,
      "learning_rate": 4.966011873900001e-06,
      "loss": 0.3776,
      "step": 9414
    },
    {
      "epoch": 1.633414295628036,
      "grad_norm": 1.1440554857254028,
      "learning_rate": 4.961519363519154e-06,
      "loss": 0.4086,
      "step": 9415
    },
    {
      "epoch": 1.6335877862595418,
      "grad_norm": 0.6254736185073853,
      "learning_rate": 4.957028598371498e-06,
      "loss": 0.5498,
      "step": 9416
    },
    {
      "epoch": 1.6337612768910479,
      "grad_norm": 1.0763905048370361,
      "learning_rate": 4.9525395789782085e-06,
      "loss": 0.4874,
      "step": 9417
    },
    {
      "epoch": 1.6339347675225537,
      "grad_norm": 0.9219430088996887,
      "learning_rate": 4.948052305860233e-06,
      "loss": 0.3763,
      "step": 9418
    },
    {
      "epoch": 1.6341082581540598,
      "grad_norm": 0.8166424632072449,
      "learning_rate": 4.943566779538327e-06,
      "loss": 0.4628,
      "step": 9419
    },
    {
      "epoch": 1.6342817487855656,
      "grad_norm": 0.7532215714454651,
      "learning_rate": 4.939083000533036e-06,
      "loss": 0.394,
      "step": 9420
    },
    {
      "epoch": 1.6344552394170715,
      "grad_norm": 1.005659818649292,
      "learning_rate": 4.934600969364722e-06,
      "loss": 0.4742,
      "step": 9421
    },
    {
      "epoch": 1.6346287300485773,
      "grad_norm": 0.8567238450050354,
      "learning_rate": 4.93012068655351e-06,
      "loss": 0.6027,
      "step": 9422
    },
    {
      "epoch": 1.6348022206800832,
      "grad_norm": 0.9618515372276306,
      "learning_rate": 4.925642152619357e-06,
      "loss": 0.4766,
      "step": 9423
    },
    {
      "epoch": 1.634975711311589,
      "grad_norm": 0.6993759274482727,
      "learning_rate": 4.92116536808199e-06,
      "loss": 0.4626,
      "step": 9424
    },
    {
      "epoch": 1.635149201943095,
      "grad_norm": 0.8960062861442566,
      "learning_rate": 4.9166903334609675e-06,
      "loss": 0.5129,
      "step": 9425
    },
    {
      "epoch": 1.635322692574601,
      "grad_norm": 0.8389983177185059,
      "learning_rate": 4.912217049275594e-06,
      "loss": 0.4299,
      "step": 9426
    },
    {
      "epoch": 1.635496183206107,
      "grad_norm": 0.9389564394950867,
      "learning_rate": 4.907745516045017e-06,
      "loss": 0.4811,
      "step": 9427
    },
    {
      "epoch": 1.6356696738376129,
      "grad_norm": 0.8085852861404419,
      "learning_rate": 4.903275734288162e-06,
      "loss": 0.5209,
      "step": 9428
    },
    {
      "epoch": 1.6358431644691187,
      "grad_norm": 0.7433094382286072,
      "learning_rate": 4.898807704523747e-06,
      "loss": 0.5425,
      "step": 9429
    },
    {
      "epoch": 1.6360166551006246,
      "grad_norm": 1.0063952207565308,
      "learning_rate": 4.8943414272702886e-06,
      "loss": 0.365,
      "step": 9430
    },
    {
      "epoch": 1.6361901457321304,
      "grad_norm": 0.7665280103683472,
      "learning_rate": 4.889876903046116e-06,
      "loss": 0.5081,
      "step": 9431
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 0.8327005505561829,
      "learning_rate": 4.885414132369335e-06,
      "loss": 0.4341,
      "step": 9432
    },
    {
      "epoch": 1.6365371269951423,
      "grad_norm": 0.8074477314949036,
      "learning_rate": 4.8809531157578495e-06,
      "loss": 0.3812,
      "step": 9433
    },
    {
      "epoch": 1.6367106176266482,
      "grad_norm": 2.5853981971740723,
      "learning_rate": 4.876493853729385e-06,
      "loss": 0.3887,
      "step": 9434
    },
    {
      "epoch": 1.6368841082581542,
      "grad_norm": 0.6790578961372375,
      "learning_rate": 4.872036346801419e-06,
      "loss": 0.4586,
      "step": 9435
    },
    {
      "epoch": 1.63705759888966,
      "grad_norm": 1.1151888370513916,
      "learning_rate": 4.867580595491268e-06,
      "loss": 0.3708,
      "step": 9436
    },
    {
      "epoch": 1.637231089521166,
      "grad_norm": 0.6660562753677368,
      "learning_rate": 4.863126600316021e-06,
      "loss": 0.4998,
      "step": 9437
    },
    {
      "epoch": 1.6374045801526718,
      "grad_norm": 0.679646372795105,
      "learning_rate": 4.858674361792571e-06,
      "loss": 0.4925,
      "step": 9438
    },
    {
      "epoch": 1.6375780707841776,
      "grad_norm": 1.0015981197357178,
      "learning_rate": 4.854223880437599e-06,
      "loss": 0.4862,
      "step": 9439
    },
    {
      "epoch": 1.6377515614156835,
      "grad_norm": 0.7962157726287842,
      "learning_rate": 4.849775156767598e-06,
      "loss": 0.4476,
      "step": 9440
    },
    {
      "epoch": 1.6379250520471893,
      "grad_norm": 0.7867410182952881,
      "learning_rate": 4.845328191298846e-06,
      "loss": 0.5082,
      "step": 9441
    },
    {
      "epoch": 1.6380985426786954,
      "grad_norm": 0.6451948881149292,
      "learning_rate": 4.840882984547415e-06,
      "loss": 0.6041,
      "step": 9442
    },
    {
      "epoch": 1.6382720333102012,
      "grad_norm": 0.6247390508651733,
      "learning_rate": 4.8364395370291715e-06,
      "loss": 0.5873,
      "step": 9443
    },
    {
      "epoch": 1.6384455239417073,
      "grad_norm": 0.6129417419433594,
      "learning_rate": 4.8319978492598e-06,
      "loss": 0.6053,
      "step": 9444
    },
    {
      "epoch": 1.6386190145732131,
      "grad_norm": 0.6954910159111023,
      "learning_rate": 4.827557921754742e-06,
      "loss": 0.5193,
      "step": 9445
    },
    {
      "epoch": 1.638792505204719,
      "grad_norm": 0.8252256512641907,
      "learning_rate": 4.823119755029271e-06,
      "loss": 0.4645,
      "step": 9446
    },
    {
      "epoch": 1.6389659958362248,
      "grad_norm": 1.2125977277755737,
      "learning_rate": 4.818683349598439e-06,
      "loss": 0.4218,
      "step": 9447
    },
    {
      "epoch": 1.6391394864677307,
      "grad_norm": 0.9081424474716187,
      "learning_rate": 4.814248705977092e-06,
      "loss": 0.3681,
      "step": 9448
    },
    {
      "epoch": 1.6393129770992365,
      "grad_norm": 0.6138148307800293,
      "learning_rate": 4.8098158246798734e-06,
      "loss": 0.4502,
      "step": 9449
    },
    {
      "epoch": 1.6394864677307426,
      "grad_norm": 0.7577382922172546,
      "learning_rate": 4.805384706221232e-06,
      "loss": 0.442,
      "step": 9450
    },
    {
      "epoch": 1.6396599583622484,
      "grad_norm": 0.8864315152168274,
      "learning_rate": 4.800955351115402e-06,
      "loss": 0.3993,
      "step": 9451
    },
    {
      "epoch": 1.6398334489937545,
      "grad_norm": 0.8313661217689514,
      "learning_rate": 4.796527759876415e-06,
      "loss": 0.4368,
      "step": 9452
    },
    {
      "epoch": 1.6400069396252603,
      "grad_norm": 0.8117265105247498,
      "learning_rate": 4.79210193301809e-06,
      "loss": 0.4047,
      "step": 9453
    },
    {
      "epoch": 1.6401804302567662,
      "grad_norm": 1.246932864189148,
      "learning_rate": 4.787677871054062e-06,
      "loss": 0.4875,
      "step": 9454
    },
    {
      "epoch": 1.640353920888272,
      "grad_norm": 0.7654351592063904,
      "learning_rate": 4.783255574497742e-06,
      "loss": 0.4834,
      "step": 9455
    },
    {
      "epoch": 1.6405274115197779,
      "grad_norm": 0.8894326686859131,
      "learning_rate": 4.7788350438623465e-06,
      "loss": 0.4487,
      "step": 9456
    },
    {
      "epoch": 1.6407009021512837,
      "grad_norm": 0.6546960473060608,
      "learning_rate": 4.774416279660879e-06,
      "loss": 0.4564,
      "step": 9457
    },
    {
      "epoch": 1.6408743927827896,
      "grad_norm": 0.6331251859664917,
      "learning_rate": 4.769999282406137e-06,
      "loss": 0.5345,
      "step": 9458
    },
    {
      "epoch": 1.6410478834142956,
      "grad_norm": 0.828997015953064,
      "learning_rate": 4.76558405261073e-06,
      "loss": 0.5928,
      "step": 9459
    },
    {
      "epoch": 1.6412213740458015,
      "grad_norm": 0.7962414622306824,
      "learning_rate": 4.7611705907870474e-06,
      "loss": 0.3768,
      "step": 9460
    },
    {
      "epoch": 1.6413948646773076,
      "grad_norm": 0.6648374795913696,
      "learning_rate": 4.7567588974472734e-06,
      "loss": 0.5637,
      "step": 9461
    },
    {
      "epoch": 1.6415683553088134,
      "grad_norm": 0.6690629124641418,
      "learning_rate": 4.7523489731033845e-06,
      "loss": 0.5367,
      "step": 9462
    },
    {
      "epoch": 1.6417418459403192,
      "grad_norm": 0.6988204121589661,
      "learning_rate": 4.747940818267178e-06,
      "loss": 0.4329,
      "step": 9463
    },
    {
      "epoch": 1.641915336571825,
      "grad_norm": 0.8069101572036743,
      "learning_rate": 4.743534433450199e-06,
      "loss": 0.4016,
      "step": 9464
    },
    {
      "epoch": 1.642088827203331,
      "grad_norm": 0.6000931262969971,
      "learning_rate": 4.739129819163832e-06,
      "loss": 0.4399,
      "step": 9465
    },
    {
      "epoch": 1.6422623178348368,
      "grad_norm": 0.7521392703056335,
      "learning_rate": 4.734726975919233e-06,
      "loss": 0.5347,
      "step": 9466
    },
    {
      "epoch": 1.6424358084663429,
      "grad_norm": 0.826542854309082,
      "learning_rate": 4.730325904227355e-06,
      "loss": 0.4824,
      "step": 9467
    },
    {
      "epoch": 1.6426092990978487,
      "grad_norm": 0.8546525239944458,
      "learning_rate": 4.725926604598942e-06,
      "loss": 0.4591,
      "step": 9468
    },
    {
      "epoch": 1.6427827897293548,
      "grad_norm": 0.6947259306907654,
      "learning_rate": 4.721529077544551e-06,
      "loss": 0.4568,
      "step": 9469
    },
    {
      "epoch": 1.6429562803608606,
      "grad_norm": 0.7343437671661377,
      "learning_rate": 4.7171333235745145e-06,
      "loss": 0.4131,
      "step": 9470
    },
    {
      "epoch": 1.6431297709923665,
      "grad_norm": 0.7316738367080688,
      "learning_rate": 4.712739343198962e-06,
      "loss": 0.3978,
      "step": 9471
    },
    {
      "epoch": 1.6433032616238723,
      "grad_norm": 0.8545839190483093,
      "learning_rate": 4.708347136927818e-06,
      "loss": 0.4841,
      "step": 9472
    },
    {
      "epoch": 1.6434767522553781,
      "grad_norm": 1.1487629413604736,
      "learning_rate": 4.703956705270818e-06,
      "loss": 0.4713,
      "step": 9473
    },
    {
      "epoch": 1.643650242886884,
      "grad_norm": 0.7485223412513733,
      "learning_rate": 4.699568048737453e-06,
      "loss": 0.4142,
      "step": 9474
    },
    {
      "epoch": 1.64382373351839,
      "grad_norm": 0.6850613355636597,
      "learning_rate": 4.695181167837051e-06,
      "loss": 0.4752,
      "step": 9475
    },
    {
      "epoch": 1.643997224149896,
      "grad_norm": 0.861230194568634,
      "learning_rate": 4.690796063078709e-06,
      "loss": 0.4969,
      "step": 9476
    },
    {
      "epoch": 1.6441707147814018,
      "grad_norm": 0.9509091973304749,
      "learning_rate": 4.686412734971322e-06,
      "loss": 0.4342,
      "step": 9477
    },
    {
      "epoch": 1.6443442054129078,
      "grad_norm": 0.9492801427841187,
      "learning_rate": 4.6820311840235745e-06,
      "loss": 0.4766,
      "step": 9478
    },
    {
      "epoch": 1.6445176960444137,
      "grad_norm": 0.8114972710609436,
      "learning_rate": 4.677651410743964e-06,
      "loss": 0.4592,
      "step": 9479
    },
    {
      "epoch": 1.6446911866759195,
      "grad_norm": 1.0280996561050415,
      "learning_rate": 4.67327341564076e-06,
      "loss": 0.5789,
      "step": 9480
    },
    {
      "epoch": 1.6448646773074254,
      "grad_norm": 0.7430195808410645,
      "learning_rate": 4.668897199222031e-06,
      "loss": 0.4691,
      "step": 9481
    },
    {
      "epoch": 1.6450381679389312,
      "grad_norm": 0.6867184638977051,
      "learning_rate": 4.6645227619956515e-06,
      "loss": 0.4368,
      "step": 9482
    },
    {
      "epoch": 1.645211658570437,
      "grad_norm": 0.9730817675590515,
      "learning_rate": 4.660150104469274e-06,
      "loss": 0.3169,
      "step": 9483
    },
    {
      "epoch": 1.6453851492019431,
      "grad_norm": 0.8736039400100708,
      "learning_rate": 4.655779227150352e-06,
      "loss": 0.4745,
      "step": 9484
    },
    {
      "epoch": 1.645558639833449,
      "grad_norm": 0.8217387795448303,
      "learning_rate": 4.6514101305461255e-06,
      "loss": 0.3871,
      "step": 9485
    },
    {
      "epoch": 1.645732130464955,
      "grad_norm": 0.6346628665924072,
      "learning_rate": 4.647042815163649e-06,
      "loss": 0.4198,
      "step": 9486
    },
    {
      "epoch": 1.6459056210964609,
      "grad_norm": 0.8603171110153198,
      "learning_rate": 4.6426772815097306e-06,
      "loss": 0.3154,
      "step": 9487
    },
    {
      "epoch": 1.6460791117279667,
      "grad_norm": 0.7276177406311035,
      "learning_rate": 4.638313530091016e-06,
      "loss": 0.4436,
      "step": 9488
    },
    {
      "epoch": 1.6462526023594726,
      "grad_norm": 0.9174731373786926,
      "learning_rate": 4.633951561413916e-06,
      "loss": 0.5026,
      "step": 9489
    },
    {
      "epoch": 1.6464260929909784,
      "grad_norm": 1.3416101932525635,
      "learning_rate": 4.629591375984641e-06,
      "loss": 0.3343,
      "step": 9490
    },
    {
      "epoch": 1.6465995836224843,
      "grad_norm": 0.9114309549331665,
      "learning_rate": 4.625232974309193e-06,
      "loss": 0.4585,
      "step": 9491
    },
    {
      "epoch": 1.6467730742539903,
      "grad_norm": 1.5631992816925049,
      "learning_rate": 4.620876356893385e-06,
      "loss": 0.3553,
      "step": 9492
    },
    {
      "epoch": 1.6469465648854962,
      "grad_norm": 0.6930509209632874,
      "learning_rate": 4.616521524242783e-06,
      "loss": 0.4706,
      "step": 9493
    },
    {
      "epoch": 1.6471200555170022,
      "grad_norm": 0.9006015658378601,
      "learning_rate": 4.612168476862789e-06,
      "loss": 0.4557,
      "step": 9494
    },
    {
      "epoch": 1.647293546148508,
      "grad_norm": 0.8068966865539551,
      "learning_rate": 4.607817215258574e-06,
      "loss": 0.4913,
      "step": 9495
    },
    {
      "epoch": 1.647467036780014,
      "grad_norm": 1.1427596807479858,
      "learning_rate": 4.603467739935108e-06,
      "loss": 0.3982,
      "step": 9496
    },
    {
      "epoch": 1.6476405274115198,
      "grad_norm": 0.8430987596511841,
      "learning_rate": 4.599120051397144e-06,
      "loss": 0.4786,
      "step": 9497
    },
    {
      "epoch": 1.6478140180430256,
      "grad_norm": 0.7175557613372803,
      "learning_rate": 4.594774150149251e-06,
      "loss": 0.4966,
      "step": 9498
    },
    {
      "epoch": 1.6479875086745315,
      "grad_norm": 1.0722674131393433,
      "learning_rate": 4.5904300366957675e-06,
      "loss": 0.4788,
      "step": 9499
    },
    {
      "epoch": 1.6481609993060373,
      "grad_norm": 2.1372292041778564,
      "learning_rate": 4.586087711540832e-06,
      "loss": 0.4724,
      "step": 9500
    },
    {
      "epoch": 1.6483344899375434,
      "grad_norm": 1.5587410926818848,
      "learning_rate": 4.581747175188376e-06,
      "loss": 0.3839,
      "step": 9501
    },
    {
      "epoch": 1.6485079805690492,
      "grad_norm": 0.8519799709320068,
      "learning_rate": 4.57740842814213e-06,
      "loss": 0.4861,
      "step": 9502
    },
    {
      "epoch": 1.6486814712005553,
      "grad_norm": 0.7175669074058533,
      "learning_rate": 4.573071470905608e-06,
      "loss": 0.4009,
      "step": 9503
    },
    {
      "epoch": 1.6488549618320612,
      "grad_norm": 0.7031004428863525,
      "learning_rate": 4.568736303982115e-06,
      "loss": 0.3728,
      "step": 9504
    },
    {
      "epoch": 1.649028452463567,
      "grad_norm": 0.7492570281028748,
      "learning_rate": 4.564402927874758e-06,
      "loss": 0.3357,
      "step": 9505
    },
    {
      "epoch": 1.6492019430950728,
      "grad_norm": 1.0007660388946533,
      "learning_rate": 4.560071343086421e-06,
      "loss": 0.3837,
      "step": 9506
    },
    {
      "epoch": 1.6493754337265787,
      "grad_norm": 0.9471139311790466,
      "learning_rate": 4.555741550119801e-06,
      "loss": 0.3647,
      "step": 9507
    },
    {
      "epoch": 1.6495489243580845,
      "grad_norm": 0.7544403076171875,
      "learning_rate": 4.55141354947737e-06,
      "loss": 0.4247,
      "step": 9508
    },
    {
      "epoch": 1.6497224149895906,
      "grad_norm": 0.7976730465888977,
      "learning_rate": 4.547087341661398e-06,
      "loss": 0.4744,
      "step": 9509
    },
    {
      "epoch": 1.6498959056210964,
      "grad_norm": 0.9931909441947937,
      "learning_rate": 4.542762927173941e-06,
      "loss": 0.4789,
      "step": 9510
    },
    {
      "epoch": 1.6500693962526025,
      "grad_norm": 0.6307627558708191,
      "learning_rate": 4.53844030651686e-06,
      "loss": 0.5188,
      "step": 9511
    },
    {
      "epoch": 1.6502428868841084,
      "grad_norm": 0.6200315952301025,
      "learning_rate": 4.534119480191801e-06,
      "loss": 0.5017,
      "step": 9512
    },
    {
      "epoch": 1.6504163775156142,
      "grad_norm": 0.7837086915969849,
      "learning_rate": 4.5298004487001966e-06,
      "loss": 0.3384,
      "step": 9513
    },
    {
      "epoch": 1.65058986814712,
      "grad_norm": 0.7162777185440063,
      "learning_rate": 4.525483212543273e-06,
      "loss": 0.5287,
      "step": 9514
    },
    {
      "epoch": 1.650763358778626,
      "grad_norm": 0.9284862279891968,
      "learning_rate": 4.521167772222064e-06,
      "loss": 0.5698,
      "step": 9515
    },
    {
      "epoch": 1.6509368494101317,
      "grad_norm": 0.8446031212806702,
      "learning_rate": 4.516854128237358e-06,
      "loss": 0.5477,
      "step": 9516
    },
    {
      "epoch": 1.6511103400416376,
      "grad_norm": 0.8378081321716309,
      "learning_rate": 4.51254228108978e-06,
      "loss": 0.36,
      "step": 9517
    },
    {
      "epoch": 1.6512838306731437,
      "grad_norm": 0.9502765536308289,
      "learning_rate": 4.5082322312797166e-06,
      "loss": 0.3824,
      "step": 9518
    },
    {
      "epoch": 1.6514573213046495,
      "grad_norm": 1.0102424621582031,
      "learning_rate": 4.503923979307352e-06,
      "loss": 0.5515,
      "step": 9519
    },
    {
      "epoch": 1.6516308119361556,
      "grad_norm": 0.9823939800262451,
      "learning_rate": 4.499617525672664e-06,
      "loss": 0.4515,
      "step": 9520
    },
    {
      "epoch": 1.6518043025676614,
      "grad_norm": 0.9717512726783752,
      "learning_rate": 4.4953128708754326e-06,
      "loss": 0.5508,
      "step": 9521
    },
    {
      "epoch": 1.6519777931991673,
      "grad_norm": 0.9089481234550476,
      "learning_rate": 4.491010015415198e-06,
      "loss": 0.4867,
      "step": 9522
    },
    {
      "epoch": 1.6521512838306731,
      "grad_norm": 1.0057765245437622,
      "learning_rate": 4.486708959791328e-06,
      "loss": 0.4462,
      "step": 9523
    },
    {
      "epoch": 1.652324774462179,
      "grad_norm": 1.0206904411315918,
      "learning_rate": 4.4824097045029615e-06,
      "loss": 0.4706,
      "step": 9524
    },
    {
      "epoch": 1.6524982650936848,
      "grad_norm": 0.6364613175392151,
      "learning_rate": 4.478112250049029e-06,
      "loss": 0.5292,
      "step": 9525
    },
    {
      "epoch": 1.6526717557251909,
      "grad_norm": 0.7264257669448853,
      "learning_rate": 4.473816596928251e-06,
      "loss": 0.59,
      "step": 9526
    },
    {
      "epoch": 1.6528452463566967,
      "grad_norm": 0.6859540939331055,
      "learning_rate": 4.469522745639154e-06,
      "loss": 0.6169,
      "step": 9527
    },
    {
      "epoch": 1.6530187369882028,
      "grad_norm": 0.7884160876274109,
      "learning_rate": 4.465230696680038e-06,
      "loss": 0.4731,
      "step": 9528
    },
    {
      "epoch": 1.6531922276197086,
      "grad_norm": 1.012871265411377,
      "learning_rate": 4.460940450548998e-06,
      "loss": 0.4489,
      "step": 9529
    },
    {
      "epoch": 1.6533657182512145,
      "grad_norm": 0.7458043694496155,
      "learning_rate": 4.456652007743929e-06,
      "loss": 0.5337,
      "step": 9530
    },
    {
      "epoch": 1.6535392088827203,
      "grad_norm": 0.8405235409736633,
      "learning_rate": 4.452365368762508e-06,
      "loss": 0.389,
      "step": 9531
    },
    {
      "epoch": 1.6537126995142262,
      "grad_norm": 0.6050131916999817,
      "learning_rate": 4.448080534102202e-06,
      "loss": 0.4435,
      "step": 9532
    },
    {
      "epoch": 1.653886190145732,
      "grad_norm": 0.6795126795768738,
      "learning_rate": 4.4437975042602635e-06,
      "loss": 0.5243,
      "step": 9533
    },
    {
      "epoch": 1.654059680777238,
      "grad_norm": 0.6958543062210083,
      "learning_rate": 4.439516279733764e-06,
      "loss": 0.5288,
      "step": 9534
    },
    {
      "epoch": 1.654233171408744,
      "grad_norm": 0.7578051090240479,
      "learning_rate": 4.435236861019521e-06,
      "loss": 0.4407,
      "step": 9535
    },
    {
      "epoch": 1.6544066620402498,
      "grad_norm": 0.9487757086753845,
      "learning_rate": 4.430959248614184e-06,
      "loss": 0.53,
      "step": 9536
    },
    {
      "epoch": 1.6545801526717558,
      "grad_norm": 0.9098021388053894,
      "learning_rate": 4.4266834430141654e-06,
      "loss": 0.3459,
      "step": 9537
    },
    {
      "epoch": 1.6547536433032617,
      "grad_norm": 0.800282895565033,
      "learning_rate": 4.42240944471568e-06,
      "loss": 0.408,
      "step": 9538
    },
    {
      "epoch": 1.6549271339347675,
      "grad_norm": 0.8329261541366577,
      "learning_rate": 4.418137254214725e-06,
      "loss": 0.3662,
      "step": 9539
    },
    {
      "epoch": 1.6551006245662734,
      "grad_norm": 0.733981192111969,
      "learning_rate": 4.413866872007104e-06,
      "loss": 0.4084,
      "step": 9540
    },
    {
      "epoch": 1.6552741151977792,
      "grad_norm": 0.7053385376930237,
      "learning_rate": 4.409598298588394e-06,
      "loss": 0.4844,
      "step": 9541
    },
    {
      "epoch": 1.655447605829285,
      "grad_norm": 1.4506841897964478,
      "learning_rate": 4.405331534453967e-06,
      "loss": 0.5758,
      "step": 9542
    },
    {
      "epoch": 1.6556210964607911,
      "grad_norm": 0.8209178447723389,
      "learning_rate": 4.4010665800989804e-06,
      "loss": 0.5397,
      "step": 9543
    },
    {
      "epoch": 1.655794587092297,
      "grad_norm": 0.6431795358657837,
      "learning_rate": 4.396803436018406e-06,
      "loss": 0.4479,
      "step": 9544
    },
    {
      "epoch": 1.655968077723803,
      "grad_norm": 0.8899828791618347,
      "learning_rate": 4.3925421027069645e-06,
      "loss": 0.4127,
      "step": 9545
    },
    {
      "epoch": 1.656141568355309,
      "grad_norm": 0.8157262802124023,
      "learning_rate": 4.3882825806592024e-06,
      "loss": 0.5422,
      "step": 9546
    },
    {
      "epoch": 1.6563150589868147,
      "grad_norm": 0.9176727533340454,
      "learning_rate": 4.3840248703694365e-06,
      "loss": 0.3475,
      "step": 9547
    },
    {
      "epoch": 1.6564885496183206,
      "grad_norm": 0.991773247718811,
      "learning_rate": 4.379768972331784e-06,
      "loss": 0.3473,
      "step": 9548
    },
    {
      "epoch": 1.6566620402498264,
      "grad_norm": 0.974721372127533,
      "learning_rate": 4.375514887040135e-06,
      "loss": 0.3997,
      "step": 9549
    },
    {
      "epoch": 1.6568355308813323,
      "grad_norm": 0.856035053730011,
      "learning_rate": 4.371262614988196e-06,
      "loss": 0.4032,
      "step": 9550
    },
    {
      "epoch": 1.6570090215128384,
      "grad_norm": 0.7112841010093689,
      "learning_rate": 4.367012156669441e-06,
      "loss": 0.5157,
      "step": 9551
    },
    {
      "epoch": 1.6571825121443442,
      "grad_norm": 0.8579465746879578,
      "learning_rate": 4.362763512577144e-06,
      "loss": 0.4253,
      "step": 9552
    },
    {
      "epoch": 1.6573560027758503,
      "grad_norm": 0.7902611494064331,
      "learning_rate": 4.358516683204355e-06,
      "loss": 0.534,
      "step": 9553
    },
    {
      "epoch": 1.6575294934073561,
      "grad_norm": 0.7621585726737976,
      "learning_rate": 4.354271669043934e-06,
      "loss": 0.4409,
      "step": 9554
    },
    {
      "epoch": 1.657702984038862,
      "grad_norm": 0.9942328333854675,
      "learning_rate": 4.350028470588521e-06,
      "loss": 0.3877,
      "step": 9555
    },
    {
      "epoch": 1.6578764746703678,
      "grad_norm": 0.6779175996780396,
      "learning_rate": 4.345787088330537e-06,
      "loss": 0.5796,
      "step": 9556
    },
    {
      "epoch": 1.6580499653018737,
      "grad_norm": 1.2362260818481445,
      "learning_rate": 4.341547522762202e-06,
      "loss": 0.5586,
      "step": 9557
    },
    {
      "epoch": 1.6582234559333795,
      "grad_norm": 0.7966768145561218,
      "learning_rate": 4.33730977437552e-06,
      "loss": 0.5646,
      "step": 9558
    },
    {
      "epoch": 1.6583969465648853,
      "grad_norm": 0.6846951842308044,
      "learning_rate": 4.333073843662292e-06,
      "loss": 0.3647,
      "step": 9559
    },
    {
      "epoch": 1.6585704371963914,
      "grad_norm": 0.6259403824806213,
      "learning_rate": 4.328839731114101e-06,
      "loss": 0.4735,
      "step": 9560
    },
    {
      "epoch": 1.6587439278278973,
      "grad_norm": 0.6614114046096802,
      "learning_rate": 4.324607437222319e-06,
      "loss": 0.4196,
      "step": 9561
    },
    {
      "epoch": 1.6589174184594033,
      "grad_norm": 0.6352244019508362,
      "learning_rate": 4.3203769624781055e-06,
      "loss": 0.3795,
      "step": 9562
    },
    {
      "epoch": 1.6590909090909092,
      "grad_norm": 1.7433141469955444,
      "learning_rate": 4.316148307372425e-06,
      "loss": 0.4055,
      "step": 9563
    },
    {
      "epoch": 1.659264399722415,
      "grad_norm": 1.985318899154663,
      "learning_rate": 4.311921472395999e-06,
      "loss": 0.3379,
      "step": 9564
    },
    {
      "epoch": 1.6594378903539209,
      "grad_norm": 0.8454378843307495,
      "learning_rate": 4.307696458039372e-06,
      "loss": 0.3607,
      "step": 9565
    },
    {
      "epoch": 1.6596113809854267,
      "grad_norm": 0.9035350680351257,
      "learning_rate": 4.303473264792857e-06,
      "loss": 0.5366,
      "step": 9566
    },
    {
      "epoch": 1.6597848716169326,
      "grad_norm": 0.7647597193717957,
      "learning_rate": 4.2992518931465566e-06,
      "loss": 0.3423,
      "step": 9567
    },
    {
      "epoch": 1.6599583622484386,
      "grad_norm": 0.8340458273887634,
      "learning_rate": 4.295032343590366e-06,
      "loss": 0.2598,
      "step": 9568
    },
    {
      "epoch": 1.6601318528799445,
      "grad_norm": 0.92600417137146,
      "learning_rate": 4.290814616613976e-06,
      "loss": 0.4066,
      "step": 9569
    },
    {
      "epoch": 1.6603053435114505,
      "grad_norm": 1.1044412851333618,
      "learning_rate": 4.286598712706858e-06,
      "loss": 0.3838,
      "step": 9570
    },
    {
      "epoch": 1.6604788341429564,
      "grad_norm": 0.8176655769348145,
      "learning_rate": 4.282384632358265e-06,
      "loss": 0.3781,
      "step": 9571
    },
    {
      "epoch": 1.6606523247744622,
      "grad_norm": 0.7669593691825867,
      "learning_rate": 4.278172376057246e-06,
      "loss": 0.4214,
      "step": 9572
    },
    {
      "epoch": 1.660825815405968,
      "grad_norm": 0.7503926753997803,
      "learning_rate": 4.2739619442926525e-06,
      "loss": 0.4418,
      "step": 9573
    },
    {
      "epoch": 1.660999306037474,
      "grad_norm": 1.3540635108947754,
      "learning_rate": 4.269753337553091e-06,
      "loss": 0.4642,
      "step": 9574
    },
    {
      "epoch": 1.6611727966689798,
      "grad_norm": 0.8061459064483643,
      "learning_rate": 4.265546556326989e-06,
      "loss": 0.3364,
      "step": 9575
    },
    {
      "epoch": 1.6613462873004856,
      "grad_norm": 0.8524975180625916,
      "learning_rate": 4.2613416011025424e-06,
      "loss": 0.4283,
      "step": 9576
    },
    {
      "epoch": 1.6615197779319917,
      "grad_norm": 0.636336624622345,
      "learning_rate": 4.257138472367737e-06,
      "loss": 0.5293,
      "step": 9577
    },
    {
      "epoch": 1.6616932685634975,
      "grad_norm": 0.6362683773040771,
      "learning_rate": 4.25293717061036e-06,
      "loss": 0.6263,
      "step": 9578
    },
    {
      "epoch": 1.6618667591950036,
      "grad_norm": 0.8132637739181519,
      "learning_rate": 4.248737696317975e-06,
      "loss": 0.3624,
      "step": 9579
    },
    {
      "epoch": 1.6620402498265094,
      "grad_norm": 0.7631382346153259,
      "learning_rate": 4.244540049977934e-06,
      "loss": 0.4004,
      "step": 9580
    },
    {
      "epoch": 1.6622137404580153,
      "grad_norm": 0.7176964282989502,
      "learning_rate": 4.240344232077373e-06,
      "loss": 0.5193,
      "step": 9581
    },
    {
      "epoch": 1.6623872310895211,
      "grad_norm": 0.6607311964035034,
      "learning_rate": 4.236150243103234e-06,
      "loss": 0.5862,
      "step": 9582
    },
    {
      "epoch": 1.662560721721027,
      "grad_norm": 0.7352809309959412,
      "learning_rate": 4.231958083542229e-06,
      "loss": 0.3513,
      "step": 9583
    },
    {
      "epoch": 1.6627342123525328,
      "grad_norm": 1.154639720916748,
      "learning_rate": 4.227767753880861e-06,
      "loss": 0.3141,
      "step": 9584
    },
    {
      "epoch": 1.662907702984039,
      "grad_norm": 0.6323665380477905,
      "learning_rate": 4.223579254605414e-06,
      "loss": 0.5624,
      "step": 9585
    },
    {
      "epoch": 1.6630811936155447,
      "grad_norm": 0.7658765912055969,
      "learning_rate": 4.2193925862019934e-06,
      "loss": 0.4747,
      "step": 9586
    },
    {
      "epoch": 1.6632546842470508,
      "grad_norm": 0.5467133522033691,
      "learning_rate": 4.2152077491564385e-06,
      "loss": 0.5022,
      "step": 9587
    },
    {
      "epoch": 1.6634281748785567,
      "grad_norm": 0.7082794904708862,
      "learning_rate": 4.211024743954424e-06,
      "loss": 0.5679,
      "step": 9588
    },
    {
      "epoch": 1.6636016655100625,
      "grad_norm": 0.6206352114677429,
      "learning_rate": 4.206843571081383e-06,
      "loss": 0.605,
      "step": 9589
    },
    {
      "epoch": 1.6637751561415683,
      "grad_norm": 0.8550276756286621,
      "learning_rate": 4.2026642310225505e-06,
      "loss": 0.5267,
      "step": 9590
    },
    {
      "epoch": 1.6639486467730742,
      "grad_norm": 0.8094297647476196,
      "learning_rate": 4.198486724262935e-06,
      "loss": 0.5392,
      "step": 9591
    },
    {
      "epoch": 1.66412213740458,
      "grad_norm": 0.7623730897903442,
      "learning_rate": 4.194311051287359e-06,
      "loss": 0.5349,
      "step": 9592
    },
    {
      "epoch": 1.664295628036086,
      "grad_norm": 0.6538176536560059,
      "learning_rate": 4.190137212580392e-06,
      "loss": 0.4421,
      "step": 9593
    },
    {
      "epoch": 1.664469118667592,
      "grad_norm": 0.8178464770317078,
      "learning_rate": 4.185965208626428e-06,
      "loss": 0.3848,
      "step": 9594
    },
    {
      "epoch": 1.6646426092990978,
      "grad_norm": 0.7703167796134949,
      "learning_rate": 4.181795039909631e-06,
      "loss": 0.4844,
      "step": 9595
    },
    {
      "epoch": 1.6648160999306039,
      "grad_norm": 0.9850491285324097,
      "learning_rate": 4.177626706913948e-06,
      "loss": 0.3661,
      "step": 9596
    },
    {
      "epoch": 1.6649895905621097,
      "grad_norm": 0.7138931751251221,
      "learning_rate": 4.173460210123119e-06,
      "loss": 0.5471,
      "step": 9597
    },
    {
      "epoch": 1.6651630811936156,
      "grad_norm": 1.267538070678711,
      "learning_rate": 4.16929555002068e-06,
      "loss": 0.4827,
      "step": 9598
    },
    {
      "epoch": 1.6653365718251214,
      "grad_norm": 0.7268630862236023,
      "learning_rate": 4.16513272708994e-06,
      "loss": 0.5835,
      "step": 9599
    },
    {
      "epoch": 1.6655100624566272,
      "grad_norm": 0.7588167786598206,
      "learning_rate": 4.160971741813995e-06,
      "loss": 0.3947,
      "step": 9600
    },
    {
      "epoch": 1.665683553088133,
      "grad_norm": 0.8451988101005554,
      "learning_rate": 4.156812594675732e-06,
      "loss": 0.5444,
      "step": 9601
    },
    {
      "epoch": 1.6658570437196392,
      "grad_norm": 0.7650555968284607,
      "learning_rate": 4.152655286157834e-06,
      "loss": 0.515,
      "step": 9602
    },
    {
      "epoch": 1.666030534351145,
      "grad_norm": 0.8789478540420532,
      "learning_rate": 4.148499816742755e-06,
      "loss": 0.5022,
      "step": 9603
    },
    {
      "epoch": 1.666204024982651,
      "grad_norm": 0.8302474021911621,
      "learning_rate": 4.144346186912738e-06,
      "loss": 0.459,
      "step": 9604
    },
    {
      "epoch": 1.666377515614157,
      "grad_norm": 0.7248766422271729,
      "learning_rate": 4.140194397149833e-06,
      "loss": 0.438,
      "step": 9605
    },
    {
      "epoch": 1.6665510062456628,
      "grad_norm": 0.8582567572593689,
      "learning_rate": 4.136044447935837e-06,
      "loss": 0.4225,
      "step": 9606
    },
    {
      "epoch": 1.6667244968771686,
      "grad_norm": 0.750859260559082,
      "learning_rate": 4.1318963397523725e-06,
      "loss": 0.4366,
      "step": 9607
    },
    {
      "epoch": 1.6668979875086745,
      "grad_norm": 0.7951454520225525,
      "learning_rate": 4.127750073080829e-06,
      "loss": 0.3839,
      "step": 9608
    },
    {
      "epoch": 1.6670714781401803,
      "grad_norm": 0.9483717679977417,
      "learning_rate": 4.123605648402385e-06,
      "loss": 0.3567,
      "step": 9609
    },
    {
      "epoch": 1.6672449687716864,
      "grad_norm": 0.8451215624809265,
      "learning_rate": 4.119463066197997e-06,
      "loss": 0.3776,
      "step": 9610
    },
    {
      "epoch": 1.6674184594031922,
      "grad_norm": 2.248885154724121,
      "learning_rate": 4.115322326948432e-06,
      "loss": 0.3769,
      "step": 9611
    },
    {
      "epoch": 1.6675919500346983,
      "grad_norm": 1.1927193403244019,
      "learning_rate": 4.111183431134223e-06,
      "loss": 0.401,
      "step": 9612
    },
    {
      "epoch": 1.6677654406662041,
      "grad_norm": 0.731179416179657,
      "learning_rate": 4.1070463792356865e-06,
      "loss": 0.511,
      "step": 9613
    },
    {
      "epoch": 1.66793893129771,
      "grad_norm": 1.3155031204223633,
      "learning_rate": 4.102911171732933e-06,
      "loss": 0.3981,
      "step": 9614
    },
    {
      "epoch": 1.6681124219292158,
      "grad_norm": 1.7868298292160034,
      "learning_rate": 4.098777809105871e-06,
      "loss": 0.4999,
      "step": 9615
    },
    {
      "epoch": 1.6682859125607217,
      "grad_norm": 0.7645481824874878,
      "learning_rate": 4.094646291834166e-06,
      "loss": 0.4681,
      "step": 9616
    },
    {
      "epoch": 1.6684594031922275,
      "grad_norm": 0.8161604404449463,
      "learning_rate": 4.090516620397294e-06,
      "loss": 0.465,
      "step": 9617
    },
    {
      "epoch": 1.6686328938237334,
      "grad_norm": 0.7998003959655762,
      "learning_rate": 4.086388795274508e-06,
      "loss": 0.3854,
      "step": 9618
    },
    {
      "epoch": 1.6688063844552394,
      "grad_norm": 0.6143330335617065,
      "learning_rate": 4.082262816944845e-06,
      "loss": 0.5396,
      "step": 9619
    },
    {
      "epoch": 1.6689798750867453,
      "grad_norm": 0.7949010729789734,
      "learning_rate": 4.078138685887125e-06,
      "loss": 0.4173,
      "step": 9620
    },
    {
      "epoch": 1.6691533657182513,
      "grad_norm": 1.2927134037017822,
      "learning_rate": 4.074016402579968e-06,
      "loss": 0.3998,
      "step": 9621
    },
    {
      "epoch": 1.6693268563497572,
      "grad_norm": 0.8502987027168274,
      "learning_rate": 4.069895967501765e-06,
      "loss": 0.3896,
      "step": 9622
    },
    {
      "epoch": 1.669500346981263,
      "grad_norm": 0.5281307101249695,
      "learning_rate": 4.065777381130698e-06,
      "loss": 0.5065,
      "step": 9623
    },
    {
      "epoch": 1.6696738376127689,
      "grad_norm": 0.7240021824836731,
      "learning_rate": 4.0616606439447315e-06,
      "loss": 0.5621,
      "step": 9624
    },
    {
      "epoch": 1.6698473282442747,
      "grad_norm": 0.9518048167228699,
      "learning_rate": 4.057545756421615e-06,
      "loss": 0.3801,
      "step": 9625
    },
    {
      "epoch": 1.6700208188757806,
      "grad_norm": 1.1966935396194458,
      "learning_rate": 4.053432719038895e-06,
      "loss": 0.4609,
      "step": 9626
    },
    {
      "epoch": 1.6701943095072866,
      "grad_norm": 0.8296990394592285,
      "learning_rate": 4.049321532273889e-06,
      "loss": 0.3398,
      "step": 9627
    },
    {
      "epoch": 1.6703678001387925,
      "grad_norm": 1.3707387447357178,
      "learning_rate": 4.045212196603705e-06,
      "loss": 0.4018,
      "step": 9628
    },
    {
      "epoch": 1.6705412907702986,
      "grad_norm": 0.9826870560646057,
      "learning_rate": 4.041104712505233e-06,
      "loss": 0.4464,
      "step": 9629
    },
    {
      "epoch": 1.6707147814018044,
      "grad_norm": 0.7787246108055115,
      "learning_rate": 4.03699908045516e-06,
      "loss": 0.4161,
      "step": 9630
    },
    {
      "epoch": 1.6708882720333103,
      "grad_norm": 0.9037700295448303,
      "learning_rate": 4.0328953009299425e-06,
      "loss": 0.5441,
      "step": 9631
    },
    {
      "epoch": 1.671061762664816,
      "grad_norm": 0.8366997241973877,
      "learning_rate": 4.028793374405833e-06,
      "loss": 0.4343,
      "step": 9632
    },
    {
      "epoch": 1.671235253296322,
      "grad_norm": 1.0335955619812012,
      "learning_rate": 4.024693301358855e-06,
      "loss": 0.4542,
      "step": 9633
    },
    {
      "epoch": 1.6714087439278278,
      "grad_norm": 0.6714066863059998,
      "learning_rate": 4.020595082264847e-06,
      "loss": 0.5325,
      "step": 9634
    },
    {
      "epoch": 1.6715822345593336,
      "grad_norm": 1.3859325647354126,
      "learning_rate": 4.016498717599387e-06,
      "loss": 0.4736,
      "step": 9635
    },
    {
      "epoch": 1.6717557251908397,
      "grad_norm": 0.7687572836875916,
      "learning_rate": 4.012404207837881e-06,
      "loss": 0.32,
      "step": 9636
    },
    {
      "epoch": 1.6719292158223455,
      "grad_norm": 0.8668257594108582,
      "learning_rate": 4.008311553455497e-06,
      "loss": 0.457,
      "step": 9637
    },
    {
      "epoch": 1.6721027064538516,
      "grad_norm": 0.7919766902923584,
      "learning_rate": 4.0042207549271905e-06,
      "loss": 0.4264,
      "step": 9638
    },
    {
      "epoch": 1.6722761970853575,
      "grad_norm": 0.951241672039032,
      "learning_rate": 4.0001318127276985e-06,
      "loss": 0.4434,
      "step": 9639
    },
    {
      "epoch": 1.6724496877168633,
      "grad_norm": 0.6608285903930664,
      "learning_rate": 3.996044727331558e-06,
      "loss": 0.472,
      "step": 9640
    },
    {
      "epoch": 1.6726231783483692,
      "grad_norm": 0.6451571583747864,
      "learning_rate": 3.991959499213076e-06,
      "loss": 0.5497,
      "step": 9641
    },
    {
      "epoch": 1.672796668979875,
      "grad_norm": 0.7177254557609558,
      "learning_rate": 3.987876128846349e-06,
      "loss": 0.5411,
      "step": 9642
    },
    {
      "epoch": 1.6729701596113808,
      "grad_norm": 1.3236669301986694,
      "learning_rate": 3.983794616705248e-06,
      "loss": 0.4778,
      "step": 9643
    },
    {
      "epoch": 1.673143650242887,
      "grad_norm": 0.9966208338737488,
      "learning_rate": 3.979714963263455e-06,
      "loss": 0.3341,
      "step": 9644
    },
    {
      "epoch": 1.6733171408743928,
      "grad_norm": 0.7492838501930237,
      "learning_rate": 3.975637168994397e-06,
      "loss": 0.4531,
      "step": 9645
    },
    {
      "epoch": 1.6734906315058988,
      "grad_norm": 0.9830329418182373,
      "learning_rate": 3.971561234371324e-06,
      "loss": 0.5131,
      "step": 9646
    },
    {
      "epoch": 1.6736641221374047,
      "grad_norm": 0.7363716959953308,
      "learning_rate": 3.967487159867245e-06,
      "loss": 0.4955,
      "step": 9647
    },
    {
      "epoch": 1.6738376127689105,
      "grad_norm": 0.6564109921455383,
      "learning_rate": 3.963414945954962e-06,
      "loss": 0.4585,
      "step": 9648
    },
    {
      "epoch": 1.6740111034004164,
      "grad_norm": 0.6819577813148499,
      "learning_rate": 3.959344593107057e-06,
      "loss": 0.5862,
      "step": 9649
    },
    {
      "epoch": 1.6741845940319222,
      "grad_norm": 0.705721914768219,
      "learning_rate": 3.955276101795908e-06,
      "loss": 0.4185,
      "step": 9650
    },
    {
      "epoch": 1.674358084663428,
      "grad_norm": 0.8432149291038513,
      "learning_rate": 3.951209472493664e-06,
      "loss": 0.4398,
      "step": 9651
    },
    {
      "epoch": 1.6745315752949341,
      "grad_norm": 1.7393169403076172,
      "learning_rate": 3.947144705672257e-06,
      "loss": 0.4418,
      "step": 9652
    },
    {
      "epoch": 1.67470506592644,
      "grad_norm": 1.0593560934066772,
      "learning_rate": 3.943081801803421e-06,
      "loss": 0.3892,
      "step": 9653
    },
    {
      "epoch": 1.6748785565579458,
      "grad_norm": 1.1183923482894897,
      "learning_rate": 3.939020761358641e-06,
      "loss": 0.5154,
      "step": 9654
    },
    {
      "epoch": 1.6750520471894519,
      "grad_norm": 0.9049838781356812,
      "learning_rate": 3.934961584809222e-06,
      "loss": 0.3672,
      "step": 9655
    },
    {
      "epoch": 1.6752255378209577,
      "grad_norm": 0.9599663615226746,
      "learning_rate": 3.930904272626226e-06,
      "loss": 0.5461,
      "step": 9656
    },
    {
      "epoch": 1.6753990284524636,
      "grad_norm": 0.94817054271698,
      "learning_rate": 3.926848825280524e-06,
      "loss": 0.4105,
      "step": 9657
    },
    {
      "epoch": 1.6755725190839694,
      "grad_norm": 1.129248857498169,
      "learning_rate": 3.922795243242734e-06,
      "loss": 0.348,
      "step": 9658
    },
    {
      "epoch": 1.6757460097154753,
      "grad_norm": 0.8643382787704468,
      "learning_rate": 3.918743526983295e-06,
      "loss": 0.3235,
      "step": 9659
    },
    {
      "epoch": 1.6759195003469811,
      "grad_norm": 0.9845125675201416,
      "learning_rate": 3.914693676972408e-06,
      "loss": 0.4991,
      "step": 9660
    },
    {
      "epoch": 1.6760929909784872,
      "grad_norm": 0.7317501306533813,
      "learning_rate": 3.9106456936800615e-06,
      "loss": 0.5369,
      "step": 9661
    },
    {
      "epoch": 1.676266481609993,
      "grad_norm": 1.0669113397598267,
      "learning_rate": 3.906599577576027e-06,
      "loss": 0.5198,
      "step": 9662
    },
    {
      "epoch": 1.676439972241499,
      "grad_norm": 0.9549260139465332,
      "learning_rate": 3.902555329129874e-06,
      "loss": 0.4857,
      "step": 9663
    },
    {
      "epoch": 1.676613462873005,
      "grad_norm": 0.8042954802513123,
      "learning_rate": 3.898512948810922e-06,
      "loss": 0.4109,
      "step": 9664
    },
    {
      "epoch": 1.6767869535045108,
      "grad_norm": 0.9220744967460632,
      "learning_rate": 3.894472437088308e-06,
      "loss": 0.353,
      "step": 9665
    },
    {
      "epoch": 1.6769604441360166,
      "grad_norm": 0.864961564540863,
      "learning_rate": 3.890433794430934e-06,
      "loss": 0.3972,
      "step": 9666
    },
    {
      "epoch": 1.6771339347675225,
      "grad_norm": 0.926474928855896,
      "learning_rate": 3.886397021307493e-06,
      "loss": 0.4734,
      "step": 9667
    },
    {
      "epoch": 1.6773074253990283,
      "grad_norm": 0.7888760566711426,
      "learning_rate": 3.882362118186445e-06,
      "loss": 0.4186,
      "step": 9668
    },
    {
      "epoch": 1.6774809160305344,
      "grad_norm": 1.256698489189148,
      "learning_rate": 3.878329085536061e-06,
      "loss": 0.3407,
      "step": 9669
    },
    {
      "epoch": 1.6776544066620402,
      "grad_norm": 1.0274312496185303,
      "learning_rate": 3.87429792382437e-06,
      "loss": 0.4354,
      "step": 9670
    },
    {
      "epoch": 1.6778278972935463,
      "grad_norm": 0.7899155020713806,
      "learning_rate": 3.870268633519198e-06,
      "loss": 0.506,
      "step": 9671
    },
    {
      "epoch": 1.6780013879250522,
      "grad_norm": 0.8052603006362915,
      "learning_rate": 3.86624121508814e-06,
      "loss": 0.3835,
      "step": 9672
    },
    {
      "epoch": 1.678174878556558,
      "grad_norm": 1.2355273962020874,
      "learning_rate": 3.862215668998592e-06,
      "loss": 0.2889,
      "step": 9673
    },
    {
      "epoch": 1.6783483691880638,
      "grad_norm": 1.290600061416626,
      "learning_rate": 3.858191995717722e-06,
      "loss": 0.407,
      "step": 9674
    },
    {
      "epoch": 1.6785218598195697,
      "grad_norm": 0.8202452659606934,
      "learning_rate": 3.854170195712479e-06,
      "loss": 0.3003,
      "step": 9675
    },
    {
      "epoch": 1.6786953504510755,
      "grad_norm": 0.9715308547019958,
      "learning_rate": 3.850150269449597e-06,
      "loss": 0.3814,
      "step": 9676
    },
    {
      "epoch": 1.6788688410825814,
      "grad_norm": 0.7318767309188843,
      "learning_rate": 3.846132217395593e-06,
      "loss": 0.4712,
      "step": 9677
    },
    {
      "epoch": 1.6790423317140875,
      "grad_norm": 0.7196683287620544,
      "learning_rate": 3.84211604001677e-06,
      "loss": 0.5443,
      "step": 9678
    },
    {
      "epoch": 1.6792158223455933,
      "grad_norm": 0.7301241159439087,
      "learning_rate": 3.83810173777921e-06,
      "loss": 0.4612,
      "step": 9679
    },
    {
      "epoch": 1.6793893129770994,
      "grad_norm": 0.8351182341575623,
      "learning_rate": 3.834089311148774e-06,
      "loss": 0.3381,
      "step": 9680
    },
    {
      "epoch": 1.6795628036086052,
      "grad_norm": 0.6372829079627991,
      "learning_rate": 3.830078760591107e-06,
      "loss": 0.4675,
      "step": 9681
    },
    {
      "epoch": 1.679736294240111,
      "grad_norm": 0.8103852868080139,
      "learning_rate": 3.826070086571651e-06,
      "loss": 0.3896,
      "step": 9682
    },
    {
      "epoch": 1.679909784871617,
      "grad_norm": 0.9043280482292175,
      "learning_rate": 3.822063289555597e-06,
      "loss": 0.4155,
      "step": 9683
    },
    {
      "epoch": 1.6800832755031228,
      "grad_norm": 0.7413236498832703,
      "learning_rate": 3.818058370007956e-06,
      "loss": 0.4149,
      "step": 9684
    },
    {
      "epoch": 1.6802567661346286,
      "grad_norm": 1.0025862455368042,
      "learning_rate": 3.814055328393491e-06,
      "loss": 0.4901,
      "step": 9685
    },
    {
      "epoch": 1.6804302567661347,
      "grad_norm": 0.679567813873291,
      "learning_rate": 3.810054165176775e-06,
      "loss": 0.5236,
      "step": 9686
    },
    {
      "epoch": 1.6806037473976405,
      "grad_norm": 0.6744206547737122,
      "learning_rate": 3.8060548808221277e-06,
      "loss": 0.4021,
      "step": 9687
    },
    {
      "epoch": 1.6807772380291466,
      "grad_norm": 0.9984070658683777,
      "learning_rate": 3.802057475793688e-06,
      "loss": 0.494,
      "step": 9688
    },
    {
      "epoch": 1.6809507286606524,
      "grad_norm": 0.649757444858551,
      "learning_rate": 3.7980619505553516e-06,
      "loss": 0.4397,
      "step": 9689
    },
    {
      "epoch": 1.6811242192921583,
      "grad_norm": 0.7462676167488098,
      "learning_rate": 3.794068305570804e-06,
      "loss": 0.4412,
      "step": 9690
    },
    {
      "epoch": 1.6812977099236641,
      "grad_norm": 0.6842268109321594,
      "learning_rate": 3.790076541303509e-06,
      "loss": 0.53,
      "step": 9691
    },
    {
      "epoch": 1.68147120055517,
      "grad_norm": 0.9055030345916748,
      "learning_rate": 3.78608665821673e-06,
      "loss": 0.4002,
      "step": 9692
    },
    {
      "epoch": 1.6816446911866758,
      "grad_norm": 0.6715995669364929,
      "learning_rate": 3.7820986567734787e-06,
      "loss": 0.4067,
      "step": 9693
    },
    {
      "epoch": 1.6818181818181817,
      "grad_norm": 0.6809877157211304,
      "learning_rate": 3.778112537436578e-06,
      "loss": 0.3868,
      "step": 9694
    },
    {
      "epoch": 1.6819916724496877,
      "grad_norm": 0.6496911644935608,
      "learning_rate": 3.7741283006686204e-06,
      "loss": 0.4899,
      "step": 9695
    },
    {
      "epoch": 1.6821651630811936,
      "grad_norm": 0.783438503742218,
      "learning_rate": 3.7701459469319824e-06,
      "loss": 0.5194,
      "step": 9696
    },
    {
      "epoch": 1.6823386537126996,
      "grad_norm": 1.6692430973052979,
      "learning_rate": 3.7661654766888124e-06,
      "loss": 0.4019,
      "step": 9697
    },
    {
      "epoch": 1.6825121443442055,
      "grad_norm": 0.7749941349029541,
      "learning_rate": 3.7621868904010585e-06,
      "loss": 0.5715,
      "step": 9698
    },
    {
      "epoch": 1.6826856349757113,
      "grad_norm": 0.7720735669136047,
      "learning_rate": 3.758210188530438e-06,
      "loss": 0.4672,
      "step": 9699
    },
    {
      "epoch": 1.6828591256072172,
      "grad_norm": 0.7236547470092773,
      "learning_rate": 3.7542353715384462e-06,
      "loss": 0.4265,
      "step": 9700
    },
    {
      "epoch": 1.683032616238723,
      "grad_norm": 0.9045554995536804,
      "learning_rate": 3.750262439886374e-06,
      "loss": 0.6182,
      "step": 9701
    },
    {
      "epoch": 1.6832061068702289,
      "grad_norm": 0.8127009272575378,
      "learning_rate": 3.7462913940352797e-06,
      "loss": 0.3979,
      "step": 9702
    },
    {
      "epoch": 1.683379597501735,
      "grad_norm": 1.0219383239746094,
      "learning_rate": 3.7423222344460096e-06,
      "loss": 0.4023,
      "step": 9703
    },
    {
      "epoch": 1.6835530881332408,
      "grad_norm": 0.6961208581924438,
      "learning_rate": 3.7383549615791826e-06,
      "loss": 0.489,
      "step": 9704
    },
    {
      "epoch": 1.6837265787647469,
      "grad_norm": 0.7545673251152039,
      "learning_rate": 3.734389575895221e-06,
      "loss": 0.5376,
      "step": 9705
    },
    {
      "epoch": 1.6839000693962527,
      "grad_norm": 0.666425347328186,
      "learning_rate": 3.7304260778542924e-06,
      "loss": 0.5713,
      "step": 9706
    },
    {
      "epoch": 1.6840735600277585,
      "grad_norm": 0.7239332795143127,
      "learning_rate": 3.726464467916382e-06,
      "loss": 0.4697,
      "step": 9707
    },
    {
      "epoch": 1.6842470506592644,
      "grad_norm": 0.8578612208366394,
      "learning_rate": 3.722504746541229e-06,
      "loss": 0.4288,
      "step": 9708
    },
    {
      "epoch": 1.6844205412907702,
      "grad_norm": 0.9409584403038025,
      "learning_rate": 3.71854691418837e-06,
      "loss": 0.428,
      "step": 9709
    },
    {
      "epoch": 1.684594031922276,
      "grad_norm": 0.6740388870239258,
      "learning_rate": 3.714590971317107e-06,
      "loss": 0.4722,
      "step": 9710
    },
    {
      "epoch": 1.6847675225537821,
      "grad_norm": 0.7901949286460876,
      "learning_rate": 3.710636918386543e-06,
      "loss": 0.4281,
      "step": 9711
    },
    {
      "epoch": 1.684941013185288,
      "grad_norm": 0.7378773093223572,
      "learning_rate": 3.706684755855545e-06,
      "loss": 0.4368,
      "step": 9712
    },
    {
      "epoch": 1.6851145038167938,
      "grad_norm": 0.9257503747940063,
      "learning_rate": 3.7027344841827684e-06,
      "loss": 0.3464,
      "step": 9713
    },
    {
      "epoch": 1.6852879944483,
      "grad_norm": 0.755611777305603,
      "learning_rate": 3.698786103826639e-06,
      "loss": 0.4982,
      "step": 9714
    },
    {
      "epoch": 1.6854614850798058,
      "grad_norm": 0.6503063440322876,
      "learning_rate": 3.694839615245387e-06,
      "loss": 0.5034,
      "step": 9715
    },
    {
      "epoch": 1.6856349757113116,
      "grad_norm": 0.7667842507362366,
      "learning_rate": 3.690895018896987e-06,
      "loss": 0.5358,
      "step": 9716
    },
    {
      "epoch": 1.6858084663428174,
      "grad_norm": 0.6968609094619751,
      "learning_rate": 3.6869523152392296e-06,
      "loss": 0.5353,
      "step": 9717
    },
    {
      "epoch": 1.6859819569743233,
      "grad_norm": 0.944675624370575,
      "learning_rate": 3.6830115047296633e-06,
      "loss": 0.413,
      "step": 9718
    },
    {
      "epoch": 1.6861554476058291,
      "grad_norm": 0.8377568125724792,
      "learning_rate": 3.679072587825625e-06,
      "loss": 0.5443,
      "step": 9719
    },
    {
      "epoch": 1.6863289382373352,
      "grad_norm": 0.794916033744812,
      "learning_rate": 3.675135564984227e-06,
      "loss": 0.3719,
      "step": 9720
    },
    {
      "epoch": 1.686502428868841,
      "grad_norm": 0.7275283336639404,
      "learning_rate": 3.671200436662372e-06,
      "loss": 0.4502,
      "step": 9721
    },
    {
      "epoch": 1.6866759195003471,
      "grad_norm": 0.7036992311477661,
      "learning_rate": 3.6672672033167333e-06,
      "loss": 0.512,
      "step": 9722
    },
    {
      "epoch": 1.686849410131853,
      "grad_norm": 0.757016122341156,
      "learning_rate": 3.6633358654037676e-06,
      "loss": 0.5358,
      "step": 9723
    },
    {
      "epoch": 1.6870229007633588,
      "grad_norm": 4.114889144897461,
      "learning_rate": 3.6594064233797123e-06,
      "loss": 0.5155,
      "step": 9724
    },
    {
      "epoch": 1.6871963913948647,
      "grad_norm": 1.4096416234970093,
      "learning_rate": 3.6554788777005758e-06,
      "loss": 0.4038,
      "step": 9725
    },
    {
      "epoch": 1.6873698820263705,
      "grad_norm": 0.6038339138031006,
      "learning_rate": 3.6515532288221646e-06,
      "loss": 0.4409,
      "step": 9726
    },
    {
      "epoch": 1.6875433726578764,
      "grad_norm": 0.6808177828788757,
      "learning_rate": 3.647629477200052e-06,
      "loss": 0.5103,
      "step": 9727
    },
    {
      "epoch": 1.6877168632893824,
      "grad_norm": 0.8416544198989868,
      "learning_rate": 3.643707623289592e-06,
      "loss": 0.4237,
      "step": 9728
    },
    {
      "epoch": 1.6878903539208883,
      "grad_norm": 0.6936036348342896,
      "learning_rate": 3.6397876675459175e-06,
      "loss": 0.3955,
      "step": 9729
    },
    {
      "epoch": 1.6880638445523943,
      "grad_norm": 0.9723979234695435,
      "learning_rate": 3.635869610423952e-06,
      "loss": 0.5072,
      "step": 9730
    },
    {
      "epoch": 1.6882373351839002,
      "grad_norm": 0.7053537368774414,
      "learning_rate": 3.6319534523783872e-06,
      "loss": 0.4186,
      "step": 9731
    },
    {
      "epoch": 1.688410825815406,
      "grad_norm": 2.3634746074676514,
      "learning_rate": 3.628039193863695e-06,
      "loss": 0.3661,
      "step": 9732
    },
    {
      "epoch": 1.6885843164469119,
      "grad_norm": 0.8253313899040222,
      "learning_rate": 3.6241268353341276e-06,
      "loss": 0.4602,
      "step": 9733
    },
    {
      "epoch": 1.6887578070784177,
      "grad_norm": 0.7930001020431519,
      "learning_rate": 3.6202163772437326e-06,
      "loss": 0.4257,
      "step": 9734
    },
    {
      "epoch": 1.6889312977099236,
      "grad_norm": 0.8238438963890076,
      "learning_rate": 3.616307820046303e-06,
      "loss": 0.4066,
      "step": 9735
    },
    {
      "epoch": 1.6891047883414294,
      "grad_norm": 0.9378172755241394,
      "learning_rate": 3.6124011641954473e-06,
      "loss": 0.5157,
      "step": 9736
    },
    {
      "epoch": 1.6892782789729355,
      "grad_norm": 0.7956205606460571,
      "learning_rate": 3.6084964101445307e-06,
      "loss": 0.3901,
      "step": 9737
    },
    {
      "epoch": 1.6894517696044413,
      "grad_norm": 0.742908239364624,
      "learning_rate": 3.6045935583467053e-06,
      "loss": 0.4529,
      "step": 9738
    },
    {
      "epoch": 1.6896252602359474,
      "grad_norm": 0.7737570405006409,
      "learning_rate": 3.6006926092548988e-06,
      "loss": 0.5082,
      "step": 9739
    },
    {
      "epoch": 1.6897987508674532,
      "grad_norm": 0.5851488709449768,
      "learning_rate": 3.5967935633218277e-06,
      "loss": 0.5751,
      "step": 9740
    },
    {
      "epoch": 1.689972241498959,
      "grad_norm": 0.9748550057411194,
      "learning_rate": 3.5928964209999784e-06,
      "loss": 0.4352,
      "step": 9741
    },
    {
      "epoch": 1.690145732130465,
      "grad_norm": 0.7415540218353271,
      "learning_rate": 3.589001182741616e-06,
      "loss": 0.3856,
      "step": 9742
    },
    {
      "epoch": 1.6903192227619708,
      "grad_norm": 0.9179946184158325,
      "learning_rate": 3.5851078489987834e-06,
      "loss": 0.4203,
      "step": 9743
    },
    {
      "epoch": 1.6904927133934766,
      "grad_norm": 0.8719099164009094,
      "learning_rate": 3.5812164202233236e-06,
      "loss": 0.4686,
      "step": 9744
    },
    {
      "epoch": 1.6906662040249827,
      "grad_norm": 0.8644099831581116,
      "learning_rate": 3.5773268968668194e-06,
      "loss": 0.5646,
      "step": 9745
    },
    {
      "epoch": 1.6908396946564885,
      "grad_norm": 1.2237269878387451,
      "learning_rate": 3.5734392793806704e-06,
      "loss": 0.3469,
      "step": 9746
    },
    {
      "epoch": 1.6910131852879946,
      "grad_norm": 0.8852053880691528,
      "learning_rate": 3.569553568216033e-06,
      "loss": 0.4629,
      "step": 9747
    },
    {
      "epoch": 1.6911866759195004,
      "grad_norm": 0.9039144515991211,
      "learning_rate": 3.5656697638238447e-06,
      "loss": 0.3802,
      "step": 9748
    },
    {
      "epoch": 1.6913601665510063,
      "grad_norm": 0.9759804010391235,
      "learning_rate": 3.5617878666548354e-06,
      "loss": 0.4327,
      "step": 9749
    },
    {
      "epoch": 1.6915336571825121,
      "grad_norm": 0.80108243227005,
      "learning_rate": 3.5579078771594988e-06,
      "loss": 0.402,
      "step": 9750
    },
    {
      "epoch": 1.691707147814018,
      "grad_norm": 0.8020294308662415,
      "learning_rate": 3.55402979578811e-06,
      "loss": 0.5035,
      "step": 9751
    },
    {
      "epoch": 1.6918806384455238,
      "grad_norm": 0.6639646887779236,
      "learning_rate": 3.550153622990724e-06,
      "loss": 0.5635,
      "step": 9752
    },
    {
      "epoch": 1.6920541290770297,
      "grad_norm": 0.697598934173584,
      "learning_rate": 3.546279359217186e-06,
      "loss": 0.4956,
      "step": 9753
    },
    {
      "epoch": 1.6922276197085357,
      "grad_norm": 0.7204028367996216,
      "learning_rate": 3.542407004917092e-06,
      "loss": 0.4203,
      "step": 9754
    },
    {
      "epoch": 1.6924011103400416,
      "grad_norm": 0.6753879189491272,
      "learning_rate": 3.538536560539847e-06,
      "loss": 0.5222,
      "step": 9755
    },
    {
      "epoch": 1.6925746009715477,
      "grad_norm": 1.0081450939178467,
      "learning_rate": 3.5346680265346113e-06,
      "loss": 0.4051,
      "step": 9756
    },
    {
      "epoch": 1.6927480916030535,
      "grad_norm": 0.9310484528541565,
      "learning_rate": 3.530801403350346e-06,
      "loss": 0.4008,
      "step": 9757
    },
    {
      "epoch": 1.6929215822345594,
      "grad_norm": 1.2470906972885132,
      "learning_rate": 3.5269366914357585e-06,
      "loss": 0.5457,
      "step": 9758
    },
    {
      "epoch": 1.6930950728660652,
      "grad_norm": 1.6500725746154785,
      "learning_rate": 3.523073891239368e-06,
      "loss": 0.6094,
      "step": 9759
    },
    {
      "epoch": 1.693268563497571,
      "grad_norm": 0.8279609084129333,
      "learning_rate": 3.5192130032094517e-06,
      "loss": 0.5032,
      "step": 9760
    },
    {
      "epoch": 1.693442054129077,
      "grad_norm": 0.6731119751930237,
      "learning_rate": 3.515354027794069e-06,
      "loss": 0.5054,
      "step": 9761
    },
    {
      "epoch": 1.693615544760583,
      "grad_norm": 1.3643908500671387,
      "learning_rate": 3.511496965441057e-06,
      "loss": 0.5199,
      "step": 9762
    },
    {
      "epoch": 1.6937890353920888,
      "grad_norm": 0.8825380802154541,
      "learning_rate": 3.507641816598044e-06,
      "loss": 0.4775,
      "step": 9763
    },
    {
      "epoch": 1.6939625260235949,
      "grad_norm": 5.034497261047363,
      "learning_rate": 3.503788581712406e-06,
      "loss": 0.4719,
      "step": 9764
    },
    {
      "epoch": 1.6941360166551007,
      "grad_norm": 0.7237112522125244,
      "learning_rate": 3.49993726123133e-06,
      "loss": 0.5272,
      "step": 9765
    },
    {
      "epoch": 1.6943095072866066,
      "grad_norm": 0.682573676109314,
      "learning_rate": 3.4960878556017597e-06,
      "loss": 0.473,
      "step": 9766
    },
    {
      "epoch": 1.6944829979181124,
      "grad_norm": 1.1272791624069214,
      "learning_rate": 3.492240365270425e-06,
      "loss": 0.3773,
      "step": 9767
    },
    {
      "epoch": 1.6946564885496183,
      "grad_norm": 0.7569313645362854,
      "learning_rate": 3.488394790683829e-06,
      "loss": 0.4498,
      "step": 9768
    },
    {
      "epoch": 1.694829979181124,
      "grad_norm": 0.8451586961746216,
      "learning_rate": 3.4845511322882587e-06,
      "loss": 0.3754,
      "step": 9769
    },
    {
      "epoch": 1.6950034698126302,
      "grad_norm": 0.864126443862915,
      "learning_rate": 3.480709390529777e-06,
      "loss": 0.3108,
      "step": 9770
    },
    {
      "epoch": 1.695176960444136,
      "grad_norm": 2.0130836963653564,
      "learning_rate": 3.476869565854217e-06,
      "loss": 0.4581,
      "step": 9771
    },
    {
      "epoch": 1.6953504510756419,
      "grad_norm": 0.8672719597816467,
      "learning_rate": 3.473031658707193e-06,
      "loss": 0.3718,
      "step": 9772
    },
    {
      "epoch": 1.695523941707148,
      "grad_norm": 0.7420392036437988,
      "learning_rate": 3.469195669534109e-06,
      "loss": 0.548,
      "step": 9773
    },
    {
      "epoch": 1.6956974323386538,
      "grad_norm": 0.6630168557167053,
      "learning_rate": 3.465361598780128e-06,
      "loss": 0.4626,
      "step": 9774
    },
    {
      "epoch": 1.6958709229701596,
      "grad_norm": 1.0279978513717651,
      "learning_rate": 3.4615294468902017e-06,
      "loss": 0.5593,
      "step": 9775
    },
    {
      "epoch": 1.6960444136016655,
      "grad_norm": 0.9407429099082947,
      "learning_rate": 3.4576992143090517e-06,
      "loss": 0.3635,
      "step": 9776
    },
    {
      "epoch": 1.6962179042331713,
      "grad_norm": 0.8444896340370178,
      "learning_rate": 3.45387090148118e-06,
      "loss": 0.5234,
      "step": 9777
    },
    {
      "epoch": 1.6963913948646772,
      "grad_norm": 0.9997003078460693,
      "learning_rate": 3.450044508850876e-06,
      "loss": 0.4099,
      "step": 9778
    },
    {
      "epoch": 1.6965648854961832,
      "grad_norm": 0.8258613348007202,
      "learning_rate": 3.446220036862191e-06,
      "loss": 0.4784,
      "step": 9779
    },
    {
      "epoch": 1.696738376127689,
      "grad_norm": 0.7915644645690918,
      "learning_rate": 3.4423974859589594e-06,
      "loss": 0.489,
      "step": 9780
    },
    {
      "epoch": 1.6969118667591951,
      "grad_norm": 1.6192443370819092,
      "learning_rate": 3.4385768565847876e-06,
      "loss": 0.3755,
      "step": 9781
    },
    {
      "epoch": 1.697085357390701,
      "grad_norm": 0.7252005934715271,
      "learning_rate": 3.4347581491830796e-06,
      "loss": 0.3982,
      "step": 9782
    },
    {
      "epoch": 1.6972588480222068,
      "grad_norm": 0.8437852263450623,
      "learning_rate": 3.4309413641969802e-06,
      "loss": 0.4167,
      "step": 9783
    },
    {
      "epoch": 1.6974323386537127,
      "grad_norm": 0.7831665277481079,
      "learning_rate": 3.427126502069449e-06,
      "loss": 0.436,
      "step": 9784
    },
    {
      "epoch": 1.6976058292852185,
      "grad_norm": 0.9040247201919556,
      "learning_rate": 3.4233135632431913e-06,
      "loss": 0.5322,
      "step": 9785
    },
    {
      "epoch": 1.6977793199167244,
      "grad_norm": 0.8263068199157715,
      "learning_rate": 3.4195025481607224e-06,
      "loss": 0.52,
      "step": 9786
    },
    {
      "epoch": 1.6979528105482304,
      "grad_norm": 0.5854077339172363,
      "learning_rate": 3.415693457264291e-06,
      "loss": 0.4957,
      "step": 9787
    },
    {
      "epoch": 1.6981263011797363,
      "grad_norm": 0.7523157000541687,
      "learning_rate": 3.411886290995965e-06,
      "loss": 0.4803,
      "step": 9788
    },
    {
      "epoch": 1.6982997918112424,
      "grad_norm": 1.736129641532898,
      "learning_rate": 3.4080810497975626e-06,
      "loss": 0.5338,
      "step": 9789
    },
    {
      "epoch": 1.6984732824427482,
      "grad_norm": 0.8919421434402466,
      "learning_rate": 3.4042777341106903e-06,
      "loss": 0.432,
      "step": 9790
    },
    {
      "epoch": 1.698646773074254,
      "grad_norm": 0.7974951863288879,
      "learning_rate": 3.4004763443767175e-06,
      "loss": 0.49,
      "step": 9791
    },
    {
      "epoch": 1.69882026370576,
      "grad_norm": 1.1867344379425049,
      "learning_rate": 3.3966768810368132e-06,
      "loss": 0.4305,
      "step": 9792
    },
    {
      "epoch": 1.6989937543372657,
      "grad_norm": 0.9479686617851257,
      "learning_rate": 3.392879344531903e-06,
      "loss": 0.3116,
      "step": 9793
    },
    {
      "epoch": 1.6991672449687716,
      "grad_norm": 0.8172270655632019,
      "learning_rate": 3.3890837353026964e-06,
      "loss": 0.5779,
      "step": 9794
    },
    {
      "epoch": 1.6993407356002774,
      "grad_norm": 0.6842443943023682,
      "learning_rate": 3.385290053789676e-06,
      "loss": 0.4427,
      "step": 9795
    },
    {
      "epoch": 1.6995142262317835,
      "grad_norm": 1.7373539209365845,
      "learning_rate": 3.3814983004331014e-06,
      "loss": 0.4446,
      "step": 9796
    },
    {
      "epoch": 1.6996877168632893,
      "grad_norm": 0.6328423619270325,
      "learning_rate": 3.3777084756730183e-06,
      "loss": 0.6019,
      "step": 9797
    },
    {
      "epoch": 1.6998612074947954,
      "grad_norm": 0.8261460661888123,
      "learning_rate": 3.373920579949237e-06,
      "loss": 0.4897,
      "step": 9798
    },
    {
      "epoch": 1.7000346981263013,
      "grad_norm": 0.8764965534210205,
      "learning_rate": 3.3701346137013435e-06,
      "loss": 0.4509,
      "step": 9799
    },
    {
      "epoch": 1.700208188757807,
      "grad_norm": 0.7964087724685669,
      "learning_rate": 3.3663505773687023e-06,
      "loss": 0.4847,
      "step": 9800
    },
    {
      "epoch": 1.700381679389313,
      "grad_norm": 0.984257698059082,
      "learning_rate": 3.3625684713904617e-06,
      "loss": 0.4481,
      "step": 9801
    },
    {
      "epoch": 1.7005551700208188,
      "grad_norm": 0.7451890110969543,
      "learning_rate": 3.3587882962055374e-06,
      "loss": 0.396,
      "step": 9802
    },
    {
      "epoch": 1.7007286606523246,
      "grad_norm": 0.7505077719688416,
      "learning_rate": 3.355010052252623e-06,
      "loss": 0.3909,
      "step": 9803
    },
    {
      "epoch": 1.7009021512838307,
      "grad_norm": 0.7706956267356873,
      "learning_rate": 3.3512337399701813e-06,
      "loss": 0.4438,
      "step": 9804
    },
    {
      "epoch": 1.7010756419153366,
      "grad_norm": 0.784045398235321,
      "learning_rate": 3.3474593597964746e-06,
      "loss": 0.6011,
      "step": 9805
    },
    {
      "epoch": 1.7012491325468426,
      "grad_norm": 0.5819199085235596,
      "learning_rate": 3.3436869121695013e-06,
      "loss": 0.5012,
      "step": 9806
    },
    {
      "epoch": 1.7014226231783485,
      "grad_norm": 1.122066855430603,
      "learning_rate": 3.3399163975270786e-06,
      "loss": 0.5092,
      "step": 9807
    },
    {
      "epoch": 1.7015961138098543,
      "grad_norm": 0.7156979441642761,
      "learning_rate": 3.3361478163067673e-06,
      "loss": 0.3828,
      "step": 9808
    },
    {
      "epoch": 1.7017696044413602,
      "grad_norm": 0.9581041932106018,
      "learning_rate": 3.332381168945922e-06,
      "loss": 0.3796,
      "step": 9809
    },
    {
      "epoch": 1.701943095072866,
      "grad_norm": 0.644724428653717,
      "learning_rate": 3.328616455881657e-06,
      "loss": 0.5094,
      "step": 9810
    },
    {
      "epoch": 1.7021165857043719,
      "grad_norm": 0.6522907018661499,
      "learning_rate": 3.324853677550888e-06,
      "loss": 0.5576,
      "step": 9811
    },
    {
      "epoch": 1.7022900763358777,
      "grad_norm": 0.9530681371688843,
      "learning_rate": 3.3210928343902716e-06,
      "loss": 0.432,
      "step": 9812
    },
    {
      "epoch": 1.7024635669673838,
      "grad_norm": 0.8103886246681213,
      "learning_rate": 3.31733392683627e-06,
      "loss": 0.429,
      "step": 9813
    },
    {
      "epoch": 1.7026370575988896,
      "grad_norm": 0.8654118180274963,
      "learning_rate": 3.3135769553251017e-06,
      "loss": 0.5895,
      "step": 9814
    },
    {
      "epoch": 1.7028105482303957,
      "grad_norm": 0.8727630376815796,
      "learning_rate": 3.3098219202927815e-06,
      "loss": 0.395,
      "step": 9815
    },
    {
      "epoch": 1.7029840388619015,
      "grad_norm": 2.0842878818511963,
      "learning_rate": 3.3060688221750637e-06,
      "loss": 0.416,
      "step": 9816
    },
    {
      "epoch": 1.7031575294934074,
      "grad_norm": 0.7195104956626892,
      "learning_rate": 3.302317661407519e-06,
      "loss": 0.5002,
      "step": 9817
    },
    {
      "epoch": 1.7033310201249132,
      "grad_norm": 0.9616917967796326,
      "learning_rate": 3.2985684384254648e-06,
      "loss": 0.4795,
      "step": 9818
    },
    {
      "epoch": 1.703504510756419,
      "grad_norm": 2.126901388168335,
      "learning_rate": 3.294821153664003e-06,
      "loss": 0.3903,
      "step": 9819
    },
    {
      "epoch": 1.703678001387925,
      "grad_norm": 0.8265792727470398,
      "learning_rate": 3.2910758075580085e-06,
      "loss": 0.4197,
      "step": 9820
    },
    {
      "epoch": 1.703851492019431,
      "grad_norm": 1.0048558712005615,
      "learning_rate": 3.28733240054214e-06,
      "loss": 0.3671,
      "step": 9821
    },
    {
      "epoch": 1.7040249826509368,
      "grad_norm": 1.233837604522705,
      "learning_rate": 3.283590933050822e-06,
      "loss": 0.4093,
      "step": 9822
    },
    {
      "epoch": 1.704198473282443,
      "grad_norm": 0.674347460269928,
      "learning_rate": 3.2798514055182486e-06,
      "loss": 0.5133,
      "step": 9823
    },
    {
      "epoch": 1.7043719639139487,
      "grad_norm": 1.1877583265304565,
      "learning_rate": 3.2761138183784126e-06,
      "loss": 0.3771,
      "step": 9824
    },
    {
      "epoch": 1.7045454545454546,
      "grad_norm": 0.7793569564819336,
      "learning_rate": 3.2723781720650473e-06,
      "loss": 0.3204,
      "step": 9825
    },
    {
      "epoch": 1.7047189451769604,
      "grad_norm": 0.6692693829536438,
      "learning_rate": 3.2686444670116878e-06,
      "loss": 0.4117,
      "step": 9826
    },
    {
      "epoch": 1.7048924358084663,
      "grad_norm": 0.8593830466270447,
      "learning_rate": 3.2649127036516325e-06,
      "loss": 0.3987,
      "step": 9827
    },
    {
      "epoch": 1.7050659264399721,
      "grad_norm": 0.677654504776001,
      "learning_rate": 3.261182882417966e-06,
      "loss": 0.5596,
      "step": 9828
    },
    {
      "epoch": 1.7052394170714782,
      "grad_norm": 0.7616249322891235,
      "learning_rate": 3.2574550037435214e-06,
      "loss": 0.4303,
      "step": 9829
    },
    {
      "epoch": 1.705412907702984,
      "grad_norm": 1.898205280303955,
      "learning_rate": 3.253729068060938e-06,
      "loss": 0.4346,
      "step": 9830
    },
    {
      "epoch": 1.7055863983344899,
      "grad_norm": 0.7222949862480164,
      "learning_rate": 3.250005075802607e-06,
      "loss": 0.4666,
      "step": 9831
    },
    {
      "epoch": 1.705759888965996,
      "grad_norm": 1.2181752920150757,
      "learning_rate": 3.2462830274007073e-06,
      "loss": 0.3671,
      "step": 9832
    },
    {
      "epoch": 1.7059333795975018,
      "grad_norm": 0.9278891086578369,
      "learning_rate": 3.242562923287178e-06,
      "loss": 0.4618,
      "step": 9833
    },
    {
      "epoch": 1.7061068702290076,
      "grad_norm": 0.6020893454551697,
      "learning_rate": 3.2388447638937583e-06,
      "loss": 0.4412,
      "step": 9834
    },
    {
      "epoch": 1.7062803608605135,
      "grad_norm": 0.7284871339797974,
      "learning_rate": 3.2351285496519247e-06,
      "loss": 0.392,
      "step": 9835
    },
    {
      "epoch": 1.7064538514920193,
      "grad_norm": 1.2857002019882202,
      "learning_rate": 3.2314142809929617e-06,
      "loss": 0.3588,
      "step": 9836
    },
    {
      "epoch": 1.7066273421235252,
      "grad_norm": 0.7840943336486816,
      "learning_rate": 3.227701958347911e-06,
      "loss": 0.4854,
      "step": 9837
    },
    {
      "epoch": 1.7068008327550312,
      "grad_norm": 1.0641329288482666,
      "learning_rate": 3.223991582147592e-06,
      "loss": 0.3261,
      "step": 9838
    },
    {
      "epoch": 1.706974323386537,
      "grad_norm": 1.1004409790039062,
      "learning_rate": 3.220283152822592e-06,
      "loss": 0.4625,
      "step": 9839
    },
    {
      "epoch": 1.7071478140180432,
      "grad_norm": 1.3531230688095093,
      "learning_rate": 3.216576670803291e-06,
      "loss": 0.3833,
      "step": 9840
    },
    {
      "epoch": 1.707321304649549,
      "grad_norm": 0.7847244739532471,
      "learning_rate": 3.2128721365198223e-06,
      "loss": 0.4813,
      "step": 9841
    },
    {
      "epoch": 1.7074947952810549,
      "grad_norm": 0.9855622053146362,
      "learning_rate": 3.2091695504021047e-06,
      "loss": 0.4823,
      "step": 9842
    },
    {
      "epoch": 1.7076682859125607,
      "grad_norm": 0.7081166505813599,
      "learning_rate": 3.205468912879821e-06,
      "loss": 0.464,
      "step": 9843
    },
    {
      "epoch": 1.7078417765440665,
      "grad_norm": 0.9425329566001892,
      "learning_rate": 3.2017702243824434e-06,
      "loss": 0.4929,
      "step": 9844
    },
    {
      "epoch": 1.7080152671755724,
      "grad_norm": 0.6763478517532349,
      "learning_rate": 3.198073485339204e-06,
      "loss": 0.3898,
      "step": 9845
    },
    {
      "epoch": 1.7081887578070785,
      "grad_norm": 0.6645935773849487,
      "learning_rate": 3.1943786961791166e-06,
      "loss": 0.5366,
      "step": 9846
    },
    {
      "epoch": 1.7083622484385843,
      "grad_norm": 0.9655393958091736,
      "learning_rate": 3.190685857330964e-06,
      "loss": 0.438,
      "step": 9847
    },
    {
      "epoch": 1.7085357390700904,
      "grad_norm": 0.9493473768234253,
      "learning_rate": 3.1869949692232982e-06,
      "loss": 0.4033,
      "step": 9848
    },
    {
      "epoch": 1.7087092297015962,
      "grad_norm": 0.7750371098518372,
      "learning_rate": 3.1833060322844633e-06,
      "loss": 0.3461,
      "step": 9849
    },
    {
      "epoch": 1.708882720333102,
      "grad_norm": 0.7906068563461304,
      "learning_rate": 3.179619046942557e-06,
      "loss": 0.5339,
      "step": 9850
    },
    {
      "epoch": 1.709056210964608,
      "grad_norm": 0.8526831865310669,
      "learning_rate": 3.1759340136254614e-06,
      "loss": 0.4376,
      "step": 9851
    },
    {
      "epoch": 1.7092297015961138,
      "grad_norm": 0.7395734786987305,
      "learning_rate": 3.172250932760823e-06,
      "loss": 0.5119,
      "step": 9852
    },
    {
      "epoch": 1.7094031922276196,
      "grad_norm": 1.161130428314209,
      "learning_rate": 3.1685698047760806e-06,
      "loss": 0.4056,
      "step": 9853
    },
    {
      "epoch": 1.7095766828591255,
      "grad_norm": 0.977079451084137,
      "learning_rate": 3.164890630098416e-06,
      "loss": 0.4141,
      "step": 9854
    },
    {
      "epoch": 1.7097501734906315,
      "grad_norm": 0.8077842593193054,
      "learning_rate": 3.1612134091548153e-06,
      "loss": 0.3906,
      "step": 9855
    },
    {
      "epoch": 1.7099236641221374,
      "grad_norm": 0.7422673106193542,
      "learning_rate": 3.1575381423720142e-06,
      "loss": 0.4556,
      "step": 9856
    },
    {
      "epoch": 1.7100971547536434,
      "grad_norm": 0.8558520078659058,
      "learning_rate": 3.153864830176547e-06,
      "loss": 0.3735,
      "step": 9857
    },
    {
      "epoch": 1.7102706453851493,
      "grad_norm": 0.9983931183815002,
      "learning_rate": 3.150193472994687e-06,
      "loss": 0.3532,
      "step": 9858
    },
    {
      "epoch": 1.7104441360166551,
      "grad_norm": 1.005107045173645,
      "learning_rate": 3.1465240712525124e-06,
      "loss": 0.3118,
      "step": 9859
    },
    {
      "epoch": 1.710617626648161,
      "grad_norm": 1.0908174514770508,
      "learning_rate": 3.142856625375856e-06,
      "loss": 0.4779,
      "step": 9860
    },
    {
      "epoch": 1.7107911172796668,
      "grad_norm": 1.317470908164978,
      "learning_rate": 3.139191135790334e-06,
      "loss": 0.3978,
      "step": 9861
    },
    {
      "epoch": 1.7109646079111727,
      "grad_norm": 0.7371059060096741,
      "learning_rate": 3.13552760292132e-06,
      "loss": 0.4893,
      "step": 9862
    },
    {
      "epoch": 1.7111380985426787,
      "grad_norm": 1.1593574285507202,
      "learning_rate": 3.131866027193988e-06,
      "loss": 0.4475,
      "step": 9863
    },
    {
      "epoch": 1.7113115891741846,
      "grad_norm": 1.3534239530563354,
      "learning_rate": 3.1282064090332522e-06,
      "loss": 0.4206,
      "step": 9864
    },
    {
      "epoch": 1.7114850798056906,
      "grad_norm": 0.6876006126403809,
      "learning_rate": 3.1245487488638247e-06,
      "loss": 0.449,
      "step": 9865
    },
    {
      "epoch": 1.7116585704371965,
      "grad_norm": 0.7555912733078003,
      "learning_rate": 3.1208930471101786e-06,
      "loss": 0.4205,
      "step": 9866
    },
    {
      "epoch": 1.7118320610687023,
      "grad_norm": 0.6664054989814758,
      "learning_rate": 3.1172393041965644e-06,
      "loss": 0.4686,
      "step": 9867
    },
    {
      "epoch": 1.7120055517002082,
      "grad_norm": 0.8035487532615662,
      "learning_rate": 3.1135875205469946e-06,
      "loss": 0.3636,
      "step": 9868
    },
    {
      "epoch": 1.712179042331714,
      "grad_norm": 1.3106471300125122,
      "learning_rate": 3.1099376965852744e-06,
      "loss": 0.4021,
      "step": 9869
    },
    {
      "epoch": 1.7123525329632199,
      "grad_norm": 0.740695059299469,
      "learning_rate": 3.1062898327349656e-06,
      "loss": 0.4259,
      "step": 9870
    },
    {
      "epoch": 1.7125260235947257,
      "grad_norm": 0.9075673222541809,
      "learning_rate": 3.102643929419402e-06,
      "loss": 0.4011,
      "step": 9871
    },
    {
      "epoch": 1.7126995142262318,
      "grad_norm": 0.7000410556793213,
      "learning_rate": 3.098999987061706e-06,
      "loss": 0.4046,
      "step": 9872
    },
    {
      "epoch": 1.7128730048577376,
      "grad_norm": 0.7891289591789246,
      "learning_rate": 3.0953580060847545e-06,
      "loss": 0.4453,
      "step": 9873
    },
    {
      "epoch": 1.7130464954892437,
      "grad_norm": 0.9291444420814514,
      "learning_rate": 3.0917179869112023e-06,
      "loss": 0.4176,
      "step": 9874
    },
    {
      "epoch": 1.7132199861207495,
      "grad_norm": 1.2433059215545654,
      "learning_rate": 3.0880799299634767e-06,
      "loss": 0.4355,
      "step": 9875
    },
    {
      "epoch": 1.7133934767522554,
      "grad_norm": 0.7945812940597534,
      "learning_rate": 3.084443835663791e-06,
      "loss": 0.4093,
      "step": 9876
    },
    {
      "epoch": 1.7135669673837612,
      "grad_norm": 0.7113139033317566,
      "learning_rate": 3.080809704434098e-06,
      "loss": 0.4581,
      "step": 9877
    },
    {
      "epoch": 1.713740458015267,
      "grad_norm": 0.7200466394424438,
      "learning_rate": 3.077177536696159e-06,
      "loss": 0.4387,
      "step": 9878
    },
    {
      "epoch": 1.713913948646773,
      "grad_norm": 0.5357228517532349,
      "learning_rate": 3.0735473328714873e-06,
      "loss": 0.5604,
      "step": 9879
    },
    {
      "epoch": 1.714087439278279,
      "grad_norm": 0.7950665950775146,
      "learning_rate": 3.0699190933813683e-06,
      "loss": 0.5728,
      "step": 9880
    },
    {
      "epoch": 1.7142609299097848,
      "grad_norm": 0.5989641547203064,
      "learning_rate": 3.066292818646863e-06,
      "loss": 0.4639,
      "step": 9881
    },
    {
      "epoch": 1.714434420541291,
      "grad_norm": 0.85533607006073,
      "learning_rate": 3.0626685090888177e-06,
      "loss": 0.4712,
      "step": 9882
    },
    {
      "epoch": 1.7146079111727968,
      "grad_norm": 0.8431836366653442,
      "learning_rate": 3.0590461651278168e-06,
      "loss": 0.5385,
      "step": 9883
    },
    {
      "epoch": 1.7147814018043026,
      "grad_norm": 4.216972827911377,
      "learning_rate": 3.0554257871842543e-06,
      "loss": 0.4375,
      "step": 9884
    },
    {
      "epoch": 1.7149548924358085,
      "grad_norm": 0.6002256870269775,
      "learning_rate": 3.0518073756782683e-06,
      "loss": 0.5396,
      "step": 9885
    },
    {
      "epoch": 1.7151283830673143,
      "grad_norm": 0.7627999186515808,
      "learning_rate": 3.0481909310297954e-06,
      "loss": 0.4133,
      "step": 9886
    },
    {
      "epoch": 1.7153018736988201,
      "grad_norm": 0.7516846656799316,
      "learning_rate": 3.0445764536585076e-06,
      "loss": 0.5165,
      "step": 9887
    },
    {
      "epoch": 1.7154753643303262,
      "grad_norm": 0.8690098524093628,
      "learning_rate": 3.0409639439838833e-06,
      "loss": 0.4739,
      "step": 9888
    },
    {
      "epoch": 1.715648854961832,
      "grad_norm": 0.7863034009933472,
      "learning_rate": 3.037353402425154e-06,
      "loss": 0.402,
      "step": 9889
    },
    {
      "epoch": 1.715822345593338,
      "grad_norm": 0.7509139180183411,
      "learning_rate": 3.0337448294013307e-06,
      "loss": 0.4033,
      "step": 9890
    },
    {
      "epoch": 1.715995836224844,
      "grad_norm": 0.8073070645332336,
      "learning_rate": 3.0301382253311828e-06,
      "loss": 0.3865,
      "step": 9891
    },
    {
      "epoch": 1.7161693268563498,
      "grad_norm": 1.1924837827682495,
      "learning_rate": 3.0265335906332717e-06,
      "loss": 0.3853,
      "step": 9892
    },
    {
      "epoch": 1.7163428174878557,
      "grad_norm": 2.007868528366089,
      "learning_rate": 3.0229309257259154e-06,
      "loss": 0.4081,
      "step": 9893
    },
    {
      "epoch": 1.7165163081193615,
      "grad_norm": 0.9190699458122253,
      "learning_rate": 3.019330231027209e-06,
      "loss": 0.3104,
      "step": 9894
    },
    {
      "epoch": 1.7166897987508674,
      "grad_norm": 0.7925846576690674,
      "learning_rate": 3.015731506955015e-06,
      "loss": 0.4226,
      "step": 9895
    },
    {
      "epoch": 1.7168632893823732,
      "grad_norm": 1.215623140335083,
      "learning_rate": 3.012134753926965e-06,
      "loss": 0.4392,
      "step": 9896
    },
    {
      "epoch": 1.7170367800138793,
      "grad_norm": 0.7585077285766602,
      "learning_rate": 3.008539972360478e-06,
      "loss": 0.4553,
      "step": 9897
    },
    {
      "epoch": 1.7172102706453851,
      "grad_norm": 0.9052568674087524,
      "learning_rate": 3.0049471626727246e-06,
      "loss": 0.5319,
      "step": 9898
    },
    {
      "epoch": 1.7173837612768912,
      "grad_norm": 0.7209457159042358,
      "learning_rate": 3.0013563252806576e-06,
      "loss": 0.4357,
      "step": 9899
    },
    {
      "epoch": 1.717557251908397,
      "grad_norm": 0.6793199777603149,
      "learning_rate": 2.997767460600991e-06,
      "loss": 0.5719,
      "step": 9900
    },
    {
      "epoch": 1.7177307425399029,
      "grad_norm": 0.8123597502708435,
      "learning_rate": 2.9941805690502246e-06,
      "loss": 0.4197,
      "step": 9901
    },
    {
      "epoch": 1.7179042331714087,
      "grad_norm": 0.9759712219238281,
      "learning_rate": 2.990595651044621e-06,
      "loss": 0.3523,
      "step": 9902
    },
    {
      "epoch": 1.7180777238029146,
      "grad_norm": 2.39614200592041,
      "learning_rate": 2.9870127070002117e-06,
      "loss": 0.3889,
      "step": 9903
    },
    {
      "epoch": 1.7182512144344204,
      "grad_norm": 1.0239982604980469,
      "learning_rate": 2.9834317373327983e-06,
      "loss": 0.4694,
      "step": 9904
    },
    {
      "epoch": 1.7184247050659265,
      "grad_norm": 0.8857870697975159,
      "learning_rate": 2.979852742457967e-06,
      "loss": 0.5314,
      "step": 9905
    },
    {
      "epoch": 1.7185981956974323,
      "grad_norm": 0.7471193671226501,
      "learning_rate": 2.976275722791051e-06,
      "loss": 0.473,
      "step": 9906
    },
    {
      "epoch": 1.7187716863289384,
      "grad_norm": 1.022916555404663,
      "learning_rate": 2.972700678747176e-06,
      "loss": 0.5603,
      "step": 9907
    },
    {
      "epoch": 1.7189451769604442,
      "grad_norm": 0.7705941200256348,
      "learning_rate": 2.9691276107412293e-06,
      "loss": 0.4449,
      "step": 9908
    },
    {
      "epoch": 1.71911866759195,
      "grad_norm": 1.0875569581985474,
      "learning_rate": 2.9655565191878668e-06,
      "loss": 0.3584,
      "step": 9909
    },
    {
      "epoch": 1.719292158223456,
      "grad_norm": 1.3497709035873413,
      "learning_rate": 2.961987404501516e-06,
      "loss": 0.3804,
      "step": 9910
    },
    {
      "epoch": 1.7194656488549618,
      "grad_norm": 0.6637280583381653,
      "learning_rate": 2.9584202670963892e-06,
      "loss": 0.5167,
      "step": 9911
    },
    {
      "epoch": 1.7196391394864676,
      "grad_norm": 0.9440620541572571,
      "learning_rate": 2.9548551073864386e-06,
      "loss": 0.4071,
      "step": 9912
    },
    {
      "epoch": 1.7198126301179735,
      "grad_norm": 0.7562226057052612,
      "learning_rate": 2.95129192578542e-06,
      "loss": 0.5015,
      "step": 9913
    },
    {
      "epoch": 1.7199861207494795,
      "grad_norm": 0.8880428075790405,
      "learning_rate": 2.947730722706832e-06,
      "loss": 0.5117,
      "step": 9914
    },
    {
      "epoch": 1.7201596113809854,
      "grad_norm": 1.0197423696517944,
      "learning_rate": 2.9441714985639747e-06,
      "loss": 0.3461,
      "step": 9915
    },
    {
      "epoch": 1.7203331020124915,
      "grad_norm": 0.7170447707176208,
      "learning_rate": 2.94061425376988e-06,
      "loss": 0.3707,
      "step": 9916
    },
    {
      "epoch": 1.7205065926439973,
      "grad_norm": 1.3667303323745728,
      "learning_rate": 2.9370589887373825e-06,
      "loss": 0.4254,
      "step": 9917
    },
    {
      "epoch": 1.7206800832755031,
      "grad_norm": 0.9315698742866516,
      "learning_rate": 2.9335057038790715e-06,
      "loss": 0.4695,
      "step": 9918
    },
    {
      "epoch": 1.720853573907009,
      "grad_norm": 0.7081165909767151,
      "learning_rate": 2.9299543996073067e-06,
      "loss": 0.4189,
      "step": 9919
    },
    {
      "epoch": 1.7210270645385148,
      "grad_norm": 0.7056809067726135,
      "learning_rate": 2.9264050763342267e-06,
      "loss": 0.5752,
      "step": 9920
    },
    {
      "epoch": 1.7212005551700207,
      "grad_norm": 0.8265276551246643,
      "learning_rate": 2.9228577344717357e-06,
      "loss": 0.4545,
      "step": 9921
    },
    {
      "epoch": 1.7213740458015268,
      "grad_norm": 0.6816700100898743,
      "learning_rate": 2.9193123744315e-06,
      "loss": 0.4418,
      "step": 9922
    },
    {
      "epoch": 1.7215475364330326,
      "grad_norm": 1.0332481861114502,
      "learning_rate": 2.9157689966249636e-06,
      "loss": 0.3752,
      "step": 9923
    },
    {
      "epoch": 1.7217210270645387,
      "grad_norm": 0.7489287257194519,
      "learning_rate": 2.912227601463351e-06,
      "loss": 0.5599,
      "step": 9924
    },
    {
      "epoch": 1.7218945176960445,
      "grad_norm": 1.5686593055725098,
      "learning_rate": 2.9086881893576267e-06,
      "loss": 0.5272,
      "step": 9925
    },
    {
      "epoch": 1.7220680083275504,
      "grad_norm": 0.6984260678291321,
      "learning_rate": 2.9051507607185603e-06,
      "loss": 0.4105,
      "step": 9926
    },
    {
      "epoch": 1.7222414989590562,
      "grad_norm": 0.6304365396499634,
      "learning_rate": 2.9016153159566607e-06,
      "loss": 0.5432,
      "step": 9927
    },
    {
      "epoch": 1.722414989590562,
      "grad_norm": 0.7879930734634399,
      "learning_rate": 2.8980818554822376e-06,
      "loss": 0.4963,
      "step": 9928
    },
    {
      "epoch": 1.722588480222068,
      "grad_norm": 0.7842549085617065,
      "learning_rate": 2.894550379705332e-06,
      "loss": 0.4963,
      "step": 9929
    },
    {
      "epoch": 1.7227619708535737,
      "grad_norm": 0.8720807433128357,
      "learning_rate": 2.8910208890357916e-06,
      "loss": 0.3947,
      "step": 9930
    },
    {
      "epoch": 1.7229354614850798,
      "grad_norm": 0.8417780995368958,
      "learning_rate": 2.8874933838832154e-06,
      "loss": 0.399,
      "step": 9931
    },
    {
      "epoch": 1.7231089521165857,
      "grad_norm": 1.0623208284378052,
      "learning_rate": 2.883967864656969e-06,
      "loss": 0.3695,
      "step": 9932
    },
    {
      "epoch": 1.7232824427480917,
      "grad_norm": 0.6795383095741272,
      "learning_rate": 2.8804443317661925e-06,
      "loss": 0.4988,
      "step": 9933
    },
    {
      "epoch": 1.7234559333795976,
      "grad_norm": 0.8090497255325317,
      "learning_rate": 2.876922785619809e-06,
      "loss": 0.4906,
      "step": 9934
    },
    {
      "epoch": 1.7236294240111034,
      "grad_norm": 0.9351137280464172,
      "learning_rate": 2.873403226626479e-06,
      "loss": 0.4972,
      "step": 9935
    },
    {
      "epoch": 1.7238029146426093,
      "grad_norm": 0.7210533022880554,
      "learning_rate": 2.8698856551946664e-06,
      "loss": 0.415,
      "step": 9936
    },
    {
      "epoch": 1.723976405274115,
      "grad_norm": 1.0768041610717773,
      "learning_rate": 2.866370071732585e-06,
      "loss": 0.3759,
      "step": 9937
    },
    {
      "epoch": 1.724149895905621,
      "grad_norm": 0.7763087749481201,
      "learning_rate": 2.8628564766482193e-06,
      "loss": 0.458,
      "step": 9938
    },
    {
      "epoch": 1.724323386537127,
      "grad_norm": 0.7806427478790283,
      "learning_rate": 2.859344870349323e-06,
      "loss": 0.4099,
      "step": 9939
    },
    {
      "epoch": 1.7244968771686329,
      "grad_norm": 0.7402631044387817,
      "learning_rate": 2.855835253243433e-06,
      "loss": 0.4496,
      "step": 9940
    },
    {
      "epoch": 1.724670367800139,
      "grad_norm": 0.6902532577514648,
      "learning_rate": 2.8523276257378406e-06,
      "loss": 0.525,
      "step": 9941
    },
    {
      "epoch": 1.7248438584316448,
      "grad_norm": 1.1441080570220947,
      "learning_rate": 2.848821988239605e-06,
      "loss": 0.47,
      "step": 9942
    },
    {
      "epoch": 1.7250173490631506,
      "grad_norm": 0.7120885252952576,
      "learning_rate": 2.8453183411555606e-06,
      "loss": 0.5054,
      "step": 9943
    },
    {
      "epoch": 1.7251908396946565,
      "grad_norm": 0.7540342211723328,
      "learning_rate": 2.8418166848923158e-06,
      "loss": 0.452,
      "step": 9944
    },
    {
      "epoch": 1.7253643303261623,
      "grad_norm": 0.5520671606063843,
      "learning_rate": 2.838317019856238e-06,
      "loss": 0.4268,
      "step": 9945
    },
    {
      "epoch": 1.7255378209576682,
      "grad_norm": 0.9635013937950134,
      "learning_rate": 2.834819346453468e-06,
      "loss": 0.4554,
      "step": 9946
    },
    {
      "epoch": 1.7257113115891742,
      "grad_norm": 0.7378911375999451,
      "learning_rate": 2.8313236650899135e-06,
      "loss": 0.3519,
      "step": 9947
    },
    {
      "epoch": 1.72588480222068,
      "grad_norm": 0.885434627532959,
      "learning_rate": 2.827829976171248e-06,
      "loss": 0.3899,
      "step": 9948
    },
    {
      "epoch": 1.726058292852186,
      "grad_norm": 0.8526363968849182,
      "learning_rate": 2.8243382801029295e-06,
      "loss": 0.434,
      "step": 9949
    },
    {
      "epoch": 1.726231783483692,
      "grad_norm": 0.9503956437110901,
      "learning_rate": 2.820848577290165e-06,
      "loss": 0.403,
      "step": 9950
    },
    {
      "epoch": 1.7264052741151978,
      "grad_norm": 1.0041067600250244,
      "learning_rate": 2.8173608681379417e-06,
      "loss": 0.4014,
      "step": 9951
    },
    {
      "epoch": 1.7265787647467037,
      "grad_norm": 1.171455979347229,
      "learning_rate": 2.8138751530510065e-06,
      "loss": 0.3354,
      "step": 9952
    },
    {
      "epoch": 1.7267522553782095,
      "grad_norm": 0.7163133025169373,
      "learning_rate": 2.8103914324338965e-06,
      "loss": 0.4609,
      "step": 9953
    },
    {
      "epoch": 1.7269257460097154,
      "grad_norm": 0.6427908539772034,
      "learning_rate": 2.806909706690881e-06,
      "loss": 0.535,
      "step": 9954
    },
    {
      "epoch": 1.7270992366412212,
      "grad_norm": 0.9116261005401611,
      "learning_rate": 2.8034299762260308e-06,
      "loss": 0.4227,
      "step": 9955
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.5795822143554688,
      "learning_rate": 2.799952241443167e-06,
      "loss": 0.4703,
      "step": 9956
    },
    {
      "epoch": 1.7274462179042331,
      "grad_norm": 0.8556695580482483,
      "learning_rate": 2.796476502745895e-06,
      "loss": 0.4191,
      "step": 9957
    },
    {
      "epoch": 1.7276197085357392,
      "grad_norm": 0.7606567144393921,
      "learning_rate": 2.7930027605375644e-06,
      "loss": 0.5043,
      "step": 9958
    },
    {
      "epoch": 1.727793199167245,
      "grad_norm": 0.8193055391311646,
      "learning_rate": 2.7895310152213163e-06,
      "loss": 0.4131,
      "step": 9959
    },
    {
      "epoch": 1.727966689798751,
      "grad_norm": 0.9150891900062561,
      "learning_rate": 2.7860612672000485e-06,
      "loss": 0.4764,
      "step": 9960
    },
    {
      "epoch": 1.7281401804302567,
      "grad_norm": 0.7192273139953613,
      "learning_rate": 2.7825935168764284e-06,
      "loss": 0.6002,
      "step": 9961
    },
    {
      "epoch": 1.7283136710617626,
      "grad_norm": 0.724804699420929,
      "learning_rate": 2.7791277646528893e-06,
      "loss": 0.4231,
      "step": 9962
    },
    {
      "epoch": 1.7284871616932684,
      "grad_norm": 0.8425358533859253,
      "learning_rate": 2.7756640109316423e-06,
      "loss": 0.5256,
      "step": 9963
    },
    {
      "epoch": 1.7286606523247745,
      "grad_norm": 1.4396371841430664,
      "learning_rate": 2.77220225611466e-06,
      "loss": 0.4001,
      "step": 9964
    },
    {
      "epoch": 1.7288341429562804,
      "grad_norm": 0.9159610271453857,
      "learning_rate": 2.768742500603678e-06,
      "loss": 0.5305,
      "step": 9965
    },
    {
      "epoch": 1.7290076335877864,
      "grad_norm": 0.6820814609527588,
      "learning_rate": 2.7652847448002074e-06,
      "loss": 0.417,
      "step": 9966
    },
    {
      "epoch": 1.7291811242192923,
      "grad_norm": 0.7331205010414124,
      "learning_rate": 2.7618289891055217e-06,
      "loss": 0.5649,
      "step": 9967
    },
    {
      "epoch": 1.729354614850798,
      "grad_norm": 1.05643892288208,
      "learning_rate": 2.7583752339206714e-06,
      "loss": 0.4611,
      "step": 9968
    },
    {
      "epoch": 1.729528105482304,
      "grad_norm": 0.6232126951217651,
      "learning_rate": 2.754923479646465e-06,
      "loss": 0.5616,
      "step": 9969
    },
    {
      "epoch": 1.7297015961138098,
      "grad_norm": 0.6674892902374268,
      "learning_rate": 2.7514737266834845e-06,
      "loss": 0.4,
      "step": 9970
    },
    {
      "epoch": 1.7298750867453156,
      "grad_norm": 0.8359811902046204,
      "learning_rate": 2.7480259754320716e-06,
      "loss": 0.4115,
      "step": 9971
    },
    {
      "epoch": 1.7300485773768215,
      "grad_norm": 0.9754602909088135,
      "learning_rate": 2.7445802262923505e-06,
      "loss": 0.426,
      "step": 9972
    },
    {
      "epoch": 1.7302220680083276,
      "grad_norm": 0.8265777230262756,
      "learning_rate": 2.7411364796642015e-06,
      "loss": 0.4432,
      "step": 9973
    },
    {
      "epoch": 1.7303955586398334,
      "grad_norm": 0.7472012042999268,
      "learning_rate": 2.737694735947276e-06,
      "loss": 0.5466,
      "step": 9974
    },
    {
      "epoch": 1.7305690492713395,
      "grad_norm": 0.790481448173523,
      "learning_rate": 2.7342549955409836e-06,
      "loss": 0.4149,
      "step": 9975
    },
    {
      "epoch": 1.7307425399028453,
      "grad_norm": 0.7875306606292725,
      "learning_rate": 2.730817258844529e-06,
      "loss": 0.473,
      "step": 9976
    },
    {
      "epoch": 1.7309160305343512,
      "grad_norm": 1.0280705690383911,
      "learning_rate": 2.727381526256845e-06,
      "loss": 0.4008,
      "step": 9977
    },
    {
      "epoch": 1.731089521165857,
      "grad_norm": 0.7789735198020935,
      "learning_rate": 2.723947798176665e-06,
      "loss": 0.4231,
      "step": 9978
    },
    {
      "epoch": 1.7312630117973629,
      "grad_norm": 0.6488164663314819,
      "learning_rate": 2.720516075002473e-06,
      "loss": 0.6138,
      "step": 9979
    },
    {
      "epoch": 1.7314365024288687,
      "grad_norm": 0.6671362519264221,
      "learning_rate": 2.7170863571325257e-06,
      "loss": 0.5438,
      "step": 9980
    },
    {
      "epoch": 1.7316099930603748,
      "grad_norm": 1.200859546661377,
      "learning_rate": 2.7136586449648407e-06,
      "loss": 0.3715,
      "step": 9981
    },
    {
      "epoch": 1.7317834836918806,
      "grad_norm": 0.6614787578582764,
      "learning_rate": 2.7102329388972215e-06,
      "loss": 0.3636,
      "step": 9982
    },
    {
      "epoch": 1.7319569743233867,
      "grad_norm": 0.8215885758399963,
      "learning_rate": 2.7068092393272082e-06,
      "loss": 0.4867,
      "step": 9983
    },
    {
      "epoch": 1.7321304649548925,
      "grad_norm": 0.7617893815040588,
      "learning_rate": 2.7033875466521363e-06,
      "loss": 0.4686,
      "step": 9984
    },
    {
      "epoch": 1.7323039555863984,
      "grad_norm": 0.7937578558921814,
      "learning_rate": 2.6999678612690907e-06,
      "loss": 0.3956,
      "step": 9985
    },
    {
      "epoch": 1.7324774462179042,
      "grad_norm": 0.9495329260826111,
      "learning_rate": 2.696550183574942e-06,
      "loss": 0.4323,
      "step": 9986
    },
    {
      "epoch": 1.73265093684941,
      "grad_norm": 1.5091499090194702,
      "learning_rate": 2.6931345139663e-06,
      "loss": 0.4475,
      "step": 9987
    },
    {
      "epoch": 1.732824427480916,
      "grad_norm": 0.668420135974884,
      "learning_rate": 2.6897208528395656e-06,
      "loss": 0.5051,
      "step": 9988
    },
    {
      "epoch": 1.7329979181124218,
      "grad_norm": 0.6214038729667664,
      "learning_rate": 2.6863092005908973e-06,
      "loss": 0.5253,
      "step": 9989
    },
    {
      "epoch": 1.7331714087439278,
      "grad_norm": 0.9628726840019226,
      "learning_rate": 2.682899557616223e-06,
      "loss": 0.4208,
      "step": 9990
    },
    {
      "epoch": 1.7333448993754337,
      "grad_norm": 0.6523237228393555,
      "learning_rate": 2.679491924311226e-06,
      "loss": 0.4209,
      "step": 9991
    },
    {
      "epoch": 1.7335183900069397,
      "grad_norm": 0.9754306077957153,
      "learning_rate": 2.676086301071381e-06,
      "loss": 0.3916,
      "step": 9992
    },
    {
      "epoch": 1.7336918806384456,
      "grad_norm": 0.8062222003936768,
      "learning_rate": 2.6726826882919055e-06,
      "loss": 0.4349,
      "step": 9993
    },
    {
      "epoch": 1.7338653712699514,
      "grad_norm": 1.4655934572219849,
      "learning_rate": 2.66928108636779e-06,
      "loss": 0.4634,
      "step": 9994
    },
    {
      "epoch": 1.7340388619014573,
      "grad_norm": 0.993941605091095,
      "learning_rate": 2.6658814956938073e-06,
      "loss": 0.4526,
      "step": 9995
    },
    {
      "epoch": 1.7342123525329631,
      "grad_norm": 0.8276480436325073,
      "learning_rate": 2.662483916664467e-06,
      "loss": 0.3713,
      "step": 9996
    },
    {
      "epoch": 1.734385843164469,
      "grad_norm": 0.7622029781341553,
      "learning_rate": 2.6590883496740727e-06,
      "loss": 0.4706,
      "step": 9997
    },
    {
      "epoch": 1.734559333795975,
      "grad_norm": 0.816926121711731,
      "learning_rate": 2.6556947951166836e-06,
      "loss": 0.3597,
      "step": 9998
    },
    {
      "epoch": 1.734732824427481,
      "grad_norm": 0.6731540560722351,
      "learning_rate": 2.6523032533861236e-06,
      "loss": 0.4859,
      "step": 9999
    },
    {
      "epoch": 1.734906315058987,
      "grad_norm": 1.0628679990768433,
      "learning_rate": 2.648913724875981e-06,
      "loss": 0.3247,
      "step": 10000
    },
    {
      "epoch": 1.7350798056904928,
      "grad_norm": 0.7253689765930176,
      "learning_rate": 2.6455262099796233e-06,
      "loss": 0.5372,
      "step": 10001
    },
    {
      "epoch": 1.7352532963219987,
      "grad_norm": 0.8236063122749329,
      "learning_rate": 2.6421407090901707e-06,
      "loss": 0.529,
      "step": 10002
    },
    {
      "epoch": 1.7354267869535045,
      "grad_norm": 1.1921359300613403,
      "learning_rate": 2.6387572226005143e-06,
      "loss": 0.3976,
      "step": 10003
    },
    {
      "epoch": 1.7356002775850103,
      "grad_norm": 1.595541000366211,
      "learning_rate": 2.635375750903306e-06,
      "loss": 0.383,
      "step": 10004
    },
    {
      "epoch": 1.7357737682165162,
      "grad_norm": 0.7787811756134033,
      "learning_rate": 2.631996294390986e-06,
      "loss": 0.5839,
      "step": 10005
    },
    {
      "epoch": 1.7359472588480223,
      "grad_norm": 0.6764829158782959,
      "learning_rate": 2.628618853455727e-06,
      "loss": 0.5826,
      "step": 10006
    },
    {
      "epoch": 1.736120749479528,
      "grad_norm": 0.8212352991104126,
      "learning_rate": 2.625243428489492e-06,
      "loss": 0.442,
      "step": 10007
    },
    {
      "epoch": 1.7362942401110342,
      "grad_norm": 0.8655128479003906,
      "learning_rate": 2.621870019884005e-06,
      "loss": 0.394,
      "step": 10008
    },
    {
      "epoch": 1.73646773074254,
      "grad_norm": 0.7016631960868835,
      "learning_rate": 2.6184986280307525e-06,
      "loss": 0.5529,
      "step": 10009
    },
    {
      "epoch": 1.7366412213740459,
      "grad_norm": 0.855135440826416,
      "learning_rate": 2.6151292533209826e-06,
      "loss": 0.6355,
      "step": 10010
    },
    {
      "epoch": 1.7368147120055517,
      "grad_norm": 0.900076150894165,
      "learning_rate": 2.6117618961457235e-06,
      "loss": 0.4408,
      "step": 10011
    },
    {
      "epoch": 1.7369882026370576,
      "grad_norm": 0.9842840433120728,
      "learning_rate": 2.6083965568957603e-06,
      "loss": 0.4554,
      "step": 10012
    },
    {
      "epoch": 1.7371616932685634,
      "grad_norm": 0.6928418874740601,
      "learning_rate": 2.6050332359616403e-06,
      "loss": 0.4578,
      "step": 10013
    },
    {
      "epoch": 1.7373351839000692,
      "grad_norm": 0.6585573554039001,
      "learning_rate": 2.601671933733678e-06,
      "loss": 0.5532,
      "step": 10014
    },
    {
      "epoch": 1.7375086745315753,
      "grad_norm": 0.6021066904067993,
      "learning_rate": 2.598312650601964e-06,
      "loss": 0.5166,
      "step": 10015
    },
    {
      "epoch": 1.7376821651630812,
      "grad_norm": 0.6157493591308594,
      "learning_rate": 2.594955386956346e-06,
      "loss": 0.5314,
      "step": 10016
    },
    {
      "epoch": 1.7378556557945872,
      "grad_norm": 0.9439457058906555,
      "learning_rate": 2.5916001431864346e-06,
      "loss": 0.3905,
      "step": 10017
    },
    {
      "epoch": 1.738029146426093,
      "grad_norm": 0.974572479724884,
      "learning_rate": 2.588246919681614e-06,
      "loss": 0.3445,
      "step": 10018
    },
    {
      "epoch": 1.738202637057599,
      "grad_norm": 1.8008818626403809,
      "learning_rate": 2.5848957168310195e-06,
      "loss": 0.3837,
      "step": 10019
    },
    {
      "epoch": 1.7383761276891048,
      "grad_norm": 0.8484526872634888,
      "learning_rate": 2.5815465350235756e-06,
      "loss": 0.5154,
      "step": 10020
    },
    {
      "epoch": 1.7385496183206106,
      "grad_norm": 0.746911346912384,
      "learning_rate": 2.5781993746479537e-06,
      "loss": 0.5085,
      "step": 10021
    },
    {
      "epoch": 1.7387231089521165,
      "grad_norm": 0.7924544811248779,
      "learning_rate": 2.5748542360925944e-06,
      "loss": 0.4045,
      "step": 10022
    },
    {
      "epoch": 1.7388965995836225,
      "grad_norm": 1.0243325233459473,
      "learning_rate": 2.5715111197457e-06,
      "loss": 0.4777,
      "step": 10023
    },
    {
      "epoch": 1.7390700902151284,
      "grad_norm": 0.621435821056366,
      "learning_rate": 2.568170025995258e-06,
      "loss": 0.531,
      "step": 10024
    },
    {
      "epoch": 1.7392435808466344,
      "grad_norm": 0.9693120718002319,
      "learning_rate": 2.5648309552289875e-06,
      "loss": 0.5636,
      "step": 10025
    },
    {
      "epoch": 1.7394170714781403,
      "grad_norm": 0.7685256600379944,
      "learning_rate": 2.561493907834405e-06,
      "loss": 0.45,
      "step": 10026
    },
    {
      "epoch": 1.7395905621096461,
      "grad_norm": 0.6334861516952515,
      "learning_rate": 2.5581588841987693e-06,
      "loss": 0.5999,
      "step": 10027
    },
    {
      "epoch": 1.739764052741152,
      "grad_norm": 1.0231965780258179,
      "learning_rate": 2.5548258847091266e-06,
      "loss": 0.464,
      "step": 10028
    },
    {
      "epoch": 1.7399375433726578,
      "grad_norm": 0.672338604927063,
      "learning_rate": 2.551494909752261e-06,
      "loss": 0.5079,
      "step": 10029
    },
    {
      "epoch": 1.7401110340041637,
      "grad_norm": 0.807580828666687,
      "learning_rate": 2.548165959714748e-06,
      "loss": 0.3598,
      "step": 10030
    },
    {
      "epoch": 1.7402845246356695,
      "grad_norm": 0.6821241974830627,
      "learning_rate": 2.544839034982909e-06,
      "loss": 0.366,
      "step": 10031
    },
    {
      "epoch": 1.7404580152671756,
      "grad_norm": 0.8526586890220642,
      "learning_rate": 2.54151413594284e-06,
      "loss": 0.4384,
      "step": 10032
    },
    {
      "epoch": 1.7406315058986814,
      "grad_norm": 0.8350969552993774,
      "learning_rate": 2.538191262980394e-06,
      "loss": 0.3817,
      "step": 10033
    },
    {
      "epoch": 1.7408049965301875,
      "grad_norm": 0.7892290949821472,
      "learning_rate": 2.534870416481208e-06,
      "loss": 0.5518,
      "step": 10034
    },
    {
      "epoch": 1.7409784871616933,
      "grad_norm": 0.6795223951339722,
      "learning_rate": 2.5315515968306503e-06,
      "loss": 0.4832,
      "step": 10035
    },
    {
      "epoch": 1.7411519777931992,
      "grad_norm": 1.3044379949569702,
      "learning_rate": 2.5282348044138915e-06,
      "loss": 0.3311,
      "step": 10036
    },
    {
      "epoch": 1.741325468424705,
      "grad_norm": 1.0554447174072266,
      "learning_rate": 2.5249200396158414e-06,
      "loss": 0.4771,
      "step": 10037
    },
    {
      "epoch": 1.7414989590562109,
      "grad_norm": 0.812095046043396,
      "learning_rate": 2.521607302821183e-06,
      "loss": 0.4941,
      "step": 10038
    },
    {
      "epoch": 1.7416724496877167,
      "grad_norm": 0.6278399229049683,
      "learning_rate": 2.51829659441436e-06,
      "loss": 0.5004,
      "step": 10039
    },
    {
      "epoch": 1.7418459403192228,
      "grad_norm": 0.9077810645103455,
      "learning_rate": 2.514987914779592e-06,
      "loss": 0.404,
      "step": 10040
    },
    {
      "epoch": 1.7420194309507286,
      "grad_norm": 0.7014707326889038,
      "learning_rate": 2.5116812643008494e-06,
      "loss": 0.2863,
      "step": 10041
    },
    {
      "epoch": 1.7421929215822347,
      "grad_norm": 0.9283987283706665,
      "learning_rate": 2.5083766433618695e-06,
      "loss": 0.352,
      "step": 10042
    },
    {
      "epoch": 1.7423664122137406,
      "grad_norm": 0.7386503219604492,
      "learning_rate": 2.5050740523461682e-06,
      "loss": 0.4692,
      "step": 10043
    },
    {
      "epoch": 1.7425399028452464,
      "grad_norm": 0.58295738697052,
      "learning_rate": 2.5017734916370073e-06,
      "loss": 0.6099,
      "step": 10044
    },
    {
      "epoch": 1.7427133934767522,
      "grad_norm": 0.9074196219444275,
      "learning_rate": 2.498474961617421e-06,
      "loss": 0.428,
      "step": 10045
    },
    {
      "epoch": 1.742886884108258,
      "grad_norm": 0.7043365240097046,
      "learning_rate": 2.495178462670207e-06,
      "loss": 0.4751,
      "step": 10046
    },
    {
      "epoch": 1.743060374739764,
      "grad_norm": 0.8902522325515747,
      "learning_rate": 2.4918839951779374e-06,
      "loss": 0.3989,
      "step": 10047
    },
    {
      "epoch": 1.7432338653712698,
      "grad_norm": 1.1538342237472534,
      "learning_rate": 2.4885915595229215e-06,
      "loss": 0.3883,
      "step": 10048
    },
    {
      "epoch": 1.7434073560027759,
      "grad_norm": 0.8350592255592346,
      "learning_rate": 2.4853011560872653e-06,
      "loss": 0.4956,
      "step": 10049
    },
    {
      "epoch": 1.7435808466342817,
      "grad_norm": 0.6407992839813232,
      "learning_rate": 2.4820127852528163e-06,
      "loss": 0.5217,
      "step": 10050
    },
    {
      "epoch": 1.7437543372657878,
      "grad_norm": 1.2211695909500122,
      "learning_rate": 2.4787264474011984e-06,
      "loss": 0.3939,
      "step": 10051
    },
    {
      "epoch": 1.7439278278972936,
      "grad_norm": 0.8848746418952942,
      "learning_rate": 2.4754421429137887e-06,
      "loss": 0.3485,
      "step": 10052
    },
    {
      "epoch": 1.7441013185287995,
      "grad_norm": 0.84989333152771,
      "learning_rate": 2.4721598721717465e-06,
      "loss": 0.4356,
      "step": 10053
    },
    {
      "epoch": 1.7442748091603053,
      "grad_norm": 0.8318312764167786,
      "learning_rate": 2.468879635555965e-06,
      "loss": 0.4052,
      "step": 10054
    },
    {
      "epoch": 1.7444482997918112,
      "grad_norm": 0.8529197573661804,
      "learning_rate": 2.4656014334471357e-06,
      "loss": 0.5172,
      "step": 10055
    },
    {
      "epoch": 1.744621790423317,
      "grad_norm": 0.7869355082511902,
      "learning_rate": 2.462325266225687e-06,
      "loss": 0.5625,
      "step": 10056
    },
    {
      "epoch": 1.744795281054823,
      "grad_norm": 0.6938173770904541,
      "learning_rate": 2.4590511342718348e-06,
      "loss": 0.448,
      "step": 10057
    },
    {
      "epoch": 1.744968771686329,
      "grad_norm": 1.1824657917022705,
      "learning_rate": 2.455779037965529e-06,
      "loss": 0.4856,
      "step": 10058
    },
    {
      "epoch": 1.745142262317835,
      "grad_norm": 1.080678939819336,
      "learning_rate": 2.452508977686514e-06,
      "loss": 0.3717,
      "step": 10059
    },
    {
      "epoch": 1.7453157529493408,
      "grad_norm": 0.6940746903419495,
      "learning_rate": 2.4492409538142803e-06,
      "loss": 0.4337,
      "step": 10060
    },
    {
      "epoch": 1.7454892435808467,
      "grad_norm": 0.7876454591751099,
      "learning_rate": 2.445974966728082e-06,
      "loss": 0.4869,
      "step": 10061
    },
    {
      "epoch": 1.7456627342123525,
      "grad_norm": 0.839835524559021,
      "learning_rate": 2.44271101680694e-06,
      "loss": 0.4744,
      "step": 10062
    },
    {
      "epoch": 1.7458362248438584,
      "grad_norm": 2.489712953567505,
      "learning_rate": 2.4394491044296474e-06,
      "loss": 0.4675,
      "step": 10063
    },
    {
      "epoch": 1.7460097154753642,
      "grad_norm": 0.7875381112098694,
      "learning_rate": 2.436189229974748e-06,
      "loss": 0.5355,
      "step": 10064
    },
    {
      "epoch": 1.7461832061068703,
      "grad_norm": 0.8252909183502197,
      "learning_rate": 2.4329313938205546e-06,
      "loss": 0.3831,
      "step": 10065
    },
    {
      "epoch": 1.7463566967383761,
      "grad_norm": 0.6469065546989441,
      "learning_rate": 2.4296755963451424e-06,
      "loss": 0.4752,
      "step": 10066
    },
    {
      "epoch": 1.7465301873698822,
      "grad_norm": 1.6683177947998047,
      "learning_rate": 2.426421837926345e-06,
      "loss": 0.4642,
      "step": 10067
    },
    {
      "epoch": 1.746703678001388,
      "grad_norm": 0.6783648729324341,
      "learning_rate": 2.423170118941778e-06,
      "loss": 0.351,
      "step": 10068
    },
    {
      "epoch": 1.7468771686328939,
      "grad_norm": 0.6830014586448669,
      "learning_rate": 2.4199204397687968e-06,
      "loss": 0.4288,
      "step": 10069
    },
    {
      "epoch": 1.7470506592643997,
      "grad_norm": 0.9472129940986633,
      "learning_rate": 2.4166728007845364e-06,
      "loss": 0.5343,
      "step": 10070
    },
    {
      "epoch": 1.7472241498959056,
      "grad_norm": 0.7599621415138245,
      "learning_rate": 2.413427202365879e-06,
      "loss": 0.432,
      "step": 10071
    },
    {
      "epoch": 1.7473976405274114,
      "grad_norm": 0.8760243058204651,
      "learning_rate": 2.4101836448894924e-06,
      "loss": 0.3851,
      "step": 10072
    },
    {
      "epoch": 1.7475711311589173,
      "grad_norm": 0.8620213270187378,
      "learning_rate": 2.406942128731791e-06,
      "loss": 0.4406,
      "step": 10073
    },
    {
      "epoch": 1.7477446217904233,
      "grad_norm": 1.1500178575515747,
      "learning_rate": 2.4037026542689555e-06,
      "loss": 0.4408,
      "step": 10074
    },
    {
      "epoch": 1.7479181124219292,
      "grad_norm": 0.8120666742324829,
      "learning_rate": 2.4004652218769244e-06,
      "loss": 0.5251,
      "step": 10075
    },
    {
      "epoch": 1.7480916030534353,
      "grad_norm": 0.8861424326896667,
      "learning_rate": 2.3972298319314224e-06,
      "loss": 0.4749,
      "step": 10076
    },
    {
      "epoch": 1.748265093684941,
      "grad_norm": 0.7357455492019653,
      "learning_rate": 2.393996484807901e-06,
      "loss": 0.3649,
      "step": 10077
    },
    {
      "epoch": 1.748438584316447,
      "grad_norm": 0.6997224688529968,
      "learning_rate": 2.3907651808816067e-06,
      "loss": 0.5558,
      "step": 10078
    },
    {
      "epoch": 1.7486120749479528,
      "grad_norm": 0.9641300439834595,
      "learning_rate": 2.3875359205275307e-06,
      "loss": 0.4082,
      "step": 10079
    },
    {
      "epoch": 1.7487855655794586,
      "grad_norm": 0.9489569067955017,
      "learning_rate": 2.384308704120435e-06,
      "loss": 0.4228,
      "step": 10080
    },
    {
      "epoch": 1.7489590562109645,
      "grad_norm": 0.8642375469207764,
      "learning_rate": 2.381083532034836e-06,
      "loss": 0.4338,
      "step": 10081
    },
    {
      "epoch": 1.7491325468424705,
      "grad_norm": 0.7411103248596191,
      "learning_rate": 2.3778604046450313e-06,
      "loss": 0.3862,
      "step": 10082
    },
    {
      "epoch": 1.7493060374739764,
      "grad_norm": 0.6525177359580994,
      "learning_rate": 2.374639322325052e-06,
      "loss": 0.4488,
      "step": 10083
    },
    {
      "epoch": 1.7494795281054825,
      "grad_norm": 0.8858721256256104,
      "learning_rate": 2.371420285448722e-06,
      "loss": 0.4323,
      "step": 10084
    },
    {
      "epoch": 1.7496530187369883,
      "grad_norm": 1.2127820253372192,
      "learning_rate": 2.368203294389606e-06,
      "loss": 0.3624,
      "step": 10085
    },
    {
      "epoch": 1.7498265093684942,
      "grad_norm": 0.7921899557113647,
      "learning_rate": 2.364988349521049e-06,
      "loss": 0.4139,
      "step": 10086
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9015601873397827,
      "learning_rate": 2.3617754512161353e-06,
      "loss": 0.4285,
      "step": 10087
    },
    {
      "epoch": 1.7501734906315058,
      "grad_norm": 0.9655046463012695,
      "learning_rate": 2.358564599847737e-06,
      "loss": 0.549,
      "step": 10088
    },
    {
      "epoch": 1.7503469812630117,
      "grad_norm": 0.7301394939422607,
      "learning_rate": 2.3553557957884744e-06,
      "loss": 0.3454,
      "step": 10089
    },
    {
      "epoch": 1.7505204718945175,
      "grad_norm": 1.331050157546997,
      "learning_rate": 2.352149039410727e-06,
      "loss": 0.4889,
      "step": 10090
    },
    {
      "epoch": 1.7506939625260236,
      "grad_norm": 0.8005419373512268,
      "learning_rate": 2.348944331086651e-06,
      "loss": 0.3352,
      "step": 10091
    },
    {
      "epoch": 1.7508674531575295,
      "grad_norm": 0.8455546498298645,
      "learning_rate": 2.345741671188153e-06,
      "loss": 0.3281,
      "step": 10092
    },
    {
      "epoch": 1.7510409437890355,
      "grad_norm": 1.028125524520874,
      "learning_rate": 2.342541060086907e-06,
      "loss": 0.4072,
      "step": 10093
    },
    {
      "epoch": 1.7512144344205414,
      "grad_norm": 0.8474204540252686,
      "learning_rate": 2.33934249815434e-06,
      "loss": 0.4888,
      "step": 10094
    },
    {
      "epoch": 1.7513879250520472,
      "grad_norm": 0.8342925906181335,
      "learning_rate": 2.336145985761664e-06,
      "loss": 0.5154,
      "step": 10095
    },
    {
      "epoch": 1.751561415683553,
      "grad_norm": 1.0096440315246582,
      "learning_rate": 2.3329515232798207e-06,
      "loss": 0.4929,
      "step": 10096
    },
    {
      "epoch": 1.751734906315059,
      "grad_norm": 0.7327522039413452,
      "learning_rate": 2.3297591110795437e-06,
      "loss": 0.4319,
      "step": 10097
    },
    {
      "epoch": 1.7519083969465647,
      "grad_norm": 0.848007082939148,
      "learning_rate": 2.3265687495313106e-06,
      "loss": 0.3483,
      "step": 10098
    },
    {
      "epoch": 1.7520818875780708,
      "grad_norm": 0.9690507650375366,
      "learning_rate": 2.323380439005367e-06,
      "loss": 0.3968,
      "step": 10099
    },
    {
      "epoch": 1.7522553782095767,
      "grad_norm": 0.6661014556884766,
      "learning_rate": 2.3201941798717176e-06,
      "loss": 0.4374,
      "step": 10100
    },
    {
      "epoch": 1.7524288688410827,
      "grad_norm": 0.764918863773346,
      "learning_rate": 2.3170099725001393e-06,
      "loss": 0.4493,
      "step": 10101
    },
    {
      "epoch": 1.7526023594725886,
      "grad_norm": 0.7615021467208862,
      "learning_rate": 2.313827817260159e-06,
      "loss": 0.4489,
      "step": 10102
    },
    {
      "epoch": 1.7527758501040944,
      "grad_norm": 0.7164032459259033,
      "learning_rate": 2.310647714521068e-06,
      "loss": 0.4794,
      "step": 10103
    },
    {
      "epoch": 1.7529493407356003,
      "grad_norm": 1.0460331439971924,
      "learning_rate": 2.307469664651918e-06,
      "loss": 0.4767,
      "step": 10104
    },
    {
      "epoch": 1.7531228313671061,
      "grad_norm": 0.9769534468650818,
      "learning_rate": 2.304293668021538e-06,
      "loss": 0.4417,
      "step": 10105
    },
    {
      "epoch": 1.753296321998612,
      "grad_norm": 0.8242330551147461,
      "learning_rate": 2.3011197249984886e-06,
      "loss": 0.4003,
      "step": 10106
    },
    {
      "epoch": 1.7534698126301178,
      "grad_norm": 1.248779058456421,
      "learning_rate": 2.2979478359511244e-06,
      "loss": 0.4025,
      "step": 10107
    },
    {
      "epoch": 1.7536433032616239,
      "grad_norm": 1.0432181358337402,
      "learning_rate": 2.2947780012475396e-06,
      "loss": 0.449,
      "step": 10108
    },
    {
      "epoch": 1.7538167938931297,
      "grad_norm": 0.8337266445159912,
      "learning_rate": 2.291610221255598e-06,
      "loss": 0.4597,
      "step": 10109
    },
    {
      "epoch": 1.7539902845246358,
      "grad_norm": 0.8977551460266113,
      "learning_rate": 2.2884444963429188e-06,
      "loss": 0.4958,
      "step": 10110
    },
    {
      "epoch": 1.7541637751561416,
      "grad_norm": 0.7688846588134766,
      "learning_rate": 2.285280826876901e-06,
      "loss": 0.3867,
      "step": 10111
    },
    {
      "epoch": 1.7543372657876475,
      "grad_norm": 0.837756872177124,
      "learning_rate": 2.282119213224683e-06,
      "loss": 0.3563,
      "step": 10112
    },
    {
      "epoch": 1.7545107564191533,
      "grad_norm": 1.8376587629318237,
      "learning_rate": 2.2789596557531766e-06,
      "loss": 0.3823,
      "step": 10113
    },
    {
      "epoch": 1.7546842470506592,
      "grad_norm": 0.7061861157417297,
      "learning_rate": 2.2758021548290478e-06,
      "loss": 0.522,
      "step": 10114
    },
    {
      "epoch": 1.754857737682165,
      "grad_norm": 0.6778156757354736,
      "learning_rate": 2.2726467108187335e-06,
      "loss": 0.5293,
      "step": 10115
    },
    {
      "epoch": 1.755031228313671,
      "grad_norm": 0.6336186528205872,
      "learning_rate": 2.2694933240884277e-06,
      "loss": 0.5385,
      "step": 10116
    },
    {
      "epoch": 1.755204718945177,
      "grad_norm": 0.6999941468238831,
      "learning_rate": 2.26634199500408e-06,
      "loss": 0.3204,
      "step": 10117
    },
    {
      "epoch": 1.755378209576683,
      "grad_norm": 1.1926878690719604,
      "learning_rate": 2.263192723931409e-06,
      "loss": 0.3995,
      "step": 10118
    },
    {
      "epoch": 1.7555517002081888,
      "grad_norm": 1.460123062133789,
      "learning_rate": 2.2600455112358843e-06,
      "loss": 0.4139,
      "step": 10119
    },
    {
      "epoch": 1.7557251908396947,
      "grad_norm": 0.7850372195243835,
      "learning_rate": 2.2569003572827543e-06,
      "loss": 0.3902,
      "step": 10120
    },
    {
      "epoch": 1.7558986814712005,
      "grad_norm": 0.8340042233467102,
      "learning_rate": 2.2537572624370107e-06,
      "loss": 0.3344,
      "step": 10121
    },
    {
      "epoch": 1.7560721721027064,
      "grad_norm": 0.9595273733139038,
      "learning_rate": 2.250616227063418e-06,
      "loss": 0.3964,
      "step": 10122
    },
    {
      "epoch": 1.7562456627342122,
      "grad_norm": 1.0658237934112549,
      "learning_rate": 2.247477251526489e-06,
      "loss": 0.3994,
      "step": 10123
    },
    {
      "epoch": 1.7564191533657183,
      "grad_norm": 0.9107815027236938,
      "learning_rate": 2.244340336190518e-06,
      "loss": 0.4043,
      "step": 10124
    },
    {
      "epoch": 1.7565926439972241,
      "grad_norm": 1.0793975591659546,
      "learning_rate": 2.2412054814195326e-06,
      "loss": 0.4739,
      "step": 10125
    },
    {
      "epoch": 1.7567661346287302,
      "grad_norm": 0.716812789440155,
      "learning_rate": 2.2380726875773507e-06,
      "loss": 0.4882,
      "step": 10126
    },
    {
      "epoch": 1.756939625260236,
      "grad_norm": 0.8273600339889526,
      "learning_rate": 2.2349419550275275e-06,
      "loss": 0.5278,
      "step": 10127
    },
    {
      "epoch": 1.757113115891742,
      "grad_norm": 0.6313033103942871,
      "learning_rate": 2.2318132841333906e-06,
      "loss": 0.446,
      "step": 10128
    },
    {
      "epoch": 1.7572866065232478,
      "grad_norm": 0.9580927491188049,
      "learning_rate": 2.228686675258025e-06,
      "loss": 0.4038,
      "step": 10129
    },
    {
      "epoch": 1.7574600971547536,
      "grad_norm": 0.8508036136627197,
      "learning_rate": 2.2255621287642805e-06,
      "loss": 0.519,
      "step": 10130
    },
    {
      "epoch": 1.7576335877862594,
      "grad_norm": 0.7911813855171204,
      "learning_rate": 2.2224396450147623e-06,
      "loss": 0.4701,
      "step": 10131
    },
    {
      "epoch": 1.7578070784177653,
      "grad_norm": 0.8794583678245544,
      "learning_rate": 2.2193192243718385e-06,
      "loss": 0.4142,
      "step": 10132
    },
    {
      "epoch": 1.7579805690492714,
      "grad_norm": 0.7201571464538574,
      "learning_rate": 2.216200867197633e-06,
      "loss": 0.5037,
      "step": 10133
    },
    {
      "epoch": 1.7581540596807772,
      "grad_norm": 0.6206756234169006,
      "learning_rate": 2.2130845738540475e-06,
      "loss": 0.5522,
      "step": 10134
    },
    {
      "epoch": 1.7583275503122833,
      "grad_norm": 1.1410282850265503,
      "learning_rate": 2.2099703447027142e-06,
      "loss": 0.4435,
      "step": 10135
    },
    {
      "epoch": 1.7585010409437891,
      "grad_norm": 0.754542350769043,
      "learning_rate": 2.2068581801050557e-06,
      "loss": 0.3986,
      "step": 10136
    },
    {
      "epoch": 1.758674531575295,
      "grad_norm": 1.0550987720489502,
      "learning_rate": 2.203748080422239e-06,
      "loss": 0.3549,
      "step": 10137
    },
    {
      "epoch": 1.7588480222068008,
      "grad_norm": 1.461365818977356,
      "learning_rate": 2.2006400460151923e-06,
      "loss": 0.3439,
      "step": 10138
    },
    {
      "epoch": 1.7590215128383067,
      "grad_norm": 0.8991324305534363,
      "learning_rate": 2.1975340772446095e-06,
      "loss": 0.3806,
      "step": 10139
    },
    {
      "epoch": 1.7591950034698125,
      "grad_norm": 0.8629834055900574,
      "learning_rate": 2.1944301744709428e-06,
      "loss": 0.4394,
      "step": 10140
    },
    {
      "epoch": 1.7593684941013186,
      "grad_norm": 0.7667680978775024,
      "learning_rate": 2.1913283380544013e-06,
      "loss": 0.5836,
      "step": 10141
    },
    {
      "epoch": 1.7595419847328244,
      "grad_norm": 0.8159300088882446,
      "learning_rate": 2.1882285683549555e-06,
      "loss": 0.4668,
      "step": 10142
    },
    {
      "epoch": 1.7597154753643305,
      "grad_norm": 0.8265716433525085,
      "learning_rate": 2.185130865732341e-06,
      "loss": 0.4355,
      "step": 10143
    },
    {
      "epoch": 1.7598889659958363,
      "grad_norm": 1.1387720108032227,
      "learning_rate": 2.1820352305460492e-06,
      "loss": 0.468,
      "step": 10144
    },
    {
      "epoch": 1.7600624566273422,
      "grad_norm": 0.7151750326156616,
      "learning_rate": 2.1789416631553294e-06,
      "loss": 0.5829,
      "step": 10145
    },
    {
      "epoch": 1.760235947258848,
      "grad_norm": 0.6007384061813354,
      "learning_rate": 2.1758501639191908e-06,
      "loss": 0.5656,
      "step": 10146
    },
    {
      "epoch": 1.7604094378903539,
      "grad_norm": 0.7706556916236877,
      "learning_rate": 2.1727607331964197e-06,
      "loss": 0.4412,
      "step": 10147
    },
    {
      "epoch": 1.7605829285218597,
      "grad_norm": 0.8431768417358398,
      "learning_rate": 2.169673371345531e-06,
      "loss": 0.4158,
      "step": 10148
    },
    {
      "epoch": 1.7607564191533656,
      "grad_norm": 0.8578419089317322,
      "learning_rate": 2.166588078724827e-06,
      "loss": 0.4647,
      "step": 10149
    },
    {
      "epoch": 1.7609299097848716,
      "grad_norm": 1.0609016418457031,
      "learning_rate": 2.1635048556923555e-06,
      "loss": 0.4402,
      "step": 10150
    },
    {
      "epoch": 1.7611034004163775,
      "grad_norm": 0.7982937693595886,
      "learning_rate": 2.1604237026059296e-06,
      "loss": 0.3978,
      "step": 10151
    },
    {
      "epoch": 1.7612768910478835,
      "grad_norm": 0.7780153155326843,
      "learning_rate": 2.1573446198231185e-06,
      "loss": 0.4018,
      "step": 10152
    },
    {
      "epoch": 1.7614503816793894,
      "grad_norm": 0.8052198886871338,
      "learning_rate": 2.154267607701259e-06,
      "loss": 0.4348,
      "step": 10153
    },
    {
      "epoch": 1.7616238723108952,
      "grad_norm": 0.752699613571167,
      "learning_rate": 2.1511926665974324e-06,
      "loss": 0.4047,
      "step": 10154
    },
    {
      "epoch": 1.761797362942401,
      "grad_norm": 0.7057964205741882,
      "learning_rate": 2.1481197968684998e-06,
      "loss": 0.5571,
      "step": 10155
    },
    {
      "epoch": 1.761970853573907,
      "grad_norm": 0.8463672995567322,
      "learning_rate": 2.1450489988710644e-06,
      "loss": 0.4365,
      "step": 10156
    },
    {
      "epoch": 1.7621443442054128,
      "grad_norm": 0.8756439685821533,
      "learning_rate": 2.1419802729614993e-06,
      "loss": 0.4567,
      "step": 10157
    },
    {
      "epoch": 1.7623178348369188,
      "grad_norm": 0.7664997577667236,
      "learning_rate": 2.138913619495928e-06,
      "loss": 0.3491,
      "step": 10158
    },
    {
      "epoch": 1.7624913254684247,
      "grad_norm": 1.0745248794555664,
      "learning_rate": 2.1358490388302466e-06,
      "loss": 0.3785,
      "step": 10159
    },
    {
      "epoch": 1.7626648160999308,
      "grad_norm": 1.0532866716384888,
      "learning_rate": 2.1327865313201015e-06,
      "loss": 0.449,
      "step": 10160
    },
    {
      "epoch": 1.7628383067314366,
      "grad_norm": 1.0335370302200317,
      "learning_rate": 2.1297260973208987e-06,
      "loss": 0.4962,
      "step": 10161
    },
    {
      "epoch": 1.7630117973629424,
      "grad_norm": 0.7184655666351318,
      "learning_rate": 2.1266677371877996e-06,
      "loss": 0.4698,
      "step": 10162
    },
    {
      "epoch": 1.7631852879944483,
      "grad_norm": 0.926429271697998,
      "learning_rate": 2.1236114512757423e-06,
      "loss": 0.5233,
      "step": 10163
    },
    {
      "epoch": 1.7633587786259541,
      "grad_norm": 0.7761516571044922,
      "learning_rate": 2.120557239939405e-06,
      "loss": 0.418,
      "step": 10164
    },
    {
      "epoch": 1.76353226925746,
      "grad_norm": 0.665821373462677,
      "learning_rate": 2.1175051035332285e-06,
      "loss": 0.5332,
      "step": 10165
    },
    {
      "epoch": 1.7637057598889658,
      "grad_norm": 1.3791216611862183,
      "learning_rate": 2.114455042411432e-06,
      "loss": 0.3474,
      "step": 10166
    },
    {
      "epoch": 1.763879250520472,
      "grad_norm": 0.8793630003929138,
      "learning_rate": 2.111407056927959e-06,
      "loss": 0.4183,
      "step": 10167
    },
    {
      "epoch": 1.7640527411519777,
      "grad_norm": 0.9198608994483948,
      "learning_rate": 2.108361147436546e-06,
      "loss": 0.3718,
      "step": 10168
    },
    {
      "epoch": 1.7642262317834838,
      "grad_norm": 1.2578948736190796,
      "learning_rate": 2.105317314290671e-06,
      "loss": 0.4615,
      "step": 10169
    },
    {
      "epoch": 1.7643997224149897,
      "grad_norm": 0.6621404886245728,
      "learning_rate": 2.1022755578435715e-06,
      "loss": 0.4683,
      "step": 10170
    },
    {
      "epoch": 1.7645732130464955,
      "grad_norm": 0.7790876030921936,
      "learning_rate": 2.099235878448247e-06,
      "loss": 0.4091,
      "step": 10171
    },
    {
      "epoch": 1.7647467036780013,
      "grad_norm": 1.2456015348434448,
      "learning_rate": 2.0961982764574597e-06,
      "loss": 0.3719,
      "step": 10172
    },
    {
      "epoch": 1.7649201943095072,
      "grad_norm": 0.8161839246749878,
      "learning_rate": 2.093162752223723e-06,
      "loss": 0.4462,
      "step": 10173
    },
    {
      "epoch": 1.765093684941013,
      "grad_norm": 1.1468185186386108,
      "learning_rate": 2.0901293060993154e-06,
      "loss": 0.4938,
      "step": 10174
    },
    {
      "epoch": 1.765267175572519,
      "grad_norm": 0.8534411191940308,
      "learning_rate": 2.087097938436269e-06,
      "loss": 0.437,
      "step": 10175
    },
    {
      "epoch": 1.765440666204025,
      "grad_norm": 0.6585747599601746,
      "learning_rate": 2.0840686495863837e-06,
      "loss": 0.4927,
      "step": 10176
    },
    {
      "epoch": 1.765614156835531,
      "grad_norm": 0.6890860199928284,
      "learning_rate": 2.0810414399012034e-06,
      "loss": 0.4344,
      "step": 10177
    },
    {
      "epoch": 1.7657876474670369,
      "grad_norm": 0.8579777479171753,
      "learning_rate": 2.078016309732047e-06,
      "loss": 0.444,
      "step": 10178
    },
    {
      "epoch": 1.7659611380985427,
      "grad_norm": 0.7780200839042664,
      "learning_rate": 2.0749932594299804e-06,
      "loss": 0.4333,
      "step": 10179
    },
    {
      "epoch": 1.7661346287300486,
      "grad_norm": 1.21971595287323,
      "learning_rate": 2.0719722893458317e-06,
      "loss": 0.347,
      "step": 10180
    },
    {
      "epoch": 1.7663081193615544,
      "grad_norm": 0.831279456615448,
      "learning_rate": 2.0689533998301868e-06,
      "loss": 0.426,
      "step": 10181
    },
    {
      "epoch": 1.7664816099930603,
      "grad_norm": 1.502861738204956,
      "learning_rate": 2.0659365912333972e-06,
      "loss": 0.4578,
      "step": 10182
    },
    {
      "epoch": 1.7666551006245663,
      "grad_norm": 0.8384041786193848,
      "learning_rate": 2.0629218639055625e-06,
      "loss": 0.4526,
      "step": 10183
    },
    {
      "epoch": 1.7668285912560722,
      "grad_norm": 0.7056971192359924,
      "learning_rate": 2.0599092181965474e-06,
      "loss": 0.3987,
      "step": 10184
    },
    {
      "epoch": 1.7670020818875782,
      "grad_norm": 0.7954768538475037,
      "learning_rate": 2.05689865445597e-06,
      "loss": 0.5176,
      "step": 10185
    },
    {
      "epoch": 1.767175572519084,
      "grad_norm": 0.6894344687461853,
      "learning_rate": 2.0538901730332128e-06,
      "loss": 0.4622,
      "step": 10186
    },
    {
      "epoch": 1.76734906315059,
      "grad_norm": 0.7572733163833618,
      "learning_rate": 2.0508837742774125e-06,
      "loss": 0.3764,
      "step": 10187
    },
    {
      "epoch": 1.7675225537820958,
      "grad_norm": 0.7938310503959656,
      "learning_rate": 2.047879458537465e-06,
      "loss": 0.4937,
      "step": 10188
    },
    {
      "epoch": 1.7676960444136016,
      "grad_norm": 0.6096211671829224,
      "learning_rate": 2.0448772261620254e-06,
      "loss": 0.5094,
      "step": 10189
    },
    {
      "epoch": 1.7678695350451075,
      "grad_norm": 0.9245356321334839,
      "learning_rate": 2.0418770774995034e-06,
      "loss": 0.536,
      "step": 10190
    },
    {
      "epoch": 1.7680430256766133,
      "grad_norm": 0.82867830991745,
      "learning_rate": 2.038879012898074e-06,
      "loss": 0.482,
      "step": 10191
    },
    {
      "epoch": 1.7682165163081194,
      "grad_norm": 1.137810468673706,
      "learning_rate": 2.0358830327056633e-06,
      "loss": 0.4565,
      "step": 10192
    },
    {
      "epoch": 1.7683900069396252,
      "grad_norm": 0.9972358345985413,
      "learning_rate": 2.03288913726996e-06,
      "loss": 0.4631,
      "step": 10193
    },
    {
      "epoch": 1.7685634975711313,
      "grad_norm": 0.6278783082962036,
      "learning_rate": 2.0298973269384037e-06,
      "loss": 0.4972,
      "step": 10194
    },
    {
      "epoch": 1.7687369882026371,
      "grad_norm": 0.6001548171043396,
      "learning_rate": 2.02690760205821e-06,
      "loss": 0.559,
      "step": 10195
    },
    {
      "epoch": 1.768910478834143,
      "grad_norm": 0.8958408236503601,
      "learning_rate": 2.023919962976324e-06,
      "loss": 0.4073,
      "step": 10196
    },
    {
      "epoch": 1.7690839694656488,
      "grad_norm": 0.938954770565033,
      "learning_rate": 2.020934410039477e-06,
      "loss": 0.4994,
      "step": 10197
    },
    {
      "epoch": 1.7692574600971547,
      "grad_norm": 0.6814796924591064,
      "learning_rate": 2.0179509435941403e-06,
      "loss": 0.5527,
      "step": 10198
    },
    {
      "epoch": 1.7694309507286605,
      "grad_norm": 0.8747894167900085,
      "learning_rate": 2.0149695639865507e-06,
      "loss": 0.3686,
      "step": 10199
    },
    {
      "epoch": 1.7696044413601666,
      "grad_norm": 1.2067766189575195,
      "learning_rate": 2.011990271562696e-06,
      "loss": 0.3603,
      "step": 10200
    },
    {
      "epoch": 1.7697779319916724,
      "grad_norm": 0.8373238444328308,
      "learning_rate": 2.0090130666683347e-06,
      "loss": 0.3881,
      "step": 10201
    },
    {
      "epoch": 1.7699514226231785,
      "grad_norm": 1.303937554359436,
      "learning_rate": 2.006037949648971e-06,
      "loss": 0.5143,
      "step": 10202
    },
    {
      "epoch": 1.7701249132546844,
      "grad_norm": 0.6505475640296936,
      "learning_rate": 2.0030649208498685e-06,
      "loss": 0.5986,
      "step": 10203
    },
    {
      "epoch": 1.7702984038861902,
      "grad_norm": 1.174577236175537,
      "learning_rate": 2.000093980616051e-06,
      "loss": 0.4208,
      "step": 10204
    },
    {
      "epoch": 1.770471894517696,
      "grad_norm": 1.0686638355255127,
      "learning_rate": 1.9971251292923076e-06,
      "loss": 0.4965,
      "step": 10205
    },
    {
      "epoch": 1.770645385149202,
      "grad_norm": 0.8603082299232483,
      "learning_rate": 1.9941583672231624e-06,
      "loss": 0.4551,
      "step": 10206
    },
    {
      "epoch": 1.7708188757807077,
      "grad_norm": 0.8211946487426758,
      "learning_rate": 1.991193694752924e-06,
      "loss": 0.4965,
      "step": 10207
    },
    {
      "epoch": 1.7709923664122136,
      "grad_norm": 0.8139055967330933,
      "learning_rate": 1.9882311122256425e-06,
      "loss": 0.4799,
      "step": 10208
    },
    {
      "epoch": 1.7711658570437196,
      "grad_norm": 0.8746566772460938,
      "learning_rate": 1.985270619985127e-06,
      "loss": 0.447,
      "step": 10209
    },
    {
      "epoch": 1.7713393476752255,
      "grad_norm": 1.1020523309707642,
      "learning_rate": 1.9823122183749443e-06,
      "loss": 0.4387,
      "step": 10210
    },
    {
      "epoch": 1.7715128383067316,
      "grad_norm": 0.7173384428024292,
      "learning_rate": 1.979355907738427e-06,
      "loss": 0.5352,
      "step": 10211
    },
    {
      "epoch": 1.7716863289382374,
      "grad_norm": 0.7200915217399597,
      "learning_rate": 1.9764016884186545e-06,
      "loss": 0.4022,
      "step": 10212
    },
    {
      "epoch": 1.7718598195697433,
      "grad_norm": 0.8327056765556335,
      "learning_rate": 1.973449560758465e-06,
      "loss": 0.4324,
      "step": 10213
    },
    {
      "epoch": 1.772033310201249,
      "grad_norm": 0.7548806667327881,
      "learning_rate": 1.9704995251004622e-06,
      "loss": 0.5532,
      "step": 10214
    },
    {
      "epoch": 1.772206800832755,
      "grad_norm": 0.927376925945282,
      "learning_rate": 1.9675515817869974e-06,
      "loss": 0.4202,
      "step": 10215
    },
    {
      "epoch": 1.7723802914642608,
      "grad_norm": 1.4509769678115845,
      "learning_rate": 1.9646057311601853e-06,
      "loss": 0.4789,
      "step": 10216
    },
    {
      "epoch": 1.7725537820957669,
      "grad_norm": 0.6847125887870789,
      "learning_rate": 1.961661973561888e-06,
      "loss": 0.5554,
      "step": 10217
    },
    {
      "epoch": 1.7727272727272727,
      "grad_norm": 0.7594535946846008,
      "learning_rate": 1.958720309333746e-06,
      "loss": 0.4485,
      "step": 10218
    },
    {
      "epoch": 1.7729007633587788,
      "grad_norm": 0.7194278836250305,
      "learning_rate": 1.9557807388171257e-06,
      "loss": 0.4448,
      "step": 10219
    },
    {
      "epoch": 1.7730742539902846,
      "grad_norm": 0.6663083434104919,
      "learning_rate": 1.952843262353181e-06,
      "loss": 0.4426,
      "step": 10220
    },
    {
      "epoch": 1.7732477446217905,
      "grad_norm": 0.731451153755188,
      "learning_rate": 1.9499078802828044e-06,
      "loss": 0.478,
      "step": 10221
    },
    {
      "epoch": 1.7734212352532963,
      "grad_norm": 1.2679458856582642,
      "learning_rate": 1.946974592946651e-06,
      "loss": 0.4308,
      "step": 10222
    },
    {
      "epoch": 1.7735947258848022,
      "grad_norm": 1.0227187871932983,
      "learning_rate": 1.9440434006851296e-06,
      "loss": 0.4109,
      "step": 10223
    },
    {
      "epoch": 1.773768216516308,
      "grad_norm": 0.8104212880134583,
      "learning_rate": 1.9411143038384163e-06,
      "loss": 0.5212,
      "step": 10224
    },
    {
      "epoch": 1.773941707147814,
      "grad_norm": 0.8245218992233276,
      "learning_rate": 1.938187302746424e-06,
      "loss": 0.4449,
      "step": 10225
    },
    {
      "epoch": 1.77411519777932,
      "grad_norm": 0.6253479719161987,
      "learning_rate": 1.935262397748845e-06,
      "loss": 0.5405,
      "step": 10226
    },
    {
      "epoch": 1.7742886884108258,
      "grad_norm": 0.8679730892181396,
      "learning_rate": 1.932339589185115e-06,
      "loss": 0.46,
      "step": 10227
    },
    {
      "epoch": 1.7744621790423318,
      "grad_norm": 0.9214310646057129,
      "learning_rate": 1.929418877394429e-06,
      "loss": 0.4501,
      "step": 10228
    },
    {
      "epoch": 1.7746356696738377,
      "grad_norm": 0.6874799728393555,
      "learning_rate": 1.9265002627157335e-06,
      "loss": 0.5813,
      "step": 10229
    },
    {
      "epoch": 1.7748091603053435,
      "grad_norm": 0.7295671105384827,
      "learning_rate": 1.923583745487747e-06,
      "loss": 0.4421,
      "step": 10230
    },
    {
      "epoch": 1.7749826509368494,
      "grad_norm": 0.7274503111839294,
      "learning_rate": 1.920669326048932e-06,
      "loss": 0.4229,
      "step": 10231
    },
    {
      "epoch": 1.7751561415683552,
      "grad_norm": 0.7916246056556702,
      "learning_rate": 1.917757004737506e-06,
      "loss": 0.486,
      "step": 10232
    },
    {
      "epoch": 1.775329632199861,
      "grad_norm": 0.7616907954216003,
      "learning_rate": 1.914846781891444e-06,
      "loss": 0.3854,
      "step": 10233
    },
    {
      "epoch": 1.7755031228313671,
      "grad_norm": 0.7439457774162292,
      "learning_rate": 1.9119386578484934e-06,
      "loss": 0.4105,
      "step": 10234
    },
    {
      "epoch": 1.775676613462873,
      "grad_norm": 0.7740806341171265,
      "learning_rate": 1.909032632946137e-06,
      "loss": 0.4697,
      "step": 10235
    },
    {
      "epoch": 1.775850104094379,
      "grad_norm": 0.9047095775604248,
      "learning_rate": 1.906128707521624e-06,
      "loss": 0.4653,
      "step": 10236
    },
    {
      "epoch": 1.776023594725885,
      "grad_norm": 0.6695199608802795,
      "learning_rate": 1.903226881911957e-06,
      "loss": 0.6056,
      "step": 10237
    },
    {
      "epoch": 1.7761970853573907,
      "grad_norm": 0.5723161697387695,
      "learning_rate": 1.900327156453896e-06,
      "loss": 0.4205,
      "step": 10238
    },
    {
      "epoch": 1.7763705759888966,
      "grad_norm": 1.0215060710906982,
      "learning_rate": 1.8974295314839609e-06,
      "loss": 0.5173,
      "step": 10239
    },
    {
      "epoch": 1.7765440666204024,
      "grad_norm": 0.7813283205032349,
      "learning_rate": 1.894534007338422e-06,
      "loss": 0.4249,
      "step": 10240
    },
    {
      "epoch": 1.7767175572519083,
      "grad_norm": 0.8839203119277954,
      "learning_rate": 1.8916405843533092e-06,
      "loss": 0.4543,
      "step": 10241
    },
    {
      "epoch": 1.7768910478834143,
      "grad_norm": 0.9386798739433289,
      "learning_rate": 1.8887492628644022e-06,
      "loss": 0.3719,
      "step": 10242
    },
    {
      "epoch": 1.7770645385149202,
      "grad_norm": 0.8831916451454163,
      "learning_rate": 1.8858600432072527e-06,
      "loss": 0.4041,
      "step": 10243
    },
    {
      "epoch": 1.7772380291464263,
      "grad_norm": 0.710510790348053,
      "learning_rate": 1.8829729257171503e-06,
      "loss": 0.6177,
      "step": 10244
    },
    {
      "epoch": 1.777411519777932,
      "grad_norm": 0.6700908541679382,
      "learning_rate": 1.8800879107291537e-06,
      "loss": 0.5376,
      "step": 10245
    },
    {
      "epoch": 1.777585010409438,
      "grad_norm": 0.991139829158783,
      "learning_rate": 1.8772049985780616e-06,
      "loss": 0.3387,
      "step": 10246
    },
    {
      "epoch": 1.7777585010409438,
      "grad_norm": 0.8388208150863647,
      "learning_rate": 1.8743241895984554e-06,
      "loss": 0.424,
      "step": 10247
    },
    {
      "epoch": 1.7779319916724496,
      "grad_norm": 0.7843843698501587,
      "learning_rate": 1.871445484124641e-06,
      "loss": 0.5238,
      "step": 10248
    },
    {
      "epoch": 1.7781054823039555,
      "grad_norm": 1.037152886390686,
      "learning_rate": 1.8685688824907044e-06,
      "loss": 0.5073,
      "step": 10249
    },
    {
      "epoch": 1.7782789729354613,
      "grad_norm": 0.7034249305725098,
      "learning_rate": 1.8656943850304765e-06,
      "loss": 0.4507,
      "step": 10250
    },
    {
      "epoch": 1.7784524635669674,
      "grad_norm": 0.9017806053161621,
      "learning_rate": 1.8628219920775481e-06,
      "loss": 0.3208,
      "step": 10251
    },
    {
      "epoch": 1.7786259541984732,
      "grad_norm": 1.3028045892715454,
      "learning_rate": 1.8599517039652548e-06,
      "loss": 0.3081,
      "step": 10252
    },
    {
      "epoch": 1.7787994448299793,
      "grad_norm": 1.6372616291046143,
      "learning_rate": 1.8570835210267125e-06,
      "loss": 0.3827,
      "step": 10253
    },
    {
      "epoch": 1.7789729354614852,
      "grad_norm": 0.6452852487564087,
      "learning_rate": 1.8542174435947614e-06,
      "loss": 0.5815,
      "step": 10254
    },
    {
      "epoch": 1.779146426092991,
      "grad_norm": 0.6727705001831055,
      "learning_rate": 1.851353472002022e-06,
      "loss": 0.5663,
      "step": 10255
    },
    {
      "epoch": 1.7793199167244969,
      "grad_norm": 0.8144544363021851,
      "learning_rate": 1.8484916065808622e-06,
      "loss": 0.4896,
      "step": 10256
    },
    {
      "epoch": 1.7794934073560027,
      "grad_norm": 1.128566861152649,
      "learning_rate": 1.8456318476634006e-06,
      "loss": 0.4705,
      "step": 10257
    },
    {
      "epoch": 1.7796668979875085,
      "grad_norm": 0.8338884711265564,
      "learning_rate": 1.8427741955815138e-06,
      "loss": 0.4083,
      "step": 10258
    },
    {
      "epoch": 1.7798403886190146,
      "grad_norm": 0.813132107257843,
      "learning_rate": 1.839918650666841e-06,
      "loss": 0.4037,
      "step": 10259
    },
    {
      "epoch": 1.7800138792505205,
      "grad_norm": 0.926805853843689,
      "learning_rate": 1.8370652132507705e-06,
      "loss": 0.3605,
      "step": 10260
    },
    {
      "epoch": 1.7801873698820265,
      "grad_norm": 0.8099647760391235,
      "learning_rate": 1.8342138836644419e-06,
      "loss": 0.3469,
      "step": 10261
    },
    {
      "epoch": 1.7803608605135324,
      "grad_norm": 0.727428674697876,
      "learning_rate": 1.8313646622387639e-06,
      "loss": 0.3259,
      "step": 10262
    },
    {
      "epoch": 1.7805343511450382,
      "grad_norm": 0.8847869038581848,
      "learning_rate": 1.8285175493043893e-06,
      "loss": 0.4856,
      "step": 10263
    },
    {
      "epoch": 1.780707841776544,
      "grad_norm": 1.1018608808517456,
      "learning_rate": 1.8256725451917233e-06,
      "loss": 0.4294,
      "step": 10264
    },
    {
      "epoch": 1.78088133240805,
      "grad_norm": 0.8565403819084167,
      "learning_rate": 1.822829650230935e-06,
      "loss": 0.4897,
      "step": 10265
    },
    {
      "epoch": 1.7810548230395558,
      "grad_norm": 1.2412939071655273,
      "learning_rate": 1.8199888647519537e-06,
      "loss": 0.3682,
      "step": 10266
    },
    {
      "epoch": 1.7812283136710616,
      "grad_norm": 1.0650684833526611,
      "learning_rate": 1.81715018908444e-06,
      "loss": 0.3518,
      "step": 10267
    },
    {
      "epoch": 1.7814018043025677,
      "grad_norm": 0.8299738764762878,
      "learning_rate": 1.8143136235578374e-06,
      "loss": 0.551,
      "step": 10268
    },
    {
      "epoch": 1.7815752949340735,
      "grad_norm": 0.6508877873420715,
      "learning_rate": 1.811479168501329e-06,
      "loss": 0.5564,
      "step": 10269
    },
    {
      "epoch": 1.7817487855655796,
      "grad_norm": 0.7891836762428284,
      "learning_rate": 1.8086468242438582e-06,
      "loss": 0.4751,
      "step": 10270
    },
    {
      "epoch": 1.7819222761970854,
      "grad_norm": 1.8561694622039795,
      "learning_rate": 1.8058165911141179e-06,
      "loss": 0.4174,
      "step": 10271
    },
    {
      "epoch": 1.7820957668285913,
      "grad_norm": 0.8580641150474548,
      "learning_rate": 1.8029884694405631e-06,
      "loss": 0.5828,
      "step": 10272
    },
    {
      "epoch": 1.7822692574600971,
      "grad_norm": 0.7157613039016724,
      "learning_rate": 1.8001624595514022e-06,
      "loss": 0.4888,
      "step": 10273
    },
    {
      "epoch": 1.782442748091603,
      "grad_norm": 0.972023606300354,
      "learning_rate": 1.7973385617745953e-06,
      "loss": 0.4709,
      "step": 10274
    },
    {
      "epoch": 1.7826162387231088,
      "grad_norm": 1.1180193424224854,
      "learning_rate": 1.7945167764378536e-06,
      "loss": 0.46,
      "step": 10275
    },
    {
      "epoch": 1.7827897293546149,
      "grad_norm": 0.731166422367096,
      "learning_rate": 1.7916971038686614e-06,
      "loss": 0.6198,
      "step": 10276
    },
    {
      "epoch": 1.7829632199861207,
      "grad_norm": 0.8009506464004517,
      "learning_rate": 1.7888795443942308e-06,
      "loss": 0.5244,
      "step": 10277
    },
    {
      "epoch": 1.7831367106176268,
      "grad_norm": 0.9906429052352905,
      "learning_rate": 1.7860640983415533e-06,
      "loss": 0.3961,
      "step": 10278
    },
    {
      "epoch": 1.7833102012491326,
      "grad_norm": 0.8273224830627441,
      "learning_rate": 1.7832507660373589e-06,
      "loss": 0.3832,
      "step": 10279
    },
    {
      "epoch": 1.7834836918806385,
      "grad_norm": 1.7725870609283447,
      "learning_rate": 1.7804395478081416e-06,
      "loss": 0.5018,
      "step": 10280
    },
    {
      "epoch": 1.7836571825121443,
      "grad_norm": 0.9515945911407471,
      "learning_rate": 1.7776304439801384e-06,
      "loss": 0.4553,
      "step": 10281
    },
    {
      "epoch": 1.7838306731436502,
      "grad_norm": 0.7654293179512024,
      "learning_rate": 1.774823454879362e-06,
      "loss": 0.4698,
      "step": 10282
    },
    {
      "epoch": 1.784004163775156,
      "grad_norm": 1.058476448059082,
      "learning_rate": 1.7720185808315583e-06,
      "loss": 0.4478,
      "step": 10283
    },
    {
      "epoch": 1.784177654406662,
      "grad_norm": 0.7763571739196777,
      "learning_rate": 1.7692158221622379e-06,
      "loss": 0.3785,
      "step": 10284
    },
    {
      "epoch": 1.784351145038168,
      "grad_norm": 0.8149800300598145,
      "learning_rate": 1.7664151791966654e-06,
      "loss": 0.4167,
      "step": 10285
    },
    {
      "epoch": 1.7845246356696738,
      "grad_norm": 0.8562837839126587,
      "learning_rate": 1.763616652259854e-06,
      "loss": 0.4088,
      "step": 10286
    },
    {
      "epoch": 1.7846981263011799,
      "grad_norm": 0.7629222273826599,
      "learning_rate": 1.760820241676584e-06,
      "loss": 0.3582,
      "step": 10287
    },
    {
      "epoch": 1.7848716169326857,
      "grad_norm": 1.0314542055130005,
      "learning_rate": 1.758025947771378e-06,
      "loss": 0.4069,
      "step": 10288
    },
    {
      "epoch": 1.7850451075641915,
      "grad_norm": 0.9086339473724365,
      "learning_rate": 1.7552337708685163e-06,
      "loss": 0.4018,
      "step": 10289
    },
    {
      "epoch": 1.7852185981956974,
      "grad_norm": 0.8123794794082642,
      "learning_rate": 1.752443711292029e-06,
      "loss": 0.3992,
      "step": 10290
    },
    {
      "epoch": 1.7853920888272032,
      "grad_norm": 0.9427492618560791,
      "learning_rate": 1.749655769365719e-06,
      "loss": 0.3154,
      "step": 10291
    },
    {
      "epoch": 1.785565579458709,
      "grad_norm": 1.0322617292404175,
      "learning_rate": 1.7468699454131211e-06,
      "loss": 0.4545,
      "step": 10292
    },
    {
      "epoch": 1.7857390700902152,
      "grad_norm": 0.9409964680671692,
      "learning_rate": 1.7440862397575343e-06,
      "loss": 0.3777,
      "step": 10293
    },
    {
      "epoch": 1.785912560721721,
      "grad_norm": 0.7909843325614929,
      "learning_rate": 1.741304652722009e-06,
      "loss": 0.4912,
      "step": 10294
    },
    {
      "epoch": 1.786086051353227,
      "grad_norm": 1.1315380334854126,
      "learning_rate": 1.7385251846293606e-06,
      "loss": 0.4037,
      "step": 10295
    },
    {
      "epoch": 1.786259541984733,
      "grad_norm": 0.7563734650611877,
      "learning_rate": 1.7357478358021374e-06,
      "loss": 0.4702,
      "step": 10296
    },
    {
      "epoch": 1.7864330326162388,
      "grad_norm": 2.4120776653289795,
      "learning_rate": 1.732972606562664e-06,
      "loss": 0.4395,
      "step": 10297
    },
    {
      "epoch": 1.7866065232477446,
      "grad_norm": 0.7617950439453125,
      "learning_rate": 1.7301994972330028e-06,
      "loss": 0.4054,
      "step": 10298
    },
    {
      "epoch": 1.7867800138792505,
      "grad_norm": 0.8544481992721558,
      "learning_rate": 1.7274285081349807e-06,
      "loss": 0.3783,
      "step": 10299
    },
    {
      "epoch": 1.7869535045107563,
      "grad_norm": 0.6865641474723816,
      "learning_rate": 1.724659639590167e-06,
      "loss": 0.3918,
      "step": 10300
    },
    {
      "epoch": 1.7871269951422624,
      "grad_norm": 1.0140076875686646,
      "learning_rate": 1.7218928919199008e-06,
      "loss": 0.4379,
      "step": 10301
    },
    {
      "epoch": 1.7873004857737682,
      "grad_norm": 0.9270845055580139,
      "learning_rate": 1.7191282654452646e-06,
      "loss": 0.6001,
      "step": 10302
    },
    {
      "epoch": 1.7874739764052743,
      "grad_norm": 0.764961838722229,
      "learning_rate": 1.7163657604870932e-06,
      "loss": 0.4532,
      "step": 10303
    },
    {
      "epoch": 1.7876474670367801,
      "grad_norm": 0.953648030757904,
      "learning_rate": 1.7136053773659766e-06,
      "loss": 0.4871,
      "step": 10304
    },
    {
      "epoch": 1.787820957668286,
      "grad_norm": 0.9284045696258545,
      "learning_rate": 1.71084711640227e-06,
      "loss": 0.5942,
      "step": 10305
    },
    {
      "epoch": 1.7879944482997918,
      "grad_norm": 0.9803312420845032,
      "learning_rate": 1.7080909779160615e-06,
      "loss": 0.4761,
      "step": 10306
    },
    {
      "epoch": 1.7881679389312977,
      "grad_norm": 0.7785720825195312,
      "learning_rate": 1.705336962227211e-06,
      "loss": 0.454,
      "step": 10307
    },
    {
      "epoch": 1.7883414295628035,
      "grad_norm": 0.9398047924041748,
      "learning_rate": 1.7025850696553248e-06,
      "loss": 0.3566,
      "step": 10308
    },
    {
      "epoch": 1.7885149201943094,
      "grad_norm": 0.910733163356781,
      "learning_rate": 1.6998353005197565e-06,
      "loss": 0.4122,
      "step": 10309
    },
    {
      "epoch": 1.7886884108258154,
      "grad_norm": 0.7384318709373474,
      "learning_rate": 1.6970876551396309e-06,
      "loss": 0.424,
      "step": 10310
    },
    {
      "epoch": 1.7888619014573213,
      "grad_norm": 0.763297975063324,
      "learning_rate": 1.6943421338338085e-06,
      "loss": 0.367,
      "step": 10311
    },
    {
      "epoch": 1.7890353920888273,
      "grad_norm": 1.9891629219055176,
      "learning_rate": 1.6915987369209142e-06,
      "loss": 0.3597,
      "step": 10312
    },
    {
      "epoch": 1.7892088827203332,
      "grad_norm": 0.8588789701461792,
      "learning_rate": 1.6888574647193157e-06,
      "loss": 0.3425,
      "step": 10313
    },
    {
      "epoch": 1.789382373351839,
      "grad_norm": 0.9315842986106873,
      "learning_rate": 1.6861183175471495e-06,
      "loss": 0.4163,
      "step": 10314
    },
    {
      "epoch": 1.7895558639833449,
      "grad_norm": 0.820051908493042,
      "learning_rate": 1.6833812957222884e-06,
      "loss": 0.4178,
      "step": 10315
    },
    {
      "epoch": 1.7897293546148507,
      "grad_norm": 2.0135653018951416,
      "learning_rate": 1.6806463995623735e-06,
      "loss": 0.4259,
      "step": 10316
    },
    {
      "epoch": 1.7899028452463566,
      "grad_norm": 0.8339837193489075,
      "learning_rate": 1.6779136293847864e-06,
      "loss": 0.4941,
      "step": 10317
    },
    {
      "epoch": 1.7900763358778626,
      "grad_norm": 1.223927617073059,
      "learning_rate": 1.6751829855066804e-06,
      "loss": 0.4333,
      "step": 10318
    },
    {
      "epoch": 1.7902498265093685,
      "grad_norm": 1.108045220375061,
      "learning_rate": 1.6724544682449328e-06,
      "loss": 0.4402,
      "step": 10319
    },
    {
      "epoch": 1.7904233171408745,
      "grad_norm": 1.0140706300735474,
      "learning_rate": 1.669728077916206e-06,
      "loss": 0.3382,
      "step": 10320
    },
    {
      "epoch": 1.7905968077723804,
      "grad_norm": 0.6455281972885132,
      "learning_rate": 1.6670038148368916e-06,
      "loss": 0.4989,
      "step": 10321
    },
    {
      "epoch": 1.7907702984038862,
      "grad_norm": 0.7716452479362488,
      "learning_rate": 1.6642816793231499e-06,
      "loss": 0.5653,
      "step": 10322
    },
    {
      "epoch": 1.790943789035392,
      "grad_norm": 1.3400412797927856,
      "learning_rate": 1.661561671690879e-06,
      "loss": 0.4108,
      "step": 10323
    },
    {
      "epoch": 1.791117279666898,
      "grad_norm": 1.5090062618255615,
      "learning_rate": 1.6588437922557533e-06,
      "loss": 0.3788,
      "step": 10324
    },
    {
      "epoch": 1.7912907702984038,
      "grad_norm": 0.7390293478965759,
      "learning_rate": 1.6561280413331672e-06,
      "loss": 0.484,
      "step": 10325
    },
    {
      "epoch": 1.7914642609299096,
      "grad_norm": 0.8134437203407288,
      "learning_rate": 1.6534144192383038e-06,
      "loss": 0.5377,
      "step": 10326
    },
    {
      "epoch": 1.7916377515614157,
      "grad_norm": 0.7178210020065308,
      "learning_rate": 1.6507029262860718e-06,
      "loss": 0.491,
      "step": 10327
    },
    {
      "epoch": 1.7918112421929215,
      "grad_norm": 0.9239500761032104,
      "learning_rate": 1.6479935627911481e-06,
      "loss": 0.584,
      "step": 10328
    },
    {
      "epoch": 1.7919847328244276,
      "grad_norm": 1.1858335733413696,
      "learning_rate": 1.6452863290679522e-06,
      "loss": 0.5474,
      "step": 10329
    },
    {
      "epoch": 1.7921582234559335,
      "grad_norm": 0.9040897488594055,
      "learning_rate": 1.6425812254306707e-06,
      "loss": 0.3726,
      "step": 10330
    },
    {
      "epoch": 1.7923317140874393,
      "grad_norm": 0.6616557240486145,
      "learning_rate": 1.6398782521932254e-06,
      "loss": 0.578,
      "step": 10331
    },
    {
      "epoch": 1.7925052047189451,
      "grad_norm": 0.6143748760223389,
      "learning_rate": 1.637177409669304e-06,
      "loss": 0.5868,
      "step": 10332
    },
    {
      "epoch": 1.792678695350451,
      "grad_norm": 1.123853325843811,
      "learning_rate": 1.6344786981723371e-06,
      "loss": 0.3966,
      "step": 10333
    },
    {
      "epoch": 1.7928521859819568,
      "grad_norm": 1.0074896812438965,
      "learning_rate": 1.6317821180155214e-06,
      "loss": 0.3824,
      "step": 10334
    },
    {
      "epoch": 1.793025676613463,
      "grad_norm": 1.1881918907165527,
      "learning_rate": 1.6290876695117951e-06,
      "loss": 0.5178,
      "step": 10335
    },
    {
      "epoch": 1.7931991672449688,
      "grad_norm": 1.177901029586792,
      "learning_rate": 1.6263953529738464e-06,
      "loss": 0.3974,
      "step": 10336
    },
    {
      "epoch": 1.7933726578764748,
      "grad_norm": 0.6227571964263916,
      "learning_rate": 1.6237051687141336e-06,
      "loss": 0.5773,
      "step": 10337
    },
    {
      "epoch": 1.7935461485079807,
      "grad_norm": 0.8060899376869202,
      "learning_rate": 1.621017117044843e-06,
      "loss": 0.4395,
      "step": 10338
    },
    {
      "epoch": 1.7937196391394865,
      "grad_norm": 0.8530584573745728,
      "learning_rate": 1.6183311982779337e-06,
      "loss": 0.4142,
      "step": 10339
    },
    {
      "epoch": 1.7938931297709924,
      "grad_norm": 0.9127979278564453,
      "learning_rate": 1.6156474127251077e-06,
      "loss": 0.4025,
      "step": 10340
    },
    {
      "epoch": 1.7940666204024982,
      "grad_norm": 0.813700795173645,
      "learning_rate": 1.6129657606978221e-06,
      "loss": 0.5027,
      "step": 10341
    },
    {
      "epoch": 1.794240111034004,
      "grad_norm": 1.3757143020629883,
      "learning_rate": 1.6102862425072818e-06,
      "loss": 0.3531,
      "step": 10342
    },
    {
      "epoch": 1.7944136016655101,
      "grad_norm": 0.8702859878540039,
      "learning_rate": 1.6076088584644534e-06,
      "loss": 0.3686,
      "step": 10343
    },
    {
      "epoch": 1.794587092297016,
      "grad_norm": 0.9329940676689148,
      "learning_rate": 1.6049336088800505e-06,
      "loss": 0.4861,
      "step": 10344
    },
    {
      "epoch": 1.7947605829285218,
      "grad_norm": 1.13565194606781,
      "learning_rate": 1.6022604940645337e-06,
      "loss": 0.5552,
      "step": 10345
    },
    {
      "epoch": 1.7949340735600279,
      "grad_norm": 0.797062337398529,
      "learning_rate": 1.5995895143281236e-06,
      "loss": 0.5052,
      "step": 10346
    },
    {
      "epoch": 1.7951075641915337,
      "grad_norm": 0.7553451061248779,
      "learning_rate": 1.596920669980797e-06,
      "loss": 0.5099,
      "step": 10347
    },
    {
      "epoch": 1.7952810548230396,
      "grad_norm": 0.8680810928344727,
      "learning_rate": 1.5942539613322638e-06,
      "loss": 0.3078,
      "step": 10348
    },
    {
      "epoch": 1.7954545454545454,
      "grad_norm": 0.8695831298828125,
      "learning_rate": 1.5915893886920098e-06,
      "loss": 0.3704,
      "step": 10349
    },
    {
      "epoch": 1.7956280360860513,
      "grad_norm": 0.7685412764549255,
      "learning_rate": 1.5889269523692541e-06,
      "loss": 0.4189,
      "step": 10350
    },
    {
      "epoch": 1.795801526717557,
      "grad_norm": 0.724926233291626,
      "learning_rate": 1.586266652672981e-06,
      "loss": 0.514,
      "step": 10351
    },
    {
      "epoch": 1.7959750173490632,
      "grad_norm": 3.163278341293335,
      "learning_rate": 1.5836084899119165e-06,
      "loss": 0.4619,
      "step": 10352
    },
    {
      "epoch": 1.796148507980569,
      "grad_norm": 2.5651533603668213,
      "learning_rate": 1.5809524643945472e-06,
      "loss": 0.4409,
      "step": 10353
    },
    {
      "epoch": 1.796321998612075,
      "grad_norm": 0.7803921103477478,
      "learning_rate": 1.5782985764291091e-06,
      "loss": 0.3702,
      "step": 10354
    },
    {
      "epoch": 1.796495489243581,
      "grad_norm": 0.798792839050293,
      "learning_rate": 1.575646826323587e-06,
      "loss": 0.3646,
      "step": 10355
    },
    {
      "epoch": 1.7966689798750868,
      "grad_norm": 0.964057981967926,
      "learning_rate": 1.5729972143857164e-06,
      "loss": 0.3783,
      "step": 10356
    },
    {
      "epoch": 1.7968424705065926,
      "grad_norm": 1.277469515800476,
      "learning_rate": 1.5703497409229896e-06,
      "loss": 0.3801,
      "step": 10357
    },
    {
      "epoch": 1.7970159611380985,
      "grad_norm": 0.7194745540618896,
      "learning_rate": 1.567704406242654e-06,
      "loss": 0.5676,
      "step": 10358
    },
    {
      "epoch": 1.7971894517696043,
      "grad_norm": 0.824563205242157,
      "learning_rate": 1.5650612106516993e-06,
      "loss": 0.3987,
      "step": 10359
    },
    {
      "epoch": 1.7973629424011104,
      "grad_norm": 0.9087297916412354,
      "learning_rate": 1.5624201544568717e-06,
      "loss": 0.3857,
      "step": 10360
    },
    {
      "epoch": 1.7975364330326162,
      "grad_norm": 0.7791262269020081,
      "learning_rate": 1.559781237964666e-06,
      "loss": 0.4685,
      "step": 10361
    },
    {
      "epoch": 1.7977099236641223,
      "grad_norm": 0.7930310964584351,
      "learning_rate": 1.557144461481337e-06,
      "loss": 0.4318,
      "step": 10362
    },
    {
      "epoch": 1.7978834142956281,
      "grad_norm": 0.7366043925285339,
      "learning_rate": 1.5545098253128843e-06,
      "loss": 0.5748,
      "step": 10363
    },
    {
      "epoch": 1.798056904927134,
      "grad_norm": 0.7938430309295654,
      "learning_rate": 1.5518773297650613e-06,
      "loss": 0.4272,
      "step": 10364
    },
    {
      "epoch": 1.7982303955586398,
      "grad_norm": 1.9784938097000122,
      "learning_rate": 1.5492469751433658e-06,
      "loss": 0.3588,
      "step": 10365
    },
    {
      "epoch": 1.7984038861901457,
      "grad_norm": 0.7573646903038025,
      "learning_rate": 1.5466187617530647e-06,
      "loss": 0.4295,
      "step": 10366
    },
    {
      "epoch": 1.7985773768216515,
      "grad_norm": 0.8965981602668762,
      "learning_rate": 1.543992689899152e-06,
      "loss": 0.3536,
      "step": 10367
    },
    {
      "epoch": 1.7987508674531574,
      "grad_norm": 0.7845249772071838,
      "learning_rate": 1.541368759886397e-06,
      "loss": 0.3402,
      "step": 10368
    },
    {
      "epoch": 1.7989243580846634,
      "grad_norm": 0.685406506061554,
      "learning_rate": 1.5387469720193048e-06,
      "loss": 0.5784,
      "step": 10369
    },
    {
      "epoch": 1.7990978487161693,
      "grad_norm": 0.6819063425064087,
      "learning_rate": 1.5361273266021392e-06,
      "loss": 0.5713,
      "step": 10370
    },
    {
      "epoch": 1.7992713393476754,
      "grad_norm": 0.6778554320335388,
      "learning_rate": 1.5335098239389102e-06,
      "loss": 0.4902,
      "step": 10371
    },
    {
      "epoch": 1.7994448299791812,
      "grad_norm": 0.6078512668609619,
      "learning_rate": 1.5308944643333857e-06,
      "loss": 0.5503,
      "step": 10372
    },
    {
      "epoch": 1.799618320610687,
      "grad_norm": 0.7713558077812195,
      "learning_rate": 1.5282812480890784e-06,
      "loss": 0.4594,
      "step": 10373
    },
    {
      "epoch": 1.799791811242193,
      "grad_norm": 0.6342697143554688,
      "learning_rate": 1.5256701755092574e-06,
      "loss": 0.5511,
      "step": 10374
    },
    {
      "epoch": 1.7999653018736987,
      "grad_norm": 0.811905026435852,
      "learning_rate": 1.5230612468969352e-06,
      "loss": 0.4613,
      "step": 10375
    },
    {
      "epoch": 1.8001387925052046,
      "grad_norm": 0.6842833161354065,
      "learning_rate": 1.5204544625548922e-06,
      "loss": 0.4199,
      "step": 10376
    },
    {
      "epoch": 1.8003122831367107,
      "grad_norm": 0.8111035823822021,
      "learning_rate": 1.5178498227856353e-06,
      "loss": 0.4482,
      "step": 10377
    },
    {
      "epoch": 1.8004857737682165,
      "grad_norm": 0.774206817150116,
      "learning_rate": 1.5152473278914447e-06,
      "loss": 0.4365,
      "step": 10378
    },
    {
      "epoch": 1.8006592643997226,
      "grad_norm": 0.9334141612052917,
      "learning_rate": 1.5126469781743436e-06,
      "loss": 0.4044,
      "step": 10379
    },
    {
      "epoch": 1.8008327550312284,
      "grad_norm": 0.7614678144454956,
      "learning_rate": 1.5100487739360993e-06,
      "loss": 0.3746,
      "step": 10380
    },
    {
      "epoch": 1.8010062456627343,
      "grad_norm": 0.7316577434539795,
      "learning_rate": 1.5074527154782393e-06,
      "loss": 0.4777,
      "step": 10381
    },
    {
      "epoch": 1.80117973629424,
      "grad_norm": 0.8212964534759521,
      "learning_rate": 1.5048588031020405e-06,
      "loss": 0.3773,
      "step": 10382
    },
    {
      "epoch": 1.801353226925746,
      "grad_norm": 2.187487840652466,
      "learning_rate": 1.5022670371085314e-06,
      "loss": 0.5046,
      "step": 10383
    },
    {
      "epoch": 1.8015267175572518,
      "grad_norm": 0.6952325701713562,
      "learning_rate": 1.4996774177984818e-06,
      "loss": 0.4368,
      "step": 10384
    },
    {
      "epoch": 1.8017002081887576,
      "grad_norm": 1.030739665031433,
      "learning_rate": 1.4970899454724319e-06,
      "loss": 0.543,
      "step": 10385
    },
    {
      "epoch": 1.8018736988202637,
      "grad_norm": 0.6696087718009949,
      "learning_rate": 1.494504620430648e-06,
      "loss": 0.4968,
      "step": 10386
    },
    {
      "epoch": 1.8020471894517696,
      "grad_norm": 0.9498783349990845,
      "learning_rate": 1.4919214429731677e-06,
      "loss": 0.3719,
      "step": 10387
    },
    {
      "epoch": 1.8022206800832756,
      "grad_norm": 0.7543936371803284,
      "learning_rate": 1.489340413399769e-06,
      "loss": 0.4169,
      "step": 10388
    },
    {
      "epoch": 1.8023941707147815,
      "grad_norm": 0.7348383069038391,
      "learning_rate": 1.4867615320099904e-06,
      "loss": 0.4058,
      "step": 10389
    },
    {
      "epoch": 1.8025676613462873,
      "grad_norm": 1.013086199760437,
      "learning_rate": 1.484184799103101e-06,
      "loss": 0.399,
      "step": 10390
    },
    {
      "epoch": 1.8027411519777932,
      "grad_norm": 0.8915300369262695,
      "learning_rate": 1.4816102149781442e-06,
      "loss": 0.3633,
      "step": 10391
    },
    {
      "epoch": 1.802914642609299,
      "grad_norm": 0.7792953252792358,
      "learning_rate": 1.4790377799339007e-06,
      "loss": 0.4714,
      "step": 10392
    },
    {
      "epoch": 1.8030881332408049,
      "grad_norm": 0.6869494318962097,
      "learning_rate": 1.476467494268905e-06,
      "loss": 0.5162,
      "step": 10393
    },
    {
      "epoch": 1.803261623872311,
      "grad_norm": 0.7749457359313965,
      "learning_rate": 1.4738993582814343e-06,
      "loss": 0.5782,
      "step": 10394
    },
    {
      "epoch": 1.8034351145038168,
      "grad_norm": 0.6259393692016602,
      "learning_rate": 1.471333372269539e-06,
      "loss": 0.4888,
      "step": 10395
    },
    {
      "epoch": 1.8036086051353228,
      "grad_norm": 0.9049027562141418,
      "learning_rate": 1.4687695365309895e-06,
      "loss": 0.4958,
      "step": 10396
    },
    {
      "epoch": 1.8037820957668287,
      "grad_norm": 0.8636259436607361,
      "learning_rate": 1.4662078513633327e-06,
      "loss": 0.4354,
      "step": 10397
    },
    {
      "epoch": 1.8039555863983345,
      "grad_norm": 0.949533998966217,
      "learning_rate": 1.4636483170638505e-06,
      "loss": 0.5353,
      "step": 10398
    },
    {
      "epoch": 1.8041290770298404,
      "grad_norm": 0.8600096702575684,
      "learning_rate": 1.4610909339295788e-06,
      "loss": 0.4138,
      "step": 10399
    },
    {
      "epoch": 1.8043025676613462,
      "grad_norm": 0.8306368589401245,
      "learning_rate": 1.4585357022573043e-06,
      "loss": 0.4678,
      "step": 10400
    },
    {
      "epoch": 1.804476058292852,
      "grad_norm": 0.7927076816558838,
      "learning_rate": 1.455982622343568e-06,
      "loss": 0.4506,
      "step": 10401
    },
    {
      "epoch": 1.8046495489243581,
      "grad_norm": 0.8654670715332031,
      "learning_rate": 1.4534316944846595e-06,
      "loss": 0.4847,
      "step": 10402
    },
    {
      "epoch": 1.804823039555864,
      "grad_norm": 0.7912794947624207,
      "learning_rate": 1.4508829189766104e-06,
      "loss": 0.424,
      "step": 10403
    },
    {
      "epoch": 1.8049965301873698,
      "grad_norm": 0.7177606821060181,
      "learning_rate": 1.4483362961152114e-06,
      "loss": 0.4616,
      "step": 10404
    },
    {
      "epoch": 1.805170020818876,
      "grad_norm": 0.8227936029434204,
      "learning_rate": 1.4457918261960057e-06,
      "loss": 0.5386,
      "step": 10405
    },
    {
      "epoch": 1.8053435114503817,
      "grad_norm": 1.0631943941116333,
      "learning_rate": 1.4432495095142796e-06,
      "loss": 0.3779,
      "step": 10406
    },
    {
      "epoch": 1.8055170020818876,
      "grad_norm": 0.9247499108314514,
      "learning_rate": 1.4407093463650679e-06,
      "loss": 0.5629,
      "step": 10407
    },
    {
      "epoch": 1.8056904927133934,
      "grad_norm": 1.0188275575637817,
      "learning_rate": 1.438171337043164e-06,
      "loss": 0.42,
      "step": 10408
    },
    {
      "epoch": 1.8058639833448993,
      "grad_norm": 0.669012725353241,
      "learning_rate": 1.4356354818431028e-06,
      "loss": 0.4211,
      "step": 10409
    },
    {
      "epoch": 1.8060374739764051,
      "grad_norm": 0.7970829606056213,
      "learning_rate": 1.4331017810591764e-06,
      "loss": 0.431,
      "step": 10410
    },
    {
      "epoch": 1.8062109646079112,
      "grad_norm": 0.9136406183242798,
      "learning_rate": 1.4305702349854245e-06,
      "loss": 0.499,
      "step": 10411
    },
    {
      "epoch": 1.806384455239417,
      "grad_norm": 0.9455069303512573,
      "learning_rate": 1.4280408439156369e-06,
      "loss": 0.3217,
      "step": 10412
    },
    {
      "epoch": 1.806557945870923,
      "grad_norm": 0.6552941799163818,
      "learning_rate": 1.425513608143343e-06,
      "loss": 0.4631,
      "step": 10413
    },
    {
      "epoch": 1.806731436502429,
      "grad_norm": 0.6552343964576721,
      "learning_rate": 1.4229885279618461e-06,
      "loss": 0.4694,
      "step": 10414
    },
    {
      "epoch": 1.8069049271339348,
      "grad_norm": 0.7940164804458618,
      "learning_rate": 1.4204656036641717e-06,
      "loss": 0.3919,
      "step": 10415
    },
    {
      "epoch": 1.8070784177654406,
      "grad_norm": 0.7920050024986267,
      "learning_rate": 1.4179448355431168e-06,
      "loss": 0.4413,
      "step": 10416
    },
    {
      "epoch": 1.8072519083969465,
      "grad_norm": 0.7787114977836609,
      "learning_rate": 1.415426223891212e-06,
      "loss": 0.4563,
      "step": 10417
    },
    {
      "epoch": 1.8074253990284523,
      "grad_norm": 0.8907957673072815,
      "learning_rate": 1.4129097690007543e-06,
      "loss": 0.5712,
      "step": 10418
    },
    {
      "epoch": 1.8075988896599584,
      "grad_norm": 0.8291707634925842,
      "learning_rate": 1.4103954711637724e-06,
      "loss": 0.3877,
      "step": 10419
    },
    {
      "epoch": 1.8077723802914643,
      "grad_norm": 1.100144863128662,
      "learning_rate": 1.4078833306720573e-06,
      "loss": 0.4199,
      "step": 10420
    },
    {
      "epoch": 1.8079458709229703,
      "grad_norm": 0.8819019198417664,
      "learning_rate": 1.4053733478171493e-06,
      "loss": 0.5428,
      "step": 10421
    },
    {
      "epoch": 1.8081193615544762,
      "grad_norm": 0.6700547933578491,
      "learning_rate": 1.4028655228903286e-06,
      "loss": 0.4471,
      "step": 10422
    },
    {
      "epoch": 1.808292852185982,
      "grad_norm": 0.7875379920005798,
      "learning_rate": 1.4003598561826337e-06,
      "loss": 0.3529,
      "step": 10423
    },
    {
      "epoch": 1.8084663428174879,
      "grad_norm": 0.9104384183883667,
      "learning_rate": 1.3978563479848538e-06,
      "loss": 0.4794,
      "step": 10424
    },
    {
      "epoch": 1.8086398334489937,
      "grad_norm": 0.9371272921562195,
      "learning_rate": 1.3953549985875148e-06,
      "loss": 0.3531,
      "step": 10425
    },
    {
      "epoch": 1.8088133240804996,
      "grad_norm": 0.7614732980728149,
      "learning_rate": 1.3928558082809107e-06,
      "loss": 0.4926,
      "step": 10426
    },
    {
      "epoch": 1.8089868147120054,
      "grad_norm": 0.8160543441772461,
      "learning_rate": 1.390358777355072e-06,
      "loss": 0.4971,
      "step": 10427
    },
    {
      "epoch": 1.8091603053435115,
      "grad_norm": 1.235662579536438,
      "learning_rate": 1.3878639060997822e-06,
      "loss": 0.3992,
      "step": 10428
    },
    {
      "epoch": 1.8093337959750173,
      "grad_norm": 0.9123250842094421,
      "learning_rate": 1.3853711948045678e-06,
      "loss": 0.4918,
      "step": 10429
    },
    {
      "epoch": 1.8095072866065234,
      "grad_norm": 0.7046394944190979,
      "learning_rate": 1.3828806437587216e-06,
      "loss": 0.4102,
      "step": 10430
    },
    {
      "epoch": 1.8096807772380292,
      "grad_norm": 0.9694055914878845,
      "learning_rate": 1.3803922532512703e-06,
      "loss": 0.4529,
      "step": 10431
    },
    {
      "epoch": 1.809854267869535,
      "grad_norm": 0.8562542200088501,
      "learning_rate": 1.3779060235709918e-06,
      "loss": 0.5111,
      "step": 10432
    },
    {
      "epoch": 1.810027758501041,
      "grad_norm": 0.7286620736122131,
      "learning_rate": 1.3754219550064196e-06,
      "loss": 0.4911,
      "step": 10433
    },
    {
      "epoch": 1.8102012491325468,
      "grad_norm": 0.9972683191299438,
      "learning_rate": 1.3729400478458322e-06,
      "loss": 0.4991,
      "step": 10434
    },
    {
      "epoch": 1.8103747397640526,
      "grad_norm": 0.7474499344825745,
      "learning_rate": 1.3704603023772567e-06,
      "loss": 0.3622,
      "step": 10435
    },
    {
      "epoch": 1.8105482303955587,
      "grad_norm": 1.534613013267517,
      "learning_rate": 1.3679827188884675e-06,
      "loss": 0.4198,
      "step": 10436
    },
    {
      "epoch": 1.8107217210270645,
      "grad_norm": 0.8257385492324829,
      "learning_rate": 1.3655072976670014e-06,
      "loss": 0.5247,
      "step": 10437
    },
    {
      "epoch": 1.8108952116585706,
      "grad_norm": 0.897869884967804,
      "learning_rate": 1.3630340390001195e-06,
      "loss": 0.3745,
      "step": 10438
    },
    {
      "epoch": 1.8110687022900764,
      "grad_norm": 0.7141605019569397,
      "learning_rate": 1.360562943174859e-06,
      "loss": 0.4653,
      "step": 10439
    },
    {
      "epoch": 1.8112421929215823,
      "grad_norm": 0.8600662350654602,
      "learning_rate": 1.358094010477986e-06,
      "loss": 0.3854,
      "step": 10440
    },
    {
      "epoch": 1.8114156835530881,
      "grad_norm": 0.5740742087364197,
      "learning_rate": 1.355627241196027e-06,
      "loss": 0.5186,
      "step": 10441
    },
    {
      "epoch": 1.811589174184594,
      "grad_norm": 0.6400471925735474,
      "learning_rate": 1.353162635615246e-06,
      "loss": 0.5026,
      "step": 10442
    },
    {
      "epoch": 1.8117626648160998,
      "grad_norm": 0.8278120160102844,
      "learning_rate": 1.3507001940216767e-06,
      "loss": 0.4049,
      "step": 10443
    },
    {
      "epoch": 1.8119361554476057,
      "grad_norm": 1.3171443939208984,
      "learning_rate": 1.3482399167010752e-06,
      "loss": 0.3921,
      "step": 10444
    },
    {
      "epoch": 1.8121096460791117,
      "grad_norm": 1.3614461421966553,
      "learning_rate": 1.345781803938968e-06,
      "loss": 0.3783,
      "step": 10445
    },
    {
      "epoch": 1.8122831367106176,
      "grad_norm": 0.786858320236206,
      "learning_rate": 1.3433258560206165e-06,
      "loss": 0.5413,
      "step": 10446
    },
    {
      "epoch": 1.8124566273421236,
      "grad_norm": 0.8558787703514099,
      "learning_rate": 1.3408720732310432e-06,
      "loss": 0.3998,
      "step": 10447
    },
    {
      "epoch": 1.8126301179736295,
      "grad_norm": 0.9111509919166565,
      "learning_rate": 1.3384204558550028e-06,
      "loss": 0.4165,
      "step": 10448
    },
    {
      "epoch": 1.8128036086051353,
      "grad_norm": 1.0603716373443604,
      "learning_rate": 1.3359710041770147e-06,
      "loss": 0.3884,
      "step": 10449
    },
    {
      "epoch": 1.8129770992366412,
      "grad_norm": 3.856734037399292,
      "learning_rate": 1.333523718481342e-06,
      "loss": 0.5634,
      "step": 10450
    },
    {
      "epoch": 1.813150589868147,
      "grad_norm": 0.672587513923645,
      "learning_rate": 1.3310785990519913e-06,
      "loss": 0.5081,
      "step": 10451
    },
    {
      "epoch": 1.8133240804996529,
      "grad_norm": 0.8601288795471191,
      "learning_rate": 1.3286356461727202e-06,
      "loss": 0.5056,
      "step": 10452
    },
    {
      "epoch": 1.813497571131159,
      "grad_norm": 0.9924600720405579,
      "learning_rate": 1.3261948601270413e-06,
      "loss": 0.3995,
      "step": 10453
    },
    {
      "epoch": 1.8136710617626648,
      "grad_norm": 0.8877877593040466,
      "learning_rate": 1.3237562411982086e-06,
      "loss": 0.4657,
      "step": 10454
    },
    {
      "epoch": 1.8138445523941709,
      "grad_norm": 0.6752092242240906,
      "learning_rate": 1.3213197896692264e-06,
      "loss": 0.6122,
      "step": 10455
    },
    {
      "epoch": 1.8140180430256767,
      "grad_norm": 0.9856737852096558,
      "learning_rate": 1.3188855058228468e-06,
      "loss": 0.4596,
      "step": 10456
    },
    {
      "epoch": 1.8141915336571826,
      "grad_norm": 0.7623448967933655,
      "learning_rate": 1.31645338994157e-06,
      "loss": 0.5175,
      "step": 10457
    },
    {
      "epoch": 1.8143650242886884,
      "grad_norm": 0.8224581480026245,
      "learning_rate": 1.3140234423076504e-06,
      "loss": 0.4199,
      "step": 10458
    },
    {
      "epoch": 1.8145385149201942,
      "grad_norm": 0.7056581974029541,
      "learning_rate": 1.3115956632030845e-06,
      "loss": 0.3624,
      "step": 10459
    },
    {
      "epoch": 1.8147120055517,
      "grad_norm": 0.7192043662071228,
      "learning_rate": 1.3091700529096186e-06,
      "loss": 0.5594,
      "step": 10460
    },
    {
      "epoch": 1.8148854961832062,
      "grad_norm": 0.6109599471092224,
      "learning_rate": 1.3067466117087424e-06,
      "loss": 0.5602,
      "step": 10461
    },
    {
      "epoch": 1.815058986814712,
      "grad_norm": 0.9726971983909607,
      "learning_rate": 1.304325339881709e-06,
      "loss": 0.5156,
      "step": 10462
    },
    {
      "epoch": 1.8152324774462179,
      "grad_norm": 1.1493715047836304,
      "learning_rate": 1.3019062377095026e-06,
      "loss": 0.293,
      "step": 10463
    },
    {
      "epoch": 1.815405968077724,
      "grad_norm": 0.8043683171272278,
      "learning_rate": 1.2994893054728653e-06,
      "loss": 0.3525,
      "step": 10464
    },
    {
      "epoch": 1.8155794587092298,
      "grad_norm": 0.821807324886322,
      "learning_rate": 1.2970745434522835e-06,
      "loss": 0.4825,
      "step": 10465
    },
    {
      "epoch": 1.8157529493407356,
      "grad_norm": 4.892745494842529,
      "learning_rate": 1.2946619519279979e-06,
      "loss": 0.3548,
      "step": 10466
    },
    {
      "epoch": 1.8159264399722415,
      "grad_norm": 1.036017894744873,
      "learning_rate": 1.2922515311799843e-06,
      "loss": 0.4144,
      "step": 10467
    },
    {
      "epoch": 1.8160999306037473,
      "grad_norm": 1.1273837089538574,
      "learning_rate": 1.2898432814879813e-06,
      "loss": 0.421,
      "step": 10468
    },
    {
      "epoch": 1.8162734212352531,
      "grad_norm": 1.5924334526062012,
      "learning_rate": 1.2874372031314675e-06,
      "loss": 0.5062,
      "step": 10469
    },
    {
      "epoch": 1.8164469118667592,
      "grad_norm": 1.2282288074493408,
      "learning_rate": 1.2850332963896706e-06,
      "loss": 0.4563,
      "step": 10470
    },
    {
      "epoch": 1.816620402498265,
      "grad_norm": 0.7977563142776489,
      "learning_rate": 1.2826315615415652e-06,
      "loss": 0.476,
      "step": 10471
    },
    {
      "epoch": 1.8167938931297711,
      "grad_norm": 0.8569274544715881,
      "learning_rate": 1.2802319988658818e-06,
      "loss": 0.3538,
      "step": 10472
    },
    {
      "epoch": 1.816967383761277,
      "grad_norm": 0.6680198907852173,
      "learning_rate": 1.277834608641082e-06,
      "loss": 0.5015,
      "step": 10473
    },
    {
      "epoch": 1.8171408743927828,
      "grad_norm": 0.7478260397911072,
      "learning_rate": 1.2754393911453944e-06,
      "loss": 0.5546,
      "step": 10474
    },
    {
      "epoch": 1.8173143650242887,
      "grad_norm": 1.1330883502960205,
      "learning_rate": 1.273046346656781e-06,
      "loss": 0.3973,
      "step": 10475
    },
    {
      "epoch": 1.8174878556557945,
      "grad_norm": 0.7933983206748962,
      "learning_rate": 1.2706554754529665e-06,
      "loss": 0.3932,
      "step": 10476
    },
    {
      "epoch": 1.8176613462873004,
      "grad_norm": 0.918095052242279,
      "learning_rate": 1.2682667778114022e-06,
      "loss": 0.4001,
      "step": 10477
    },
    {
      "epoch": 1.8178348369188064,
      "grad_norm": 0.7663379907608032,
      "learning_rate": 1.2658802540093084e-06,
      "loss": 0.3914,
      "step": 10478
    },
    {
      "epoch": 1.8180083275503123,
      "grad_norm": 0.7536807060241699,
      "learning_rate": 1.2634959043236395e-06,
      "loss": 0.537,
      "step": 10479
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 1.1602418422698975,
      "learning_rate": 1.2611137290311003e-06,
      "loss": 0.449,
      "step": 10480
    },
    {
      "epoch": 1.8183553088133242,
      "grad_norm": 0.8546183705329895,
      "learning_rate": 1.2587337284081524e-06,
      "loss": 0.3546,
      "step": 10481
    },
    {
      "epoch": 1.81852879944483,
      "grad_norm": 1.0874425172805786,
      "learning_rate": 1.2563559027309925e-06,
      "loss": 0.4035,
      "step": 10482
    },
    {
      "epoch": 1.8187022900763359,
      "grad_norm": 0.9559531807899475,
      "learning_rate": 1.2539802522755707e-06,
      "loss": 0.4085,
      "step": 10483
    },
    {
      "epoch": 1.8188757807078417,
      "grad_norm": 0.7305978536605835,
      "learning_rate": 1.2516067773175822e-06,
      "loss": 0.5343,
      "step": 10484
    },
    {
      "epoch": 1.8190492713393476,
      "grad_norm": 0.7242780923843384,
      "learning_rate": 1.2492354781324778e-06,
      "loss": 0.35,
      "step": 10485
    },
    {
      "epoch": 1.8192227619708534,
      "grad_norm": 0.8772955536842346,
      "learning_rate": 1.2468663549954397e-06,
      "loss": 0.3341,
      "step": 10486
    },
    {
      "epoch": 1.8193962526023595,
      "grad_norm": 0.790827751159668,
      "learning_rate": 1.2444994081814165e-06,
      "loss": 0.4374,
      "step": 10487
    },
    {
      "epoch": 1.8195697432338653,
      "grad_norm": 0.7314450144767761,
      "learning_rate": 1.2421346379650868e-06,
      "loss": 0.3947,
      "step": 10488
    },
    {
      "epoch": 1.8197432338653714,
      "grad_norm": 0.8860965371131897,
      "learning_rate": 1.2397720446208972e-06,
      "loss": 0.3487,
      "step": 10489
    },
    {
      "epoch": 1.8199167244968772,
      "grad_norm": 0.8088721036911011,
      "learning_rate": 1.2374116284230153e-06,
      "loss": 0.3801,
      "step": 10490
    },
    {
      "epoch": 1.820090215128383,
      "grad_norm": 0.8200245499610901,
      "learning_rate": 1.2350533896453799e-06,
      "loss": 0.3249,
      "step": 10491
    },
    {
      "epoch": 1.820263705759889,
      "grad_norm": 1.028024673461914,
      "learning_rate": 1.2326973285616628e-06,
      "loss": 0.4276,
      "step": 10492
    },
    {
      "epoch": 1.8204371963913948,
      "grad_norm": 0.7007224559783936,
      "learning_rate": 1.2303434454452901e-06,
      "loss": 0.5033,
      "step": 10493
    },
    {
      "epoch": 1.8206106870229006,
      "grad_norm": 0.7374883890151978,
      "learning_rate": 1.2279917405694298e-06,
      "loss": 0.4574,
      "step": 10494
    },
    {
      "epoch": 1.8207841776544067,
      "grad_norm": 0.7309583425521851,
      "learning_rate": 1.2256422142070057e-06,
      "loss": 0.4271,
      "step": 10495
    },
    {
      "epoch": 1.8209576682859125,
      "grad_norm": 0.7799156904220581,
      "learning_rate": 1.2232948666306732e-06,
      "loss": 0.4905,
      "step": 10496
    },
    {
      "epoch": 1.8211311589174186,
      "grad_norm": 1.1447699069976807,
      "learning_rate": 1.2209496981128543e-06,
      "loss": 0.5212,
      "step": 10497
    },
    {
      "epoch": 1.8213046495489245,
      "grad_norm": 0.6853039264678955,
      "learning_rate": 1.218606708925707e-06,
      "loss": 0.6538,
      "step": 10498
    },
    {
      "epoch": 1.8214781401804303,
      "grad_norm": 1.4950239658355713,
      "learning_rate": 1.2162658993411336e-06,
      "loss": 0.557,
      "step": 10499
    },
    {
      "epoch": 1.8216516308119362,
      "grad_norm": 0.827681303024292,
      "learning_rate": 1.2139272696307857e-06,
      "loss": 0.5206,
      "step": 10500
    },
    {
      "epoch": 1.821825121443442,
      "grad_norm": 1.179308295249939,
      "learning_rate": 1.2115908200660731e-06,
      "loss": 0.4045,
      "step": 10501
    },
    {
      "epoch": 1.8219986120749478,
      "grad_norm": 1.6559895277023315,
      "learning_rate": 1.2092565509181386e-06,
      "loss": 0.4258,
      "step": 10502
    },
    {
      "epoch": 1.8221721027064537,
      "grad_norm": 1.2037452459335327,
      "learning_rate": 1.2069244624578768e-06,
      "loss": 0.3815,
      "step": 10503
    },
    {
      "epoch": 1.8223455933379598,
      "grad_norm": 0.9145764112472534,
      "learning_rate": 1.2045945549559269e-06,
      "loss": 0.3381,
      "step": 10504
    },
    {
      "epoch": 1.8225190839694656,
      "grad_norm": 0.7107357382774353,
      "learning_rate": 1.202266828682681e-06,
      "loss": 0.4226,
      "step": 10505
    },
    {
      "epoch": 1.8226925746009717,
      "grad_norm": 1.0881927013397217,
      "learning_rate": 1.1999412839082748e-06,
      "loss": 0.5344,
      "step": 10506
    },
    {
      "epoch": 1.8228660652324775,
      "grad_norm": 0.9871724843978882,
      "learning_rate": 1.1976179209025896e-06,
      "loss": 0.4009,
      "step": 10507
    },
    {
      "epoch": 1.8230395558639834,
      "grad_norm": 0.9611091613769531,
      "learning_rate": 1.1952967399352522e-06,
      "loss": 0.3506,
      "step": 10508
    },
    {
      "epoch": 1.8232130464954892,
      "grad_norm": 0.9632605314254761,
      "learning_rate": 1.1929777412756382e-06,
      "loss": 0.3148,
      "step": 10509
    },
    {
      "epoch": 1.823386537126995,
      "grad_norm": 1.171610713005066,
      "learning_rate": 1.1906609251928746e-06,
      "loss": 0.4294,
      "step": 10510
    },
    {
      "epoch": 1.823560027758501,
      "grad_norm": 0.9603515863418579,
      "learning_rate": 1.1883462919558263e-06,
      "loss": 0.4021,
      "step": 10511
    },
    {
      "epoch": 1.823733518390007,
      "grad_norm": 0.8278272151947021,
      "learning_rate": 1.186033841833112e-06,
      "loss": 0.4159,
      "step": 10512
    },
    {
      "epoch": 1.8239070090215128,
      "grad_norm": 0.9903392791748047,
      "learning_rate": 1.1837235750930898e-06,
      "loss": 0.4825,
      "step": 10513
    },
    {
      "epoch": 1.8240804996530189,
      "grad_norm": 0.928013801574707,
      "learning_rate": 1.1814154920038789e-06,
      "loss": 0.3651,
      "step": 10514
    },
    {
      "epoch": 1.8242539902845247,
      "grad_norm": 0.636833667755127,
      "learning_rate": 1.179109592833323e-06,
      "loss": 0.4077,
      "step": 10515
    },
    {
      "epoch": 1.8244274809160306,
      "grad_norm": 1.2930148839950562,
      "learning_rate": 1.17680587784903e-06,
      "loss": 0.5465,
      "step": 10516
    },
    {
      "epoch": 1.8246009715475364,
      "grad_norm": 0.766405463218689,
      "learning_rate": 1.174504347318346e-06,
      "loss": 0.5189,
      "step": 10517
    },
    {
      "epoch": 1.8247744621790423,
      "grad_norm": 0.7657548189163208,
      "learning_rate": 1.1722050015083752e-06,
      "loss": 0.4747,
      "step": 10518
    },
    {
      "epoch": 1.8249479528105481,
      "grad_norm": 0.7444841861724854,
      "learning_rate": 1.1699078406859466e-06,
      "loss": 0.3865,
      "step": 10519
    },
    {
      "epoch": 1.8251214434420542,
      "grad_norm": 0.8649503588676453,
      "learning_rate": 1.1676128651176578e-06,
      "loss": 0.6057,
      "step": 10520
    },
    {
      "epoch": 1.82529493407356,
      "grad_norm": 0.674697756767273,
      "learning_rate": 1.1653200750698402e-06,
      "loss": 0.4941,
      "step": 10521
    },
    {
      "epoch": 1.8254684247050659,
      "grad_norm": 0.7950411438941956,
      "learning_rate": 1.163029470808572e-06,
      "loss": 0.4839,
      "step": 10522
    },
    {
      "epoch": 1.825641915336572,
      "grad_norm": 1.2704163789749146,
      "learning_rate": 1.1607410525996832e-06,
      "loss": 0.3766,
      "step": 10523
    },
    {
      "epoch": 1.8258154059680778,
      "grad_norm": 0.9411687850952148,
      "learning_rate": 1.1584548207087498e-06,
      "loss": 0.3929,
      "step": 10524
    },
    {
      "epoch": 1.8259888965995836,
      "grad_norm": 0.7350007891654968,
      "learning_rate": 1.1561707754010887e-06,
      "loss": 0.5835,
      "step": 10525
    },
    {
      "epoch": 1.8261623872310895,
      "grad_norm": 0.7564200758934021,
      "learning_rate": 1.1538889169417654e-06,
      "loss": 0.6437,
      "step": 10526
    },
    {
      "epoch": 1.8263358778625953,
      "grad_norm": 1.0063823461532593,
      "learning_rate": 1.1516092455955951e-06,
      "loss": 0.4537,
      "step": 10527
    },
    {
      "epoch": 1.8265093684941012,
      "grad_norm": 2.187542200088501,
      "learning_rate": 1.1493317616271327e-06,
      "loss": 0.3118,
      "step": 10528
    },
    {
      "epoch": 1.8266828591256072,
      "grad_norm": 0.7954789400100708,
      "learning_rate": 1.1470564653006844e-06,
      "loss": 0.42,
      "step": 10529
    },
    {
      "epoch": 1.826856349757113,
      "grad_norm": 0.8000457286834717,
      "learning_rate": 1.1447833568803036e-06,
      "loss": 0.5223,
      "step": 10530
    },
    {
      "epoch": 1.8270298403886192,
      "grad_norm": 0.8941777944564819,
      "learning_rate": 1.1425124366297858e-06,
      "loss": 0.4643,
      "step": 10531
    },
    {
      "epoch": 1.827203331020125,
      "grad_norm": 0.8192468285560608,
      "learning_rate": 1.140243704812667e-06,
      "loss": 0.5059,
      "step": 10532
    },
    {
      "epoch": 1.8273768216516308,
      "grad_norm": 0.8780628442764282,
      "learning_rate": 1.1379771616922474e-06,
      "loss": 0.4836,
      "step": 10533
    },
    {
      "epoch": 1.8275503122831367,
      "grad_norm": 1.4004952907562256,
      "learning_rate": 1.1357128075315572e-06,
      "loss": 0.458,
      "step": 10534
    },
    {
      "epoch": 1.8277238029146425,
      "grad_norm": 0.8280038833618164,
      "learning_rate": 1.1334506425933745e-06,
      "loss": 0.4286,
      "step": 10535
    },
    {
      "epoch": 1.8278972935461484,
      "grad_norm": 1.0875632762908936,
      "learning_rate": 1.1311906671402274e-06,
      "loss": 0.381,
      "step": 10536
    },
    {
      "epoch": 1.8280707841776545,
      "grad_norm": 1.107812762260437,
      "learning_rate": 1.1289328814343969e-06,
      "loss": 0.358,
      "step": 10537
    },
    {
      "epoch": 1.8282442748091603,
      "grad_norm": 0.6956422924995422,
      "learning_rate": 1.126677285737887e-06,
      "loss": 0.5237,
      "step": 10538
    },
    {
      "epoch": 1.8284177654406664,
      "grad_norm": 0.6917555332183838,
      "learning_rate": 1.124423880312473e-06,
      "loss": 0.4739,
      "step": 10539
    },
    {
      "epoch": 1.8285912560721722,
      "grad_norm": 0.9574361443519592,
      "learning_rate": 1.122172665419663e-06,
      "loss": 0.3891,
      "step": 10540
    },
    {
      "epoch": 1.828764746703678,
      "grad_norm": 1.0273165702819824,
      "learning_rate": 1.1199236413207127e-06,
      "loss": 0.3956,
      "step": 10541
    },
    {
      "epoch": 1.828938237335184,
      "grad_norm": 0.8338664770126343,
      "learning_rate": 1.1176768082766177e-06,
      "loss": 0.4883,
      "step": 10542
    },
    {
      "epoch": 1.8291117279666897,
      "grad_norm": 0.733159601688385,
      "learning_rate": 1.1154321665481404e-06,
      "loss": 0.412,
      "step": 10543
    },
    {
      "epoch": 1.8292852185981956,
      "grad_norm": 0.7883723974227905,
      "learning_rate": 1.1131897163957573e-06,
      "loss": 0.578,
      "step": 10544
    },
    {
      "epoch": 1.8294587092297014,
      "grad_norm": 0.8465474843978882,
      "learning_rate": 1.1109494580797175e-06,
      "loss": 0.4261,
      "step": 10545
    },
    {
      "epoch": 1.8296321998612075,
      "grad_norm": 0.6504952311515808,
      "learning_rate": 1.1087113918600023e-06,
      "loss": 0.5467,
      "step": 10546
    },
    {
      "epoch": 1.8298056904927134,
      "grad_norm": 0.9745897054672241,
      "learning_rate": 1.1064755179963483e-06,
      "loss": 0.4567,
      "step": 10547
    },
    {
      "epoch": 1.8299791811242194,
      "grad_norm": 0.9075683951377869,
      "learning_rate": 1.1042418367482188e-06,
      "loss": 0.4359,
      "step": 10548
    },
    {
      "epoch": 1.8301526717557253,
      "grad_norm": 0.8048148155212402,
      "learning_rate": 1.1020103483748469e-06,
      "loss": 0.5377,
      "step": 10549
    },
    {
      "epoch": 1.8303261623872311,
      "grad_norm": 0.8447327613830566,
      "learning_rate": 1.0997810531351916e-06,
      "loss": 0.5651,
      "step": 10550
    },
    {
      "epoch": 1.830499653018737,
      "grad_norm": 0.7904649376869202,
      "learning_rate": 1.0975539512879707e-06,
      "loss": 0.398,
      "step": 10551
    },
    {
      "epoch": 1.8306731436502428,
      "grad_norm": 0.7916994094848633,
      "learning_rate": 1.0953290430916353e-06,
      "loss": 0.5654,
      "step": 10552
    },
    {
      "epoch": 1.8308466342817487,
      "grad_norm": 0.932354211807251,
      "learning_rate": 1.0931063288043963e-06,
      "loss": 0.4965,
      "step": 10553
    },
    {
      "epoch": 1.8310201249132547,
      "grad_norm": 1.4943174123764038,
      "learning_rate": 1.0908858086841989e-06,
      "loss": 0.4603,
      "step": 10554
    },
    {
      "epoch": 1.8311936155447606,
      "grad_norm": 0.7021231055259705,
      "learning_rate": 1.0886674829887323e-06,
      "loss": 0.4224,
      "step": 10555
    },
    {
      "epoch": 1.8313671061762666,
      "grad_norm": 0.7502918839454651,
      "learning_rate": 1.0864513519754484e-06,
      "loss": 0.5375,
      "step": 10556
    },
    {
      "epoch": 1.8315405968077725,
      "grad_norm": 0.7647824883460999,
      "learning_rate": 1.0842374159015167e-06,
      "loss": 0.5237,
      "step": 10557
    },
    {
      "epoch": 1.8317140874392783,
      "grad_norm": 0.8950784802436829,
      "learning_rate": 1.082025675023879e-06,
      "loss": 0.4702,
      "step": 10558
    },
    {
      "epoch": 1.8318875780707842,
      "grad_norm": 0.7453234791755676,
      "learning_rate": 1.0798161295992004e-06,
      "loss": 0.5563,
      "step": 10559
    },
    {
      "epoch": 1.83206106870229,
      "grad_norm": 0.9420710802078247,
      "learning_rate": 1.077608779883912e-06,
      "loss": 0.4962,
      "step": 10560
    },
    {
      "epoch": 1.8322345593337959,
      "grad_norm": 2.714823007583618,
      "learning_rate": 1.0754036261341683e-06,
      "loss": 0.436,
      "step": 10561
    },
    {
      "epoch": 1.8324080499653017,
      "grad_norm": 0.8746506571769714,
      "learning_rate": 1.0732006686058893e-06,
      "loss": 0.3875,
      "step": 10562
    },
    {
      "epoch": 1.8325815405968078,
      "grad_norm": 1.7194308042526245,
      "learning_rate": 1.070999907554726e-06,
      "loss": 0.4247,
      "step": 10563
    },
    {
      "epoch": 1.8327550312283136,
      "grad_norm": 1.1054048538208008,
      "learning_rate": 1.0688013432360811e-06,
      "loss": 0.368,
      "step": 10564
    },
    {
      "epoch": 1.8329285218598197,
      "grad_norm": 1.1475105285644531,
      "learning_rate": 1.0666049759050945e-06,
      "loss": 0.3995,
      "step": 10565
    },
    {
      "epoch": 1.8331020124913255,
      "grad_norm": 0.7717334628105164,
      "learning_rate": 1.0644108058166692e-06,
      "loss": 0.4082,
      "step": 10566
    },
    {
      "epoch": 1.8332755031228314,
      "grad_norm": 0.8031683564186096,
      "learning_rate": 1.0622188332254302e-06,
      "loss": 0.5031,
      "step": 10567
    },
    {
      "epoch": 1.8334489937543372,
      "grad_norm": 0.905417799949646,
      "learning_rate": 1.0600290583857631e-06,
      "loss": 0.4075,
      "step": 10568
    },
    {
      "epoch": 1.833622484385843,
      "grad_norm": 0.850563108921051,
      "learning_rate": 1.0578414815517936e-06,
      "loss": 0.4917,
      "step": 10569
    },
    {
      "epoch": 1.833795975017349,
      "grad_norm": 0.6751877665519714,
      "learning_rate": 1.0556561029773914e-06,
      "loss": 0.3687,
      "step": 10570
    },
    {
      "epoch": 1.833969465648855,
      "grad_norm": 0.6808153986930847,
      "learning_rate": 1.0534729229161722e-06,
      "loss": 0.4904,
      "step": 10571
    },
    {
      "epoch": 1.8341429562803608,
      "grad_norm": 0.90809565782547,
      "learning_rate": 1.0512919416214995e-06,
      "loss": 0.4065,
      "step": 10572
    },
    {
      "epoch": 1.834316446911867,
      "grad_norm": 0.7460927367210388,
      "learning_rate": 1.0491131593464755e-06,
      "loss": 0.4163,
      "step": 10573
    },
    {
      "epoch": 1.8344899375433728,
      "grad_norm": 0.8532546162605286,
      "learning_rate": 1.0469365763439532e-06,
      "loss": 0.4336,
      "step": 10574
    },
    {
      "epoch": 1.8346634281748786,
      "grad_norm": 1.418191909790039,
      "learning_rate": 1.044762192866522e-06,
      "loss": 0.4834,
      "step": 10575
    },
    {
      "epoch": 1.8348369188063844,
      "grad_norm": 0.8263697028160095,
      "learning_rate": 1.0425900091665286e-06,
      "loss": 0.3685,
      "step": 10576
    },
    {
      "epoch": 1.8350104094378903,
      "grad_norm": 0.9749839305877686,
      "learning_rate": 1.040420025496054e-06,
      "loss": 0.4115,
      "step": 10577
    },
    {
      "epoch": 1.8351839000693961,
      "grad_norm": 0.9161428213119507,
      "learning_rate": 1.0382522421069274e-06,
      "loss": 0.3691,
      "step": 10578
    },
    {
      "epoch": 1.8353573907009022,
      "grad_norm": 0.8972198367118835,
      "learning_rate": 1.0360866592507236e-06,
      "loss": 0.5196,
      "step": 10579
    },
    {
      "epoch": 1.835530881332408,
      "grad_norm": 0.6500658392906189,
      "learning_rate": 1.033923277178759e-06,
      "loss": 0.5386,
      "step": 10580
    },
    {
      "epoch": 1.835704371963914,
      "grad_norm": 0.6305949687957764,
      "learning_rate": 1.0317620961420993e-06,
      "loss": 0.4969,
      "step": 10581
    },
    {
      "epoch": 1.83587786259542,
      "grad_norm": 0.6953607201576233,
      "learning_rate": 1.029603116391551e-06,
      "loss": 0.5178,
      "step": 10582
    },
    {
      "epoch": 1.8360513532269258,
      "grad_norm": 0.8355174660682678,
      "learning_rate": 1.0274463381776646e-06,
      "loss": 0.4069,
      "step": 10583
    },
    {
      "epoch": 1.8362248438584317,
      "grad_norm": 1.1380466222763062,
      "learning_rate": 1.0252917617507374e-06,
      "loss": 0.4872,
      "step": 10584
    },
    {
      "epoch": 1.8363983344899375,
      "grad_norm": 0.7222457528114319,
      "learning_rate": 1.0231393873608164e-06,
      "loss": 0.4034,
      "step": 10585
    },
    {
      "epoch": 1.8365718251214433,
      "grad_norm": 0.7789379358291626,
      "learning_rate": 1.020989215257675e-06,
      "loss": 0.4271,
      "step": 10586
    },
    {
      "epoch": 1.8367453157529492,
      "grad_norm": 0.8136478662490845,
      "learning_rate": 1.0188412456908537e-06,
      "loss": 0.4878,
      "step": 10587
    },
    {
      "epoch": 1.8369188063844553,
      "grad_norm": 0.7922045588493347,
      "learning_rate": 1.0166954789096194e-06,
      "loss": 0.4371,
      "step": 10588
    },
    {
      "epoch": 1.837092297015961,
      "grad_norm": 0.9087977409362793,
      "learning_rate": 1.0145519151630023e-06,
      "loss": 0.4856,
      "step": 10589
    },
    {
      "epoch": 1.8372657876474672,
      "grad_norm": 1.0198791027069092,
      "learning_rate": 1.0124105546997521e-06,
      "loss": 0.4105,
      "step": 10590
    },
    {
      "epoch": 1.837439278278973,
      "grad_norm": 0.6983714699745178,
      "learning_rate": 1.0102713977683832e-06,
      "loss": 0.4863,
      "step": 10591
    },
    {
      "epoch": 1.8376127689104789,
      "grad_norm": 1.7546323537826538,
      "learning_rate": 1.008134444617146e-06,
      "loss": 0.3851,
      "step": 10592
    },
    {
      "epoch": 1.8377862595419847,
      "grad_norm": 1.195565104484558,
      "learning_rate": 1.0059996954940377e-06,
      "loss": 0.4205,
      "step": 10593
    },
    {
      "epoch": 1.8379597501734906,
      "grad_norm": 1.0703043937683105,
      "learning_rate": 1.0038671506467934e-06,
      "loss": 0.3286,
      "step": 10594
    },
    {
      "epoch": 1.8381332408049964,
      "grad_norm": 1.0536229610443115,
      "learning_rate": 1.0017368103229086e-06,
      "loss": 0.3837,
      "step": 10595
    },
    {
      "epoch": 1.8383067314365025,
      "grad_norm": 0.873356819152832,
      "learning_rate": 9.996086747695966e-07,
      "loss": 0.3619,
      "step": 10596
    },
    {
      "epoch": 1.8384802220680083,
      "grad_norm": 0.8494886755943298,
      "learning_rate": 9.97482744233842e-07,
      "loss": 0.4105,
      "step": 10597
    },
    {
      "epoch": 1.8386537126995144,
      "grad_norm": 0.9837471842765808,
      "learning_rate": 9.953590189623563e-07,
      "loss": 0.4213,
      "step": 10598
    },
    {
      "epoch": 1.8388272033310202,
      "grad_norm": 0.7704948782920837,
      "learning_rate": 9.932374992016002e-07,
      "loss": 0.4456,
      "step": 10599
    },
    {
      "epoch": 1.839000693962526,
      "grad_norm": 0.7444039583206177,
      "learning_rate": 9.911181851977792e-07,
      "loss": 0.5393,
      "step": 10600
    },
    {
      "epoch": 1.839174184594032,
      "grad_norm": 0.9906989932060242,
      "learning_rate": 9.890010771968428e-07,
      "loss": 0.3533,
      "step": 10601
    },
    {
      "epoch": 1.8393476752255378,
      "grad_norm": 1.106204867362976,
      "learning_rate": 9.868861754444858e-07,
      "loss": 0.322,
      "step": 10602
    },
    {
      "epoch": 1.8395211658570436,
      "grad_norm": 1.4313024282455444,
      "learning_rate": 9.847734801861387e-07,
      "loss": 0.4437,
      "step": 10603
    },
    {
      "epoch": 1.8396946564885495,
      "grad_norm": 1.0267696380615234,
      "learning_rate": 9.826629916669917e-07,
      "loss": 0.4823,
      "step": 10604
    },
    {
      "epoch": 1.8398681471200555,
      "grad_norm": 0.7928594350814819,
      "learning_rate": 9.805547101319601e-07,
      "loss": 0.4402,
      "step": 10605
    },
    {
      "epoch": 1.8400416377515614,
      "grad_norm": 0.8010715842247009,
      "learning_rate": 9.784486358257194e-07,
      "loss": 0.4567,
      "step": 10606
    },
    {
      "epoch": 1.8402151283830674,
      "grad_norm": 0.7973601222038269,
      "learning_rate": 9.763447689926763e-07,
      "loss": 0.3728,
      "step": 10607
    },
    {
      "epoch": 1.8403886190145733,
      "grad_norm": 1.1704373359680176,
      "learning_rate": 9.742431098769933e-07,
      "loss": 0.3536,
      "step": 10608
    },
    {
      "epoch": 1.8405621096460791,
      "grad_norm": 0.74921715259552,
      "learning_rate": 9.721436587225618e-07,
      "loss": 0.4862,
      "step": 10609
    },
    {
      "epoch": 1.840735600277585,
      "grad_norm": 1.003538966178894,
      "learning_rate": 9.700464157730338e-07,
      "loss": 0.3729,
      "step": 10610
    },
    {
      "epoch": 1.8409090909090908,
      "grad_norm": 0.8335089087486267,
      "learning_rate": 9.679513812717945e-07,
      "loss": 0.4648,
      "step": 10611
    },
    {
      "epoch": 1.8410825815405967,
      "grad_norm": 0.7531124949455261,
      "learning_rate": 9.658585554619737e-07,
      "loss": 0.3383,
      "step": 10612
    },
    {
      "epoch": 1.8412560721721027,
      "grad_norm": 0.68199223279953,
      "learning_rate": 9.637679385864417e-07,
      "loss": 0.4976,
      "step": 10613
    },
    {
      "epoch": 1.8414295628036086,
      "grad_norm": 1.115538239479065,
      "learning_rate": 9.616795308878313e-07,
      "loss": 0.3687,
      "step": 10614
    },
    {
      "epoch": 1.8416030534351147,
      "grad_norm": 1.10149347782135,
      "learning_rate": 9.59593332608486e-07,
      "loss": 0.3217,
      "step": 10615
    },
    {
      "epoch": 1.8417765440666205,
      "grad_norm": 0.9675643444061279,
      "learning_rate": 9.575093439905259e-07,
      "loss": 0.5491,
      "step": 10616
    },
    {
      "epoch": 1.8419500346981263,
      "grad_norm": 0.7751714587211609,
      "learning_rate": 9.554275652757928e-07,
      "loss": 0.5919,
      "step": 10617
    },
    {
      "epoch": 1.8421235253296322,
      "grad_norm": 0.9060567617416382,
      "learning_rate": 9.533479967058867e-07,
      "loss": 0.4501,
      "step": 10618
    },
    {
      "epoch": 1.842297015961138,
      "grad_norm": 0.73239666223526,
      "learning_rate": 9.512706385221348e-07,
      "loss": 0.4877,
      "step": 10619
    },
    {
      "epoch": 1.8424705065926439,
      "grad_norm": 1.1301404237747192,
      "learning_rate": 9.491954909656242e-07,
      "loss": 0.4669,
      "step": 10620
    },
    {
      "epoch": 1.8426439972241497,
      "grad_norm": 0.5790345668792725,
      "learning_rate": 9.471225542771734e-07,
      "loss": 0.4295,
      "step": 10621
    },
    {
      "epoch": 1.8428174878556558,
      "grad_norm": 0.8333771824836731,
      "learning_rate": 9.450518286973542e-07,
      "loss": 0.4653,
      "step": 10622
    },
    {
      "epoch": 1.8429909784871616,
      "grad_norm": 1.0313693284988403,
      "learning_rate": 9.42983314466468e-07,
      "loss": 0.4835,
      "step": 10623
    },
    {
      "epoch": 1.8431644691186677,
      "grad_norm": 0.8666651248931885,
      "learning_rate": 9.409170118245803e-07,
      "loss": 0.5321,
      "step": 10624
    },
    {
      "epoch": 1.8433379597501736,
      "grad_norm": 0.7337196469306946,
      "learning_rate": 9.388529210114794e-07,
      "loss": 0.6223,
      "step": 10625
    },
    {
      "epoch": 1.8435114503816794,
      "grad_norm": 0.6616241335868835,
      "learning_rate": 9.36791042266707e-07,
      "loss": 0.4851,
      "step": 10626
    },
    {
      "epoch": 1.8436849410131853,
      "grad_norm": 0.7307387590408325,
      "learning_rate": 9.347313758295473e-07,
      "loss": 0.451,
      "step": 10627
    },
    {
      "epoch": 1.843858431644691,
      "grad_norm": 0.9073920249938965,
      "learning_rate": 9.326739219390246e-07,
      "loss": 0.5051,
      "step": 10628
    },
    {
      "epoch": 1.844031922276197,
      "grad_norm": 1.2759959697723389,
      "learning_rate": 9.306186808339146e-07,
      "loss": 0.4465,
      "step": 10629
    },
    {
      "epoch": 1.844205412907703,
      "grad_norm": 0.6395655274391174,
      "learning_rate": 9.285656527527264e-07,
      "loss": 0.6111,
      "step": 10630
    },
    {
      "epoch": 1.8443789035392089,
      "grad_norm": 0.8483899831771851,
      "learning_rate": 9.265148379337164e-07,
      "loss": 0.4337,
      "step": 10631
    },
    {
      "epoch": 1.844552394170715,
      "grad_norm": 0.9316275119781494,
      "learning_rate": 9.244662366148826e-07,
      "loss": 0.399,
      "step": 10632
    },
    {
      "epoch": 1.8447258848022208,
      "grad_norm": 1.5879032611846924,
      "learning_rate": 9.224198490339731e-07,
      "loss": 0.4324,
      "step": 10633
    },
    {
      "epoch": 1.8448993754337266,
      "grad_norm": 1.099420428276062,
      "learning_rate": 9.203756754284665e-07,
      "loss": 0.4224,
      "step": 10634
    },
    {
      "epoch": 1.8450728660652325,
      "grad_norm": 0.6880114078521729,
      "learning_rate": 9.183337160355976e-07,
      "loss": 0.5272,
      "step": 10635
    },
    {
      "epoch": 1.8452463566967383,
      "grad_norm": 0.8261739015579224,
      "learning_rate": 9.162939710923324e-07,
      "loss": 0.5243,
      "step": 10636
    },
    {
      "epoch": 1.8454198473282442,
      "grad_norm": 0.9456450343132019,
      "learning_rate": 9.142564408353949e-07,
      "loss": 0.2906,
      "step": 10637
    },
    {
      "epoch": 1.8455933379597502,
      "grad_norm": 0.7066915035247803,
      "learning_rate": 9.122211255012292e-07,
      "loss": 0.4951,
      "step": 10638
    },
    {
      "epoch": 1.845766828591256,
      "grad_norm": 0.825546145439148,
      "learning_rate": 9.101880253260487e-07,
      "loss": 0.3624,
      "step": 10639
    },
    {
      "epoch": 1.845940319222762,
      "grad_norm": 0.9259288311004639,
      "learning_rate": 9.081571405457912e-07,
      "loss": 0.3451,
      "step": 10640
    },
    {
      "epoch": 1.846113809854268,
      "grad_norm": 0.8269305229187012,
      "learning_rate": 9.061284713961416e-07,
      "loss": 0.3447,
      "step": 10641
    },
    {
      "epoch": 1.8462873004857738,
      "grad_norm": 0.779420018196106,
      "learning_rate": 9.041020181125315e-07,
      "loss": 0.4613,
      "step": 10642
    },
    {
      "epoch": 1.8464607911172797,
      "grad_norm": 0.8219154477119446,
      "learning_rate": 9.020777809301396e-07,
      "loss": 0.4539,
      "step": 10643
    },
    {
      "epoch": 1.8466342817487855,
      "grad_norm": 0.8316527605056763,
      "learning_rate": 9.000557600838666e-07,
      "loss": 0.494,
      "step": 10644
    },
    {
      "epoch": 1.8468077723802914,
      "grad_norm": 0.8427203297615051,
      "learning_rate": 8.980359558083828e-07,
      "loss": 0.3285,
      "step": 10645
    },
    {
      "epoch": 1.8469812630117972,
      "grad_norm": 0.9683478474617004,
      "learning_rate": 8.960183683380807e-07,
      "loss": 0.3295,
      "step": 10646
    },
    {
      "epoch": 1.8471547536433033,
      "grad_norm": 0.7695950865745544,
      "learning_rate": 8.940029979071152e-07,
      "loss": 0.4235,
      "step": 10647
    },
    {
      "epoch": 1.8473282442748091,
      "grad_norm": 1.1187565326690674,
      "learning_rate": 8.919898447493569e-07,
      "loss": 0.4458,
      "step": 10648
    },
    {
      "epoch": 1.8475017349063152,
      "grad_norm": 0.9226999878883362,
      "learning_rate": 8.899789090984457e-07,
      "loss": 0.4841,
      "step": 10649
    },
    {
      "epoch": 1.847675225537821,
      "grad_norm": 0.8049643635749817,
      "learning_rate": 8.879701911877503e-07,
      "loss": 0.4939,
      "step": 10650
    },
    {
      "epoch": 1.8478487161693269,
      "grad_norm": 0.8780126571655273,
      "learning_rate": 8.859636912503822e-07,
      "loss": 0.3636,
      "step": 10651
    },
    {
      "epoch": 1.8480222068008327,
      "grad_norm": 1.5426934957504272,
      "learning_rate": 8.839594095191995e-07,
      "loss": 0.3875,
      "step": 10652
    },
    {
      "epoch": 1.8481956974323386,
      "grad_norm": 0.9760283827781677,
      "learning_rate": 8.81957346226805e-07,
      "loss": 0.379,
      "step": 10653
    },
    {
      "epoch": 1.8483691880638444,
      "grad_norm": 1.1850718259811401,
      "learning_rate": 8.799575016055373e-07,
      "loss": 0.4696,
      "step": 10654
    },
    {
      "epoch": 1.8485426786953505,
      "grad_norm": 0.9418070316314697,
      "learning_rate": 8.779598758874774e-07,
      "loss": 0.429,
      "step": 10655
    },
    {
      "epoch": 1.8487161693268563,
      "grad_norm": 0.9057166576385498,
      "learning_rate": 8.7596446930446e-07,
      "loss": 0.4333,
      "step": 10656
    },
    {
      "epoch": 1.8488896599583624,
      "grad_norm": 1.0457319021224976,
      "learning_rate": 8.739712820880441e-07,
      "loss": 0.4059,
      "step": 10657
    },
    {
      "epoch": 1.8490631505898683,
      "grad_norm": 1.0274351835250854,
      "learning_rate": 8.719803144695516e-07,
      "loss": 0.3586,
      "step": 10658
    },
    {
      "epoch": 1.849236641221374,
      "grad_norm": 0.9383352398872375,
      "learning_rate": 8.699915666800285e-07,
      "loss": 0.4126,
      "step": 10659
    },
    {
      "epoch": 1.84941013185288,
      "grad_norm": 1.4329230785369873,
      "learning_rate": 8.680050389502814e-07,
      "loss": 0.4045,
      "step": 10660
    },
    {
      "epoch": 1.8495836224843858,
      "grad_norm": 1.739837408065796,
      "learning_rate": 8.660207315108371e-07,
      "loss": 0.3916,
      "step": 10661
    },
    {
      "epoch": 1.8497571131158916,
      "grad_norm": 1.2653837203979492,
      "learning_rate": 8.640386445919847e-07,
      "loss": 0.313,
      "step": 10662
    },
    {
      "epoch": 1.8499306037473975,
      "grad_norm": 1.0398705005645752,
      "learning_rate": 8.620587784237444e-07,
      "loss": 0.4147,
      "step": 10663
    },
    {
      "epoch": 1.8501040943789036,
      "grad_norm": 1.1222190856933594,
      "learning_rate": 8.600811332358861e-07,
      "loss": 0.4729,
      "step": 10664
    },
    {
      "epoch": 1.8502775850104094,
      "grad_norm": 0.9458642601966858,
      "learning_rate": 8.581057092579081e-07,
      "loss": 0.531,
      "step": 10665
    },
    {
      "epoch": 1.8504510756419155,
      "grad_norm": 1.0276833772659302,
      "learning_rate": 8.561325067190762e-07,
      "loss": 0.373,
      "step": 10666
    },
    {
      "epoch": 1.8506245662734213,
      "grad_norm": 1.9846408367156982,
      "learning_rate": 8.541615258483671e-07,
      "loss": 0.3614,
      "step": 10667
    },
    {
      "epoch": 1.8507980569049272,
      "grad_norm": 1.0455187559127808,
      "learning_rate": 8.521927668745244e-07,
      "loss": 0.4457,
      "step": 10668
    },
    {
      "epoch": 1.850971547536433,
      "grad_norm": 0.6620354652404785,
      "learning_rate": 8.502262300260234e-07,
      "loss": 0.5068,
      "step": 10669
    },
    {
      "epoch": 1.8511450381679388,
      "grad_norm": 0.9360253810882568,
      "learning_rate": 8.482619155310812e-07,
      "loss": 0.3726,
      "step": 10670
    },
    {
      "epoch": 1.8513185287994447,
      "grad_norm": 1.065224528312683,
      "learning_rate": 8.462998236176578e-07,
      "loss": 0.3748,
      "step": 10671
    },
    {
      "epoch": 1.8514920194309508,
      "grad_norm": 0.7790229916572571,
      "learning_rate": 8.443399545134623e-07,
      "loss": 0.5011,
      "step": 10672
    },
    {
      "epoch": 1.8516655100624566,
      "grad_norm": 7.807602405548096,
      "learning_rate": 8.423823084459349e-07,
      "loss": 0.4644,
      "step": 10673
    },
    {
      "epoch": 1.8518390006939627,
      "grad_norm": 0.8364564180374146,
      "learning_rate": 8.404268856422626e-07,
      "loss": 0.4195,
      "step": 10674
    },
    {
      "epoch": 1.8520124913254685,
      "grad_norm": 0.9146015048027039,
      "learning_rate": 8.384736863293729e-07,
      "loss": 0.3894,
      "step": 10675
    },
    {
      "epoch": 1.8521859819569744,
      "grad_norm": 1.106425404548645,
      "learning_rate": 8.365227107339447e-07,
      "loss": 0.4592,
      "step": 10676
    },
    {
      "epoch": 1.8523594725884802,
      "grad_norm": 0.6851017475128174,
      "learning_rate": 8.345739590823832e-07,
      "loss": 0.5092,
      "step": 10677
    },
    {
      "epoch": 1.852532963219986,
      "grad_norm": 1.0080454349517822,
      "learning_rate": 8.326274316008475e-07,
      "loss": 0.3457,
      "step": 10678
    },
    {
      "epoch": 1.852706453851492,
      "grad_norm": 1.062127947807312,
      "learning_rate": 8.306831285152328e-07,
      "loss": 0.5371,
      "step": 10679
    },
    {
      "epoch": 1.8528799444829978,
      "grad_norm": 1.255875587463379,
      "learning_rate": 8.287410500511739e-07,
      "loss": 0.4425,
      "step": 10680
    },
    {
      "epoch": 1.8530534351145038,
      "grad_norm": 0.6575207710266113,
      "learning_rate": 8.268011964340595e-07,
      "loss": 0.4111,
      "step": 10681
    },
    {
      "epoch": 1.8532269257460097,
      "grad_norm": 0.9395532011985779,
      "learning_rate": 8.248635678890049e-07,
      "loss": 0.3969,
      "step": 10682
    },
    {
      "epoch": 1.8534004163775157,
      "grad_norm": 1.6537333726882935,
      "learning_rate": 8.229281646408793e-07,
      "loss": 0.5719,
      "step": 10683
    },
    {
      "epoch": 1.8535739070090216,
      "grad_norm": 0.6713128089904785,
      "learning_rate": 8.209949869142808e-07,
      "loss": 0.6195,
      "step": 10684
    },
    {
      "epoch": 1.8537473976405274,
      "grad_norm": 0.8929294347763062,
      "learning_rate": 8.190640349335699e-07,
      "loss": 0.4669,
      "step": 10685
    },
    {
      "epoch": 1.8539208882720333,
      "grad_norm": 0.7345629334449768,
      "learning_rate": 8.171353089228206e-07,
      "loss": 0.5186,
      "step": 10686
    },
    {
      "epoch": 1.8540943789035391,
      "grad_norm": 0.961094856262207,
      "learning_rate": 8.152088091058741e-07,
      "loss": 0.39,
      "step": 10687
    },
    {
      "epoch": 1.854267869535045,
      "grad_norm": 1.1825238466262817,
      "learning_rate": 8.132845357062979e-07,
      "loss": 0.4435,
      "step": 10688
    },
    {
      "epoch": 1.854441360166551,
      "grad_norm": 0.9082005620002747,
      "learning_rate": 8.113624889474136e-07,
      "loss": 0.3444,
      "step": 10689
    },
    {
      "epoch": 1.8546148507980569,
      "grad_norm": 0.8415492177009583,
      "learning_rate": 8.094426690522672e-07,
      "loss": 0.3608,
      "step": 10690
    },
    {
      "epoch": 1.854788341429563,
      "grad_norm": 0.5826301574707031,
      "learning_rate": 8.075250762436626e-07,
      "loss": 0.4982,
      "step": 10691
    },
    {
      "epoch": 1.8549618320610688,
      "grad_norm": 0.8230922818183899,
      "learning_rate": 8.056097107441352e-07,
      "loss": 0.4164,
      "step": 10692
    },
    {
      "epoch": 1.8551353226925746,
      "grad_norm": 1.342071771621704,
      "learning_rate": 8.036965727759693e-07,
      "loss": 0.3766,
      "step": 10693
    },
    {
      "epoch": 1.8553088133240805,
      "grad_norm": 0.7745541334152222,
      "learning_rate": 8.017856625611809e-07,
      "loss": 0.353,
      "step": 10694
    },
    {
      "epoch": 1.8554823039555863,
      "grad_norm": 0.8241654634475708,
      "learning_rate": 7.998769803215389e-07,
      "loss": 0.3567,
      "step": 10695
    },
    {
      "epoch": 1.8556557945870922,
      "grad_norm": 1.0320768356323242,
      "learning_rate": 7.979705262785442e-07,
      "loss": 0.3619,
      "step": 10696
    },
    {
      "epoch": 1.8558292852185982,
      "grad_norm": 0.924490749835968,
      "learning_rate": 7.960663006534464e-07,
      "loss": 0.4421,
      "step": 10697
    },
    {
      "epoch": 1.856002775850104,
      "grad_norm": 0.9098055362701416,
      "learning_rate": 7.941643036672309e-07,
      "loss": 0.5029,
      "step": 10698
    },
    {
      "epoch": 1.85617626648161,
      "grad_norm": 0.7238054871559143,
      "learning_rate": 7.922645355406256e-07,
      "loss": 0.4734,
      "step": 10699
    },
    {
      "epoch": 1.856349757113116,
      "grad_norm": 1.7676187753677368,
      "learning_rate": 7.903669964941052e-07,
      "loss": 0.4181,
      "step": 10700
    },
    {
      "epoch": 1.8565232477446219,
      "grad_norm": 1.4307605028152466,
      "learning_rate": 7.884716867478782e-07,
      "loss": 0.4248,
      "step": 10701
    },
    {
      "epoch": 1.8566967383761277,
      "grad_norm": 0.7783779501914978,
      "learning_rate": 7.865786065218973e-07,
      "loss": 0.5836,
      "step": 10702
    },
    {
      "epoch": 1.8568702290076335,
      "grad_norm": 0.778427004814148,
      "learning_rate": 7.846877560358535e-07,
      "loss": 0.3923,
      "step": 10703
    },
    {
      "epoch": 1.8570437196391394,
      "grad_norm": 1.9426189661026,
      "learning_rate": 7.827991355091891e-07,
      "loss": 0.4073,
      "step": 10704
    },
    {
      "epoch": 1.8572172102706452,
      "grad_norm": 1.0605652332305908,
      "learning_rate": 7.809127451610776e-07,
      "loss": 0.4138,
      "step": 10705
    },
    {
      "epoch": 1.8573907009021513,
      "grad_norm": 1.0573015213012695,
      "learning_rate": 7.790285852104373e-07,
      "loss": 0.3669,
      "step": 10706
    },
    {
      "epoch": 1.8575641915336571,
      "grad_norm": 0.973831057548523,
      "learning_rate": 7.771466558759244e-07,
      "loss": 0.4637,
      "step": 10707
    },
    {
      "epoch": 1.8577376821651632,
      "grad_norm": 0.8816852569580078,
      "learning_rate": 7.752669573759464e-07,
      "loss": 0.417,
      "step": 10708
    },
    {
      "epoch": 1.857911172796669,
      "grad_norm": 1.2028803825378418,
      "learning_rate": 7.733894899286332e-07,
      "loss": 0.417,
      "step": 10709
    },
    {
      "epoch": 1.858084663428175,
      "grad_norm": 0.7927412986755371,
      "learning_rate": 7.715142537518771e-07,
      "loss": 0.4883,
      "step": 10710
    },
    {
      "epoch": 1.8582581540596808,
      "grad_norm": 1.1162534952163696,
      "learning_rate": 7.696412490632954e-07,
      "loss": 0.4596,
      "step": 10711
    },
    {
      "epoch": 1.8584316446911866,
      "grad_norm": 0.8915563821792603,
      "learning_rate": 7.677704760802562e-07,
      "loss": 0.4584,
      "step": 10712
    },
    {
      "epoch": 1.8586051353226924,
      "grad_norm": 0.8360008001327515,
      "learning_rate": 7.659019350198593e-07,
      "loss": 0.5146,
      "step": 10713
    },
    {
      "epoch": 1.8587786259541985,
      "grad_norm": 0.9344514012336731,
      "learning_rate": 7.640356260989601e-07,
      "loss": 0.3469,
      "step": 10714
    },
    {
      "epoch": 1.8589521165857044,
      "grad_norm": 0.9535170793533325,
      "learning_rate": 7.621715495341364e-07,
      "loss": 0.42,
      "step": 10715
    },
    {
      "epoch": 1.8591256072172104,
      "grad_norm": 0.7365037798881531,
      "learning_rate": 7.603097055417242e-07,
      "loss": 0.5088,
      "step": 10716
    },
    {
      "epoch": 1.8592990978487163,
      "grad_norm": 0.8689652681350708,
      "learning_rate": 7.58450094337786e-07,
      "loss": 0.5477,
      "step": 10717
    },
    {
      "epoch": 1.8594725884802221,
      "grad_norm": 0.8216843605041504,
      "learning_rate": 7.565927161381403e-07,
      "loss": 0.3713,
      "step": 10718
    },
    {
      "epoch": 1.859646079111728,
      "grad_norm": 0.8870757818222046,
      "learning_rate": 7.547375711583282e-07,
      "loss": 0.3389,
      "step": 10719
    },
    {
      "epoch": 1.8598195697432338,
      "grad_norm": 1.0497157573699951,
      "learning_rate": 7.528846596136485e-07,
      "loss": 0.4905,
      "step": 10720
    },
    {
      "epoch": 1.8599930603747397,
      "grad_norm": 0.6659301519393921,
      "learning_rate": 7.510339817191314e-07,
      "loss": 0.5076,
      "step": 10721
    },
    {
      "epoch": 1.8601665510062455,
      "grad_norm": 0.7435609698295593,
      "learning_rate": 7.491855376895519e-07,
      "loss": 0.4567,
      "step": 10722
    },
    {
      "epoch": 1.8603400416377516,
      "grad_norm": 0.8212759494781494,
      "learning_rate": 7.473393277394181e-07,
      "loss": 0.4438,
      "step": 10723
    },
    {
      "epoch": 1.8605135322692574,
      "grad_norm": 0.7654137015342712,
      "learning_rate": 7.454953520829899e-07,
      "loss": 0.563,
      "step": 10724
    },
    {
      "epoch": 1.8606870229007635,
      "grad_norm": 1.0974982976913452,
      "learning_rate": 7.436536109342651e-07,
      "loss": 0.4906,
      "step": 10725
    },
    {
      "epoch": 1.8608605135322693,
      "grad_norm": 0.7998368144035339,
      "learning_rate": 7.418141045069727e-07,
      "loss": 0.363,
      "step": 10726
    },
    {
      "epoch": 1.8610340041637752,
      "grad_norm": 0.9827405214309692,
      "learning_rate": 7.399768330145995e-07,
      "loss": 0.3574,
      "step": 10727
    },
    {
      "epoch": 1.861207494795281,
      "grad_norm": 0.8307923078536987,
      "learning_rate": 7.381417966703508e-07,
      "loss": 0.5055,
      "step": 10728
    },
    {
      "epoch": 1.8613809854267869,
      "grad_norm": 1.0587093830108643,
      "learning_rate": 7.363089956871961e-07,
      "loss": 0.4659,
      "step": 10729
    },
    {
      "epoch": 1.8615544760582927,
      "grad_norm": 0.9478718042373657,
      "learning_rate": 7.344784302778274e-07,
      "loss": 0.3449,
      "step": 10730
    },
    {
      "epoch": 1.8617279666897988,
      "grad_norm": 0.7864168286323547,
      "learning_rate": 7.32650100654686e-07,
      "loss": 0.4905,
      "step": 10731
    },
    {
      "epoch": 1.8619014573213046,
      "grad_norm": 0.7862717509269714,
      "learning_rate": 7.308240070299489e-07,
      "loss": 0.4059,
      "step": 10732
    },
    {
      "epoch": 1.8620749479528107,
      "grad_norm": 0.7209656834602356,
      "learning_rate": 7.290001496155418e-07,
      "loss": 0.5417,
      "step": 10733
    },
    {
      "epoch": 1.8622484385843165,
      "grad_norm": 1.1701611280441284,
      "learning_rate": 7.271785286231204e-07,
      "loss": 0.4844,
      "step": 10734
    },
    {
      "epoch": 1.8624219292158224,
      "grad_norm": 0.6251929998397827,
      "learning_rate": 7.253591442640906e-07,
      "loss": 0.5845,
      "step": 10735
    },
    {
      "epoch": 1.8625954198473282,
      "grad_norm": 0.725080132484436,
      "learning_rate": 7.235419967495883e-07,
      "loss": 0.5399,
      "step": 10736
    },
    {
      "epoch": 1.862768910478834,
      "grad_norm": 0.704907238483429,
      "learning_rate": 7.217270862905023e-07,
      "loss": 0.463,
      "step": 10737
    },
    {
      "epoch": 1.86294240111034,
      "grad_norm": 0.8071426153182983,
      "learning_rate": 7.199144130974489e-07,
      "loss": 0.4295,
      "step": 10738
    },
    {
      "epoch": 1.8631158917418458,
      "grad_norm": 2.144774913787842,
      "learning_rate": 7.181039773807952e-07,
      "loss": 0.3875,
      "step": 10739
    },
    {
      "epoch": 1.8632893823733518,
      "grad_norm": 0.7036840319633484,
      "learning_rate": 7.16295779350642e-07,
      "loss": 0.4941,
      "step": 10740
    },
    {
      "epoch": 1.8634628730048577,
      "grad_norm": 0.6882051229476929,
      "learning_rate": 7.144898192168348e-07,
      "loss": 0.4265,
      "step": 10741
    },
    {
      "epoch": 1.8636363636363638,
      "grad_norm": 0.798976719379425,
      "learning_rate": 7.126860971889527e-07,
      "loss": 0.3959,
      "step": 10742
    },
    {
      "epoch": 1.8638098542678696,
      "grad_norm": 0.7891309261322021,
      "learning_rate": 7.108846134763258e-07,
      "loss": 0.494,
      "step": 10743
    },
    {
      "epoch": 1.8639833448993754,
      "grad_norm": 0.6543323397636414,
      "learning_rate": 7.090853682880161e-07,
      "loss": 0.4839,
      "step": 10744
    },
    {
      "epoch": 1.8641568355308813,
      "grad_norm": 1.4919569492340088,
      "learning_rate": 7.072883618328275e-07,
      "loss": 0.3905,
      "step": 10745
    },
    {
      "epoch": 1.8643303261623871,
      "grad_norm": 0.8552188277244568,
      "learning_rate": 7.054935943193042e-07,
      "loss": 0.3601,
      "step": 10746
    },
    {
      "epoch": 1.864503816793893,
      "grad_norm": 1.8037493228912354,
      "learning_rate": 7.037010659557309e-07,
      "loss": 0.3279,
      "step": 10747
    },
    {
      "epoch": 1.864677307425399,
      "grad_norm": 0.9860910773277283,
      "learning_rate": 7.019107769501366e-07,
      "loss": 0.3587,
      "step": 10748
    },
    {
      "epoch": 1.864850798056905,
      "grad_norm": 0.6967023611068726,
      "learning_rate": 7.001227275102818e-07,
      "loss": 0.3929,
      "step": 10749
    },
    {
      "epoch": 1.865024288688411,
      "grad_norm": 1.349442481994629,
      "learning_rate": 6.983369178436739e-07,
      "loss": 0.4524,
      "step": 10750
    },
    {
      "epoch": 1.8651977793199168,
      "grad_norm": 0.9055932760238647,
      "learning_rate": 6.96553348157556e-07,
      "loss": 0.363,
      "step": 10751
    },
    {
      "epoch": 1.8653712699514227,
      "grad_norm": 1.137967586517334,
      "learning_rate": 6.947720186589158e-07,
      "loss": 0.4176,
      "step": 10752
    },
    {
      "epoch": 1.8655447605829285,
      "grad_norm": 0.9633527994155884,
      "learning_rate": 6.92992929554479e-07,
      "loss": 0.5557,
      "step": 10753
    },
    {
      "epoch": 1.8657182512144344,
      "grad_norm": 0.6777358055114746,
      "learning_rate": 6.912160810507096e-07,
      "loss": 0.4574,
      "step": 10754
    },
    {
      "epoch": 1.8658917418459402,
      "grad_norm": 2.1395766735076904,
      "learning_rate": 6.894414733538113e-07,
      "loss": 0.4868,
      "step": 10755
    },
    {
      "epoch": 1.8660652324774463,
      "grad_norm": 0.7489055395126343,
      "learning_rate": 6.876691066697349e-07,
      "loss": 0.4636,
      "step": 10756
    },
    {
      "epoch": 1.8662387231089521,
      "grad_norm": 1.3545432090759277,
      "learning_rate": 6.858989812041583e-07,
      "loss": 0.4578,
      "step": 10757
    },
    {
      "epoch": 1.8664122137404582,
      "grad_norm": 1.128914475440979,
      "learning_rate": 6.841310971625103e-07,
      "loss": 0.3976,
      "step": 10758
    },
    {
      "epoch": 1.866585704371964,
      "grad_norm": 1.8644378185272217,
      "learning_rate": 6.823654547499581e-07,
      "loss": 0.3995,
      "step": 10759
    },
    {
      "epoch": 1.8667591950034699,
      "grad_norm": 0.9472402334213257,
      "learning_rate": 6.806020541714042e-07,
      "loss": 0.4569,
      "step": 10760
    },
    {
      "epoch": 1.8669326856349757,
      "grad_norm": 0.7936835885047913,
      "learning_rate": 6.788408956314918e-07,
      "loss": 0.4657,
      "step": 10761
    },
    {
      "epoch": 1.8671061762664816,
      "grad_norm": 0.8923001885414124,
      "learning_rate": 6.770819793346084e-07,
      "loss": 0.4745,
      "step": 10762
    },
    {
      "epoch": 1.8672796668979874,
      "grad_norm": 0.9739224910736084,
      "learning_rate": 6.753253054848774e-07,
      "loss": 0.397,
      "step": 10763
    },
    {
      "epoch": 1.8674531575294933,
      "grad_norm": 0.834381639957428,
      "learning_rate": 6.735708742861624e-07,
      "loss": 0.4258,
      "step": 10764
    },
    {
      "epoch": 1.8676266481609993,
      "grad_norm": 0.8976614475250244,
      "learning_rate": 6.71818685942065e-07,
      "loss": 0.4316,
      "step": 10765
    },
    {
      "epoch": 1.8678001387925052,
      "grad_norm": 0.8833864331245422,
      "learning_rate": 6.700687406559359e-07,
      "loss": 0.4022,
      "step": 10766
    },
    {
      "epoch": 1.8679736294240112,
      "grad_norm": 0.7446100115776062,
      "learning_rate": 6.683210386308481e-07,
      "loss": 0.4232,
      "step": 10767
    },
    {
      "epoch": 1.868147120055517,
      "grad_norm": 1.0145565271377563,
      "learning_rate": 6.665755800696305e-07,
      "loss": 0.4338,
      "step": 10768
    },
    {
      "epoch": 1.868320610687023,
      "grad_norm": 0.8664763569831848,
      "learning_rate": 6.648323651748457e-07,
      "loss": 0.3817,
      "step": 10769
    },
    {
      "epoch": 1.8684941013185288,
      "grad_norm": 0.8469116687774658,
      "learning_rate": 6.63091394148796e-07,
      "loss": 0.4214,
      "step": 10770
    },
    {
      "epoch": 1.8686675919500346,
      "grad_norm": 0.8851863741874695,
      "learning_rate": 6.6135266719352e-07,
      "loss": 0.5551,
      "step": 10771
    },
    {
      "epoch": 1.8688410825815405,
      "grad_norm": 0.9367919564247131,
      "learning_rate": 6.596161845108006e-07,
      "loss": 0.511,
      "step": 10772
    },
    {
      "epoch": 1.8690145732130465,
      "grad_norm": 0.6908042430877686,
      "learning_rate": 6.578819463021612e-07,
      "loss": 0.5587,
      "step": 10773
    },
    {
      "epoch": 1.8691880638445524,
      "grad_norm": 0.7423384785652161,
      "learning_rate": 6.561499527688586e-07,
      "loss": 0.5619,
      "step": 10774
    },
    {
      "epoch": 1.8693615544760585,
      "grad_norm": 0.6981567740440369,
      "learning_rate": 6.544202041118941e-07,
      "loss": 0.5239,
      "step": 10775
    },
    {
      "epoch": 1.8695350451075643,
      "grad_norm": 0.6232300400733948,
      "learning_rate": 6.526927005320072e-07,
      "loss": 0.5656,
      "step": 10776
    },
    {
      "epoch": 1.8697085357390701,
      "grad_norm": 0.9474436044692993,
      "learning_rate": 6.509674422296775e-07,
      "loss": 0.4579,
      "step": 10777
    },
    {
      "epoch": 1.869882026370576,
      "grad_norm": 0.9173471331596375,
      "learning_rate": 6.492444294051204e-07,
      "loss": 0.4402,
      "step": 10778
    },
    {
      "epoch": 1.8700555170020818,
      "grad_norm": 0.9363657832145691,
      "learning_rate": 6.47523662258298e-07,
      "loss": 0.3816,
      "step": 10779
    },
    {
      "epoch": 1.8702290076335877,
      "grad_norm": 0.9286483526229858,
      "learning_rate": 6.458051409889021e-07,
      "loss": 0.3766,
      "step": 10780
    },
    {
      "epoch": 1.8704024982650935,
      "grad_norm": 0.7287443280220032,
      "learning_rate": 6.440888657963729e-07,
      "loss": 0.3782,
      "step": 10781
    },
    {
      "epoch": 1.8705759888965996,
      "grad_norm": 0.7181859612464905,
      "learning_rate": 6.423748368798843e-07,
      "loss": 0.4519,
      "step": 10782
    },
    {
      "epoch": 1.8707494795281054,
      "grad_norm": 0.7785375714302063,
      "learning_rate": 6.40663054438353e-07,
      "loss": 0.3562,
      "step": 10783
    },
    {
      "epoch": 1.8709229701596115,
      "grad_norm": 0.8017449975013733,
      "learning_rate": 6.38953518670431e-07,
      "loss": 0.4777,
      "step": 10784
    },
    {
      "epoch": 1.8710964607911174,
      "grad_norm": 1.0029230117797852,
      "learning_rate": 6.372462297745152e-07,
      "loss": 0.5099,
      "step": 10785
    },
    {
      "epoch": 1.8712699514226232,
      "grad_norm": 0.9567710161209106,
      "learning_rate": 6.355411879487339e-07,
      "loss": 0.3909,
      "step": 10786
    },
    {
      "epoch": 1.871443442054129,
      "grad_norm": 0.7558448910713196,
      "learning_rate": 6.338383933909642e-07,
      "loss": 0.3772,
      "step": 10787
    },
    {
      "epoch": 1.871616932685635,
      "grad_norm": 1.0863302946090698,
      "learning_rate": 6.321378462988148e-07,
      "loss": 0.3905,
      "step": 10788
    },
    {
      "epoch": 1.8717904233171407,
      "grad_norm": 0.874183177947998,
      "learning_rate": 6.304395468696345e-07,
      "loss": 0.5513,
      "step": 10789
    },
    {
      "epoch": 1.8719639139486468,
      "grad_norm": 0.8077466487884521,
      "learning_rate": 6.287434953005145e-07,
      "loss": 0.4374,
      "step": 10790
    },
    {
      "epoch": 1.8721374045801527,
      "grad_norm": 0.8777197003364563,
      "learning_rate": 6.270496917882863e-07,
      "loss": 0.4331,
      "step": 10791
    },
    {
      "epoch": 1.8723108952116587,
      "grad_norm": 0.9433843493461609,
      "learning_rate": 6.253581365295148e-07,
      "loss": 0.4448,
      "step": 10792
    },
    {
      "epoch": 1.8724843858431646,
      "grad_norm": 0.7966040968894958,
      "learning_rate": 6.236688297205074e-07,
      "loss": 0.4144,
      "step": 10793
    },
    {
      "epoch": 1.8726578764746704,
      "grad_norm": 0.6521954536437988,
      "learning_rate": 6.219817715573073e-07,
      "loss": 0.4707,
      "step": 10794
    },
    {
      "epoch": 1.8728313671061763,
      "grad_norm": 0.8022371530532837,
      "learning_rate": 6.202969622357069e-07,
      "loss": 0.5208,
      "step": 10795
    },
    {
      "epoch": 1.873004857737682,
      "grad_norm": 0.8339186310768127,
      "learning_rate": 6.18614401951223e-07,
      "loss": 0.5664,
      "step": 10796
    },
    {
      "epoch": 1.873178348369188,
      "grad_norm": 0.6526886820793152,
      "learning_rate": 6.16934090899124e-07,
      "loss": 0.5718,
      "step": 10797
    },
    {
      "epoch": 1.8733518390006938,
      "grad_norm": 0.6890487670898438,
      "learning_rate": 6.152560292744091e-07,
      "loss": 0.5109,
      "step": 10798
    },
    {
      "epoch": 1.8735253296321999,
      "grad_norm": 3.252506971359253,
      "learning_rate": 6.135802172718164e-07,
      "loss": 0.4457,
      "step": 10799
    },
    {
      "epoch": 1.8736988202637057,
      "grad_norm": 1.3863199949264526,
      "learning_rate": 6.119066550858321e-07,
      "loss": 0.337,
      "step": 10800
    },
    {
      "epoch": 1.8738723108952118,
      "grad_norm": 0.8325015902519226,
      "learning_rate": 6.102353429106722e-07,
      "loss": 0.5292,
      "step": 10801
    },
    {
      "epoch": 1.8740458015267176,
      "grad_norm": 0.6851802468299866,
      "learning_rate": 6.085662809402926e-07,
      "loss": 0.4647,
      "step": 10802
    },
    {
      "epoch": 1.8742192921582235,
      "grad_norm": 0.8349834680557251,
      "learning_rate": 6.068994693683916e-07,
      "loss": 0.4414,
      "step": 10803
    },
    {
      "epoch": 1.8743927827897293,
      "grad_norm": 0.7992815375328064,
      "learning_rate": 6.052349083884057e-07,
      "loss": 0.5262,
      "step": 10804
    },
    {
      "epoch": 1.8745662734212352,
      "grad_norm": 0.7497966885566711,
      "learning_rate": 6.035725981935092e-07,
      "loss": 0.4694,
      "step": 10805
    },
    {
      "epoch": 1.874739764052741,
      "grad_norm": 1.340006709098816,
      "learning_rate": 6.019125389766123e-07,
      "loss": 0.5042,
      "step": 10806
    },
    {
      "epoch": 1.874913254684247,
      "grad_norm": 0.6926025152206421,
      "learning_rate": 6.002547309303674e-07,
      "loss": 0.4989,
      "step": 10807
    },
    {
      "epoch": 1.875086745315753,
      "grad_norm": 0.8088293671607971,
      "learning_rate": 5.985991742471698e-07,
      "loss": 0.3966,
      "step": 10808
    },
    {
      "epoch": 1.875260235947259,
      "grad_norm": 0.703403651714325,
      "learning_rate": 5.969458691191432e-07,
      "loss": 0.4469,
      "step": 10809
    },
    {
      "epoch": 1.8754337265787648,
      "grad_norm": 1.162053108215332,
      "learning_rate": 5.952948157381566e-07,
      "loss": 0.5541,
      "step": 10810
    },
    {
      "epoch": 1.8756072172102707,
      "grad_norm": 0.9857968091964722,
      "learning_rate": 5.936460142958189e-07,
      "loss": 0.4306,
      "step": 10811
    },
    {
      "epoch": 1.8757807078417765,
      "grad_norm": 0.7314509153366089,
      "learning_rate": 5.919994649834748e-07,
      "loss": 0.5096,
      "step": 10812
    },
    {
      "epoch": 1.8759541984732824,
      "grad_norm": 1.1454625129699707,
      "learning_rate": 5.903551679922049e-07,
      "loss": 0.3865,
      "step": 10813
    },
    {
      "epoch": 1.8761276891047882,
      "grad_norm": 0.7919958233833313,
      "learning_rate": 5.887131235128385e-07,
      "loss": 0.3586,
      "step": 10814
    },
    {
      "epoch": 1.8763011797362943,
      "grad_norm": 0.9079128503799438,
      "learning_rate": 5.870733317359278e-07,
      "loss": 0.4118,
      "step": 10815
    },
    {
      "epoch": 1.8764746703678001,
      "grad_norm": 0.908237099647522,
      "learning_rate": 5.854357928517806e-07,
      "loss": 0.364,
      "step": 10816
    },
    {
      "epoch": 1.8766481609993062,
      "grad_norm": 0.8470625877380371,
      "learning_rate": 5.838005070504293e-07,
      "loss": 0.3947,
      "step": 10817
    },
    {
      "epoch": 1.876821651630812,
      "grad_norm": 0.6253703236579895,
      "learning_rate": 5.821674745216599e-07,
      "loss": 0.449,
      "step": 10818
    },
    {
      "epoch": 1.876995142262318,
      "grad_norm": 0.6411759853363037,
      "learning_rate": 5.80536695454974e-07,
      "loss": 0.4144,
      "step": 10819
    },
    {
      "epoch": 1.8771686328938237,
      "grad_norm": 0.745097815990448,
      "learning_rate": 5.789081700396381e-07,
      "loss": 0.462,
      "step": 10820
    },
    {
      "epoch": 1.8773421235253296,
      "grad_norm": 0.760269284248352,
      "learning_rate": 5.772818984646367e-07,
      "loss": 0.4585,
      "step": 10821
    },
    {
      "epoch": 1.8775156141568354,
      "grad_norm": 1.3237879276275635,
      "learning_rate": 5.756578809187008e-07,
      "loss": 0.5519,
      "step": 10822
    },
    {
      "epoch": 1.8776891047883413,
      "grad_norm": 0.8203462958335876,
      "learning_rate": 5.740361175903042e-07,
      "loss": 0.3639,
      "step": 10823
    },
    {
      "epoch": 1.8778625954198473,
      "grad_norm": 1.1125695705413818,
      "learning_rate": 5.724166086676542e-07,
      "loss": 0.3593,
      "step": 10824
    },
    {
      "epoch": 1.8780360860513532,
      "grad_norm": 0.8598060011863708,
      "learning_rate": 5.707993543386914e-07,
      "loss": 0.3606,
      "step": 10825
    },
    {
      "epoch": 1.8782095766828593,
      "grad_norm": 0.7895324230194092,
      "learning_rate": 5.691843547911013e-07,
      "loss": 0.4056,
      "step": 10826
    },
    {
      "epoch": 1.878383067314365,
      "grad_norm": 0.8858323097229004,
      "learning_rate": 5.675716102123141e-07,
      "loss": 0.5073,
      "step": 10827
    },
    {
      "epoch": 1.878556557945871,
      "grad_norm": 0.6717455387115479,
      "learning_rate": 5.6596112078948e-07,
      "loss": 0.4431,
      "step": 10828
    },
    {
      "epoch": 1.8787300485773768,
      "grad_norm": 0.706802248954773,
      "learning_rate": 5.643528867095049e-07,
      "loss": 0.4803,
      "step": 10829
    },
    {
      "epoch": 1.8789035392088826,
      "grad_norm": 0.7239381670951843,
      "learning_rate": 5.627469081590242e-07,
      "loss": 0.4407,
      "step": 10830
    },
    {
      "epoch": 1.8790770298403885,
      "grad_norm": 0.9374978542327881,
      "learning_rate": 5.611431853244132e-07,
      "loss": 0.3917,
      "step": 10831
    },
    {
      "epoch": 1.8792505204718946,
      "grad_norm": 0.6650722026824951,
      "learning_rate": 5.595417183917851e-07,
      "loss": 0.6089,
      "step": 10832
    },
    {
      "epoch": 1.8794240111034004,
      "grad_norm": 0.8580365777015686,
      "learning_rate": 5.579425075469936e-07,
      "loss": 0.4304,
      "step": 10833
    },
    {
      "epoch": 1.8795975017349065,
      "grad_norm": 0.6936100721359253,
      "learning_rate": 5.563455529756301e-07,
      "loss": 0.493,
      "step": 10834
    },
    {
      "epoch": 1.8797709923664123,
      "grad_norm": 0.7324767112731934,
      "learning_rate": 5.547508548630221e-07,
      "loss": 0.3958,
      "step": 10835
    },
    {
      "epoch": 1.8799444829979182,
      "grad_norm": 0.7372058629989624,
      "learning_rate": 5.531584133942325e-07,
      "loss": 0.3337,
      "step": 10836
    },
    {
      "epoch": 1.880117973629424,
      "grad_norm": 0.79083251953125,
      "learning_rate": 5.515682287540736e-07,
      "loss": 0.4375,
      "step": 10837
    },
    {
      "epoch": 1.8802914642609299,
      "grad_norm": 0.9389005899429321,
      "learning_rate": 5.499803011270776e-07,
      "loss": 0.3457,
      "step": 10838
    },
    {
      "epoch": 1.8804649548924357,
      "grad_norm": 2.623318910598755,
      "learning_rate": 5.483946306975374e-07,
      "loss": 0.3108,
      "step": 10839
    },
    {
      "epoch": 1.8806384455239415,
      "grad_norm": 1.211706280708313,
      "learning_rate": 5.468112176494633e-07,
      "loss": 0.4312,
      "step": 10840
    },
    {
      "epoch": 1.8808119361554476,
      "grad_norm": 0.8657251596450806,
      "learning_rate": 5.452300621666151e-07,
      "loss": 0.4696,
      "step": 10841
    },
    {
      "epoch": 1.8809854267869535,
      "grad_norm": 0.7004914879798889,
      "learning_rate": 5.43651164432486e-07,
      "loss": 0.4318,
      "step": 10842
    },
    {
      "epoch": 1.8811589174184595,
      "grad_norm": 1.0329854488372803,
      "learning_rate": 5.420745246303116e-07,
      "loss": 0.5156,
      "step": 10843
    },
    {
      "epoch": 1.8813324080499654,
      "grad_norm": 0.9545468091964722,
      "learning_rate": 5.405001429430634e-07,
      "loss": 0.3884,
      "step": 10844
    },
    {
      "epoch": 1.8815058986814712,
      "grad_norm": 1.3794701099395752,
      "learning_rate": 5.389280195534463e-07,
      "loss": 0.515,
      "step": 10845
    },
    {
      "epoch": 1.881679389312977,
      "grad_norm": 0.7027184963226318,
      "learning_rate": 5.373581546439077e-07,
      "loss": 0.5974,
      "step": 10846
    },
    {
      "epoch": 1.881852879944483,
      "grad_norm": 0.8707473874092102,
      "learning_rate": 5.357905483966375e-07,
      "loss": 0.4139,
      "step": 10847
    },
    {
      "epoch": 1.8820263705759888,
      "grad_norm": 0.8357599973678589,
      "learning_rate": 5.342252009935522e-07,
      "loss": 0.392,
      "step": 10848
    },
    {
      "epoch": 1.8821998612074948,
      "grad_norm": 1.2941255569458008,
      "learning_rate": 5.326621126163157e-07,
      "loss": 0.4586,
      "step": 10849
    },
    {
      "epoch": 1.8823733518390007,
      "grad_norm": 0.8082398176193237,
      "learning_rate": 5.311012834463247e-07,
      "loss": 0.4844,
      "step": 10850
    },
    {
      "epoch": 1.8825468424705067,
      "grad_norm": 1.1110472679138184,
      "learning_rate": 5.295427136647124e-07,
      "loss": 0.4257,
      "step": 10851
    },
    {
      "epoch": 1.8827203331020126,
      "grad_norm": 0.774032711982727,
      "learning_rate": 5.279864034523586e-07,
      "loss": 0.4403,
      "step": 10852
    },
    {
      "epoch": 1.8828938237335184,
      "grad_norm": 0.9818394184112549,
      "learning_rate": 5.26432352989874e-07,
      "loss": 0.4296,
      "step": 10853
    },
    {
      "epoch": 1.8830673143650243,
      "grad_norm": 0.7783525586128235,
      "learning_rate": 5.248805624576037e-07,
      "loss": 0.4307,
      "step": 10854
    },
    {
      "epoch": 1.8832408049965301,
      "grad_norm": 0.9560015201568604,
      "learning_rate": 5.233310320356366e-07,
      "loss": 0.5295,
      "step": 10855
    },
    {
      "epoch": 1.883414295628036,
      "grad_norm": 0.689906656742096,
      "learning_rate": 5.217837619038002e-07,
      "loss": 0.4695,
      "step": 10856
    },
    {
      "epoch": 1.8835877862595418,
      "grad_norm": 0.826659083366394,
      "learning_rate": 5.20238752241653e-07,
      "loss": 0.4233,
      "step": 10857
    },
    {
      "epoch": 1.8837612768910479,
      "grad_norm": 0.8049799799919128,
      "learning_rate": 5.186960032284983e-07,
      "loss": 0.4156,
      "step": 10858
    },
    {
      "epoch": 1.8839347675225537,
      "grad_norm": 0.9805461168289185,
      "learning_rate": 5.171555150433705e-07,
      "loss": 0.3573,
      "step": 10859
    },
    {
      "epoch": 1.8841082581540598,
      "grad_norm": 0.6907187700271606,
      "learning_rate": 5.156172878650489e-07,
      "loss": 0.4215,
      "step": 10860
    },
    {
      "epoch": 1.8842817487855656,
      "grad_norm": 0.7992402911186218,
      "learning_rate": 5.140813218720442e-07,
      "loss": 0.3917,
      "step": 10861
    },
    {
      "epoch": 1.8844552394170715,
      "grad_norm": 1.0171290636062622,
      "learning_rate": 5.125476172426092e-07,
      "loss": 0.4378,
      "step": 10862
    },
    {
      "epoch": 1.8846287300485773,
      "grad_norm": 0.6632357835769653,
      "learning_rate": 5.110161741547281e-07,
      "loss": 0.5093,
      "step": 10863
    },
    {
      "epoch": 1.8848022206800832,
      "grad_norm": 1.0032912492752075,
      "learning_rate": 5.094869927861323e-07,
      "loss": 0.4222,
      "step": 10864
    },
    {
      "epoch": 1.884975711311589,
      "grad_norm": 0.8233975172042847,
      "learning_rate": 5.079600733142775e-07,
      "loss": 0.5587,
      "step": 10865
    },
    {
      "epoch": 1.885149201943095,
      "grad_norm": 0.8972030878067017,
      "learning_rate": 5.064354159163754e-07,
      "loss": 0.5225,
      "step": 10866
    },
    {
      "epoch": 1.885322692574601,
      "grad_norm": 1.1515147686004639,
      "learning_rate": 5.049130207693509e-07,
      "loss": 0.3785,
      "step": 10867
    },
    {
      "epoch": 1.885496183206107,
      "grad_norm": 0.9869181513786316,
      "learning_rate": 5.033928880498917e-07,
      "loss": 0.4395,
      "step": 10868
    },
    {
      "epoch": 1.8856696738376129,
      "grad_norm": 0.8651929497718811,
      "learning_rate": 5.018750179344034e-07,
      "loss": 0.4138,
      "step": 10869
    },
    {
      "epoch": 1.8858431644691187,
      "grad_norm": 0.750634491443634,
      "learning_rate": 5.003594105990384e-07,
      "loss": 0.3713,
      "step": 10870
    },
    {
      "epoch": 1.8860166551006246,
      "grad_norm": 0.8486379981040955,
      "learning_rate": 4.98846066219687e-07,
      "loss": 0.4789,
      "step": 10871
    },
    {
      "epoch": 1.8861901457321304,
      "grad_norm": 0.8471236228942871,
      "learning_rate": 4.973349849719733e-07,
      "loss": 0.4246,
      "step": 10872
    },
    {
      "epoch": 1.8863636363636362,
      "grad_norm": 0.8552502393722534,
      "learning_rate": 4.958261670312591e-07,
      "loss": 0.4172,
      "step": 10873
    },
    {
      "epoch": 1.8865371269951423,
      "grad_norm": 0.675193727016449,
      "learning_rate": 4.943196125726446e-07,
      "loss": 0.4707,
      "step": 10874
    },
    {
      "epoch": 1.8867106176266482,
      "grad_norm": 0.7469701170921326,
      "learning_rate": 4.928153217709674e-07,
      "loss": 0.3881,
      "step": 10875
    },
    {
      "epoch": 1.8868841082581542,
      "grad_norm": 0.9664652943611145,
      "learning_rate": 4.913132948008037e-07,
      "loss": 0.3854,
      "step": 10876
    },
    {
      "epoch": 1.88705759888966,
      "grad_norm": 0.8605236411094666,
      "learning_rate": 4.89813531836465e-07,
      "loss": 0.3905,
      "step": 10877
    },
    {
      "epoch": 1.887231089521166,
      "grad_norm": 0.7831318974494934,
      "learning_rate": 4.883160330519965e-07,
      "loss": 0.4537,
      "step": 10878
    },
    {
      "epoch": 1.8874045801526718,
      "grad_norm": 0.8109844326972961,
      "learning_rate": 4.868207986211926e-07,
      "loss": 0.3942,
      "step": 10879
    },
    {
      "epoch": 1.8875780707841776,
      "grad_norm": 0.7134442925453186,
      "learning_rate": 4.853278287175677e-07,
      "loss": 0.4084,
      "step": 10880
    },
    {
      "epoch": 1.8877515614156835,
      "grad_norm": 0.8275427222251892,
      "learning_rate": 4.838371235143902e-07,
      "loss": 0.3734,
      "step": 10881
    },
    {
      "epoch": 1.8879250520471893,
      "grad_norm": 0.6749014854431152,
      "learning_rate": 4.823486831846547e-07,
      "loss": 0.4794,
      "step": 10882
    },
    {
      "epoch": 1.8880985426786954,
      "grad_norm": 0.7942231297492981,
      "learning_rate": 4.808625079010987e-07,
      "loss": 0.509,
      "step": 10883
    },
    {
      "epoch": 1.8882720333102012,
      "grad_norm": 0.7638766169548035,
      "learning_rate": 4.793785978361887e-07,
      "loss": 0.5704,
      "step": 10884
    },
    {
      "epoch": 1.8884455239417073,
      "grad_norm": 0.7045732140541077,
      "learning_rate": 4.778969531621447e-07,
      "loss": 0.5597,
      "step": 10885
    },
    {
      "epoch": 1.8886190145732131,
      "grad_norm": 0.8598750829696655,
      "learning_rate": 4.764175740509025e-07,
      "loss": 0.4932,
      "step": 10886
    },
    {
      "epoch": 1.888792505204719,
      "grad_norm": 0.7648863792419434,
      "learning_rate": 4.749404606741514e-07,
      "loss": 0.5145,
      "step": 10887
    },
    {
      "epoch": 1.8889659958362248,
      "grad_norm": 1.0697370767593384,
      "learning_rate": 4.7346561320330997e-07,
      "loss": 0.4833,
      "step": 10888
    },
    {
      "epoch": 1.8891394864677307,
      "grad_norm": 1.54634690284729,
      "learning_rate": 4.7199303180953894e-07,
      "loss": 0.3649,
      "step": 10889
    },
    {
      "epoch": 1.8893129770992365,
      "grad_norm": 0.7695502042770386,
      "learning_rate": 4.7052271666373053e-07,
      "loss": 0.4333,
      "step": 10890
    },
    {
      "epoch": 1.8894864677307426,
      "grad_norm": 1.0159564018249512,
      "learning_rate": 4.6905466793651713e-07,
      "loss": 0.3546,
      "step": 10891
    },
    {
      "epoch": 1.8896599583622484,
      "grad_norm": 0.9595746397972107,
      "learning_rate": 4.675888857982669e-07,
      "loss": 0.3325,
      "step": 10892
    },
    {
      "epoch": 1.8898334489937545,
      "grad_norm": 0.86198890209198,
      "learning_rate": 4.6612537041908823e-07,
      "loss": 0.3857,
      "step": 10893
    },
    {
      "epoch": 1.8900069396252603,
      "grad_norm": 0.6959157586097717,
      "learning_rate": 4.646641219688186e-07,
      "loss": 0.4095,
      "step": 10894
    },
    {
      "epoch": 1.8901804302567662,
      "grad_norm": 1.132491111755371,
      "learning_rate": 4.6320514061704236e-07,
      "loss": 0.5416,
      "step": 10895
    },
    {
      "epoch": 1.890353920888272,
      "grad_norm": 1.1667431592941284,
      "learning_rate": 4.617484265330752e-07,
      "loss": 0.3961,
      "step": 10896
    },
    {
      "epoch": 1.8905274115197779,
      "grad_norm": 0.7082688808441162,
      "learning_rate": 4.602939798859685e-07,
      "loss": 0.483,
      "step": 10897
    },
    {
      "epoch": 1.8907009021512837,
      "grad_norm": 0.8543420433998108,
      "learning_rate": 4.588418008445161e-07,
      "loss": 0.4236,
      "step": 10898
    },
    {
      "epoch": 1.8908743927827896,
      "grad_norm": 0.6347614526748657,
      "learning_rate": 4.573918895772389e-07,
      "loss": 0.4264,
      "step": 10899
    },
    {
      "epoch": 1.8910478834142956,
      "grad_norm": 0.9070220589637756,
      "learning_rate": 4.5594424625240887e-07,
      "loss": 0.3589,
      "step": 10900
    },
    {
      "epoch": 1.8912213740458015,
      "grad_norm": 3.528459310531616,
      "learning_rate": 4.544988710380205e-07,
      "loss": 0.3573,
      "step": 10901
    },
    {
      "epoch": 1.8913948646773076,
      "grad_norm": 1.0848753452301025,
      "learning_rate": 4.5305576410181293e-07,
      "loss": 0.3687,
      "step": 10902
    },
    {
      "epoch": 1.8915683553088134,
      "grad_norm": 1.0708909034729004,
      "learning_rate": 4.51614925611259e-07,
      "loss": 0.4875,
      "step": 10903
    },
    {
      "epoch": 1.8917418459403192,
      "grad_norm": 0.761223554611206,
      "learning_rate": 4.5017635573357366e-07,
      "loss": 0.5276,
      "step": 10904
    },
    {
      "epoch": 1.891915336571825,
      "grad_norm": 0.9501824378967285,
      "learning_rate": 4.4874005463570126e-07,
      "loss": 0.4119,
      "step": 10905
    },
    {
      "epoch": 1.892088827203331,
      "grad_norm": 0.8690057992935181,
      "learning_rate": 4.4730602248432843e-07,
      "loss": 0.3707,
      "step": 10906
    },
    {
      "epoch": 1.8922623178348368,
      "grad_norm": 0.9825139045715332,
      "learning_rate": 4.4587425944587317e-07,
      "loss": 0.5173,
      "step": 10907
    },
    {
      "epoch": 1.8924358084663429,
      "grad_norm": 0.7432997226715088,
      "learning_rate": 4.444447656864981e-07,
      "loss": 0.4092,
      "step": 10908
    },
    {
      "epoch": 1.8926092990978487,
      "grad_norm": 0.7270981073379517,
      "learning_rate": 4.4301754137209277e-07,
      "loss": 0.5334,
      "step": 10909
    },
    {
      "epoch": 1.8927827897293548,
      "grad_norm": 1.37782883644104,
      "learning_rate": 4.4159258666828907e-07,
      "loss": 0.4818,
      "step": 10910
    },
    {
      "epoch": 1.8929562803608606,
      "grad_norm": 0.8546935319900513,
      "learning_rate": 4.401699017404593e-07,
      "loss": 0.5096,
      "step": 10911
    },
    {
      "epoch": 1.8931297709923665,
      "grad_norm": 1.0219379663467407,
      "learning_rate": 4.3874948675370233e-07,
      "loss": 0.3737,
      "step": 10912
    },
    {
      "epoch": 1.8933032616238723,
      "grad_norm": 0.9887694716453552,
      "learning_rate": 4.373313418728575e-07,
      "loss": 0.3583,
      "step": 10913
    },
    {
      "epoch": 1.8934767522553781,
      "grad_norm": 1.070763349533081,
      "learning_rate": 4.3591546726250877e-07,
      "loss": 0.574,
      "step": 10914
    },
    {
      "epoch": 1.893650242886884,
      "grad_norm": 0.7176077961921692,
      "learning_rate": 4.3450186308696685e-07,
      "loss": 0.5289,
      "step": 10915
    },
    {
      "epoch": 1.89382373351839,
      "grad_norm": 1.0414750576019287,
      "learning_rate": 4.3309052951028275e-07,
      "loss": 0.3427,
      "step": 10916
    },
    {
      "epoch": 1.893997224149896,
      "grad_norm": 0.8068662881851196,
      "learning_rate": 4.31681466696241e-07,
      "loss": 0.3973,
      "step": 10917
    },
    {
      "epoch": 1.8941707147814018,
      "grad_norm": 0.8175247311592102,
      "learning_rate": 4.302746748083664e-07,
      "loss": 0.5437,
      "step": 10918
    },
    {
      "epoch": 1.8943442054129078,
      "grad_norm": 0.9541627764701843,
      "learning_rate": 4.2887015400991937e-07,
      "loss": 0.3667,
      "step": 10919
    },
    {
      "epoch": 1.8945176960444137,
      "grad_norm": 0.7084746360778809,
      "learning_rate": 4.2746790446389853e-07,
      "loss": 0.5047,
      "step": 10920
    },
    {
      "epoch": 1.8946911866759195,
      "grad_norm": 0.8773970603942871,
      "learning_rate": 4.260679263330314e-07,
      "loss": 0.4802,
      "step": 10921
    },
    {
      "epoch": 1.8948646773074254,
      "grad_norm": 0.8791499733924866,
      "learning_rate": 4.2467021977978806e-07,
      "loss": 0.3822,
      "step": 10922
    },
    {
      "epoch": 1.8950381679389312,
      "grad_norm": 0.791195273399353,
      "learning_rate": 4.2327478496637877e-07,
      "loss": 0.3689,
      "step": 10923
    },
    {
      "epoch": 1.895211658570437,
      "grad_norm": 0.9519442915916443,
      "learning_rate": 4.218816220547406e-07,
      "loss": 0.5925,
      "step": 10924
    },
    {
      "epoch": 1.8953851492019431,
      "grad_norm": 0.6581597924232483,
      "learning_rate": 4.2049073120655315e-07,
      "loss": 0.4708,
      "step": 10925
    },
    {
      "epoch": 1.895558639833449,
      "grad_norm": 0.8855475783348083,
      "learning_rate": 4.1910211258322954e-07,
      "loss": 0.4625,
      "step": 10926
    },
    {
      "epoch": 1.895732130464955,
      "grad_norm": 0.7376269102096558,
      "learning_rate": 4.1771576634592524e-07,
      "loss": 0.381,
      "step": 10927
    },
    {
      "epoch": 1.8959056210964609,
      "grad_norm": 1.098069190979004,
      "learning_rate": 4.1633169265552274e-07,
      "loss": 0.4945,
      "step": 10928
    },
    {
      "epoch": 1.8960791117279667,
      "grad_norm": 0.8522683382034302,
      "learning_rate": 4.149498916726469e-07,
      "loss": 0.4251,
      "step": 10929
    },
    {
      "epoch": 1.8962526023594726,
      "grad_norm": 0.7908493280410767,
      "learning_rate": 4.13570363557656e-07,
      "loss": 0.3824,
      "step": 10930
    },
    {
      "epoch": 1.8964260929909784,
      "grad_norm": 0.9189895987510681,
      "learning_rate": 4.1219310847064876e-07,
      "loss": 0.3522,
      "step": 10931
    },
    {
      "epoch": 1.8965995836224843,
      "grad_norm": 0.8488624095916748,
      "learning_rate": 4.108181265714528e-07,
      "loss": 0.4468,
      "step": 10932
    },
    {
      "epoch": 1.8967730742539903,
      "grad_norm": 1.0427064895629883,
      "learning_rate": 4.0944541801964275e-07,
      "loss": 0.3856,
      "step": 10933
    },
    {
      "epoch": 1.8969465648854962,
      "grad_norm": 0.8867191672325134,
      "learning_rate": 4.0807498297451786e-07,
      "loss": 0.3552,
      "step": 10934
    },
    {
      "epoch": 1.8971200555170022,
      "grad_norm": 0.7627947330474854,
      "learning_rate": 4.06706821595122e-07,
      "loss": 0.43,
      "step": 10935
    },
    {
      "epoch": 1.897293546148508,
      "grad_norm": 0.803941011428833,
      "learning_rate": 4.053409340402259e-07,
      "loss": 0.5505,
      "step": 10936
    },
    {
      "epoch": 1.897467036780014,
      "grad_norm": 1.2127147912979126,
      "learning_rate": 4.039773204683517e-07,
      "loss": 0.4297,
      "step": 10937
    },
    {
      "epoch": 1.8976405274115198,
      "grad_norm": 0.8328952789306641,
      "learning_rate": 4.026159810377417e-07,
      "loss": 0.401,
      "step": 10938
    },
    {
      "epoch": 1.8978140180430256,
      "grad_norm": 0.720675528049469,
      "learning_rate": 4.012569159063806e-07,
      "loss": 0.4507,
      "step": 10939
    },
    {
      "epoch": 1.8979875086745315,
      "grad_norm": 0.7106125354766846,
      "learning_rate": 3.999001252319934e-07,
      "loss": 0.4246,
      "step": 10940
    },
    {
      "epoch": 1.8981609993060373,
      "grad_norm": 0.8404808640480042,
      "learning_rate": 3.9854560917203635e-07,
      "loss": 0.4801,
      "step": 10941
    },
    {
      "epoch": 1.8983344899375434,
      "grad_norm": 0.8275275230407715,
      "learning_rate": 3.971933678836992e-07,
      "loss": 0.3975,
      "step": 10942
    },
    {
      "epoch": 1.8985079805690492,
      "grad_norm": 1.1109490394592285,
      "learning_rate": 3.958434015239143e-07,
      "loss": 0.3765,
      "step": 10943
    },
    {
      "epoch": 1.8986814712005553,
      "grad_norm": 0.8118730783462524,
      "learning_rate": 3.944957102493474e-07,
      "loss": 0.3887,
      "step": 10944
    },
    {
      "epoch": 1.8988549618320612,
      "grad_norm": 1.2894953489303589,
      "learning_rate": 3.931502942163956e-07,
      "loss": 0.5012,
      "step": 10945
    },
    {
      "epoch": 1.899028452463567,
      "grad_norm": 0.985551655292511,
      "learning_rate": 3.918071535812007e-07,
      "loss": 0.5458,
      "step": 10946
    },
    {
      "epoch": 1.8992019430950728,
      "grad_norm": 0.6166860461235046,
      "learning_rate": 3.904662884996335e-07,
      "loss": 0.5833,
      "step": 10947
    },
    {
      "epoch": 1.8993754337265787,
      "grad_norm": 0.9930065870285034,
      "learning_rate": 3.8912769912730297e-07,
      "loss": 0.3824,
      "step": 10948
    },
    {
      "epoch": 1.8995489243580845,
      "grad_norm": 0.7751320004463196,
      "learning_rate": 3.8779138561955145e-07,
      "loss": 0.3983,
      "step": 10949
    },
    {
      "epoch": 1.8997224149895906,
      "grad_norm": 0.8659856915473938,
      "learning_rate": 3.864573481314682e-07,
      "loss": 0.3284,
      "step": 10950
    },
    {
      "epoch": 1.8998959056210964,
      "grad_norm": 0.9467388391494751,
      "learning_rate": 3.8512558681785826e-07,
      "loss": 0.5446,
      "step": 10951
    },
    {
      "epoch": 1.9000693962526025,
      "grad_norm": 0.8677664399147034,
      "learning_rate": 3.837961018332825e-07,
      "loss": 0.4783,
      "step": 10952
    },
    {
      "epoch": 1.9002428868841084,
      "grad_norm": 1.2335377931594849,
      "learning_rate": 3.824688933320264e-07,
      "loss": 0.4369,
      "step": 10953
    },
    {
      "epoch": 1.9004163775156142,
      "grad_norm": 0.9301946759223938,
      "learning_rate": 3.811439614681156e-07,
      "loss": 0.356,
      "step": 10954
    },
    {
      "epoch": 1.90058986814712,
      "grad_norm": 0.9937767386436462,
      "learning_rate": 3.798213063953049e-07,
      "loss": 0.3503,
      "step": 10955
    },
    {
      "epoch": 1.900763358778626,
      "grad_norm": 0.7710212469100952,
      "learning_rate": 3.7850092826709817e-07,
      "loss": 0.5227,
      "step": 10956
    },
    {
      "epoch": 1.9009368494101317,
      "grad_norm": 0.8241571187973022,
      "learning_rate": 3.771828272367195e-07,
      "loss": 0.4879,
      "step": 10957
    },
    {
      "epoch": 1.9011103400416376,
      "grad_norm": 0.8966371417045593,
      "learning_rate": 3.758670034571399e-07,
      "loss": 0.4624,
      "step": 10958
    },
    {
      "epoch": 1.9012838306731437,
      "grad_norm": 0.925626277923584,
      "learning_rate": 3.745534570810616e-07,
      "loss": 0.4044,
      "step": 10959
    },
    {
      "epoch": 1.9014573213046495,
      "grad_norm": 0.7923619747161865,
      "learning_rate": 3.7324218826092053e-07,
      "loss": 0.5719,
      "step": 10960
    },
    {
      "epoch": 1.9016308119361556,
      "grad_norm": 0.9957678914070129,
      "learning_rate": 3.7193319714889487e-07,
      "loss": 0.4647,
      "step": 10961
    },
    {
      "epoch": 1.9018043025676614,
      "grad_norm": 0.9195384383201599,
      "learning_rate": 3.7062648389689204e-07,
      "loss": 0.5049,
      "step": 10962
    },
    {
      "epoch": 1.9019777931991673,
      "grad_norm": 0.9187933802604675,
      "learning_rate": 3.6932204865655963e-07,
      "loss": 0.6035,
      "step": 10963
    },
    {
      "epoch": 1.9021512838306731,
      "grad_norm": 1.0127263069152832,
      "learning_rate": 3.680198915792765e-07,
      "loss": 0.3387,
      "step": 10964
    },
    {
      "epoch": 1.902324774462179,
      "grad_norm": 0.7908935546875,
      "learning_rate": 3.6672001281616186e-07,
      "loss": 0.3312,
      "step": 10965
    },
    {
      "epoch": 1.9024982650936848,
      "grad_norm": 0.7882046699523926,
      "learning_rate": 3.654224125180661e-07,
      "loss": 0.3206,
      "step": 10966
    },
    {
      "epoch": 1.9026717557251909,
      "grad_norm": 1.0293998718261719,
      "learning_rate": 3.6412709083557984e-07,
      "loss": 0.4872,
      "step": 10967
    },
    {
      "epoch": 1.9028452463566967,
      "grad_norm": 1.3950729370117188,
      "learning_rate": 3.628340479190229e-07,
      "loss": 0.4183,
      "step": 10968
    },
    {
      "epoch": 1.9030187369882028,
      "grad_norm": 0.7494522929191589,
      "learning_rate": 3.6154328391845963e-07,
      "loss": 0.4952,
      "step": 10969
    },
    {
      "epoch": 1.9031922276197086,
      "grad_norm": 0.7665684223175049,
      "learning_rate": 3.602547989836769e-07,
      "loss": 0.4425,
      "step": 10970
    },
    {
      "epoch": 1.9033657182512145,
      "grad_norm": 0.7976112365722656,
      "learning_rate": 3.5896859326421284e-07,
      "loss": 0.5042,
      "step": 10971
    },
    {
      "epoch": 1.9035392088827203,
      "grad_norm": 0.8296693563461304,
      "learning_rate": 3.5768466690933036e-07,
      "loss": 0.3983,
      "step": 10972
    },
    {
      "epoch": 1.9037126995142262,
      "grad_norm": 0.8233347535133362,
      "learning_rate": 3.564030200680302e-07,
      "loss": 0.4572,
      "step": 10973
    },
    {
      "epoch": 1.903886190145732,
      "grad_norm": 1.0415624380111694,
      "learning_rate": 3.551236528890445e-07,
      "loss": 0.4337,
      "step": 10974
    },
    {
      "epoch": 1.904059680777238,
      "grad_norm": 0.7998337745666504,
      "learning_rate": 3.538465655208545e-07,
      "loss": 0.369,
      "step": 10975
    },
    {
      "epoch": 1.904233171408744,
      "grad_norm": 0.6770361065864563,
      "learning_rate": 3.5257175811166166e-07,
      "loss": 0.5106,
      "step": 10976
    },
    {
      "epoch": 1.9044066620402498,
      "grad_norm": 0.9979817271232605,
      "learning_rate": 3.5129923080940985e-07,
      "loss": 0.3523,
      "step": 10977
    },
    {
      "epoch": 1.9045801526717558,
      "grad_norm": 1.1292340755462646,
      "learning_rate": 3.500289837617765e-07,
      "loss": 0.4637,
      "step": 10978
    },
    {
      "epoch": 1.9047536433032617,
      "grad_norm": 0.9591286182403564,
      "learning_rate": 3.4876101711617924e-07,
      "loss": 0.3842,
      "step": 10979
    },
    {
      "epoch": 1.9049271339347675,
      "grad_norm": 0.8513092398643494,
      "learning_rate": 3.474953310197604e-07,
      "loss": 0.3371,
      "step": 10980
    },
    {
      "epoch": 1.9051006245662734,
      "grad_norm": 0.6872183680534363,
      "learning_rate": 3.462319256194113e-07,
      "loss": 0.5271,
      "step": 10981
    },
    {
      "epoch": 1.9052741151977792,
      "grad_norm": 0.7488598227500916,
      "learning_rate": 3.4497080106174806e-07,
      "loss": 0.4376,
      "step": 10982
    },
    {
      "epoch": 1.905447605829285,
      "grad_norm": 0.8067551255226135,
      "learning_rate": 3.437119574931247e-07,
      "loss": 0.4055,
      "step": 10983
    },
    {
      "epoch": 1.9056210964607911,
      "grad_norm": 0.7539083957672119,
      "learning_rate": 3.424553950596332e-07,
      "loss": 0.473,
      "step": 10984
    },
    {
      "epoch": 1.905794587092297,
      "grad_norm": 1.280732274055481,
      "learning_rate": 3.412011139070992e-07,
      "loss": 0.486,
      "step": 10985
    },
    {
      "epoch": 1.905968077723803,
      "grad_norm": 0.9366377592086792,
      "learning_rate": 3.3994911418108176e-07,
      "loss": 0.315,
      "step": 10986
    },
    {
      "epoch": 1.906141568355309,
      "grad_norm": 0.8942596316337585,
      "learning_rate": 3.3869939602687806e-07,
      "loss": 0.412,
      "step": 10987
    },
    {
      "epoch": 1.9063150589868147,
      "grad_norm": 0.8256127834320068,
      "learning_rate": 3.374519595895209e-07,
      "loss": 0.4327,
      "step": 10988
    },
    {
      "epoch": 1.9064885496183206,
      "grad_norm": 1.0570706129074097,
      "learning_rate": 3.362068050137768e-07,
      "loss": 0.5269,
      "step": 10989
    },
    {
      "epoch": 1.9066620402498264,
      "grad_norm": 1.0103695392608643,
      "learning_rate": 3.3496393244414114e-07,
      "loss": 0.4835,
      "step": 10990
    },
    {
      "epoch": 1.9068355308813323,
      "grad_norm": 0.7182628512382507,
      "learning_rate": 3.3372334202485867e-07,
      "loss": 0.5101,
      "step": 10991
    },
    {
      "epoch": 1.9070090215128384,
      "grad_norm": 0.8422587513923645,
      "learning_rate": 3.324850338998964e-07,
      "loss": 0.4139,
      "step": 10992
    },
    {
      "epoch": 1.9071825121443442,
      "grad_norm": 0.8667024970054626,
      "learning_rate": 3.312490082129638e-07,
      "loss": 0.4093,
      "step": 10993
    },
    {
      "epoch": 1.9073560027758503,
      "grad_norm": 0.7709800004959106,
      "learning_rate": 3.300152651075039e-07,
      "loss": 0.5382,
      "step": 10994
    },
    {
      "epoch": 1.9075294934073561,
      "grad_norm": 0.799622654914856,
      "learning_rate": 3.2878380472669116e-07,
      "loss": 0.4028,
      "step": 10995
    },
    {
      "epoch": 1.907702984038862,
      "grad_norm": 0.8478907942771912,
      "learning_rate": 3.2755462721344e-07,
      "loss": 0.4365,
      "step": 10996
    },
    {
      "epoch": 1.9078764746703678,
      "grad_norm": 1.0206315517425537,
      "learning_rate": 3.2632773271039644e-07,
      "loss": 0.3174,
      "step": 10997
    },
    {
      "epoch": 1.9080499653018737,
      "grad_norm": 0.9754865765571594,
      "learning_rate": 3.251031213599465e-07,
      "loss": 0.4686,
      "step": 10998
    },
    {
      "epoch": 1.9082234559333795,
      "grad_norm": 0.9375216960906982,
      "learning_rate": 3.2388079330420095e-07,
      "loss": 0.4592,
      "step": 10999
    },
    {
      "epoch": 1.9083969465648853,
      "grad_norm": 0.9150240421295166,
      "learning_rate": 3.2266074868501976e-07,
      "loss": 0.4347,
      "step": 11000
    },
    {
      "epoch": 1.9085704371963914,
      "grad_norm": 0.8404751420021057,
      "learning_rate": 3.2144298764398505e-07,
      "loss": 0.3708,
      "step": 11001
    },
    {
      "epoch": 1.9087439278278973,
      "grad_norm": 0.8384784460067749,
      "learning_rate": 3.2022751032242396e-07,
      "loss": 0.4595,
      "step": 11002
    },
    {
      "epoch": 1.9089174184594033,
      "grad_norm": 0.7938383221626282,
      "learning_rate": 3.190143168613902e-07,
      "loss": 0.4695,
      "step": 11003
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.985153079032898,
      "learning_rate": 3.178034074016778e-07,
      "loss": 0.465,
      "step": 11004
    },
    {
      "epoch": 1.909264399722415,
      "grad_norm": 0.9028046131134033,
      "learning_rate": 3.1659478208381665e-07,
      "loss": 0.3541,
      "step": 11005
    },
    {
      "epoch": 1.9094378903539209,
      "grad_norm": 0.8445605635643005,
      "learning_rate": 3.1538844104806343e-07,
      "loss": 0.535,
      "step": 11006
    },
    {
      "epoch": 1.9096113809854267,
      "grad_norm": 0.921342134475708,
      "learning_rate": 3.141843844344195e-07,
      "loss": 0.433,
      "step": 11007
    },
    {
      "epoch": 1.9097848716169326,
      "grad_norm": 0.746217668056488,
      "learning_rate": 3.1298261238261964e-07,
      "loss": 0.5485,
      "step": 11008
    },
    {
      "epoch": 1.9099583622484386,
      "grad_norm": 1.069266676902771,
      "learning_rate": 3.1178312503212347e-07,
      "loss": 0.4598,
      "step": 11009
    },
    {
      "epoch": 1.9101318528799445,
      "grad_norm": 0.6470270156860352,
      "learning_rate": 3.105859225221397e-07,
      "loss": 0.5859,
      "step": 11010
    },
    {
      "epoch": 1.9103053435114505,
      "grad_norm": 0.9771518707275391,
      "learning_rate": 3.0939100499160155e-07,
      "loss": 0.458,
      "step": 11011
    },
    {
      "epoch": 1.9104788341429564,
      "grad_norm": 0.9310368299484253,
      "learning_rate": 3.0819837257918037e-07,
      "loss": 0.392,
      "step": 11012
    },
    {
      "epoch": 1.9106523247744622,
      "grad_norm": 0.9798033833503723,
      "learning_rate": 3.0700802542328325e-07,
      "loss": 0.4277,
      "step": 11013
    },
    {
      "epoch": 1.910825815405968,
      "grad_norm": 0.7718014121055603,
      "learning_rate": 3.058199636620529e-07,
      "loss": 0.4647,
      "step": 11014
    },
    {
      "epoch": 1.910999306037474,
      "grad_norm": 0.5992761254310608,
      "learning_rate": 3.046341874333636e-07,
      "loss": 0.4934,
      "step": 11015
    },
    {
      "epoch": 1.9111727966689798,
      "grad_norm": 0.8946365118026733,
      "learning_rate": 3.034506968748274e-07,
      "loss": 0.413,
      "step": 11016
    },
    {
      "epoch": 1.9113462873004856,
      "grad_norm": 0.7185291051864624,
      "learning_rate": 3.0226949212378786e-07,
      "loss": 0.5156,
      "step": 11017
    },
    {
      "epoch": 1.9115197779319917,
      "grad_norm": 0.7932426929473877,
      "learning_rate": 3.010905733173264e-07,
      "loss": 0.4785,
      "step": 11018
    },
    {
      "epoch": 1.9116932685634975,
      "grad_norm": 1.4811848402023315,
      "learning_rate": 2.9991394059225797e-07,
      "loss": 0.4618,
      "step": 11019
    },
    {
      "epoch": 1.9118667591950036,
      "grad_norm": 0.8622974753379822,
      "learning_rate": 2.987395940851312e-07,
      "loss": 0.3826,
      "step": 11020
    },
    {
      "epoch": 1.9120402498265094,
      "grad_norm": 0.939264714717865,
      "learning_rate": 2.975675339322326e-07,
      "loss": 0.3724,
      "step": 11021
    },
    {
      "epoch": 1.9122137404580153,
      "grad_norm": 0.7853683829307556,
      "learning_rate": 2.9639776026957777e-07,
      "loss": 0.4312,
      "step": 11022
    },
    {
      "epoch": 1.9123872310895211,
      "grad_norm": 1.148753046989441,
      "learning_rate": 2.9523027323292264e-07,
      "loss": 0.4413,
      "step": 11023
    },
    {
      "epoch": 1.912560721721027,
      "grad_norm": 0.7600244283676147,
      "learning_rate": 2.9406507295775657e-07,
      "loss": 0.472,
      "step": 11024
    },
    {
      "epoch": 1.9127342123525328,
      "grad_norm": 0.8786101937294006,
      "learning_rate": 2.9290215957929804e-07,
      "loss": 0.3672,
      "step": 11025
    },
    {
      "epoch": 1.912907702984039,
      "grad_norm": 0.6643133163452148,
      "learning_rate": 2.91741533232508e-07,
      "loss": 0.4315,
      "step": 11026
    },
    {
      "epoch": 1.9130811936155447,
      "grad_norm": 1.05028235912323,
      "learning_rate": 2.905831940520809e-07,
      "loss": 0.4463,
      "step": 11027
    },
    {
      "epoch": 1.9132546842470508,
      "grad_norm": 0.9317830801010132,
      "learning_rate": 2.894271421724359e-07,
      "loss": 0.3679,
      "step": 11028
    },
    {
      "epoch": 1.9134281748785567,
      "grad_norm": 1.1145577430725098,
      "learning_rate": 2.882733777277391e-07,
      "loss": 0.4349,
      "step": 11029
    },
    {
      "epoch": 1.9136016655100625,
      "grad_norm": 0.8806248307228088,
      "learning_rate": 2.871219008518877e-07,
      "loss": 0.47,
      "step": 11030
    },
    {
      "epoch": 1.9137751561415683,
      "grad_norm": 0.7149972915649414,
      "learning_rate": 2.859727116785083e-07,
      "loss": 0.5299,
      "step": 11031
    },
    {
      "epoch": 1.9139486467730742,
      "grad_norm": 0.6683125495910645,
      "learning_rate": 2.8482581034096733e-07,
      "loss": 0.5088,
      "step": 11032
    },
    {
      "epoch": 1.91412213740458,
      "grad_norm": 1.0784515142440796,
      "learning_rate": 2.8368119697236297e-07,
      "loss": 0.4847,
      "step": 11033
    },
    {
      "epoch": 1.914295628036086,
      "grad_norm": 1.723268747329712,
      "learning_rate": 2.825388717055311e-07,
      "loss": 0.3745,
      "step": 11034
    },
    {
      "epoch": 1.914469118667592,
      "grad_norm": 0.8399494290351868,
      "learning_rate": 2.8139883467303896e-07,
      "loss": 0.4497,
      "step": 11035
    },
    {
      "epoch": 1.9146426092990978,
      "grad_norm": 0.7521690726280212,
      "learning_rate": 2.8026108600718746e-07,
      "loss": 0.5481,
      "step": 11036
    },
    {
      "epoch": 1.9148160999306039,
      "grad_norm": 2.2653141021728516,
      "learning_rate": 2.7912562584001766e-07,
      "loss": 0.3365,
      "step": 11037
    },
    {
      "epoch": 1.9149895905621097,
      "grad_norm": 0.9587934017181396,
      "learning_rate": 2.7799245430329526e-07,
      "loss": 0.4149,
      "step": 11038
    },
    {
      "epoch": 1.9151630811936156,
      "grad_norm": 0.7059314846992493,
      "learning_rate": 2.768615715285283e-07,
      "loss": 0.4384,
      "step": 11039
    },
    {
      "epoch": 1.9153365718251214,
      "grad_norm": 0.7510185837745667,
      "learning_rate": 2.7573297764696085e-07,
      "loss": 0.4666,
      "step": 11040
    },
    {
      "epoch": 1.9155100624566272,
      "grad_norm": 1.0781872272491455,
      "learning_rate": 2.7460667278956355e-07,
      "loss": 0.4735,
      "step": 11041
    },
    {
      "epoch": 1.915683553088133,
      "grad_norm": 1.0659220218658447,
      "learning_rate": 2.7348265708704745e-07,
      "loss": 0.4026,
      "step": 11042
    },
    {
      "epoch": 1.9158570437196392,
      "grad_norm": 0.7296978831291199,
      "learning_rate": 2.723609306698527e-07,
      "loss": 0.5192,
      "step": 11043
    },
    {
      "epoch": 1.916030534351145,
      "grad_norm": 1.140160083770752,
      "learning_rate": 2.7124149366816177e-07,
      "loss": 0.4902,
      "step": 11044
    },
    {
      "epoch": 1.916204024982651,
      "grad_norm": 0.7639879584312439,
      "learning_rate": 2.701243462118819e-07,
      "loss": 0.4689,
      "step": 11045
    },
    {
      "epoch": 1.916377515614157,
      "grad_norm": 0.7788089513778687,
      "learning_rate": 2.690094884306649e-07,
      "loss": 0.447,
      "step": 11046
    },
    {
      "epoch": 1.9165510062456628,
      "grad_norm": 0.6959072947502136,
      "learning_rate": 2.678969204538828e-07,
      "loss": 0.5604,
      "step": 11047
    },
    {
      "epoch": 1.9167244968771686,
      "grad_norm": 0.7567552924156189,
      "learning_rate": 2.667866424106591e-07,
      "loss": 0.4889,
      "step": 11048
    },
    {
      "epoch": 1.9168979875086745,
      "grad_norm": 0.9657365679740906,
      "learning_rate": 2.656786544298373e-07,
      "loss": 0.4136,
      "step": 11049
    },
    {
      "epoch": 1.9170714781401803,
      "grad_norm": 0.9236763715744019,
      "learning_rate": 2.6457295664000573e-07,
      "loss": 0.427,
      "step": 11050
    },
    {
      "epoch": 1.9172449687716864,
      "grad_norm": 0.8116747140884399,
      "learning_rate": 2.634695491694772e-07,
      "loss": 0.3677,
      "step": 11051
    },
    {
      "epoch": 1.9174184594031922,
      "grad_norm": 0.860243558883667,
      "learning_rate": 2.623684321463049e-07,
      "loss": 0.3685,
      "step": 11052
    },
    {
      "epoch": 1.9175919500346983,
      "grad_norm": 0.7028695344924927,
      "learning_rate": 2.6126960569827554e-07,
      "loss": 0.5715,
      "step": 11053
    },
    {
      "epoch": 1.9177654406662041,
      "grad_norm": 0.8180177807807922,
      "learning_rate": 2.6017306995290926e-07,
      "loss": 0.4062,
      "step": 11054
    },
    {
      "epoch": 1.91793893129771,
      "grad_norm": 1.0677820444107056,
      "learning_rate": 2.5907882503745764e-07,
      "loss": 0.4395,
      "step": 11055
    },
    {
      "epoch": 1.9181124219292158,
      "grad_norm": 1.0341719388961792,
      "learning_rate": 2.579868710789124e-07,
      "loss": 0.513,
      "step": 11056
    },
    {
      "epoch": 1.9182859125607217,
      "grad_norm": 0.8403821587562561,
      "learning_rate": 2.5689720820399445e-07,
      "loss": 0.374,
      "step": 11057
    },
    {
      "epoch": 1.9184594031922275,
      "grad_norm": 0.813058614730835,
      "learning_rate": 2.5580983653916035e-07,
      "loss": 0.4329,
      "step": 11058
    },
    {
      "epoch": 1.9186328938237334,
      "grad_norm": 0.6556667685508728,
      "learning_rate": 2.5472475621060255e-07,
      "loss": 0.5389,
      "step": 11059
    },
    {
      "epoch": 1.9188063844552394,
      "grad_norm": 1.0568166971206665,
      "learning_rate": 2.5364196734424475e-07,
      "loss": 0.4509,
      "step": 11060
    },
    {
      "epoch": 1.9189798750867453,
      "grad_norm": 0.6956100463867188,
      "learning_rate": 2.5256147006574195e-07,
      "loss": 0.5348,
      "step": 11061
    },
    {
      "epoch": 1.9191533657182513,
      "grad_norm": 0.6176671981811523,
      "learning_rate": 2.514832645004939e-07,
      "loss": 0.5112,
      "step": 11062
    },
    {
      "epoch": 1.9193268563497572,
      "grad_norm": 1.0088751316070557,
      "learning_rate": 2.504073507736249e-07,
      "loss": 0.5662,
      "step": 11063
    },
    {
      "epoch": 1.919500346981263,
      "grad_norm": 0.7418835759162903,
      "learning_rate": 2.493337290099973e-07,
      "loss": 0.6219,
      "step": 11064
    },
    {
      "epoch": 1.9196738376127689,
      "grad_norm": 0.7149689793586731,
      "learning_rate": 2.482623993342004e-07,
      "loss": 0.4348,
      "step": 11065
    },
    {
      "epoch": 1.9198473282442747,
      "grad_norm": 0.9855237007141113,
      "learning_rate": 2.471933618705702e-07,
      "loss": 0.3461,
      "step": 11066
    },
    {
      "epoch": 1.9200208188757806,
      "grad_norm": 0.8616670370101929,
      "learning_rate": 2.4612661674316527e-07,
      "loss": 0.421,
      "step": 11067
    },
    {
      "epoch": 1.9201943095072866,
      "grad_norm": 0.8050004839897156,
      "learning_rate": 2.4506216407578665e-07,
      "loss": 0.3878,
      "step": 11068
    },
    {
      "epoch": 1.9203678001387925,
      "grad_norm": 0.8894696831703186,
      "learning_rate": 2.440000039919621e-07,
      "loss": 0.3459,
      "step": 11069
    },
    {
      "epoch": 1.9205412907702986,
      "grad_norm": 1.3463835716247559,
      "learning_rate": 2.429401366149553e-07,
      "loss": 0.2943,
      "step": 11070
    },
    {
      "epoch": 1.9207147814018044,
      "grad_norm": 0.8622847199440002,
      "learning_rate": 2.4188256206776785e-07,
      "loss": 0.4121,
      "step": 11071
    },
    {
      "epoch": 1.9208882720333103,
      "grad_norm": 0.9659159779548645,
      "learning_rate": 2.4082728047313487e-07,
      "loss": 0.5378,
      "step": 11072
    },
    {
      "epoch": 1.921061762664816,
      "grad_norm": 0.9911413788795471,
      "learning_rate": 2.397742919535162e-07,
      "loss": 0.3511,
      "step": 11073
    },
    {
      "epoch": 1.921235253296322,
      "grad_norm": 0.736795961856842,
      "learning_rate": 2.3872359663111856e-07,
      "loss": 0.4901,
      "step": 11074
    },
    {
      "epoch": 1.9214087439278278,
      "grad_norm": 0.8180474042892456,
      "learning_rate": 2.3767519462787326e-07,
      "loss": 0.5552,
      "step": 11075
    },
    {
      "epoch": 1.9215822345593336,
      "grad_norm": 1.6108638048171997,
      "learning_rate": 2.3662908606544964e-07,
      "loss": 0.4609,
      "step": 11076
    },
    {
      "epoch": 1.9217557251908397,
      "grad_norm": 0.8285250663757324,
      "learning_rate": 2.355852710652484e-07,
      "loss": 0.5109,
      "step": 11077
    },
    {
      "epoch": 1.9219292158223455,
      "grad_norm": 3.328786611557007,
      "learning_rate": 2.34543749748406e-07,
      "loss": 0.3688,
      "step": 11078
    },
    {
      "epoch": 1.9221027064538516,
      "grad_norm": 0.9429089426994324,
      "learning_rate": 2.3350452223579678e-07,
      "loss": 0.3843,
      "step": 11079
    },
    {
      "epoch": 1.9222761970853575,
      "grad_norm": 0.771192193031311,
      "learning_rate": 2.3246758864801544e-07,
      "loss": 0.3145,
      "step": 11080
    },
    {
      "epoch": 1.9224496877168633,
      "grad_norm": 0.9310546517372131,
      "learning_rate": 2.3143294910540794e-07,
      "loss": 0.4742,
      "step": 11081
    },
    {
      "epoch": 1.9226231783483692,
      "grad_norm": 0.8659986853599548,
      "learning_rate": 2.304006037280404e-07,
      "loss": 0.4091,
      "step": 11082
    },
    {
      "epoch": 1.922796668979875,
      "grad_norm": 0.8585099577903748,
      "learning_rate": 2.2937055263571928e-07,
      "loss": 0.5405,
      "step": 11083
    },
    {
      "epoch": 1.9229701596113808,
      "grad_norm": 0.7499017119407654,
      "learning_rate": 2.2834279594798002e-07,
      "loss": 0.3819,
      "step": 11084
    },
    {
      "epoch": 1.923143650242887,
      "grad_norm": 0.8291475772857666,
      "learning_rate": 2.2731733378410058e-07,
      "loss": 0.463,
      "step": 11085
    },
    {
      "epoch": 1.9233171408743928,
      "grad_norm": 0.5632716417312622,
      "learning_rate": 2.2629416626308353e-07,
      "loss": 0.5469,
      "step": 11086
    },
    {
      "epoch": 1.9234906315058988,
      "grad_norm": 0.7836756110191345,
      "learning_rate": 2.2527329350367166e-07,
      "loss": 0.4356,
      "step": 11087
    },
    {
      "epoch": 1.9236641221374047,
      "grad_norm": 0.9071162939071655,
      "learning_rate": 2.2425471562433466e-07,
      "loss": 0.4471,
      "step": 11088
    },
    {
      "epoch": 1.9238376127689105,
      "grad_norm": 1.5258830785751343,
      "learning_rate": 2.2323843274327793e-07,
      "loss": 0.4359,
      "step": 11089
    },
    {
      "epoch": 1.9240111034004164,
      "grad_norm": 0.8846318125724792,
      "learning_rate": 2.222244449784494e-07,
      "loss": 0.351,
      "step": 11090
    },
    {
      "epoch": 1.9241845940319222,
      "grad_norm": 0.9541126489639282,
      "learning_rate": 2.2121275244751939e-07,
      "loss": 0.4517,
      "step": 11091
    },
    {
      "epoch": 1.924358084663428,
      "grad_norm": 0.6848516464233398,
      "learning_rate": 2.2020335526789616e-07,
      "loss": 0.5078,
      "step": 11092
    },
    {
      "epoch": 1.9245315752949341,
      "grad_norm": 0.7995764017105103,
      "learning_rate": 2.1919625355671936e-07,
      "loss": 0.4564,
      "step": 11093
    },
    {
      "epoch": 1.92470506592644,
      "grad_norm": 1.459402084350586,
      "learning_rate": 2.1819144743086883e-07,
      "loss": 0.3871,
      "step": 11094
    },
    {
      "epoch": 1.9248785565579458,
      "grad_norm": 1.27466881275177,
      "learning_rate": 2.1718893700695132e-07,
      "loss": 0.452,
      "step": 11095
    },
    {
      "epoch": 1.9250520471894519,
      "grad_norm": 0.7997653484344482,
      "learning_rate": 2.1618872240130928e-07,
      "loss": 0.3805,
      "step": 11096
    },
    {
      "epoch": 1.9252255378209577,
      "grad_norm": 1.0554280281066895,
      "learning_rate": 2.1519080373001655e-07,
      "loss": 0.3438,
      "step": 11097
    },
    {
      "epoch": 1.9253990284524636,
      "grad_norm": 0.6617966890335083,
      "learning_rate": 2.1419518110888938e-07,
      "loss": 0.5869,
      "step": 11098
    },
    {
      "epoch": 1.9255725190839694,
      "grad_norm": 0.6871652007102966,
      "learning_rate": 2.13201854653462e-07,
      "loss": 0.4417,
      "step": 11099
    },
    {
      "epoch": 1.9257460097154753,
      "grad_norm": 0.6930230259895325,
      "learning_rate": 2.1221082447901774e-07,
      "loss": 0.5223,
      "step": 11100
    },
    {
      "epoch": 1.9259195003469811,
      "grad_norm": 0.887649416923523,
      "learning_rate": 2.1122209070056466e-07,
      "loss": 0.3975,
      "step": 11101
    },
    {
      "epoch": 1.9260929909784872,
      "grad_norm": 0.9196067452430725,
      "learning_rate": 2.1023565343284425e-07,
      "loss": 0.3744,
      "step": 11102
    },
    {
      "epoch": 1.926266481609993,
      "grad_norm": 1.1521224975585938,
      "learning_rate": 2.0925151279033828e-07,
      "loss": 0.4348,
      "step": 11103
    },
    {
      "epoch": 1.926439972241499,
      "grad_norm": 0.8633833527565002,
      "learning_rate": 2.082696688872554e-07,
      "loss": 0.5502,
      "step": 11104
    },
    {
      "epoch": 1.926613462873005,
      "grad_norm": 1.1526999473571777,
      "learning_rate": 2.0729012183753783e-07,
      "loss": 0.3765,
      "step": 11105
    },
    {
      "epoch": 1.9267869535045108,
      "grad_norm": 0.8256582021713257,
      "learning_rate": 2.063128717548657e-07,
      "loss": 0.4686,
      "step": 11106
    },
    {
      "epoch": 1.9269604441360166,
      "grad_norm": 0.8180612921714783,
      "learning_rate": 2.0533791875264608e-07,
      "loss": 0.3721,
      "step": 11107
    },
    {
      "epoch": 1.9271339347675225,
      "grad_norm": 0.7702810764312744,
      "learning_rate": 2.043652629440307e-07,
      "loss": 0.5107,
      "step": 11108
    },
    {
      "epoch": 1.9273074253990283,
      "grad_norm": 1.0001927614212036,
      "learning_rate": 2.0339490444188925e-07,
      "loss": 0.5558,
      "step": 11109
    },
    {
      "epoch": 1.9274809160305344,
      "grad_norm": 1.0472002029418945,
      "learning_rate": 2.0242684335884056e-07,
      "loss": 0.3996,
      "step": 11110
    },
    {
      "epoch": 1.9276544066620402,
      "grad_norm": 0.9137578010559082,
      "learning_rate": 2.0146107980722362e-07,
      "loss": 0.4105,
      "step": 11111
    },
    {
      "epoch": 1.9278278972935463,
      "grad_norm": 1.6406512260437012,
      "learning_rate": 2.0049761389911772e-07,
      "loss": 0.384,
      "step": 11112
    },
    {
      "epoch": 1.9280013879250522,
      "grad_norm": 0.7500235438346863,
      "learning_rate": 1.9953644574633335e-07,
      "loss": 0.4729,
      "step": 11113
    },
    {
      "epoch": 1.928174878556558,
      "grad_norm": 0.95838862657547,
      "learning_rate": 1.9857757546041912e-07,
      "loss": 0.3784,
      "step": 11114
    },
    {
      "epoch": 1.9283483691880638,
      "grad_norm": 1.3327245712280273,
      "learning_rate": 1.9762100315265043e-07,
      "loss": 0.5729,
      "step": 11115
    },
    {
      "epoch": 1.9285218598195697,
      "grad_norm": 0.7426334619522095,
      "learning_rate": 1.9666672893403627e-07,
      "loss": 0.4587,
      "step": 11116
    },
    {
      "epoch": 1.9286953504510755,
      "grad_norm": 1.5321580171585083,
      "learning_rate": 1.9571475291532805e-07,
      "loss": 0.4171,
      "step": 11117
    },
    {
      "epoch": 1.9288688410825814,
      "grad_norm": 0.6488795280456543,
      "learning_rate": 1.9476507520699518e-07,
      "loss": 0.4987,
      "step": 11118
    },
    {
      "epoch": 1.9290423317140875,
      "grad_norm": 0.7648763656616211,
      "learning_rate": 1.9381769591925614e-07,
      "loss": 0.5037,
      "step": 11119
    },
    {
      "epoch": 1.9292158223455933,
      "grad_norm": 0.8581758737564087,
      "learning_rate": 1.928726151620497e-07,
      "loss": 0.4137,
      "step": 11120
    },
    {
      "epoch": 1.9293893129770994,
      "grad_norm": 0.6748529076576233,
      "learning_rate": 1.9192983304505697e-07,
      "loss": 0.4775,
      "step": 11121
    },
    {
      "epoch": 1.9295628036086052,
      "grad_norm": 0.9109181761741638,
      "learning_rate": 1.9098934967768823e-07,
      "loss": 0.5974,
      "step": 11122
    },
    {
      "epoch": 1.929736294240111,
      "grad_norm": 0.6578526496887207,
      "learning_rate": 1.9005116516908729e-07,
      "loss": 0.5822,
      "step": 11123
    },
    {
      "epoch": 1.929909784871617,
      "grad_norm": 0.7644575834274292,
      "learning_rate": 1.891152796281337e-07,
      "loss": 0.4944,
      "step": 11124
    },
    {
      "epoch": 1.9300832755031228,
      "grad_norm": 0.693666398525238,
      "learning_rate": 1.8818169316343393e-07,
      "loss": 0.5334,
      "step": 11125
    },
    {
      "epoch": 1.9302567661346286,
      "grad_norm": 0.6019750833511353,
      "learning_rate": 1.8725040588333466e-07,
      "loss": 0.5242,
      "step": 11126
    },
    {
      "epoch": 1.9304302567661347,
      "grad_norm": 0.8340058326721191,
      "learning_rate": 1.8632141789591384e-07,
      "loss": 0.3843,
      "step": 11127
    },
    {
      "epoch": 1.9306037473976405,
      "grad_norm": 0.7593865990638733,
      "learning_rate": 1.853947293089764e-07,
      "loss": 0.4756,
      "step": 11128
    },
    {
      "epoch": 1.9307772380291466,
      "grad_norm": 1.217259407043457,
      "learning_rate": 1.8447034023007183e-07,
      "loss": 0.5166,
      "step": 11129
    },
    {
      "epoch": 1.9309507286606524,
      "grad_norm": 0.8378265500068665,
      "learning_rate": 1.8354825076647432e-07,
      "loss": 0.5759,
      "step": 11130
    },
    {
      "epoch": 1.9311242192921583,
      "grad_norm": 0.7213740944862366,
      "learning_rate": 1.826284610251916e-07,
      "loss": 0.5657,
      "step": 11131
    },
    {
      "epoch": 1.9312977099236641,
      "grad_norm": 0.7749907374382019,
      "learning_rate": 1.817109711129672e-07,
      "loss": 0.4541,
      "step": 11132
    },
    {
      "epoch": 1.93147120055517,
      "grad_norm": 0.7630119323730469,
      "learning_rate": 1.8079578113627815e-07,
      "loss": 0.4769,
      "step": 11133
    },
    {
      "epoch": 1.9316446911866758,
      "grad_norm": 1.7729418277740479,
      "learning_rate": 1.7988289120133507e-07,
      "loss": 0.3316,
      "step": 11134
    },
    {
      "epoch": 1.9318181818181817,
      "grad_norm": 0.7646145820617676,
      "learning_rate": 1.789723014140754e-07,
      "loss": 0.5846,
      "step": 11135
    },
    {
      "epoch": 1.9319916724496877,
      "grad_norm": 0.9360604882240295,
      "learning_rate": 1.7806401188017463e-07,
      "loss": 0.5652,
      "step": 11136
    },
    {
      "epoch": 1.9321651630811936,
      "grad_norm": 0.8564902544021606,
      "learning_rate": 1.77158022705044e-07,
      "loss": 0.3467,
      "step": 11137
    },
    {
      "epoch": 1.9323386537126996,
      "grad_norm": 1.0362969636917114,
      "learning_rate": 1.7625433399382386e-07,
      "loss": 0.4261,
      "step": 11138
    },
    {
      "epoch": 1.9325121443442055,
      "grad_norm": 0.7522450089454651,
      "learning_rate": 1.7535294585138808e-07,
      "loss": 0.5601,
      "step": 11139
    },
    {
      "epoch": 1.9326856349757113,
      "grad_norm": 0.6849558353424072,
      "learning_rate": 1.7445385838234185e-07,
      "loss": 0.4843,
      "step": 11140
    },
    {
      "epoch": 1.9328591256072172,
      "grad_norm": 0.7014449238777161,
      "learning_rate": 1.735570716910262e-07,
      "loss": 0.5789,
      "step": 11141
    },
    {
      "epoch": 1.933032616238723,
      "grad_norm": 0.7415981292724609,
      "learning_rate": 1.7266258588151562e-07,
      "loss": 0.4229,
      "step": 11142
    },
    {
      "epoch": 1.9332061068702289,
      "grad_norm": 0.8598527908325195,
      "learning_rate": 1.71770401057616e-07,
      "loss": 0.4771,
      "step": 11143
    },
    {
      "epoch": 1.933379597501735,
      "grad_norm": 0.6886983513832092,
      "learning_rate": 1.7088051732286448e-07,
      "loss": 0.5484,
      "step": 11144
    },
    {
      "epoch": 1.9335530881332408,
      "grad_norm": 1.1314243078231812,
      "learning_rate": 1.6999293478053404e-07,
      "loss": 0.4929,
      "step": 11145
    },
    {
      "epoch": 1.9337265787647469,
      "grad_norm": 0.7523395419120789,
      "learning_rate": 1.6910765353363334e-07,
      "loss": 0.4861,
      "step": 11146
    },
    {
      "epoch": 1.9339000693962527,
      "grad_norm": 0.8565650582313538,
      "learning_rate": 1.682246736848936e-07,
      "loss": 0.5773,
      "step": 11147
    },
    {
      "epoch": 1.9340735600277585,
      "grad_norm": 0.7338639497756958,
      "learning_rate": 1.6734399533679057e-07,
      "loss": 0.4894,
      "step": 11148
    },
    {
      "epoch": 1.9342470506592644,
      "grad_norm": 1.2362143993377686,
      "learning_rate": 1.6646561859152476e-07,
      "loss": 0.4126,
      "step": 11149
    },
    {
      "epoch": 1.9344205412907702,
      "grad_norm": 0.9865709543228149,
      "learning_rate": 1.6558954355103686e-07,
      "loss": 0.3748,
      "step": 11150
    },
    {
      "epoch": 1.934594031922276,
      "grad_norm": 0.7517378330230713,
      "learning_rate": 1.6471577031699214e-07,
      "loss": 0.4597,
      "step": 11151
    },
    {
      "epoch": 1.9347675225537821,
      "grad_norm": 0.8060457706451416,
      "learning_rate": 1.6384429899079624e-07,
      "loss": 0.3928,
      "step": 11152
    },
    {
      "epoch": 1.934941013185288,
      "grad_norm": 0.8719451427459717,
      "learning_rate": 1.6297512967358374e-07,
      "loss": 0.348,
      "step": 11153
    },
    {
      "epoch": 1.9351145038167938,
      "grad_norm": 1.1927706003189087,
      "learning_rate": 1.6210826246622068e-07,
      "loss": 0.364,
      "step": 11154
    },
    {
      "epoch": 1.9352879944483,
      "grad_norm": 1.2058417797088623,
      "learning_rate": 1.6124369746931102e-07,
      "loss": 0.5424,
      "step": 11155
    },
    {
      "epoch": 1.9354614850798058,
      "grad_norm": 0.6051502227783203,
      "learning_rate": 1.603814347831856e-07,
      "loss": 0.4691,
      "step": 11156
    },
    {
      "epoch": 1.9356349757113116,
      "grad_norm": 0.7322439551353455,
      "learning_rate": 1.595214745079132e-07,
      "loss": 0.5056,
      "step": 11157
    },
    {
      "epoch": 1.9358084663428174,
      "grad_norm": 0.6667351722717285,
      "learning_rate": 1.586638167432919e-07,
      "loss": 0.5811,
      "step": 11158
    },
    {
      "epoch": 1.9359819569743233,
      "grad_norm": 0.7149832248687744,
      "learning_rate": 1.5780846158885533e-07,
      "loss": 0.5526,
      "step": 11159
    },
    {
      "epoch": 1.9361554476058291,
      "grad_norm": 1.4899497032165527,
      "learning_rate": 1.5695540914386632e-07,
      "loss": 0.5562,
      "step": 11160
    },
    {
      "epoch": 1.9363289382373352,
      "grad_norm": 0.8246138691902161,
      "learning_rate": 1.5610465950732569e-07,
      "loss": 0.5623,
      "step": 11161
    },
    {
      "epoch": 1.936502428868841,
      "grad_norm": 2.793271541595459,
      "learning_rate": 1.552562127779611e-07,
      "loss": 0.4957,
      "step": 11162
    },
    {
      "epoch": 1.9366759195003471,
      "grad_norm": 0.9781505465507507,
      "learning_rate": 1.5441006905423605e-07,
      "loss": 0.4288,
      "step": 11163
    },
    {
      "epoch": 1.936849410131853,
      "grad_norm": 0.9941627383232117,
      "learning_rate": 1.5356622843434533e-07,
      "loss": 0.5642,
      "step": 11164
    },
    {
      "epoch": 1.9370229007633588,
      "grad_norm": 0.7078642845153809,
      "learning_rate": 1.527246910162239e-07,
      "loss": 0.4048,
      "step": 11165
    },
    {
      "epoch": 1.9371963913948647,
      "grad_norm": 0.6741459965705872,
      "learning_rate": 1.51885456897527e-07,
      "loss": 0.4824,
      "step": 11166
    },
    {
      "epoch": 1.9373698820263705,
      "grad_norm": 0.8584177494049072,
      "learning_rate": 1.5104852617565004e-07,
      "loss": 0.4513,
      "step": 11167
    },
    {
      "epoch": 1.9375433726578764,
      "grad_norm": 0.5705842971801758,
      "learning_rate": 1.5021389894771753e-07,
      "loss": 0.5319,
      "step": 11168
    },
    {
      "epoch": 1.9377168632893824,
      "grad_norm": 0.7550567388534546,
      "learning_rate": 1.493815753105965e-07,
      "loss": 0.4417,
      "step": 11169
    },
    {
      "epoch": 1.9378903539208883,
      "grad_norm": 0.9085505604743958,
      "learning_rate": 1.4855155536087184e-07,
      "loss": 0.3988,
      "step": 11170
    },
    {
      "epoch": 1.9380638445523943,
      "grad_norm": 2.0384068489074707,
      "learning_rate": 1.47723839194871e-07,
      "loss": 0.3557,
      "step": 11171
    },
    {
      "epoch": 1.9382373351839002,
      "grad_norm": 0.9320572018623352,
      "learning_rate": 1.4689842690865042e-07,
      "loss": 0.3676,
      "step": 11172
    },
    {
      "epoch": 1.938410825815406,
      "grad_norm": 0.8271452188491821,
      "learning_rate": 1.4607531859800238e-07,
      "loss": 0.4604,
      "step": 11173
    },
    {
      "epoch": 1.9385843164469119,
      "grad_norm": 0.8859689235687256,
      "learning_rate": 1.4525451435844608e-07,
      "loss": 0.3803,
      "step": 11174
    },
    {
      "epoch": 1.9387578070784177,
      "grad_norm": 0.7108444571495056,
      "learning_rate": 1.444360142852408e-07,
      "loss": 0.3311,
      "step": 11175
    },
    {
      "epoch": 1.9389312977099236,
      "grad_norm": 1.1863503456115723,
      "learning_rate": 1.436198184733706e-07,
      "loss": 0.401,
      "step": 11176
    },
    {
      "epoch": 1.9391047883414294,
      "grad_norm": 1.4052577018737793,
      "learning_rate": 1.428059270175597e-07,
      "loss": 0.3954,
      "step": 11177
    },
    {
      "epoch": 1.9392782789729355,
      "grad_norm": 0.7990099191665649,
      "learning_rate": 1.4199434001225697e-07,
      "loss": 0.4471,
      "step": 11178
    },
    {
      "epoch": 1.9394517696044413,
      "grad_norm": 0.9306536316871643,
      "learning_rate": 1.411850575516538e-07,
      "loss": 0.4528,
      "step": 11179
    },
    {
      "epoch": 1.9396252602359474,
      "grad_norm": 0.9495396614074707,
      "learning_rate": 1.4037807972966167e-07,
      "loss": 0.5072,
      "step": 11180
    },
    {
      "epoch": 1.9397987508674532,
      "grad_norm": 1.8335951566696167,
      "learning_rate": 1.3957340663993458e-07,
      "loss": 0.583,
      "step": 11181
    },
    {
      "epoch": 1.939972241498959,
      "grad_norm": 0.869001567363739,
      "learning_rate": 1.387710383758556e-07,
      "loss": 0.4868,
      "step": 11182
    },
    {
      "epoch": 1.940145732130465,
      "grad_norm": 0.9316543340682983,
      "learning_rate": 1.3797097503054136e-07,
      "loss": 0.3813,
      "step": 11183
    },
    {
      "epoch": 1.9403192227619708,
      "grad_norm": 1.5570003986358643,
      "learning_rate": 1.3717321669683981e-07,
      "loss": 0.4044,
      "step": 11184
    },
    {
      "epoch": 1.9404927133934766,
      "grad_norm": 0.6643952131271362,
      "learning_rate": 1.3637776346733022e-07,
      "loss": 0.3754,
      "step": 11185
    },
    {
      "epoch": 1.9406662040249827,
      "grad_norm": 1.1451090574264526,
      "learning_rate": 1.3558461543432767e-07,
      "loss": 0.3491,
      "step": 11186
    },
    {
      "epoch": 1.9408396946564885,
      "grad_norm": 0.9554349780082703,
      "learning_rate": 1.3479377268987626e-07,
      "loss": 0.5867,
      "step": 11187
    },
    {
      "epoch": 1.9410131852879946,
      "grad_norm": 0.7808243036270142,
      "learning_rate": 1.3400523532575592e-07,
      "loss": 0.3839,
      "step": 11188
    },
    {
      "epoch": 1.9411866759195004,
      "grad_norm": 0.8965674638748169,
      "learning_rate": 1.332190034334757e-07,
      "loss": 0.3585,
      "step": 11189
    },
    {
      "epoch": 1.9413601665510063,
      "grad_norm": 0.901248574256897,
      "learning_rate": 1.324350771042804e-07,
      "loss": 0.3152,
      "step": 11190
    },
    {
      "epoch": 1.9415336571825121,
      "grad_norm": 7.298800468444824,
      "learning_rate": 1.3165345642914385e-07,
      "loss": 0.3949,
      "step": 11191
    },
    {
      "epoch": 1.941707147814018,
      "grad_norm": 1.6301578283309937,
      "learning_rate": 1.3087414149877574e-07,
      "loss": 0.4851,
      "step": 11192
    },
    {
      "epoch": 1.9418806384455238,
      "grad_norm": 1.0139745473861694,
      "learning_rate": 1.3009713240361488e-07,
      "loss": 0.5591,
      "step": 11193
    },
    {
      "epoch": 1.9420541290770297,
      "grad_norm": 1.1971397399902344,
      "learning_rate": 1.2932242923383575e-07,
      "loss": 0.4487,
      "step": 11194
    },
    {
      "epoch": 1.9422276197085357,
      "grad_norm": 1.0054748058319092,
      "learning_rate": 1.2855003207934203e-07,
      "loss": 0.3332,
      "step": 11195
    },
    {
      "epoch": 1.9424011103400416,
      "grad_norm": 0.9131165146827698,
      "learning_rate": 1.277799410297731e-07,
      "loss": 0.4636,
      "step": 11196
    },
    {
      "epoch": 1.9425746009715477,
      "grad_norm": 1.2870208024978638,
      "learning_rate": 1.2701215617449526e-07,
      "loss": 0.5453,
      "step": 11197
    },
    {
      "epoch": 1.9427480916030535,
      "grad_norm": 0.7163132429122925,
      "learning_rate": 1.26246677602615e-07,
      "loss": 0.4519,
      "step": 11198
    },
    {
      "epoch": 1.9429215822345594,
      "grad_norm": 0.9892769455909729,
      "learning_rate": 1.2548350540296573e-07,
      "loss": 0.4975,
      "step": 11199
    },
    {
      "epoch": 1.9430950728660652,
      "grad_norm": 1.3368390798568726,
      "learning_rate": 1.2472263966411214e-07,
      "loss": 0.3402,
      "step": 11200
    },
    {
      "epoch": 1.943268563497571,
      "grad_norm": 1.0240529775619507,
      "learning_rate": 1.2396408047435694e-07,
      "loss": 0.3857,
      "step": 11201
    },
    {
      "epoch": 1.943442054129077,
      "grad_norm": 0.7207836508750916,
      "learning_rate": 1.2320782792173192e-07,
      "loss": 0.4889,
      "step": 11202
    },
    {
      "epoch": 1.943615544760583,
      "grad_norm": 1.571078896522522,
      "learning_rate": 1.224538820939958e-07,
      "loss": 0.4353,
      "step": 11203
    },
    {
      "epoch": 1.9437890353920888,
      "grad_norm": 0.9303411841392517,
      "learning_rate": 1.2170224307865185e-07,
      "loss": 0.4258,
      "step": 11204
    },
    {
      "epoch": 1.9439625260235949,
      "grad_norm": 1.0531816482543945,
      "learning_rate": 1.2095291096292373e-07,
      "loss": 0.3763,
      "step": 11205
    },
    {
      "epoch": 1.9441360166551007,
      "grad_norm": 0.9104992747306824,
      "learning_rate": 1.2020588583377513e-07,
      "loss": 0.4343,
      "step": 11206
    },
    {
      "epoch": 1.9443095072866066,
      "grad_norm": 0.9288526773452759,
      "learning_rate": 1.1946116777789673e-07,
      "loss": 0.4338,
      "step": 11207
    },
    {
      "epoch": 1.9444829979181124,
      "grad_norm": 0.7803632020950317,
      "learning_rate": 1.187187568817172e-07,
      "loss": 0.5033,
      "step": 11208
    },
    {
      "epoch": 1.9446564885496183,
      "grad_norm": 0.7046752572059631,
      "learning_rate": 1.179786532313898e-07,
      "loss": 0.5062,
      "step": 11209
    },
    {
      "epoch": 1.944829979181124,
      "grad_norm": 0.913530707359314,
      "learning_rate": 1.1724085691280806e-07,
      "loss": 0.4622,
      "step": 11210
    },
    {
      "epoch": 1.9450034698126302,
      "grad_norm": 1.06773042678833,
      "learning_rate": 1.165053680115924e-07,
      "loss": 0.4644,
      "step": 11211
    },
    {
      "epoch": 1.945176960444136,
      "grad_norm": 0.8160414099693298,
      "learning_rate": 1.1577218661309896e-07,
      "loss": 0.4015,
      "step": 11212
    },
    {
      "epoch": 1.9453504510756419,
      "grad_norm": 0.8438186645507812,
      "learning_rate": 1.1504131280241083e-07,
      "loss": 0.4371,
      "step": 11213
    },
    {
      "epoch": 1.945523941707148,
      "grad_norm": 0.822489857673645,
      "learning_rate": 1.1431274666435121e-07,
      "loss": 0.5104,
      "step": 11214
    },
    {
      "epoch": 1.9456974323386538,
      "grad_norm": 0.7588160634040833,
      "learning_rate": 1.1358648828346808e-07,
      "loss": 0.4696,
      "step": 11215
    },
    {
      "epoch": 1.9458709229701596,
      "grad_norm": 0.8216853737831116,
      "learning_rate": 1.1286253774404288e-07,
      "loss": 0.4314,
      "step": 11216
    },
    {
      "epoch": 1.9460444136016655,
      "grad_norm": 1.0485059022903442,
      "learning_rate": 1.121408951300973e-07,
      "loss": 0.3275,
      "step": 11217
    },
    {
      "epoch": 1.9462179042331713,
      "grad_norm": 0.6730453968048096,
      "learning_rate": 1.11421560525371e-07,
      "loss": 0.432,
      "step": 11218
    },
    {
      "epoch": 1.9463913948646772,
      "grad_norm": 1.148043155670166,
      "learning_rate": 1.1070453401335058e-07,
      "loss": 0.3652,
      "step": 11219
    },
    {
      "epoch": 1.9465648854961832,
      "grad_norm": 1.0786750316619873,
      "learning_rate": 1.0998981567724276e-07,
      "loss": 0.4781,
      "step": 11220
    },
    {
      "epoch": 1.946738376127689,
      "grad_norm": 1.0060077905654907,
      "learning_rate": 1.0927740559999455e-07,
      "loss": 0.4407,
      "step": 11221
    },
    {
      "epoch": 1.9469118667591951,
      "grad_norm": 0.7473181486129761,
      "learning_rate": 1.0856730386427983e-07,
      "loss": 0.4492,
      "step": 11222
    },
    {
      "epoch": 1.947085357390701,
      "grad_norm": 1.0398026704788208,
      "learning_rate": 1.078595105525082e-07,
      "loss": 0.4271,
      "step": 11223
    },
    {
      "epoch": 1.9472588480222068,
      "grad_norm": 0.624709963798523,
      "learning_rate": 1.0715402574681843e-07,
      "loss": 0.5225,
      "step": 11224
    },
    {
      "epoch": 1.9474323386537127,
      "grad_norm": 0.8374506235122681,
      "learning_rate": 1.0645084952908502e-07,
      "loss": 0.3796,
      "step": 11225
    },
    {
      "epoch": 1.9476058292852185,
      "grad_norm": 1.1964772939682007,
      "learning_rate": 1.0574998198090935e-07,
      "loss": 0.334,
      "step": 11226
    },
    {
      "epoch": 1.9477793199167244,
      "grad_norm": 0.7303481101989746,
      "learning_rate": 1.0505142318363082e-07,
      "loss": 0.5492,
      "step": 11227
    },
    {
      "epoch": 1.9479528105482304,
      "grad_norm": 1.1537619829177856,
      "learning_rate": 1.0435517321831568e-07,
      "loss": 0.4274,
      "step": 11228
    },
    {
      "epoch": 1.9481263011797363,
      "grad_norm": 0.7701857686042786,
      "learning_rate": 1.0366123216576817e-07,
      "loss": 0.3414,
      "step": 11229
    },
    {
      "epoch": 1.9482997918112424,
      "grad_norm": 1.5516411066055298,
      "learning_rate": 1.0296960010651725e-07,
      "loss": 0.3502,
      "step": 11230
    },
    {
      "epoch": 1.9484732824427482,
      "grad_norm": 0.9262827634811401,
      "learning_rate": 1.0228027712082755e-07,
      "loss": 0.3613,
      "step": 11231
    },
    {
      "epoch": 1.948646773074254,
      "grad_norm": 1.0155408382415771,
      "learning_rate": 1.0159326328869734e-07,
      "loss": 0.5461,
      "step": 11232
    },
    {
      "epoch": 1.94882026370576,
      "grad_norm": 0.8864052295684814,
      "learning_rate": 1.0090855868985616e-07,
      "loss": 0.4274,
      "step": 11233
    },
    {
      "epoch": 1.9489937543372657,
      "grad_norm": 0.7725752592086792,
      "learning_rate": 1.0022616340376489e-07,
      "loss": 0.5043,
      "step": 11234
    },
    {
      "epoch": 1.9491672449687716,
      "grad_norm": 0.7712228298187256,
      "learning_rate": 9.954607750961353e-08,
      "loss": 0.5599,
      "step": 11235
    },
    {
      "epoch": 1.9493407356002774,
      "grad_norm": 1.3105132579803467,
      "learning_rate": 9.886830108632784e-08,
      "loss": 0.5045,
      "step": 11236
    },
    {
      "epoch": 1.9495142262317835,
      "grad_norm": 0.7010358572006226,
      "learning_rate": 9.819283421256709e-08,
      "loss": 0.4967,
      "step": 11237
    },
    {
      "epoch": 1.9496877168632893,
      "grad_norm": 0.8237800002098083,
      "learning_rate": 9.751967696671749e-08,
      "loss": 0.4004,
      "step": 11238
    },
    {
      "epoch": 1.9498612074947954,
      "grad_norm": 1.1764156818389893,
      "learning_rate": 9.684882942690099e-08,
      "loss": 0.5132,
      "step": 11239
    },
    {
      "epoch": 1.9500346981263013,
      "grad_norm": 0.6724979877471924,
      "learning_rate": 9.618029167096865e-08,
      "loss": 0.5677,
      "step": 11240
    },
    {
      "epoch": 1.950208188757807,
      "grad_norm": 0.788466215133667,
      "learning_rate": 9.551406377650507e-08,
      "loss": 0.3838,
      "step": 11241
    },
    {
      "epoch": 1.950381679389313,
      "grad_norm": 0.6369549036026001,
      "learning_rate": 9.485014582083063e-08,
      "loss": 0.5577,
      "step": 11242
    },
    {
      "epoch": 1.9505551700208188,
      "grad_norm": 0.7695828080177307,
      "learning_rate": 9.418853788098814e-08,
      "loss": 0.541,
      "step": 11243
    },
    {
      "epoch": 1.9507286606523246,
      "grad_norm": 0.9789599180221558,
      "learning_rate": 9.352924003376285e-08,
      "loss": 0.3652,
      "step": 11244
    },
    {
      "epoch": 1.9509021512838307,
      "grad_norm": 0.75556480884552,
      "learning_rate": 9.287225235566244e-08,
      "loss": 0.5649,
      "step": 11245
    },
    {
      "epoch": 1.9510756419153366,
      "grad_norm": 0.7612143158912659,
      "learning_rate": 9.221757492293704e-08,
      "loss": 0.6423,
      "step": 11246
    },
    {
      "epoch": 1.9512491325468426,
      "grad_norm": 0.9684509634971619,
      "learning_rate": 9.156520781155698e-08,
      "loss": 0.4524,
      "step": 11247
    },
    {
      "epoch": 1.9514226231783485,
      "grad_norm": 0.9864277243614197,
      "learning_rate": 9.091515109723281e-08,
      "loss": 0.3969,
      "step": 11248
    },
    {
      "epoch": 1.9515961138098543,
      "grad_norm": 0.8819375038146973,
      "learning_rate": 9.026740485540197e-08,
      "loss": 0.5066,
      "step": 11249
    },
    {
      "epoch": 1.9517696044413602,
      "grad_norm": 0.841587483882904,
      "learning_rate": 8.96219691612421e-08,
      "loss": 0.4746,
      "step": 11250
    },
    {
      "epoch": 1.951943095072866,
      "grad_norm": 0.8239153027534485,
      "learning_rate": 8.897884408964885e-08,
      "loss": 0.3904,
      "step": 11251
    },
    {
      "epoch": 1.9521165857043719,
      "grad_norm": 0.8045653700828552,
      "learning_rate": 8.833802971526472e-08,
      "loss": 0.4621,
      "step": 11252
    },
    {
      "epoch": 1.9522900763358777,
      "grad_norm": 0.9592332243919373,
      "learning_rate": 8.769952611245248e-08,
      "loss": 0.4333,
      "step": 11253
    },
    {
      "epoch": 1.9524635669673838,
      "grad_norm": 0.8656499981880188,
      "learning_rate": 8.706333335531503e-08,
      "loss": 0.571,
      "step": 11254
    },
    {
      "epoch": 1.9526370575988896,
      "grad_norm": 1.0844917297363281,
      "learning_rate": 8.642945151767779e-08,
      "loss": 0.4554,
      "step": 11255
    },
    {
      "epoch": 1.9528105482303957,
      "grad_norm": 0.8129265904426575,
      "learning_rate": 8.579788067310858e-08,
      "loss": 0.4396,
      "step": 11256
    },
    {
      "epoch": 1.9529840388619015,
      "grad_norm": 1.2656227350234985,
      "learning_rate": 8.516862089489986e-08,
      "loss": 0.4579,
      "step": 11257
    },
    {
      "epoch": 1.9531575294934074,
      "grad_norm": 0.8240922093391418,
      "learning_rate": 8.454167225607768e-08,
      "loss": 0.4358,
      "step": 11258
    },
    {
      "epoch": 1.9533310201249132,
      "grad_norm": 1.1505444049835205,
      "learning_rate": 8.391703482939939e-08,
      "loss": 0.4935,
      "step": 11259
    },
    {
      "epoch": 1.953504510756419,
      "grad_norm": 0.8059971928596497,
      "learning_rate": 8.32947086873559e-08,
      "loss": 0.5243,
      "step": 11260
    },
    {
      "epoch": 1.953678001387925,
      "grad_norm": 0.8615079522132874,
      "learning_rate": 8.267469390217164e-08,
      "loss": 0.3658,
      "step": 11261
    },
    {
      "epoch": 1.953851492019431,
      "grad_norm": 1.0338557958602905,
      "learning_rate": 8.205699054579575e-08,
      "loss": 0.386,
      "step": 11262
    },
    {
      "epoch": 1.9540249826509368,
      "grad_norm": 0.7419182658195496,
      "learning_rate": 8.14415986899153e-08,
      "loss": 0.4982,
      "step": 11263
    },
    {
      "epoch": 1.954198473282443,
      "grad_norm": 0.9103652834892273,
      "learning_rate": 8.082851840594652e-08,
      "loss": 0.5511,
      "step": 11264
    },
    {
      "epoch": 1.9543719639139487,
      "grad_norm": 0.8418128490447998,
      "learning_rate": 8.021774976503915e-08,
      "loss": 0.4381,
      "step": 11265
    },
    {
      "epoch": 1.9545454545454546,
      "grad_norm": 0.7194525599479675,
      "learning_rate": 7.960929283807429e-08,
      "loss": 0.5033,
      "step": 11266
    },
    {
      "epoch": 1.9547189451769604,
      "grad_norm": 0.9392405152320862,
      "learning_rate": 7.900314769566208e-08,
      "loss": 0.5702,
      "step": 11267
    },
    {
      "epoch": 1.9548924358084663,
      "grad_norm": 0.6881068348884583,
      "learning_rate": 7.83993144081463e-08,
      "loss": 0.5391,
      "step": 11268
    },
    {
      "epoch": 1.9550659264399721,
      "grad_norm": 0.7253534197807312,
      "learning_rate": 7.77977930456042e-08,
      "loss": 0.4403,
      "step": 11269
    },
    {
      "epoch": 1.9552394170714782,
      "grad_norm": 0.9656354188919067,
      "learning_rate": 7.719858367784216e-08,
      "loss": 0.5089,
      "step": 11270
    },
    {
      "epoch": 1.955412907702984,
      "grad_norm": 0.573002278804779,
      "learning_rate": 7.66016863743979e-08,
      "loss": 0.5603,
      "step": 11271
    },
    {
      "epoch": 1.9555863983344899,
      "grad_norm": 0.6548886299133301,
      "learning_rate": 7.600710120454491e-08,
      "loss": 0.5315,
      "step": 11272
    },
    {
      "epoch": 1.955759888965996,
      "grad_norm": 3.2776379585266113,
      "learning_rate": 7.541482823728352e-08,
      "loss": 0.4942,
      "step": 11273
    },
    {
      "epoch": 1.9559333795975018,
      "grad_norm": 1.1634551286697388,
      "learning_rate": 7.482486754134765e-08,
      "loss": 0.4874,
      "step": 11274
    },
    {
      "epoch": 1.9561068702290076,
      "grad_norm": 0.9876819849014282,
      "learning_rate": 7.423721918520477e-08,
      "loss": 0.5181,
      "step": 11275
    },
    {
      "epoch": 1.9562803608605135,
      "grad_norm": 0.7193216681480408,
      "learning_rate": 7.365188323704919e-08,
      "loss": 0.4337,
      "step": 11276
    },
    {
      "epoch": 1.9564538514920193,
      "grad_norm": 0.8667590618133545,
      "learning_rate": 7.306885976481104e-08,
      "loss": 0.4097,
      "step": 11277
    },
    {
      "epoch": 1.9566273421235252,
      "grad_norm": 0.9627739787101746,
      "learning_rate": 7.248814883615174e-08,
      "loss": 0.5001,
      "step": 11278
    },
    {
      "epoch": 1.9568008327550312,
      "grad_norm": 0.7978907823562622,
      "learning_rate": 7.190975051846406e-08,
      "loss": 0.4882,
      "step": 11279
    },
    {
      "epoch": 1.956974323386537,
      "grad_norm": 0.685856282711029,
      "learning_rate": 7.133366487886762e-08,
      "loss": 0.6051,
      "step": 11280
    },
    {
      "epoch": 1.9571478140180432,
      "grad_norm": 0.6509477496147156,
      "learning_rate": 7.07598919842245e-08,
      "loss": 0.6182,
      "step": 11281
    },
    {
      "epoch": 1.957321304649549,
      "grad_norm": 0.8358609676361084,
      "learning_rate": 7.018843190111479e-08,
      "loss": 0.3975,
      "step": 11282
    },
    {
      "epoch": 1.9574947952810549,
      "grad_norm": 1.0334913730621338,
      "learning_rate": 6.961928469586321e-08,
      "loss": 0.5237,
      "step": 11283
    },
    {
      "epoch": 1.9576682859125607,
      "grad_norm": 0.5888184309005737,
      "learning_rate": 6.90524504345147e-08,
      "loss": 0.4191,
      "step": 11284
    },
    {
      "epoch": 1.9578417765440665,
      "grad_norm": 0.9598404169082642,
      "learning_rate": 6.848792918285663e-08,
      "loss": 0.4172,
      "step": 11285
    },
    {
      "epoch": 1.9580152671755724,
      "grad_norm": 1.0853744745254517,
      "learning_rate": 6.792572100639661e-08,
      "loss": 0.344,
      "step": 11286
    },
    {
      "epoch": 1.9581887578070785,
      "grad_norm": 0.9685314893722534,
      "learning_rate": 6.736582597038243e-08,
      "loss": 0.4727,
      "step": 11287
    },
    {
      "epoch": 1.9583622484385843,
      "grad_norm": 0.7487372159957886,
      "learning_rate": 6.680824413979103e-08,
      "loss": 0.4193,
      "step": 11288
    },
    {
      "epoch": 1.9585357390700904,
      "grad_norm": 0.9230625629425049,
      "learning_rate": 6.625297557932842e-08,
      "loss": 0.3921,
      "step": 11289
    },
    {
      "epoch": 1.9587092297015962,
      "grad_norm": 0.8867302536964417,
      "learning_rate": 6.570002035343636e-08,
      "loss": 0.3885,
      "step": 11290
    },
    {
      "epoch": 1.958882720333102,
      "grad_norm": 1.5723206996917725,
      "learning_rate": 6.514937852628578e-08,
      "loss": 0.3677,
      "step": 11291
    },
    {
      "epoch": 1.959056210964608,
      "grad_norm": 2.336076498031616,
      "learning_rate": 6.460105016177887e-08,
      "loss": 0.439,
      "step": 11292
    },
    {
      "epoch": 1.9592297015961138,
      "grad_norm": 0.865355908870697,
      "learning_rate": 6.405503532354695e-08,
      "loss": 0.402,
      "step": 11293
    },
    {
      "epoch": 1.9594031922276196,
      "grad_norm": 1.979523777961731,
      "learning_rate": 6.351133407495936e-08,
      "loss": 0.5083,
      "step": 11294
    },
    {
      "epoch": 1.9595766828591255,
      "grad_norm": 0.8357073068618774,
      "learning_rate": 6.296994647911448e-08,
      "loss": 0.4095,
      "step": 11295
    },
    {
      "epoch": 1.9597501734906315,
      "grad_norm": 0.9894707202911377,
      "learning_rate": 6.24308725988354e-08,
      "loss": 0.4575,
      "step": 11296
    },
    {
      "epoch": 1.9599236641221374,
      "grad_norm": 0.8860669136047363,
      "learning_rate": 6.189411249668542e-08,
      "loss": 0.3809,
      "step": 11297
    },
    {
      "epoch": 1.9600971547536434,
      "grad_norm": 0.9638963937759399,
      "learning_rate": 6.135966623495915e-08,
      "loss": 0.3666,
      "step": 11298
    },
    {
      "epoch": 1.9602706453851493,
      "grad_norm": 1.0723718404769897,
      "learning_rate": 6.082753387567364e-08,
      "loss": 0.496,
      "step": 11299
    },
    {
      "epoch": 1.9604441360166551,
      "grad_norm": 1.1697533130645752,
      "learning_rate": 6.029771548058838e-08,
      "loss": 0.4438,
      "step": 11300
    },
    {
      "epoch": 1.960617626648161,
      "grad_norm": 0.9623770117759705,
      "learning_rate": 5.977021111118752e-08,
      "loss": 0.348,
      "step": 11301
    },
    {
      "epoch": 1.9607911172796668,
      "grad_norm": 1.0842669010162354,
      "learning_rate": 5.924502082868655e-08,
      "loss": 0.3619,
      "step": 11302
    },
    {
      "epoch": 1.9609646079111727,
      "grad_norm": 1.1347887516021729,
      "learning_rate": 5.872214469403892e-08,
      "loss": 0.4243,
      "step": 11303
    },
    {
      "epoch": 1.9611380985426787,
      "grad_norm": 0.8801885843276978,
      "learning_rate": 5.820158276792054e-08,
      "loss": 0.4603,
      "step": 11304
    },
    {
      "epoch": 1.9613115891741846,
      "grad_norm": 0.6171122193336487,
      "learning_rate": 5.768333511074753e-08,
      "loss": 0.5493,
      "step": 11305
    },
    {
      "epoch": 1.9614850798056906,
      "grad_norm": 0.8176255822181702,
      "learning_rate": 5.716740178266067e-08,
      "loss": 0.4808,
      "step": 11306
    },
    {
      "epoch": 1.9616585704371965,
      "grad_norm": 0.7006524205207825,
      "learning_rate": 5.665378284353207e-08,
      "loss": 0.5588,
      "step": 11307
    },
    {
      "epoch": 1.9618320610687023,
      "grad_norm": 0.9156227707862854,
      "learning_rate": 5.614247835297404e-08,
      "loss": 0.5604,
      "step": 11308
    },
    {
      "epoch": 1.9620055517002082,
      "grad_norm": 0.8872699737548828,
      "learning_rate": 5.563348837031912e-08,
      "loss": 0.4641,
      "step": 11309
    },
    {
      "epoch": 1.962179042331714,
      "grad_norm": 0.9895948171615601,
      "learning_rate": 5.512681295463784e-08,
      "loss": 0.449,
      "step": 11310
    },
    {
      "epoch": 1.9623525329632199,
      "grad_norm": 1.2851054668426514,
      "learning_rate": 5.462245216472983e-08,
      "loss": 0.4772,
      "step": 11311
    },
    {
      "epoch": 1.9625260235947257,
      "grad_norm": 0.9638526439666748,
      "learning_rate": 5.4120406059128274e-08,
      "loss": 0.4779,
      "step": 11312
    },
    {
      "epoch": 1.9626995142262318,
      "grad_norm": 0.9218172430992126,
      "learning_rate": 5.362067469609322e-08,
      "loss": 0.4899,
      "step": 11313
    },
    {
      "epoch": 1.9628730048577376,
      "grad_norm": 0.8112190365791321,
      "learning_rate": 5.312325813362274e-08,
      "loss": 0.4637,
      "step": 11314
    },
    {
      "epoch": 1.9630464954892437,
      "grad_norm": 0.9659833908081055,
      "learning_rate": 5.262815642943953e-08,
      "loss": 0.4358,
      "step": 11315
    },
    {
      "epoch": 1.9632199861207495,
      "grad_norm": 0.9530385732650757,
      "learning_rate": 5.21353696410043e-08,
      "loss": 0.5176,
      "step": 11316
    },
    {
      "epoch": 1.9633934767522554,
      "grad_norm": 0.8164271712303162,
      "learning_rate": 5.1644897825502416e-08,
      "loss": 0.4286,
      "step": 11317
    },
    {
      "epoch": 1.9635669673837612,
      "grad_norm": 0.9133117198944092,
      "learning_rate": 5.11567410398528e-08,
      "loss": 0.525,
      "step": 11318
    },
    {
      "epoch": 1.963740458015267,
      "grad_norm": 0.8649409413337708,
      "learning_rate": 5.0670899340710124e-08,
      "loss": 0.4792,
      "step": 11319
    },
    {
      "epoch": 1.963913948646773,
      "grad_norm": 0.8076494932174683,
      "learning_rate": 5.0187372784453734e-08,
      "loss": 0.4442,
      "step": 11320
    },
    {
      "epoch": 1.964087439278279,
      "grad_norm": 1.2337809801101685,
      "learning_rate": 4.970616142720097e-08,
      "loss": 0.4258,
      "step": 11321
    },
    {
      "epoch": 1.9642609299097848,
      "grad_norm": 0.6409690380096436,
      "learning_rate": 4.922726532479383e-08,
      "loss": 0.5856,
      "step": 11322
    },
    {
      "epoch": 1.964434420541291,
      "grad_norm": 0.9598265886306763,
      "learning_rate": 4.8750684532810064e-08,
      "loss": 0.4957,
      "step": 11323
    },
    {
      "epoch": 1.9646079111727968,
      "grad_norm": 1.5102465152740479,
      "learning_rate": 4.827641910655656e-08,
      "loss": 0.4661,
      "step": 11324
    },
    {
      "epoch": 1.9647814018043026,
      "grad_norm": 0.7760123610496521,
      "learning_rate": 4.7804469101073725e-08,
      "loss": 0.4912,
      "step": 11325
    },
    {
      "epoch": 1.9649548924358085,
      "grad_norm": 0.7980136871337891,
      "learning_rate": 4.7334834571128866e-08,
      "loss": 0.4211,
      "step": 11326
    },
    {
      "epoch": 1.9651283830673143,
      "grad_norm": 0.8376246094703674,
      "learning_rate": 4.6867515571229485e-08,
      "loss": 0.3892,
      "step": 11327
    },
    {
      "epoch": 1.9653018736988201,
      "grad_norm": 0.8894677758216858,
      "learning_rate": 4.640251215560332e-08,
      "loss": 0.384,
      "step": 11328
    },
    {
      "epoch": 1.9654753643303262,
      "grad_norm": 0.8758121728897095,
      "learning_rate": 4.593982437821609e-08,
      "loss": 0.3887,
      "step": 11329
    },
    {
      "epoch": 1.965648854961832,
      "grad_norm": 0.7407442927360535,
      "learning_rate": 4.547945229276263e-08,
      "loss": 0.4206,
      "step": 11330
    },
    {
      "epoch": 1.965822345593338,
      "grad_norm": 0.7268120646476746,
      "learning_rate": 4.5021395952671297e-08,
      "loss": 0.5168,
      "step": 11331
    },
    {
      "epoch": 1.965995836224844,
      "grad_norm": 0.9915668368339539,
      "learning_rate": 4.456565541109958e-08,
      "loss": 0.4204,
      "step": 11332
    },
    {
      "epoch": 1.9661693268563498,
      "grad_norm": 1.1841089725494385,
      "learning_rate": 4.411223072093629e-08,
      "loss": 0.4003,
      "step": 11333
    },
    {
      "epoch": 1.9663428174878557,
      "grad_norm": 0.8532261848449707,
      "learning_rate": 4.366112193480154e-08,
      "loss": 0.4045,
      "step": 11334
    },
    {
      "epoch": 1.9665163081193615,
      "grad_norm": 0.8033369183540344,
      "learning_rate": 4.32123291050468e-08,
      "loss": 0.5156,
      "step": 11335
    },
    {
      "epoch": 1.9666897987508674,
      "grad_norm": 0.7514310479164124,
      "learning_rate": 4.276585228375485e-08,
      "loss": 0.4547,
      "step": 11336
    },
    {
      "epoch": 1.9668632893823732,
      "grad_norm": 0.7093128561973572,
      "learning_rate": 4.2321691522742013e-08,
      "loss": 0.543,
      "step": 11337
    },
    {
      "epoch": 1.9670367800138793,
      "grad_norm": 0.8851273059844971,
      "learning_rate": 4.187984687355151e-08,
      "loss": 0.4689,
      "step": 11338
    },
    {
      "epoch": 1.9672102706453851,
      "grad_norm": 1.343516230583191,
      "learning_rate": 4.14403183874601e-08,
      "loss": 0.4718,
      "step": 11339
    },
    {
      "epoch": 1.9673837612768912,
      "grad_norm": 0.8390950560569763,
      "learning_rate": 4.100310611547809e-08,
      "loss": 0.5067,
      "step": 11340
    },
    {
      "epoch": 1.967557251908397,
      "grad_norm": 0.7784774303436279,
      "learning_rate": 4.056821010834045e-08,
      "loss": 0.3678,
      "step": 11341
    },
    {
      "epoch": 1.9677307425399029,
      "grad_norm": 0.8302329778671265,
      "learning_rate": 4.013563041652013e-08,
      "loss": 0.5221,
      "step": 11342
    },
    {
      "epoch": 1.9679042331714087,
      "grad_norm": 1.199031114578247,
      "learning_rate": 3.9705367090216995e-08,
      "loss": 0.3467,
      "step": 11343
    },
    {
      "epoch": 1.9680777238029146,
      "grad_norm": 0.9485912919044495,
      "learning_rate": 3.927742017936664e-08,
      "loss": 0.3676,
      "step": 11344
    },
    {
      "epoch": 1.9682512144344204,
      "grad_norm": 0.9357582926750183,
      "learning_rate": 3.885178973362713e-08,
      "loss": 0.454,
      "step": 11345
    },
    {
      "epoch": 1.9684247050659265,
      "grad_norm": 0.74022376537323,
      "learning_rate": 3.8428475802398944e-08,
      "loss": 0.5356,
      "step": 11346
    },
    {
      "epoch": 1.9685981956974323,
      "grad_norm": 0.6848518252372742,
      "learning_rate": 3.800747843480501e-08,
      "loss": 0.5144,
      "step": 11347
    },
    {
      "epoch": 1.9687716863289384,
      "grad_norm": 0.937227189540863,
      "learning_rate": 3.7588797679706245e-08,
      "loss": 0.399,
      "step": 11348
    },
    {
      "epoch": 1.9689451769604442,
      "grad_norm": 0.6459150314331055,
      "learning_rate": 3.717243358568379e-08,
      "loss": 0.5599,
      "step": 11349
    },
    {
      "epoch": 1.96911866759195,
      "grad_norm": 0.8304569125175476,
      "learning_rate": 3.6758386201065645e-08,
      "loss": 0.45,
      "step": 11350
    },
    {
      "epoch": 1.969292158223456,
      "grad_norm": 1.2463147640228271,
      "learning_rate": 3.6346655573897823e-08,
      "loss": 0.3424,
      "step": 11351
    },
    {
      "epoch": 1.9694656488549618,
      "grad_norm": 0.9651402831077576,
      "learning_rate": 3.5937241751962115e-08,
      "loss": 0.532,
      "step": 11352
    },
    {
      "epoch": 1.9696391394864676,
      "grad_norm": 0.8306807279586792,
      "learning_rate": 3.5530144782771616e-08,
      "loss": 0.4624,
      "step": 11353
    },
    {
      "epoch": 1.9698126301179735,
      "grad_norm": 1.2534830570220947,
      "learning_rate": 3.5125364713572976e-08,
      "loss": 0.353,
      "step": 11354
    },
    {
      "epoch": 1.9699861207494795,
      "grad_norm": 1.3988927602767944,
      "learning_rate": 3.472290159133751e-08,
      "loss": 0.4821,
      "step": 11355
    },
    {
      "epoch": 1.9701596113809854,
      "grad_norm": 0.7488888502120972,
      "learning_rate": 3.4322755462774525e-08,
      "loss": 0.604,
      "step": 11356
    },
    {
      "epoch": 1.9703331020124915,
      "grad_norm": 1.0150195360183716,
      "learning_rate": 3.392492637432021e-08,
      "loss": 0.4247,
      "step": 11357
    },
    {
      "epoch": 1.9705065926439973,
      "grad_norm": 0.9276250600814819,
      "learning_rate": 3.3529414372142074e-08,
      "loss": 0.3502,
      "step": 11358
    },
    {
      "epoch": 1.9706800832755031,
      "grad_norm": 1.1141239404678345,
      "learning_rate": 3.3136219502143406e-08,
      "loss": 0.3543,
      "step": 11359
    },
    {
      "epoch": 1.970853573907009,
      "grad_norm": 0.6158331632614136,
      "learning_rate": 3.2745341809949923e-08,
      "loss": 0.5022,
      "step": 11360
    },
    {
      "epoch": 1.9710270645385148,
      "grad_norm": 0.9471350312232971,
      "learning_rate": 3.235678134092757e-08,
      "loss": 0.3702,
      "step": 11361
    },
    {
      "epoch": 1.9712005551700207,
      "grad_norm": 0.8030177354812622,
      "learning_rate": 3.1970538140166927e-08,
      "loss": 0.601,
      "step": 11362
    },
    {
      "epoch": 1.9713740458015268,
      "grad_norm": 0.9062731266021729,
      "learning_rate": 3.158661225249437e-08,
      "loss": 0.4815,
      "step": 11363
    },
    {
      "epoch": 1.9715475364330326,
      "grad_norm": 0.655491054058075,
      "learning_rate": 3.1205003722460935e-08,
      "loss": 0.481,
      "step": 11364
    },
    {
      "epoch": 1.9717210270645387,
      "grad_norm": 0.81975257396698,
      "learning_rate": 3.082571259435563e-08,
      "loss": 0.4901,
      "step": 11365
    },
    {
      "epoch": 1.9718945176960445,
      "grad_norm": 0.8341805338859558,
      "learning_rate": 3.0448738912196574e-08,
      "loss": 0.344,
      "step": 11366
    },
    {
      "epoch": 1.9720680083275504,
      "grad_norm": 1.1064727306365967,
      "learning_rate": 3.007408271972878e-08,
      "loss": 0.3383,
      "step": 11367
    },
    {
      "epoch": 1.9722414989590562,
      "grad_norm": 1.14534592628479,
      "learning_rate": 2.9701744060435246e-08,
      "loss": 0.3644,
      "step": 11368
    },
    {
      "epoch": 1.972414989590562,
      "grad_norm": 1.072205901145935,
      "learning_rate": 2.9331722977523625e-08,
      "loss": 0.3726,
      "step": 11369
    },
    {
      "epoch": 1.972588480222068,
      "grad_norm": 0.8797577619552612,
      "learning_rate": 2.8964019513935126e-08,
      "loss": 0.4996,
      "step": 11370
    },
    {
      "epoch": 1.9727619708535737,
      "grad_norm": 1.1882641315460205,
      "learning_rate": 2.8598633712342283e-08,
      "loss": 0.4749,
      "step": 11371
    },
    {
      "epoch": 1.9729354614850798,
      "grad_norm": 0.8657835721969604,
      "learning_rate": 2.8235565615151172e-08,
      "loss": 0.4869,
      "step": 11372
    },
    {
      "epoch": 1.9731089521165857,
      "grad_norm": 1.03721284866333,
      "learning_rate": 2.787481526449476e-08,
      "loss": 0.3392,
      "step": 11373
    },
    {
      "epoch": 1.9732824427480917,
      "grad_norm": 0.7597808241844177,
      "learning_rate": 2.7516382702235112e-08,
      "loss": 0.4977,
      "step": 11374
    },
    {
      "epoch": 1.9734559333795976,
      "grad_norm": 0.8189690113067627,
      "learning_rate": 2.7160267969974508e-08,
      "loss": 0.3325,
      "step": 11375
    },
    {
      "epoch": 1.9736294240111034,
      "grad_norm": 0.7845630645751953,
      "learning_rate": 2.6806471109037668e-08,
      "loss": 0.4585,
      "step": 11376
    },
    {
      "epoch": 1.9738029146426093,
      "grad_norm": 0.7243348956108093,
      "learning_rate": 2.645499216048286e-08,
      "loss": 0.4583,
      "step": 11377
    },
    {
      "epoch": 1.973976405274115,
      "grad_norm": 1.1351598501205444,
      "learning_rate": 2.6105831165099683e-08,
      "loss": 0.3558,
      "step": 11378
    },
    {
      "epoch": 1.974149895905621,
      "grad_norm": 1.3864905834197998,
      "learning_rate": 2.575898816340905e-08,
      "loss": 0.3636,
      "step": 11379
    },
    {
      "epoch": 1.974323386537127,
      "grad_norm": 0.6895972490310669,
      "learning_rate": 2.541446319566321e-08,
      "loss": 0.4867,
      "step": 11380
    },
    {
      "epoch": 1.9744968771686329,
      "grad_norm": 0.7314673066139221,
      "learning_rate": 2.5072256301843513e-08,
      "loss": 0.4059,
      "step": 11381
    },
    {
      "epoch": 1.974670367800139,
      "grad_norm": 1.2502267360687256,
      "learning_rate": 2.473236752166264e-08,
      "loss": 0.3456,
      "step": 11382
    },
    {
      "epoch": 1.9748438584316448,
      "grad_norm": 0.7512307167053223,
      "learning_rate": 2.4394796894566807e-08,
      "loss": 0.4225,
      "step": 11383
    },
    {
      "epoch": 1.9750173490631506,
      "grad_norm": 0.8592346906661987,
      "learning_rate": 2.4059544459731356e-08,
      "loss": 0.4131,
      "step": 11384
    },
    {
      "epoch": 1.9751908396946565,
      "grad_norm": 1.1116094589233398,
      "learning_rate": 2.3726610256062933e-08,
      "loss": 0.4628,
      "step": 11385
    },
    {
      "epoch": 1.9753643303261623,
      "grad_norm": 1.1277997493743896,
      "learning_rate": 2.3395994322199522e-08,
      "loss": 0.429,
      "step": 11386
    },
    {
      "epoch": 1.9755378209576682,
      "grad_norm": 0.8579531908035278,
      "learning_rate": 2.3067696696505992e-08,
      "loss": 0.5604,
      "step": 11387
    },
    {
      "epoch": 1.9757113115891742,
      "grad_norm": 1.0007001161575317,
      "learning_rate": 2.2741717417085196e-08,
      "loss": 0.4213,
      "step": 11388
    },
    {
      "epoch": 1.97588480222068,
      "grad_norm": 0.8642428517341614,
      "learning_rate": 2.2418056521764653e-08,
      "loss": 0.4944,
      "step": 11389
    },
    {
      "epoch": 1.976058292852186,
      "grad_norm": 1.2023932933807373,
      "learning_rate": 2.2096714048109867e-08,
      "loss": 0.5009,
      "step": 11390
    },
    {
      "epoch": 1.976231783483692,
      "grad_norm": 0.8968191146850586,
      "learning_rate": 2.1777690033408795e-08,
      "loss": 0.4225,
      "step": 11391
    },
    {
      "epoch": 1.9764052741151978,
      "grad_norm": 0.76925128698349,
      "learning_rate": 2.1460984514685145e-08,
      "loss": 0.5323,
      "step": 11392
    },
    {
      "epoch": 1.9765787647467037,
      "grad_norm": 0.8122766613960266,
      "learning_rate": 2.114659752869619e-08,
      "loss": 0.3684,
      "step": 11393
    },
    {
      "epoch": 1.9767522553782095,
      "grad_norm": 0.7954847812652588,
      "learning_rate": 2.083452911192163e-08,
      "loss": 0.6046,
      "step": 11394
    },
    {
      "epoch": 1.9769257460097154,
      "grad_norm": 0.7282482981681824,
      "learning_rate": 2.0524779300581386e-08,
      "loss": 0.4869,
      "step": 11395
    },
    {
      "epoch": 1.9770992366412212,
      "grad_norm": 0.833308219909668,
      "learning_rate": 2.021734813062226e-08,
      "loss": 0.5249,
      "step": 11396
    },
    {
      "epoch": 1.9772727272727273,
      "grad_norm": 0.7510433793067932,
      "learning_rate": 1.9912235637720156e-08,
      "loss": 0.5702,
      "step": 11397
    },
    {
      "epoch": 1.9774462179042331,
      "grad_norm": 0.7454752326011658,
      "learning_rate": 1.960944185728675e-08,
      "loss": 0.5729,
      "step": 11398
    },
    {
      "epoch": 1.9776197085357392,
      "grad_norm": 0.7768095135688782,
      "learning_rate": 1.9308966824456154e-08,
      "loss": 0.4779,
      "step": 11399
    },
    {
      "epoch": 1.977793199167245,
      "grad_norm": 0.8578220009803772,
      "learning_rate": 1.9010810574102702e-08,
      "loss": 0.4636,
      "step": 11400
    },
    {
      "epoch": 1.977966689798751,
      "grad_norm": 0.8072434067726135,
      "learning_rate": 1.8714973140827596e-08,
      "loss": 0.4553,
      "step": 11401
    },
    {
      "epoch": 1.9781401804302567,
      "grad_norm": 0.8005666732788086,
      "learning_rate": 1.842145455896338e-08,
      "loss": 0.3922,
      "step": 11402
    },
    {
      "epoch": 1.9783136710617626,
      "grad_norm": 0.8736897706985474,
      "learning_rate": 1.8130254862571693e-08,
      "loss": 0.4479,
      "step": 11403
    },
    {
      "epoch": 1.9784871616932684,
      "grad_norm": 1.0670888423919678,
      "learning_rate": 1.7841374085447728e-08,
      "loss": 0.3694,
      "step": 11404
    },
    {
      "epoch": 1.9786606523247745,
      "grad_norm": 0.860889196395874,
      "learning_rate": 1.7554812261117994e-08,
      "loss": 0.3245,
      "step": 11405
    },
    {
      "epoch": 1.9788341429562804,
      "grad_norm": 0.9610465168952942,
      "learning_rate": 1.727056942283367e-08,
      "loss": 0.4326,
      "step": 11406
    },
    {
      "epoch": 1.9790076335877864,
      "grad_norm": 0.8125248551368713,
      "learning_rate": 1.6988645603586153e-08,
      "loss": 0.5042,
      "step": 11407
    },
    {
      "epoch": 1.9791811242192923,
      "grad_norm": 0.7917909026145935,
      "learning_rate": 1.6709040836089262e-08,
      "loss": 0.5035,
      "step": 11408
    },
    {
      "epoch": 1.979354614850798,
      "grad_norm": 0.9587294459342957,
      "learning_rate": 1.6431755152794827e-08,
      "loss": 0.3547,
      "step": 11409
    },
    {
      "epoch": 1.979528105482304,
      "grad_norm": 1.0159237384796143,
      "learning_rate": 1.6156788585879325e-08,
      "loss": 0.5345,
      "step": 11410
    },
    {
      "epoch": 1.9797015961138098,
      "grad_norm": 0.8712301850318909,
      "learning_rate": 1.5884141167255007e-08,
      "loss": 0.4755,
      "step": 11411
    },
    {
      "epoch": 1.9798750867453156,
      "grad_norm": 0.683390736579895,
      "learning_rate": 1.5613812928563233e-08,
      "loss": 0.4174,
      "step": 11412
    },
    {
      "epoch": 1.9800485773768215,
      "grad_norm": 0.6609227061271667,
      "learning_rate": 1.534580390117446e-08,
      "loss": 0.5688,
      "step": 11413
    },
    {
      "epoch": 1.9802220680083276,
      "grad_norm": 0.8728233575820923,
      "learning_rate": 1.508011411619048e-08,
      "loss": 0.4512,
      "step": 11414
    },
    {
      "epoch": 1.9803955586398334,
      "grad_norm": 1.1786134243011475,
      "learning_rate": 1.4816743604448846e-08,
      "loss": 0.4167,
      "step": 11415
    },
    {
      "epoch": 1.9805690492713395,
      "grad_norm": 0.8179705739021301,
      "learning_rate": 1.4555692396509557e-08,
      "loss": 0.4727,
      "step": 11416
    },
    {
      "epoch": 1.9807425399028453,
      "grad_norm": 0.765632688999176,
      "learning_rate": 1.42969605226706e-08,
      "loss": 0.5857,
      "step": 11417
    },
    {
      "epoch": 1.9809160305343512,
      "grad_norm": 0.9371681809425354,
      "learning_rate": 1.4040548012956844e-08,
      "loss": 0.4243,
      "step": 11418
    },
    {
      "epoch": 1.981089521165857,
      "grad_norm": 0.8259038329124451,
      "learning_rate": 1.378645489712449e-08,
      "loss": 0.3293,
      "step": 11419
    },
    {
      "epoch": 1.9812630117973629,
      "grad_norm": 1.594635009765625,
      "learning_rate": 1.3534681204665502e-08,
      "loss": 0.5088,
      "step": 11420
    },
    {
      "epoch": 1.9814365024288687,
      "grad_norm": 0.6444791555404663,
      "learning_rate": 1.3285226964794284e-08,
      "loss": 0.5004,
      "step": 11421
    },
    {
      "epoch": 1.9816099930603748,
      "grad_norm": 1.0436303615570068,
      "learning_rate": 1.3038092206461017e-08,
      "loss": 0.3192,
      "step": 11422
    },
    {
      "epoch": 1.9817834836918806,
      "grad_norm": 0.865827739238739,
      "learning_rate": 1.2793276958347201e-08,
      "loss": 0.4277,
      "step": 11423
    },
    {
      "epoch": 1.9819569743233867,
      "grad_norm": 1.2612078189849854,
      "learning_rate": 1.2550781248863442e-08,
      "loss": 0.3894,
      "step": 11424
    },
    {
      "epoch": 1.9821304649548925,
      "grad_norm": 0.7467960715293884,
      "learning_rate": 1.2310605106149454e-08,
      "loss": 0.4304,
      "step": 11425
    },
    {
      "epoch": 1.9823039555863984,
      "grad_norm": 1.9402074813842773,
      "learning_rate": 1.2072748558082936e-08,
      "loss": 0.4438,
      "step": 11426
    },
    {
      "epoch": 1.9824774462179042,
      "grad_norm": 0.782226026058197,
      "learning_rate": 1.1837211632264034e-08,
      "loss": 0.363,
      "step": 11427
    },
    {
      "epoch": 1.98265093684941,
      "grad_norm": 0.7636884450912476,
      "learning_rate": 1.1603994356026437e-08,
      "loss": 0.433,
      "step": 11428
    },
    {
      "epoch": 1.982824427480916,
      "grad_norm": 0.9060726165771484,
      "learning_rate": 1.1373096756437385e-08,
      "loss": 0.3583,
      "step": 11429
    },
    {
      "epoch": 1.9829979181124218,
      "grad_norm": 1.1688307523727417,
      "learning_rate": 1.1144518860290998e-08,
      "loss": 0.3875,
      "step": 11430
    },
    {
      "epoch": 1.9831714087439278,
      "grad_norm": 1.0081478357315063,
      "learning_rate": 1.0918260694114947e-08,
      "loss": 0.3859,
      "step": 11431
    },
    {
      "epoch": 1.9833448993754337,
      "grad_norm": 0.7923986315727234,
      "learning_rate": 1.0694322284166003e-08,
      "loss": 0.4458,
      "step": 11432
    },
    {
      "epoch": 1.9835183900069397,
      "grad_norm": 1.339667797088623,
      "learning_rate": 1.047270365643449e-08,
      "loss": 0.4237,
      "step": 11433
    },
    {
      "epoch": 1.9836918806384456,
      "grad_norm": 0.7551453709602356,
      "learning_rate": 1.0253404836637615e-08,
      "loss": 0.4138,
      "step": 11434
    },
    {
      "epoch": 1.9838653712699514,
      "grad_norm": 0.7832826375961304,
      "learning_rate": 1.0036425850226129e-08,
      "loss": 0.5472,
      "step": 11435
    },
    {
      "epoch": 1.9840388619014573,
      "grad_norm": 0.7386212944984436,
      "learning_rate": 9.821766722379888e-09,
      "loss": 0.5157,
      "step": 11436
    },
    {
      "epoch": 1.9842123525329631,
      "grad_norm": 1.4132112264633179,
      "learning_rate": 9.609427478010081e-09,
      "loss": 0.4332,
      "step": 11437
    },
    {
      "epoch": 1.984385843164469,
      "grad_norm": 0.6954006552696228,
      "learning_rate": 9.399408141761434e-09,
      "loss": 0.4255,
      "step": 11438
    },
    {
      "epoch": 1.984559333795975,
      "grad_norm": 0.8655229210853577,
      "learning_rate": 9.191708738003346e-09,
      "loss": 0.4097,
      "step": 11439
    },
    {
      "epoch": 1.984732824427481,
      "grad_norm": 1.0122206211090088,
      "learning_rate": 8.9863292908432e-09,
      "loss": 0.3265,
      "step": 11440
    },
    {
      "epoch": 1.984906315058987,
      "grad_norm": 0.6448938250541687,
      "learning_rate": 8.78326982411304e-09,
      "loss": 0.5307,
      "step": 11441
    },
    {
      "epoch": 1.9850798056904928,
      "grad_norm": 1.0658587217330933,
      "learning_rate": 8.58253036137846e-09,
      "loss": 0.4203,
      "step": 11442
    },
    {
      "epoch": 1.9852532963219987,
      "grad_norm": 0.8777505159378052,
      "learning_rate": 8.384110925936384e-09,
      "loss": 0.3995,
      "step": 11443
    },
    {
      "epoch": 1.9854267869535045,
      "grad_norm": 0.9446192979812622,
      "learning_rate": 8.188011540812834e-09,
      "loss": 0.4147,
      "step": 11444
    },
    {
      "epoch": 1.9856002775850103,
      "grad_norm": 0.7639160752296448,
      "learning_rate": 7.994232228765164e-09,
      "loss": 0.4045,
      "step": 11445
    },
    {
      "epoch": 1.9857737682165162,
      "grad_norm": 0.81900554895401,
      "learning_rate": 7.80277301228205e-09,
      "loss": 0.4487,
      "step": 11446
    },
    {
      "epoch": 1.9859472588480223,
      "grad_norm": 0.8122382760047913,
      "learning_rate": 7.613633913583495e-09,
      "loss": 0.5093,
      "step": 11447
    },
    {
      "epoch": 1.986120749479528,
      "grad_norm": 0.8263910412788391,
      "learning_rate": 7.426814954618611e-09,
      "loss": 0.5033,
      "step": 11448
    },
    {
      "epoch": 1.9862942401110342,
      "grad_norm": 0.7352614998817444,
      "learning_rate": 7.24231615706561e-09,
      "loss": 0.5255,
      "step": 11449
    },
    {
      "epoch": 1.98646773074254,
      "grad_norm": 0.7796475887298584,
      "learning_rate": 7.060137542340695e-09,
      "loss": 0.4988,
      "step": 11450
    },
    {
      "epoch": 1.9866412213740459,
      "grad_norm": 0.816645085811615,
      "learning_rate": 6.8802791315825125e-09,
      "loss": 0.5663,
      "step": 11451
    },
    {
      "epoch": 1.9868147120055517,
      "grad_norm": 3.847252607345581,
      "learning_rate": 6.702740945663256e-09,
      "loss": 0.6187,
      "step": 11452
    },
    {
      "epoch": 1.9869882026370576,
      "grad_norm": 0.8500548601150513,
      "learning_rate": 6.527523005188663e-09,
      "loss": 0.5137,
      "step": 11453
    },
    {
      "epoch": 1.9871616932685634,
      "grad_norm": 0.9805374145507812,
      "learning_rate": 6.35462533049358e-09,
      "loss": 0.4279,
      "step": 11454
    },
    {
      "epoch": 1.9873351839000692,
      "grad_norm": 0.7661542892456055,
      "learning_rate": 6.184047941639737e-09,
      "loss": 0.5093,
      "step": 11455
    },
    {
      "epoch": 1.9875086745315753,
      "grad_norm": 0.9448789954185486,
      "learning_rate": 6.0157908584246305e-09,
      "loss": 0.3073,
      "step": 11456
    },
    {
      "epoch": 1.9876821651630812,
      "grad_norm": 1.0407168865203857,
      "learning_rate": 5.8498541003748634e-09,
      "loss": 0.4161,
      "step": 11457
    },
    {
      "epoch": 1.9878556557945872,
      "grad_norm": 0.884189784526825,
      "learning_rate": 5.686237686746143e-09,
      "loss": 0.4435,
      "step": 11458
    },
    {
      "epoch": 1.988029146426093,
      "grad_norm": 1.0521981716156006,
      "learning_rate": 5.5249416365299455e-09,
      "loss": 0.3564,
      "step": 11459
    },
    {
      "epoch": 1.988202637057599,
      "grad_norm": 1.0453505516052246,
      "learning_rate": 5.36596596844019e-09,
      "loss": 0.3783,
      "step": 11460
    },
    {
      "epoch": 1.9883761276891048,
      "grad_norm": 0.8650076389312744,
      "learning_rate": 5.209310700931003e-09,
      "loss": 0.4122,
      "step": 11461
    },
    {
      "epoch": 1.9885496183206106,
      "grad_norm": 1.120578408241272,
      "learning_rate": 5.054975852176736e-09,
      "loss": 0.3675,
      "step": 11462
    },
    {
      "epoch": 1.9887231089521165,
      "grad_norm": 0.8298072218894958,
      "learning_rate": 4.9029614400941675e-09,
      "loss": 0.5121,
      "step": 11463
    },
    {
      "epoch": 1.9888965995836225,
      "grad_norm": 0.7727584838867188,
      "learning_rate": 4.7532674823203e-09,
      "loss": 0.4401,
      "step": 11464
    },
    {
      "epoch": 1.9890700902151284,
      "grad_norm": 0.7380238175392151,
      "learning_rate": 4.605893996227906e-09,
      "loss": 0.3819,
      "step": 11465
    },
    {
      "epoch": 1.9892435808466344,
      "grad_norm": 0.8323608040809631,
      "learning_rate": 4.4608409989232995e-09,
      "loss": 0.4757,
      "step": 11466
    },
    {
      "epoch": 1.9894170714781403,
      "grad_norm": 0.8822579979896545,
      "learning_rate": 4.318108507235241e-09,
      "loss": 0.5186,
      "step": 11467
    },
    {
      "epoch": 1.9895905621096461,
      "grad_norm": 0.8925347924232483,
      "learning_rate": 4.17769653773048e-09,
      "loss": 0.3621,
      "step": 11468
    },
    {
      "epoch": 1.989764052741152,
      "grad_norm": 1.0496636629104614,
      "learning_rate": 4.039605106704869e-09,
      "loss": 0.4544,
      "step": 11469
    },
    {
      "epoch": 1.9899375433726578,
      "grad_norm": 0.8532253503799438,
      "learning_rate": 3.903834230183368e-09,
      "loss": 0.3895,
      "step": 11470
    },
    {
      "epoch": 1.9901110340041637,
      "grad_norm": 0.8864930868148804,
      "learning_rate": 3.77038392392004e-09,
      "loss": 0.4286,
      "step": 11471
    },
    {
      "epoch": 1.9902845246356695,
      "grad_norm": 0.8147814273834229,
      "learning_rate": 3.639254203406939e-09,
      "loss": 0.3987,
      "step": 11472
    },
    {
      "epoch": 1.9904580152671756,
      "grad_norm": 0.9215081930160522,
      "learning_rate": 3.5104450838563396e-09,
      "loss": 0.4346,
      "step": 11473
    },
    {
      "epoch": 1.9906315058986814,
      "grad_norm": 0.771024763584137,
      "learning_rate": 3.383956580218506e-09,
      "loss": 0.4662,
      "step": 11474
    },
    {
      "epoch": 1.9908049965301875,
      "grad_norm": 0.9461677074432373,
      "learning_rate": 3.2597887071750266e-09,
      "loss": 0.4031,
      "step": 11475
    },
    {
      "epoch": 1.9909784871616933,
      "grad_norm": 0.8339713215827942,
      "learning_rate": 3.1379414791343766e-09,
      "loss": 0.428,
      "step": 11476
    },
    {
      "epoch": 1.9911519777931992,
      "grad_norm": 1.1781772375106812,
      "learning_rate": 3.0184149102341354e-09,
      "loss": 0.4286,
      "step": 11477
    },
    {
      "epoch": 1.991325468424705,
      "grad_norm": 0.877213180065155,
      "learning_rate": 2.9012090143498704e-09,
      "loss": 0.4603,
      "step": 11478
    },
    {
      "epoch": 1.9914989590562109,
      "grad_norm": 0.7394884824752808,
      "learning_rate": 2.786323805081814e-09,
      "loss": 0.6095,
      "step": 11479
    },
    {
      "epoch": 1.9916724496877167,
      "grad_norm": 1.008466362953186,
      "learning_rate": 2.6737592957615243e-09,
      "loss": 0.5907,
      "step": 11480
    },
    {
      "epoch": 1.9918459403192228,
      "grad_norm": 0.7919140458106995,
      "learning_rate": 2.563515499451885e-09,
      "loss": 0.3691,
      "step": 11481
    },
    {
      "epoch": 1.9920194309507286,
      "grad_norm": 0.7679170966148376,
      "learning_rate": 2.4555924289493272e-09,
      "loss": 0.5347,
      "step": 11482
    },
    {
      "epoch": 1.9921929215822347,
      "grad_norm": 0.7232471704483032,
      "learning_rate": 2.349990096777166e-09,
      "loss": 0.489,
      "step": 11483
    },
    {
      "epoch": 1.9923664122137406,
      "grad_norm": 3.296504497528076,
      "learning_rate": 2.2467085151900436e-09,
      "loss": 0.4176,
      "step": 11484
    },
    {
      "epoch": 1.9925399028452464,
      "grad_norm": 1.5242438316345215,
      "learning_rate": 2.145747696173928e-09,
      "loss": 0.3864,
      "step": 11485
    },
    {
      "epoch": 1.9927133934767522,
      "grad_norm": 0.8411022424697876,
      "learning_rate": 2.047107651446112e-09,
      "loss": 0.386,
      "step": 11486
    },
    {
      "epoch": 1.992886884108258,
      "grad_norm": 0.8437713980674744,
      "learning_rate": 1.950788392455216e-09,
      "loss": 0.4785,
      "step": 11487
    },
    {
      "epoch": 1.993060374739764,
      "grad_norm": 0.7413724064826965,
      "learning_rate": 1.8567899303767457e-09,
      "loss": 0.5667,
      "step": 11488
    },
    {
      "epoch": 1.9932338653712698,
      "grad_norm": 0.7066528797149658,
      "learning_rate": 1.7651122761197515e-09,
      "loss": 0.4302,
      "step": 11489
    },
    {
      "epoch": 1.9934073560027759,
      "grad_norm": 0.7304266691207886,
      "learning_rate": 1.6757554403223907e-09,
      "loss": 0.4046,
      "step": 11490
    },
    {
      "epoch": 1.9935808466342817,
      "grad_norm": 0.9667731523513794,
      "learning_rate": 1.5887194333585876e-09,
      "loss": 0.4823,
      "step": 11491
    },
    {
      "epoch": 1.9937543372657878,
      "grad_norm": 1.013989806175232,
      "learning_rate": 1.5040042653269304e-09,
      "loss": 0.4267,
      "step": 11492
    },
    {
      "epoch": 1.9939278278972936,
      "grad_norm": 0.7241480946540833,
      "learning_rate": 1.421609946057334e-09,
      "loss": 0.4301,
      "step": 11493
    },
    {
      "epoch": 1.9941013185287995,
      "grad_norm": 0.9055463075637817,
      "learning_rate": 1.3415364851132595e-09,
      "loss": 0.5103,
      "step": 11494
    },
    {
      "epoch": 1.9942748091603053,
      "grad_norm": 1.3344109058380127,
      "learning_rate": 1.2637838917872735e-09,
      "loss": 0.5143,
      "step": 11495
    },
    {
      "epoch": 1.9944482997918112,
      "grad_norm": 0.8314899802207947,
      "learning_rate": 1.188352175103269e-09,
      "loss": 0.4003,
      "step": 11496
    },
    {
      "epoch": 1.994621790423317,
      "grad_norm": 0.9117047786712646,
      "learning_rate": 1.1152413438120235e-09,
      "loss": 0.3704,
      "step": 11497
    },
    {
      "epoch": 1.994795281054823,
      "grad_norm": 1.4010546207427979,
      "learning_rate": 1.0444514064023025e-09,
      "loss": 0.4529,
      "step": 11498
    },
    {
      "epoch": 1.994968771686329,
      "grad_norm": 0.9583075642585754,
      "learning_rate": 9.759823710853155e-10,
      "loss": 0.3901,
      "step": 11499
    },
    {
      "epoch": 1.995142262317835,
      "grad_norm": 0.8378259539604187,
      "learning_rate": 9.098342458102593e-10,
      "loss": 0.3929,
      "step": 11500
    },
    {
      "epoch": 1.9953157529493408,
      "grad_norm": 0.7189657688140869,
      "learning_rate": 8.460070382532159e-10,
      "loss": 0.4873,
      "step": 11501
    },
    {
      "epoch": 1.9954892435808467,
      "grad_norm": 0.7251210808753967,
      "learning_rate": 7.845007558193729e-10,
      "loss": 0.3544,
      "step": 11502
    },
    {
      "epoch": 1.9956627342123525,
      "grad_norm": 0.9614545702934265,
      "learning_rate": 7.253154056474643e-10,
      "loss": 0.401,
      "step": 11503
    },
    {
      "epoch": 1.9958362248438584,
      "grad_norm": 0.6946287155151367,
      "learning_rate": 6.684509946075501e-10,
      "loss": 0.4194,
      "step": 11504
    },
    {
      "epoch": 1.9960097154753642,
      "grad_norm": 0.8468782305717468,
      "learning_rate": 6.139075292987962e-10,
      "loss": 0.4404,
      "step": 11505
    },
    {
      "epoch": 1.9961832061068703,
      "grad_norm": 0.777607262134552,
      "learning_rate": 5.616850160494736e-10,
      "loss": 0.5247,
      "step": 11506
    },
    {
      "epoch": 1.9963566967383761,
      "grad_norm": 0.7831170558929443,
      "learning_rate": 5.117834609191796e-10,
      "loss": 0.4486,
      "step": 11507
    },
    {
      "epoch": 1.9965301873698822,
      "grad_norm": 0.990483283996582,
      "learning_rate": 4.642028697010581e-10,
      "loss": 0.4615,
      "step": 11508
    },
    {
      "epoch": 1.996703678001388,
      "grad_norm": 0.735356330871582,
      "learning_rate": 4.189432479173583e-10,
      "loss": 0.4634,
      "step": 11509
    },
    {
      "epoch": 1.9968771686328939,
      "grad_norm": 0.8335587978363037,
      "learning_rate": 3.760046008172147e-10,
      "loss": 0.4609,
      "step": 11510
    },
    {
      "epoch": 1.9970506592643997,
      "grad_norm": 8.307694435119629,
      "learning_rate": 3.353869333877491e-10,
      "loss": 0.4355,
      "step": 11511
    },
    {
      "epoch": 1.9972241498959056,
      "grad_norm": 0.9904146194458008,
      "learning_rate": 2.97090250340748e-10,
      "loss": 0.4745,
      "step": 11512
    },
    {
      "epoch": 1.9973976405274114,
      "grad_norm": 1.9969080686569214,
      "learning_rate": 2.611145561215445e-10,
      "loss": 0.4222,
      "step": 11513
    },
    {
      "epoch": 1.9975711311589173,
      "grad_norm": 0.8702035546302795,
      "learning_rate": 2.2745985490235656e-10,
      "loss": 0.3834,
      "step": 11514
    },
    {
      "epoch": 1.9977446217904233,
      "grad_norm": 1.0072624683380127,
      "learning_rate": 1.9612615059116935e-10,
      "loss": 0.3547,
      "step": 11515
    },
    {
      "epoch": 1.9979181124219292,
      "grad_norm": 0.9296289086341858,
      "learning_rate": 1.6711344682507346e-10,
      "loss": 0.3597,
      "step": 11516
    },
    {
      "epoch": 1.9980916030534353,
      "grad_norm": 1.0401052236557007,
      "learning_rate": 1.4042174696804467e-10,
      "loss": 0.3784,
      "step": 11517
    },
    {
      "epoch": 1.998265093684941,
      "grad_norm": 0.9357911348342896,
      "learning_rate": 1.160510541220461e-10,
      "loss": 0.4485,
      "step": 11518
    },
    {
      "epoch": 1.998438584316447,
      "grad_norm": 0.9865544438362122,
      "learning_rate": 9.400137111148511e-11,
      "loss": 0.4338,
      "step": 11519
    },
    {
      "epoch": 1.9986120749479528,
      "grad_norm": 1.2822304964065552,
      "learning_rate": 7.427270049653601e-11,
      "loss": 0.411,
      "step": 11520
    },
    {
      "epoch": 1.9987855655794586,
      "grad_norm": 0.9479796886444092,
      "learning_rate": 5.6865044566478676e-11,
      "loss": 0.3289,
      "step": 11521
    },
    {
      "epoch": 1.9989590562109645,
      "grad_norm": 0.7394636273384094,
      "learning_rate": 4.1778405341919016e-11,
      "loss": 0.4054,
      "step": 11522
    },
    {
      "epoch": 1.9991325468424705,
      "grad_norm": 0.8287288546562195,
      "learning_rate": 2.9012784574788954e-11,
      "loss": 0.5789,
      "step": 11523
    },
    {
      "epoch": 1.9993060374739764,
      "grad_norm": 0.9048954248428345,
      "learning_rate": 1.8568183743905566e-11,
      "loss": 0.4268,
      "step": 11524
    },
    {
      "epoch": 1.9994795281054825,
      "grad_norm": 0.8199928998947144,
      "learning_rate": 1.0444604063852837e-11,
      "loss": 0.432,
      "step": 11525
    },
    {
      "epoch": 1.9996530187369883,
      "grad_norm": 0.8431583046913147,
      "learning_rate": 4.6420464738794465e-12,
      "loss": 0.499,
      "step": 11526
    },
    {
      "epoch": 1.9998265093684942,
      "grad_norm": 0.8587999939918518,
      "learning_rate": 1.160511651221441e-12,
      "loss": 0.5066,
      "step": 11527
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8854231834411621,
      "learning_rate": 0.0,
      "loss": 0.3571,
      "step": 11528
    },
    {
      "epoch": 2.0,
      "step": 11528,
      "total_flos": 0.0,
      "train_loss": 0.0,
      "train_runtime": 1.8833,
      "train_samples_per_second": 195885.996,
      "train_steps_per_second": 6121.238
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 11528,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}