{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.985401459854015,
  "eval_steps": 500,
  "global_step": 912,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004379562043795621,
      "grad_norm": 34.64235305786133,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 2.6583,
      "step": 1
    },
    {
      "epoch": 0.008759124087591242,
      "grad_norm": 33.89678192138672,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 2.5074,
      "step": 2
    },
    {
      "epoch": 0.013138686131386862,
      "grad_norm": 35.2148551940918,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 2.7094,
      "step": 3
    },
    {
      "epoch": 0.017518248175182483,
      "grad_norm": 35.11457061767578,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.7266,
      "step": 4
    },
    {
      "epoch": 0.021897810218978103,
      "grad_norm": 35.70753479003906,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 2.7442,
      "step": 5
    },
    {
      "epoch": 0.026277372262773723,
      "grad_norm": 34.34943771362305,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 2.5578,
      "step": 6
    },
    {
      "epoch": 0.030656934306569343,
      "grad_norm": 34.31540298461914,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 2.5893,
      "step": 7
    },
    {
      "epoch": 0.035036496350364967,
      "grad_norm": 32.545223236083984,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.5039,
      "step": 8
    },
    {
      "epoch": 0.03941605839416058,
      "grad_norm": 35.70431137084961,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 2.6719,
      "step": 9
    },
    {
      "epoch": 0.043795620437956206,
      "grad_norm": 34.14265441894531,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.5764,
      "step": 10
    },
    {
      "epoch": 0.04817518248175182,
      "grad_norm": 32.08097839355469,
      "learning_rate": 5.5e-07,
      "loss": 2.4564,
      "step": 11
    },
    {
      "epoch": 0.052554744525547446,
      "grad_norm": 32.66060256958008,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.458,
      "step": 12
    },
    {
      "epoch": 0.05693430656934306,
      "grad_norm": 33.21636962890625,
      "learning_rate": 6.5e-07,
      "loss": 2.4835,
      "step": 13
    },
    {
      "epoch": 0.061313868613138686,
      "grad_norm": 33.92257308959961,
      "learning_rate": 7.000000000000001e-07,
      "loss": 2.4288,
      "step": 14
    },
    {
      "epoch": 0.06569343065693431,
      "grad_norm": 32.19805145263672,
      "learning_rate": 7.5e-07,
      "loss": 2.2411,
      "step": 15
    },
    {
      "epoch": 0.07007299270072993,
      "grad_norm": 32.355220794677734,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.1597,
      "step": 16
    },
    {
      "epoch": 0.07445255474452554,
      "grad_norm": 33.08480453491211,
      "learning_rate": 8.500000000000001e-07,
      "loss": 2.1377,
      "step": 17
    },
    {
      "epoch": 0.07883211678832117,
      "grad_norm": 33.459957122802734,
      "learning_rate": 9.000000000000001e-07,
      "loss": 2.0306,
      "step": 18
    },
    {
      "epoch": 0.08321167883211679,
      "grad_norm": 32.897315979003906,
      "learning_rate": 9.500000000000001e-07,
      "loss": 1.8697,
      "step": 19
    },
    {
      "epoch": 0.08759124087591241,
      "grad_norm": 33.81785202026367,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.8147,
      "step": 20
    },
    {
      "epoch": 0.09197080291970802,
      "grad_norm": 32.52595520019531,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 1.6526,
      "step": 21
    },
    {
      "epoch": 0.09635036496350365,
      "grad_norm": 34.09442138671875,
      "learning_rate": 1.1e-06,
      "loss": 1.6127,
      "step": 22
    },
    {
      "epoch": 0.10072992700729927,
      "grad_norm": 30.89822769165039,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 1.3872,
      "step": 23
    },
    {
      "epoch": 0.10510948905109489,
      "grad_norm": 29.566524505615234,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.2755,
      "step": 24
    },
    {
      "epoch": 0.10948905109489052,
      "grad_norm": 28.26628875732422,
      "learning_rate": 1.25e-06,
      "loss": 1.1409,
      "step": 25
    },
    {
      "epoch": 0.11386861313868613,
      "grad_norm": 30.7103328704834,
      "learning_rate": 1.3e-06,
      "loss": 0.966,
      "step": 26
    },
    {
      "epoch": 0.11824817518248175,
      "grad_norm": 28.975385665893555,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 0.7579,
      "step": 27
    },
    {
      "epoch": 0.12262773722627737,
      "grad_norm": 26.821529388427734,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.6013,
      "step": 28
    },
    {
      "epoch": 0.12700729927007298,
      "grad_norm": 23.804439544677734,
      "learning_rate": 1.45e-06,
      "loss": 0.4978,
      "step": 29
    },
    {
      "epoch": 0.13138686131386862,
      "grad_norm": 21.404451370239258,
      "learning_rate": 1.5e-06,
      "loss": 0.3926,
      "step": 30
    },
    {
      "epoch": 0.13576642335766423,
      "grad_norm": 17.63161849975586,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 0.2568,
      "step": 31
    },
    {
      "epoch": 0.14014598540145987,
      "grad_norm": 10.998854637145996,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.2373,
      "step": 32
    },
    {
      "epoch": 0.14452554744525548,
      "grad_norm": 6.9544997215271,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 0.1689,
      "step": 33
    },
    {
      "epoch": 0.14890510948905109,
      "grad_norm": 5.1013102531433105,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.1471,
      "step": 34
    },
    {
      "epoch": 0.15328467153284672,
      "grad_norm": 4.501709461212158,
      "learning_rate": 1.75e-06,
      "loss": 0.132,
      "step": 35
    },
    {
      "epoch": 0.15766423357664233,
      "grad_norm": 3.198529005050659,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.1065,
      "step": 36
    },
    {
      "epoch": 0.16204379562043797,
      "grad_norm": 3.2325005531311035,
      "learning_rate": 1.85e-06,
      "loss": 0.0907,
      "step": 37
    },
    {
      "epoch": 0.16642335766423358,
      "grad_norm": 1.5125375986099243,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.0782,
      "step": 38
    },
    {
      "epoch": 0.1708029197080292,
      "grad_norm": 1.9160635471343994,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.0852,
      "step": 39
    },
    {
      "epoch": 0.17518248175182483,
      "grad_norm": 1.6062333583831787,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.074,
      "step": 40
    },
    {
      "epoch": 0.17956204379562044,
      "grad_norm": 1.5675855875015259,
      "learning_rate": 2.05e-06,
      "loss": 0.0704,
      "step": 41
    },
    {
      "epoch": 0.18394160583941604,
      "grad_norm": 1.440182089805603,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.0674,
      "step": 42
    },
    {
      "epoch": 0.18832116788321168,
      "grad_norm": 1.1466726064682007,
      "learning_rate": 2.15e-06,
      "loss": 0.0702,
      "step": 43
    },
    {
      "epoch": 0.1927007299270073,
      "grad_norm": 1.2195515632629395,
      "learning_rate": 2.2e-06,
      "loss": 0.0723,
      "step": 44
    },
    {
      "epoch": 0.19708029197080293,
      "grad_norm": 1.743561029434204,
      "learning_rate": 2.25e-06,
      "loss": 0.0875,
      "step": 45
    },
    {
      "epoch": 0.20145985401459854,
      "grad_norm": 0.9764343500137329,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.062,
      "step": 46
    },
    {
      "epoch": 0.20583941605839415,
      "grad_norm": 0.8891277313232422,
      "learning_rate": 2.35e-06,
      "loss": 0.0576,
      "step": 47
    },
    {
      "epoch": 0.21021897810218979,
      "grad_norm": 0.9648666977882385,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.0656,
      "step": 48
    },
    {
      "epoch": 0.2145985401459854,
      "grad_norm": 0.784566342830658,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.0548,
      "step": 49
    },
    {
      "epoch": 0.21897810218978103,
      "grad_norm": 0.9402966499328613,
      "learning_rate": 2.5e-06,
      "loss": 0.0626,
      "step": 50
    },
    {
      "epoch": 0.22335766423357664,
      "grad_norm": 1.3284685611724854,
      "learning_rate": 2.55e-06,
      "loss": 0.0632,
      "step": 51
    },
    {
      "epoch": 0.22773722627737225,
      "grad_norm": 1.0913968086242676,
      "learning_rate": 2.6e-06,
      "loss": 0.0675,
      "step": 52
    },
    {
      "epoch": 0.2321167883211679,
      "grad_norm": 1.1069140434265137,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.0541,
      "step": 53
    },
    {
      "epoch": 0.2364963503649635,
      "grad_norm": 0.8529757857322693,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.0657,
      "step": 54
    },
    {
      "epoch": 0.24087591240875914,
      "grad_norm": 0.7182446718215942,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0607,
      "step": 55
    },
    {
      "epoch": 0.24525547445255474,
      "grad_norm": 1.0538653135299683,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.0556,
      "step": 56
    },
    {
      "epoch": 0.24963503649635035,
      "grad_norm": 1.2083594799041748,
      "learning_rate": 2.85e-06,
      "loss": 0.0532,
      "step": 57
    },
    {
      "epoch": 0.25401459854014596,
      "grad_norm": 0.8183572888374329,
      "learning_rate": 2.9e-06,
      "loss": 0.0529,
      "step": 58
    },
    {
      "epoch": 0.2583941605839416,
      "grad_norm": 0.9014842510223389,
      "learning_rate": 2.95e-06,
      "loss": 0.0601,
      "step": 59
    },
    {
      "epoch": 0.26277372262773724,
      "grad_norm": 0.9017247557640076,
      "learning_rate": 3e-06,
      "loss": 0.0584,
      "step": 60
    },
    {
      "epoch": 0.2671532846715328,
      "grad_norm": 1.1078683137893677,
      "learning_rate": 3.05e-06,
      "loss": 0.0635,
      "step": 61
    },
    {
      "epoch": 0.27153284671532846,
      "grad_norm": 1.174526572227478,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.0523,
      "step": 62
    },
    {
      "epoch": 0.2759124087591241,
      "grad_norm": 0.9296770095825195,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.0588,
      "step": 63
    },
    {
      "epoch": 0.28029197080291973,
      "grad_norm": 0.8549372553825378,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.0639,
      "step": 64
    },
    {
      "epoch": 0.2846715328467153,
      "grad_norm": 0.8956279158592224,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.059,
      "step": 65
    },
    {
      "epoch": 0.28905109489051095,
      "grad_norm": 0.7937710285186768,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.0579,
      "step": 66
    },
    {
      "epoch": 0.2934306569343066,
      "grad_norm": 0.7786620855331421,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.0586,
      "step": 67
    },
    {
      "epoch": 0.29781021897810217,
      "grad_norm": 0.7562637329101562,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.046,
      "step": 68
    },
    {
      "epoch": 0.3021897810218978,
      "grad_norm": 0.8958250880241394,
      "learning_rate": 3.45e-06,
      "loss": 0.0566,
      "step": 69
    },
    {
      "epoch": 0.30656934306569344,
      "grad_norm": 0.9434528946876526,
      "learning_rate": 3.5e-06,
      "loss": 0.0548,
      "step": 70
    },
    {
      "epoch": 0.310948905109489,
      "grad_norm": 1.0564453601837158,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.0529,
      "step": 71
    },
    {
      "epoch": 0.31532846715328466,
      "grad_norm": 0.896443247795105,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.0517,
      "step": 72
    },
    {
      "epoch": 0.3197080291970803,
      "grad_norm": 1.1364223957061768,
      "learning_rate": 3.65e-06,
      "loss": 0.0489,
      "step": 73
    },
    {
      "epoch": 0.32408759124087594,
      "grad_norm": 1.1319010257720947,
      "learning_rate": 3.7e-06,
      "loss": 0.0548,
      "step": 74
    },
    {
      "epoch": 0.3284671532846715,
      "grad_norm": 0.9694503545761108,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0525,
      "step": 75
    },
    {
      "epoch": 0.33284671532846716,
      "grad_norm": 0.8128111958503723,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.0566,
      "step": 76
    },
    {
      "epoch": 0.3372262773722628,
      "grad_norm": 0.9068273901939392,
      "learning_rate": 3.85e-06,
      "loss": 0.0475,
      "step": 77
    },
    {
      "epoch": 0.3416058394160584,
      "grad_norm": 0.9689438343048096,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.048,
      "step": 78
    },
    {
      "epoch": 0.345985401459854,
      "grad_norm": 0.940131664276123,
      "learning_rate": 3.95e-06,
      "loss": 0.0567,
      "step": 79
    },
    {
      "epoch": 0.35036496350364965,
      "grad_norm": 0.8836082220077515,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0542,
      "step": 80
    },
    {
      "epoch": 0.35474452554744523,
      "grad_norm": 0.9325949549674988,
      "learning_rate": 4.05e-06,
      "loss": 0.0551,
      "step": 81
    },
    {
      "epoch": 0.35912408759124087,
      "grad_norm": 0.8954764008522034,
      "learning_rate": 4.1e-06,
      "loss": 0.0517,
      "step": 82
    },
    {
      "epoch": 0.3635036496350365,
      "grad_norm": 0.6444959044456482,
      "learning_rate": 4.15e-06,
      "loss": 0.0434,
      "step": 83
    },
    {
      "epoch": 0.3678832116788321,
      "grad_norm": 0.9097581505775452,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.0471,
      "step": 84
    },
    {
      "epoch": 0.3722627737226277,
      "grad_norm": 0.849006712436676,
      "learning_rate": 4.25e-06,
      "loss": 0.0529,
      "step": 85
    },
    {
      "epoch": 0.37664233576642336,
      "grad_norm": 0.8611392378807068,
      "learning_rate": 4.3e-06,
      "loss": 0.0513,
      "step": 86
    },
    {
      "epoch": 0.381021897810219,
      "grad_norm": 0.7885357737541199,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.0523,
      "step": 87
    },
    {
      "epoch": 0.3854014598540146,
      "grad_norm": 0.7642116546630859,
      "learning_rate": 4.4e-06,
      "loss": 0.0407,
      "step": 88
    },
    {
      "epoch": 0.3897810218978102,
      "grad_norm": 0.8920945525169373,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.0485,
      "step": 89
    },
    {
      "epoch": 0.39416058394160586,
      "grad_norm": 0.9801046848297119,
      "learning_rate": 4.5e-06,
      "loss": 0.0404,
      "step": 90
    },
    {
      "epoch": 0.39854014598540144,
      "grad_norm": 1.0874953269958496,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.0588,
      "step": 91
    },
    {
      "epoch": 0.4029197080291971,
      "grad_norm": 0.9019029140472412,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.0466,
      "step": 92
    },
    {
      "epoch": 0.4072992700729927,
      "grad_norm": 0.7258988618850708,
      "learning_rate": 4.65e-06,
      "loss": 0.0493,
      "step": 93
    },
    {
      "epoch": 0.4116788321167883,
      "grad_norm": 1.103407859802246,
      "learning_rate": 4.7e-06,
      "loss": 0.0495,
      "step": 94
    },
    {
      "epoch": 0.41605839416058393,
      "grad_norm": 0.751805305480957,
      "learning_rate": 4.75e-06,
      "loss": 0.0484,
      "step": 95
    },
    {
      "epoch": 0.42043795620437957,
      "grad_norm": 0.7717764973640442,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.0447,
      "step": 96
    },
    {
      "epoch": 0.4248175182481752,
      "grad_norm": 0.7147190570831299,
      "learning_rate": 4.85e-06,
      "loss": 0.0523,
      "step": 97
    },
    {
      "epoch": 0.4291970802919708,
      "grad_norm": 0.9990110993385315,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.0454,
      "step": 98
    },
    {
      "epoch": 0.4335766423357664,
      "grad_norm": 0.7766187191009521,
      "learning_rate": 4.95e-06,
      "loss": 0.0472,
      "step": 99
    },
    {
      "epoch": 0.43795620437956206,
      "grad_norm": 0.7124347686767578,
      "learning_rate": 5e-06,
      "loss": 0.0473,
      "step": 100
    },
    {
      "epoch": 0.44233576642335765,
      "grad_norm": 0.9340270757675171,
      "learning_rate": 4.99999232689698e-06,
      "loss": 0.0499,
      "step": 101
    },
    {
      "epoch": 0.4467153284671533,
      "grad_norm": 0.7429985404014587,
      "learning_rate": 4.999969307635021e-06,
      "loss": 0.042,
      "step": 102
    },
    {
      "epoch": 0.4510948905109489,
      "grad_norm": 0.9131317138671875,
      "learning_rate": 4.999930942355425e-06,
      "loss": 0.0519,
      "step": 103
    },
    {
      "epoch": 0.4554744525547445,
      "grad_norm": 0.9970843195915222,
      "learning_rate": 4.999877231293698e-06,
      "loss": 0.0428,
      "step": 104
    },
    {
      "epoch": 0.45985401459854014,
      "grad_norm": 0.7625145316123962,
      "learning_rate": 4.999808174779543e-06,
      "loss": 0.0442,
      "step": 105
    },
    {
      "epoch": 0.4642335766423358,
      "grad_norm": 0.6059474945068359,
      "learning_rate": 4.999723773236865e-06,
      "loss": 0.0456,
      "step": 106
    },
    {
      "epoch": 0.4686131386861314,
      "grad_norm": 0.6798833608627319,
      "learning_rate": 4.999624027183758e-06,
      "loss": 0.0408,
      "step": 107
    },
    {
      "epoch": 0.472992700729927,
      "grad_norm": 1.0250803232192993,
      "learning_rate": 4.999508937232514e-06,
      "loss": 0.0471,
      "step": 108
    },
    {
      "epoch": 0.47737226277372263,
      "grad_norm": 0.8457198739051819,
      "learning_rate": 4.999378504089609e-06,
      "loss": 0.0425,
      "step": 109
    },
    {
      "epoch": 0.48175182481751827,
      "grad_norm": 0.9417868852615356,
      "learning_rate": 4.999232728555705e-06,
      "loss": 0.0388,
      "step": 110
    },
    {
      "epoch": 0.48613138686131385,
      "grad_norm": 0.8558921813964844,
      "learning_rate": 4.999071611525643e-06,
      "loss": 0.0423,
      "step": 111
    },
    {
      "epoch": 0.4905109489051095,
      "grad_norm": 0.7070104479789734,
      "learning_rate": 4.998895153988437e-06,
      "loss": 0.0354,
      "step": 112
    },
    {
      "epoch": 0.4948905109489051,
      "grad_norm": 0.8162719011306763,
      "learning_rate": 4.998703357027268e-06,
      "loss": 0.0465,
      "step": 113
    },
    {
      "epoch": 0.4992700729927007,
      "grad_norm": 0.9140358567237854,
      "learning_rate": 4.998496221819479e-06,
      "loss": 0.0457,
      "step": 114
    },
    {
      "epoch": 0.5036496350364964,
      "grad_norm": 0.6447531580924988,
      "learning_rate": 4.998273749636564e-06,
      "loss": 0.039,
      "step": 115
    },
    {
      "epoch": 0.5080291970802919,
      "grad_norm": 0.9157156944274902,
      "learning_rate": 4.998035941844167e-06,
      "loss": 0.0469,
      "step": 116
    },
    {
      "epoch": 0.5124087591240876,
      "grad_norm": 0.7706230878829956,
      "learning_rate": 4.997782799902065e-06,
      "loss": 0.0325,
      "step": 117
    },
    {
      "epoch": 0.5167883211678832,
      "grad_norm": 0.9391443729400635,
      "learning_rate": 4.997514325364168e-06,
      "loss": 0.0397,
      "step": 118
    },
    {
      "epoch": 0.5211678832116788,
      "grad_norm": 1.0085054636001587,
      "learning_rate": 4.997230519878499e-06,
      "loss": 0.0403,
      "step": 119
    },
    {
      "epoch": 0.5255474452554745,
      "grad_norm": 1.8318824768066406,
      "learning_rate": 4.996931385187195e-06,
      "loss": 0.0463,
      "step": 120
    },
    {
      "epoch": 0.5299270072992701,
      "grad_norm": 1.0216630697250366,
      "learning_rate": 4.9966169231264885e-06,
      "loss": 0.0406,
      "step": 121
    },
    {
      "epoch": 0.5343065693430656,
      "grad_norm": 1.4819082021713257,
      "learning_rate": 4.9962871356267e-06,
      "loss": 0.0485,
      "step": 122
    },
    {
      "epoch": 0.5386861313868613,
      "grad_norm": 0.9435060024261475,
      "learning_rate": 4.995942024712222e-06,
      "loss": 0.04,
      "step": 123
    },
    {
      "epoch": 0.5430656934306569,
      "grad_norm": 0.7887905240058899,
      "learning_rate": 4.995581592501514e-06,
      "loss": 0.0397,
      "step": 124
    },
    {
      "epoch": 0.5474452554744526,
      "grad_norm": 0.8321148753166199,
      "learning_rate": 4.995205841207082e-06,
      "loss": 0.0413,
      "step": 125
    },
    {
      "epoch": 0.5518248175182482,
      "grad_norm": 1.0303553342819214,
      "learning_rate": 4.99481477313547e-06,
      "loss": 0.0422,
      "step": 126
    },
    {
      "epoch": 0.5562043795620438,
      "grad_norm": 0.7056427001953125,
      "learning_rate": 4.994408390687241e-06,
      "loss": 0.0362,
      "step": 127
    },
    {
      "epoch": 0.5605839416058395,
      "grad_norm": 0.9762740135192871,
      "learning_rate": 4.993986696356966e-06,
      "loss": 0.0385,
      "step": 128
    },
    {
      "epoch": 0.564963503649635,
      "grad_norm": 0.9447624683380127,
      "learning_rate": 4.9935496927332095e-06,
      "loss": 0.0402,
      "step": 129
    },
    {
      "epoch": 0.5693430656934306,
      "grad_norm": 0.6106760501861572,
      "learning_rate": 4.993097382498511e-06,
      "loss": 0.0319,
      "step": 130
    },
    {
      "epoch": 0.5737226277372263,
      "grad_norm": 1.0554594993591309,
      "learning_rate": 4.992629768429367e-06,
      "loss": 0.0437,
      "step": 131
    },
    {
      "epoch": 0.5781021897810219,
      "grad_norm": 1.066218376159668,
      "learning_rate": 4.992146853396219e-06,
      "loss": 0.0382,
      "step": 132
    },
    {
      "epoch": 0.5824817518248175,
      "grad_norm": 0.7517623901367188,
      "learning_rate": 4.991648640363434e-06,
      "loss": 0.0317,
      "step": 133
    },
    {
      "epoch": 0.5868613138686132,
      "grad_norm": 0.8136976957321167,
      "learning_rate": 4.991135132389282e-06,
      "loss": 0.0339,
      "step": 134
    },
    {
      "epoch": 0.5912408759124088,
      "grad_norm": 0.9254240989685059,
      "learning_rate": 4.990606332625923e-06,
      "loss": 0.0413,
      "step": 135
    },
    {
      "epoch": 0.5956204379562043,
      "grad_norm": 0.6778447031974792,
      "learning_rate": 4.990062244319387e-06,
      "loss": 0.0377,
      "step": 136
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1036059856414795,
      "learning_rate": 4.989502870809547e-06,
      "loss": 0.0376,
      "step": 137
    },
    {
      "epoch": 0.6043795620437956,
      "grad_norm": 0.8054158091545105,
      "learning_rate": 4.988928215530111e-06,
      "loss": 0.0367,
      "step": 138
    },
    {
      "epoch": 0.6087591240875913,
      "grad_norm": 0.9227175116539001,
      "learning_rate": 4.988338282008588e-06,
      "loss": 0.0374,
      "step": 139
    },
    {
      "epoch": 0.6131386861313869,
      "grad_norm": 0.8502228260040283,
      "learning_rate": 4.9877330738662755e-06,
      "loss": 0.0384,
      "step": 140
    },
    {
      "epoch": 0.6175182481751825,
      "grad_norm": 0.684752881526947,
      "learning_rate": 4.987112594818232e-06,
      "loss": 0.0366,
      "step": 141
    },
    {
      "epoch": 0.621897810218978,
      "grad_norm": 0.7456391453742981,
      "learning_rate": 4.9864768486732585e-06,
      "loss": 0.037,
      "step": 142
    },
    {
      "epoch": 0.6262773722627737,
      "grad_norm": 0.6797431111335754,
      "learning_rate": 4.985825839333872e-06,
      "loss": 0.0325,
      "step": 143
    },
    {
      "epoch": 0.6306569343065693,
      "grad_norm": 0.8098205924034119,
      "learning_rate": 4.985159570796279e-06,
      "loss": 0.0343,
      "step": 144
    },
    {
      "epoch": 0.635036496350365,
      "grad_norm": 0.8089592456817627,
      "learning_rate": 4.984478047150361e-06,
      "loss": 0.026,
      "step": 145
    },
    {
      "epoch": 0.6394160583941606,
      "grad_norm": 0.9282512664794922,
      "learning_rate": 4.983781272579637e-06,
      "loss": 0.0334,
      "step": 146
    },
    {
      "epoch": 0.6437956204379562,
      "grad_norm": 0.802608072757721,
      "learning_rate": 4.9830692513612445e-06,
      "loss": 0.0259,
      "step": 147
    },
    {
      "epoch": 0.6481751824817519,
      "grad_norm": 1.3046361207962036,
      "learning_rate": 4.982341987865914e-06,
      "loss": 0.045,
      "step": 148
    },
    {
      "epoch": 0.6525547445255474,
      "grad_norm": 1.0812411308288574,
      "learning_rate": 4.9815994865579405e-06,
      "loss": 0.0329,
      "step": 149
    },
    {
      "epoch": 0.656934306569343,
      "grad_norm": 0.7856137156486511,
      "learning_rate": 4.980841751995155e-06,
      "loss": 0.0341,
      "step": 150
    },
    {
      "epoch": 0.6613138686131387,
      "grad_norm": 1.0517083406448364,
      "learning_rate": 4.980068788828897e-06,
      "loss": 0.0299,
      "step": 151
    },
    {
      "epoch": 0.6656934306569343,
      "grad_norm": 0.6148231029510498,
      "learning_rate": 4.979280601803988e-06,
      "loss": 0.0304,
      "step": 152
    },
    {
      "epoch": 0.67007299270073,
      "grad_norm": 0.7572031021118164,
      "learning_rate": 4.9784771957586995e-06,
      "loss": 0.0309,
      "step": 153
    },
    {
      "epoch": 0.6744525547445256,
      "grad_norm": 2.0948777198791504,
      "learning_rate": 4.977658575624727e-06,
      "loss": 0.0307,
      "step": 154
    },
    {
      "epoch": 0.6788321167883211,
      "grad_norm": 0.624940037727356,
      "learning_rate": 4.976824746427153e-06,
      "loss": 0.03,
      "step": 155
    },
    {
      "epoch": 0.6832116788321168,
      "grad_norm": 0.8346346616744995,
      "learning_rate": 4.975975713284426e-06,
      "loss": 0.036,
      "step": 156
    },
    {
      "epoch": 0.6875912408759124,
      "grad_norm": 0.742098867893219,
      "learning_rate": 4.975111481408319e-06,
      "loss": 0.0325,
      "step": 157
    },
    {
      "epoch": 0.691970802919708,
      "grad_norm": 0.8000304102897644,
      "learning_rate": 4.9742320561039055e-06,
      "loss": 0.0332,
      "step": 158
    },
    {
      "epoch": 0.6963503649635037,
      "grad_norm": 1.063854694366455,
      "learning_rate": 4.973337442769523e-06,
      "loss": 0.0366,
      "step": 159
    },
    {
      "epoch": 0.7007299270072993,
      "grad_norm": 0.965560257434845,
      "learning_rate": 4.972427646896738e-06,
      "loss": 0.0331,
      "step": 160
    },
    {
      "epoch": 0.7051094890510949,
      "grad_norm": 1.5070244073867798,
      "learning_rate": 4.971502674070317e-06,
      "loss": 0.0446,
      "step": 161
    },
    {
      "epoch": 0.7094890510948905,
      "grad_norm": 0.8810545206069946,
      "learning_rate": 4.970562529968189e-06,
      "loss": 0.0299,
      "step": 162
    },
    {
      "epoch": 0.7138686131386861,
      "grad_norm": 0.7683446407318115,
      "learning_rate": 4.969607220361414e-06,
      "loss": 0.0244,
      "step": 163
    },
    {
      "epoch": 0.7182481751824817,
      "grad_norm": 0.7444891929626465,
      "learning_rate": 4.968636751114141e-06,
      "loss": 0.0338,
      "step": 164
    },
    {
      "epoch": 0.7226277372262774,
      "grad_norm": 0.7077688574790955,
      "learning_rate": 4.96765112818358e-06,
      "loss": 0.0285,
      "step": 165
    },
    {
      "epoch": 0.727007299270073,
      "grad_norm": 0.5648500919342041,
      "learning_rate": 4.9666503576199574e-06,
      "loss": 0.026,
      "step": 166
    },
    {
      "epoch": 0.7313868613138687,
      "grad_norm": 0.763556718826294,
      "learning_rate": 4.965634445566489e-06,
      "loss": 0.0299,
      "step": 167
    },
    {
      "epoch": 0.7357664233576642,
      "grad_norm": 0.6892725825309753,
      "learning_rate": 4.9646033982593315e-06,
      "loss": 0.023,
      "step": 168
    },
    {
      "epoch": 0.7401459854014598,
      "grad_norm": 1.0332573652267456,
      "learning_rate": 4.963557222027551e-06,
      "loss": 0.0313,
      "step": 169
    },
    {
      "epoch": 0.7445255474452555,
      "grad_norm": 1.214428424835205,
      "learning_rate": 4.962495923293081e-06,
      "loss": 0.027,
      "step": 170
    },
    {
      "epoch": 0.7489051094890511,
      "grad_norm": 0.9823130965232849,
      "learning_rate": 4.961419508570686e-06,
      "loss": 0.0231,
      "step": 171
    },
    {
      "epoch": 0.7532846715328467,
      "grad_norm": 1.2535115480422974,
      "learning_rate": 4.960327984467919e-06,
      "loss": 0.0326,
      "step": 172
    },
    {
      "epoch": 0.7576642335766424,
      "grad_norm": 0.9383441209793091,
      "learning_rate": 4.959221357685081e-06,
      "loss": 0.0286,
      "step": 173
    },
    {
      "epoch": 0.762043795620438,
      "grad_norm": 1.0426976680755615,
      "learning_rate": 4.958099635015182e-06,
      "loss": 0.0298,
      "step": 174
    },
    {
      "epoch": 0.7664233576642335,
      "grad_norm": 0.9159742593765259,
      "learning_rate": 4.956962823343895e-06,
      "loss": 0.025,
      "step": 175
    },
    {
      "epoch": 0.7708029197080292,
      "grad_norm": 0.8746912479400635,
      "learning_rate": 4.95581092964952e-06,
      "loss": 0.0299,
      "step": 176
    },
    {
      "epoch": 0.7751824817518248,
      "grad_norm": 0.9875199198722839,
      "learning_rate": 4.954643961002936e-06,
      "loss": 0.0309,
      "step": 177
    },
    {
      "epoch": 0.7795620437956204,
      "grad_norm": 0.7389516234397888,
      "learning_rate": 4.953461924567559e-06,
      "loss": 0.0291,
      "step": 178
    },
    {
      "epoch": 0.7839416058394161,
      "grad_norm": 0.790238082408905,
      "learning_rate": 4.952264827599299e-06,
      "loss": 0.0236,
      "step": 179
    },
    {
      "epoch": 0.7883211678832117,
      "grad_norm": 0.6766819953918457,
      "learning_rate": 4.951052677446515e-06,
      "loss": 0.0238,
      "step": 180
    },
    {
      "epoch": 0.7927007299270074,
      "grad_norm": 0.8832846283912659,
      "learning_rate": 4.94982548154997e-06,
      "loss": 0.0259,
      "step": 181
    },
    {
      "epoch": 0.7970802919708029,
      "grad_norm": 0.7298055291175842,
      "learning_rate": 4.948583247442783e-06,
      "loss": 0.023,
      "step": 182
    },
    {
      "epoch": 0.8014598540145985,
      "grad_norm": 0.911920428276062,
      "learning_rate": 4.947325982750387e-06,
      "loss": 0.0272,
      "step": 183
    },
    {
      "epoch": 0.8058394160583942,
      "grad_norm": 0.9145316481590271,
      "learning_rate": 4.946053695190479e-06,
      "loss": 0.0248,
      "step": 184
    },
    {
      "epoch": 0.8102189781021898,
      "grad_norm": 0.8759565353393555,
      "learning_rate": 4.9447663925729735e-06,
      "loss": 0.0263,
      "step": 185
    },
    {
      "epoch": 0.8145985401459854,
      "grad_norm": 1.1927592754364014,
      "learning_rate": 4.943464082799956e-06,
      "loss": 0.0305,
      "step": 186
    },
    {
      "epoch": 0.8189781021897811,
      "grad_norm": 0.752566933631897,
      "learning_rate": 4.942146773865631e-06,
      "loss": 0.0247,
      "step": 187
    },
    {
      "epoch": 0.8233576642335766,
      "grad_norm": 1.1121447086334229,
      "learning_rate": 4.940814473856278e-06,
      "loss": 0.0293,
      "step": 188
    },
    {
      "epoch": 0.8277372262773722,
      "grad_norm": 1.0319955348968506,
      "learning_rate": 4.939467190950195e-06,
      "loss": 0.0247,
      "step": 189
    },
    {
      "epoch": 0.8321167883211679,
      "grad_norm": 0.7960589528083801,
      "learning_rate": 4.938104933417655e-06,
      "loss": 0.0232,
      "step": 190
    },
    {
      "epoch": 0.8364963503649635,
      "grad_norm": 0.593197226524353,
      "learning_rate": 4.936727709620853e-06,
      "loss": 0.0232,
      "step": 191
    },
    {
      "epoch": 0.8408759124087591,
      "grad_norm": 0.6710584759712219,
      "learning_rate": 4.9353355280138525e-06,
      "loss": 0.0278,
      "step": 192
    },
    {
      "epoch": 0.8452554744525548,
      "grad_norm": 0.7627159357070923,
      "learning_rate": 4.933928397142535e-06,
      "loss": 0.0291,
      "step": 193
    },
    {
      "epoch": 0.8496350364963504,
      "grad_norm": 0.4998359680175781,
      "learning_rate": 4.93250632564455e-06,
      "loss": 0.018,
      "step": 194
    },
    {
      "epoch": 0.8540145985401459,
      "grad_norm": 0.8028760552406311,
      "learning_rate": 4.931069322249258e-06,
      "loss": 0.0193,
      "step": 195
    },
    {
      "epoch": 0.8583941605839416,
      "grad_norm": 0.6061640977859497,
      "learning_rate": 4.929617395777678e-06,
      "loss": 0.0142,
      "step": 196
    },
    {
      "epoch": 0.8627737226277372,
      "grad_norm": 0.5901748538017273,
      "learning_rate": 4.928150555142436e-06,
      "loss": 0.0177,
      "step": 197
    },
    {
      "epoch": 0.8671532846715329,
      "grad_norm": 0.7800254225730896,
      "learning_rate": 4.926668809347707e-06,
      "loss": 0.0264,
      "step": 198
    },
    {
      "epoch": 0.8715328467153285,
      "grad_norm": 0.9308339357376099,
      "learning_rate": 4.925172167489162e-06,
      "loss": 0.0247,
      "step": 199
    },
    {
      "epoch": 0.8759124087591241,
      "grad_norm": 0.9651213884353638,
      "learning_rate": 4.923660638753911e-06,
      "loss": 0.0216,
      "step": 200
    },
    {
      "epoch": 0.8802919708029197,
      "grad_norm": 1.1258251667022705,
      "learning_rate": 4.9221342324204455e-06,
      "loss": 0.0249,
      "step": 201
    },
    {
      "epoch": 0.8846715328467153,
      "grad_norm": 1.0175387859344482,
      "learning_rate": 4.9205929578585845e-06,
      "loss": 0.0201,
      "step": 202
    },
    {
      "epoch": 0.8890510948905109,
      "grad_norm": 1.5190610885620117,
      "learning_rate": 4.9190368245294155e-06,
      "loss": 0.0319,
      "step": 203
    },
    {
      "epoch": 0.8934306569343066,
      "grad_norm": 0.9947767853736877,
      "learning_rate": 4.917465841985234e-06,
      "loss": 0.0228,
      "step": 204
    },
    {
      "epoch": 0.8978102189781022,
      "grad_norm": 0.6416967511177063,
      "learning_rate": 4.91588001986949e-06,
      "loss": 0.0198,
      "step": 205
    },
    {
      "epoch": 0.9021897810218978,
      "grad_norm": 0.6980161666870117,
      "learning_rate": 4.914279367916724e-06,
      "loss": 0.0172,
      "step": 206
    },
    {
      "epoch": 0.9065693430656935,
      "grad_norm": 0.5301483869552612,
      "learning_rate": 4.912663895952511e-06,
      "loss": 0.0208,
      "step": 207
    },
    {
      "epoch": 0.910948905109489,
      "grad_norm": 0.6047857999801636,
      "learning_rate": 4.911033613893397e-06,
      "loss": 0.0227,
      "step": 208
    },
    {
      "epoch": 0.9153284671532846,
      "grad_norm": 0.6069537997245789,
      "learning_rate": 4.909388531746837e-06,
      "loss": 0.0195,
      "step": 209
    },
    {
      "epoch": 0.9197080291970803,
      "grad_norm": 0.6859843730926514,
      "learning_rate": 4.907728659611143e-06,
      "loss": 0.0244,
      "step": 210
    },
    {
      "epoch": 0.9240875912408759,
      "grad_norm": 0.6074005365371704,
      "learning_rate": 4.906054007675408e-06,
      "loss": 0.0195,
      "step": 211
    },
    {
      "epoch": 0.9284671532846716,
      "grad_norm": 1.1983692646026611,
      "learning_rate": 4.9043645862194545e-06,
      "loss": 0.023,
      "step": 212
    },
    {
      "epoch": 0.9328467153284672,
      "grad_norm": 0.8806214928627014,
      "learning_rate": 4.902660405613767e-06,
      "loss": 0.0243,
      "step": 213
    },
    {
      "epoch": 0.9372262773722628,
      "grad_norm": 0.6523962616920471,
      "learning_rate": 4.900941476319426e-06,
      "loss": 0.016,
      "step": 214
    },
    {
      "epoch": 0.9416058394160584,
      "grad_norm": 0.5673899054527283,
      "learning_rate": 4.899207808888051e-06,
      "loss": 0.0158,
      "step": 215
    },
    {
      "epoch": 0.945985401459854,
      "grad_norm": 0.9643133282661438,
      "learning_rate": 4.897459413961729e-06,
      "loss": 0.0194,
      "step": 216
    },
    {
      "epoch": 0.9503649635036496,
      "grad_norm": 0.6007612347602844,
      "learning_rate": 4.8956963022729495e-06,
      "loss": 0.0187,
      "step": 217
    },
    {
      "epoch": 0.9547445255474453,
      "grad_norm": 0.968173623085022,
      "learning_rate": 4.893918484644545e-06,
      "loss": 0.0223,
      "step": 218
    },
    {
      "epoch": 0.9591240875912409,
      "grad_norm": 0.6649457216262817,
      "learning_rate": 4.892125971989616e-06,
      "loss": 0.0205,
      "step": 219
    },
    {
      "epoch": 0.9635036496350365,
      "grad_norm": 0.48259082436561584,
      "learning_rate": 4.890318775311471e-06,
      "loss": 0.0121,
      "step": 220
    },
    {
      "epoch": 0.9678832116788321,
      "grad_norm": 0.8284991383552551,
      "learning_rate": 4.888496905703554e-06,
      "loss": 0.0176,
      "step": 221
    },
    {
      "epoch": 0.9722627737226277,
      "grad_norm": 0.5141683220863342,
      "learning_rate": 4.8866603743493805e-06,
      "loss": 0.0154,
      "step": 222
    },
    {
      "epoch": 0.9766423357664233,
      "grad_norm": 1.0223891735076904,
      "learning_rate": 4.884809192522466e-06,
      "loss": 0.0111,
      "step": 223
    },
    {
      "epoch": 0.981021897810219,
      "grad_norm": 0.844782292842865,
      "learning_rate": 4.882943371586256e-06,
      "loss": 0.016,
      "step": 224
    },
    {
      "epoch": 0.9854014598540146,
      "grad_norm": 0.6978311538696289,
      "learning_rate": 4.881062922994061e-06,
      "loss": 0.0129,
      "step": 225
    },
    {
      "epoch": 0.9897810218978103,
      "grad_norm": 0.8764100074768066,
      "learning_rate": 4.879167858288982e-06,
      "loss": 0.0213,
      "step": 226
    },
    {
      "epoch": 0.9941605839416059,
      "grad_norm": 1.0449023246765137,
      "learning_rate": 4.877258189103839e-06,
      "loss": 0.015,
      "step": 227
    },
    {
      "epoch": 0.9985401459854014,
      "grad_norm": 0.7534664869308472,
      "learning_rate": 4.875333927161104e-06,
      "loss": 0.0144,
      "step": 228
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7534664869308472,
      "learning_rate": 4.8733950842728236e-06,
      "loss": 0.0186,
      "step": 229
    },
    {
      "epoch": 1.0043795620437956,
      "grad_norm": 1.4982736110687256,
      "learning_rate": 4.871441672340551e-06,
      "loss": 0.0126,
      "step": 230
    },
    {
      "epoch": 1.0087591240875913,
      "grad_norm": 1.206292986869812,
      "learning_rate": 4.869473703355273e-06,
      "loss": 0.0165,
      "step": 231
    },
    {
      "epoch": 1.013138686131387,
      "grad_norm": 0.4586186408996582,
      "learning_rate": 4.867491189397331e-06,
      "loss": 0.0089,
      "step": 232
    },
    {
      "epoch": 1.0175182481751825,
      "grad_norm": 0.5647240281105042,
      "learning_rate": 4.8654941426363525e-06,
      "loss": 0.0122,
      "step": 233
    },
    {
      "epoch": 1.0218978102189782,
      "grad_norm": 0.6478530764579773,
      "learning_rate": 4.863482575331173e-06,
      "loss": 0.012,
      "step": 234
    },
    {
      "epoch": 1.0262773722627738,
      "grad_norm": 0.48696213960647583,
      "learning_rate": 4.861456499829764e-06,
      "loss": 0.0092,
      "step": 235
    },
    {
      "epoch": 1.0306569343065692,
      "grad_norm": 0.6736640334129333,
      "learning_rate": 4.859415928569154e-06,
      "loss": 0.0149,
      "step": 236
    },
    {
      "epoch": 1.0350364963503649,
      "grad_norm": 0.6518754363059998,
      "learning_rate": 4.857360874075355e-06,
      "loss": 0.0085,
      "step": 237
    },
    {
      "epoch": 1.0394160583941605,
      "grad_norm": 0.5145443677902222,
      "learning_rate": 4.855291348963281e-06,
      "loss": 0.0102,
      "step": 238
    },
    {
      "epoch": 1.0437956204379562,
      "grad_norm": 0.5647151470184326,
      "learning_rate": 4.853207365936676e-06,
      "loss": 0.0065,
      "step": 239
    },
    {
      "epoch": 1.0481751824817518,
      "grad_norm": 0.46668219566345215,
      "learning_rate": 4.8511089377880334e-06,
      "loss": 0.0081,
      "step": 240
    },
    {
      "epoch": 1.0525547445255474,
      "grad_norm": 0.9103809595108032,
      "learning_rate": 4.848996077398518e-06,
      "loss": 0.0107,
      "step": 241
    },
    {
      "epoch": 1.056934306569343,
      "grad_norm": 0.5947101712226868,
      "learning_rate": 4.8468687977378855e-06,
      "loss": 0.0095,
      "step": 242
    },
    {
      "epoch": 1.0613138686131387,
      "grad_norm": 0.7154219150543213,
      "learning_rate": 4.844727111864405e-06,
      "loss": 0.0097,
      "step": 243
    },
    {
      "epoch": 1.0656934306569343,
      "grad_norm": 0.9023681282997131,
      "learning_rate": 4.842571032924778e-06,
      "loss": 0.0105,
      "step": 244
    },
    {
      "epoch": 1.07007299270073,
      "grad_norm": 0.6020027995109558,
      "learning_rate": 4.840400574154056e-06,
      "loss": 0.0065,
      "step": 245
    },
    {
      "epoch": 1.0744525547445256,
      "grad_norm": 0.7602945566177368,
      "learning_rate": 4.838215748875562e-06,
      "loss": 0.0121,
      "step": 246
    },
    {
      "epoch": 1.0788321167883212,
      "grad_norm": 0.8768120408058167,
      "learning_rate": 4.83601657050081e-06,
      "loss": 0.0146,
      "step": 247
    },
    {
      "epoch": 1.0832116788321169,
      "grad_norm": 0.7482877373695374,
      "learning_rate": 4.833803052529414e-06,
      "loss": 0.0076,
      "step": 248
    },
    {
      "epoch": 1.0875912408759123,
      "grad_norm": 0.4619101881980896,
      "learning_rate": 4.831575208549018e-06,
      "loss": 0.0114,
      "step": 249
    },
    {
      "epoch": 1.091970802919708,
      "grad_norm": 0.7442188262939453,
      "learning_rate": 4.829333052235202e-06,
      "loss": 0.0119,
      "step": 250
    },
    {
      "epoch": 1.0963503649635036,
      "grad_norm": 0.754559338092804,
      "learning_rate": 4.827076597351403e-06,
      "loss": 0.011,
      "step": 251
    },
    {
      "epoch": 1.1007299270072992,
      "grad_norm": 0.8147054314613342,
      "learning_rate": 4.824805857748831e-06,
      "loss": 0.0098,
      "step": 252
    },
    {
      "epoch": 1.1051094890510949,
      "grad_norm": 0.814437985420227,
      "learning_rate": 4.82252084736638e-06,
      "loss": 0.0077,
      "step": 253
    },
    {
      "epoch": 1.1094890510948905,
      "grad_norm": 0.7731255888938904,
      "learning_rate": 4.820221580230545e-06,
      "loss": 0.0129,
      "step": 254
    },
    {
      "epoch": 1.1138686131386861,
      "grad_norm": 0.7589200139045715,
      "learning_rate": 4.8179080704553386e-06,
      "loss": 0.0095,
      "step": 255
    },
    {
      "epoch": 1.1182481751824818,
      "grad_norm": 0.455625057220459,
      "learning_rate": 4.815580332242199e-06,
      "loss": 0.0088,
      "step": 256
    },
    {
      "epoch": 1.1226277372262774,
      "grad_norm": 0.51591956615448,
      "learning_rate": 4.8132383798799075e-06,
      "loss": 0.0071,
      "step": 257
    },
    {
      "epoch": 1.127007299270073,
      "grad_norm": 0.6024675965309143,
      "learning_rate": 4.810882227744495e-06,
      "loss": 0.0108,
      "step": 258
    },
    {
      "epoch": 1.1313868613138687,
      "grad_norm": 0.6686123609542847,
      "learning_rate": 4.808511890299163e-06,
      "loss": 0.0139,
      "step": 259
    },
    {
      "epoch": 1.1357664233576643,
      "grad_norm": 0.7872790694236755,
      "learning_rate": 4.806127382094184e-06,
      "loss": 0.0113,
      "step": 260
    },
    {
      "epoch": 1.14014598540146,
      "grad_norm": 0.6551967263221741,
      "learning_rate": 4.803728717766822e-06,
      "loss": 0.0069,
      "step": 261
    },
    {
      "epoch": 1.1445255474452556,
      "grad_norm": 0.7421084642410278,
      "learning_rate": 4.801315912041232e-06,
      "loss": 0.0083,
      "step": 262
    },
    {
      "epoch": 1.148905109489051,
      "grad_norm": 0.6349561810493469,
      "learning_rate": 4.798888979728382e-06,
      "loss": 0.0097,
      "step": 263
    },
    {
      "epoch": 1.1532846715328466,
      "grad_norm": 0.6274579167366028,
      "learning_rate": 4.796447935725954e-06,
      "loss": 0.0089,
      "step": 264
    },
    {
      "epoch": 1.1576642335766423,
      "grad_norm": 0.5055127739906311,
      "learning_rate": 4.793992795018253e-06,
      "loss": 0.0062,
      "step": 265
    },
    {
      "epoch": 1.162043795620438,
      "grad_norm": 1.1284935474395752,
      "learning_rate": 4.791523572676115e-06,
      "loss": 0.0118,
      "step": 266
    },
    {
      "epoch": 1.1664233576642336,
      "grad_norm": 0.6343486905097961,
      "learning_rate": 4.789040283856822e-06,
      "loss": 0.0058,
      "step": 267
    },
    {
      "epoch": 1.1708029197080292,
      "grad_norm": 0.9384168982505798,
      "learning_rate": 4.7865429438039955e-06,
      "loss": 0.0096,
      "step": 268
    },
    {
      "epoch": 1.1751824817518248,
      "grad_norm": 0.879307746887207,
      "learning_rate": 4.784031567847515e-06,
      "loss": 0.0129,
      "step": 269
    },
    {
      "epoch": 1.1795620437956205,
      "grad_norm": 0.5268783569335938,
      "learning_rate": 4.781506171403416e-06,
      "loss": 0.0073,
      "step": 270
    },
    {
      "epoch": 1.183941605839416,
      "grad_norm": 1.332766056060791,
      "learning_rate": 4.778966769973802e-06,
      "loss": 0.0129,
      "step": 271
    },
    {
      "epoch": 1.1883211678832117,
      "grad_norm": 0.7192438244819641,
      "learning_rate": 4.7764133791467434e-06,
      "loss": 0.0083,
      "step": 272
    },
    {
      "epoch": 1.1927007299270074,
      "grad_norm": 0.5047981142997742,
      "learning_rate": 4.773846014596185e-06,
      "loss": 0.0057,
      "step": 273
    },
    {
      "epoch": 1.197080291970803,
      "grad_norm": 0.5075733661651611,
      "learning_rate": 4.7712646920818486e-06,
      "loss": 0.0098,
      "step": 274
    },
    {
      "epoch": 1.2014598540145984,
      "grad_norm": 0.5874909162521362,
      "learning_rate": 4.7686694274491375e-06,
      "loss": 0.0072,
      "step": 275
    },
    {
      "epoch": 1.205839416058394,
      "grad_norm": 0.511114239692688,
      "learning_rate": 4.766060236629037e-06,
      "loss": 0.0058,
      "step": 276
    },
    {
      "epoch": 1.2102189781021897,
      "grad_norm": 0.5427272915840149,
      "learning_rate": 4.763437135638021e-06,
      "loss": 0.0094,
      "step": 277
    },
    {
      "epoch": 1.2145985401459853,
      "grad_norm": 0.6207345724105835,
      "learning_rate": 4.760800140577947e-06,
      "loss": 0.0117,
      "step": 278
    },
    {
      "epoch": 1.218978102189781,
      "grad_norm": 0.9132710695266724,
      "learning_rate": 4.758149267635963e-06,
      "loss": 0.0085,
      "step": 279
    },
    {
      "epoch": 1.2233576642335766,
      "grad_norm": 0.500217080116272,
      "learning_rate": 4.755484533084407e-06,
      "loss": 0.01,
      "step": 280
    },
    {
      "epoch": 1.2277372262773723,
      "grad_norm": 0.38535866141319275,
      "learning_rate": 4.7528059532807045e-06,
      "loss": 0.0038,
      "step": 281
    },
    {
      "epoch": 1.2321167883211679,
      "grad_norm": 0.5505772233009338,
      "learning_rate": 4.750113544667271e-06,
      "loss": 0.0064,
      "step": 282
    },
    {
      "epoch": 1.2364963503649635,
      "grad_norm": 0.5370091795921326,
      "learning_rate": 4.747407323771408e-06,
      "loss": 0.0083,
      "step": 283
    },
    {
      "epoch": 1.2408759124087592,
      "grad_norm": 0.6680497527122498,
      "learning_rate": 4.744687307205207e-06,
      "loss": 0.006,
      "step": 284
    },
    {
      "epoch": 1.2452554744525548,
      "grad_norm": 0.5799117088317871,
      "learning_rate": 4.74195351166544e-06,
      "loss": 0.0067,
      "step": 285
    },
    {
      "epoch": 1.2496350364963504,
      "grad_norm": 0.3809143304824829,
      "learning_rate": 4.739205953933464e-06,
      "loss": 0.0081,
      "step": 286
    },
    {
      "epoch": 1.254014598540146,
      "grad_norm": 0.8633838891983032,
      "learning_rate": 4.736444650875114e-06,
      "loss": 0.0083,
      "step": 287
    },
    {
      "epoch": 1.2583941605839417,
      "grad_norm": 0.4796256124973297,
      "learning_rate": 4.7336696194405995e-06,
      "loss": 0.0083,
      "step": 288
    },
    {
      "epoch": 1.2627737226277373,
      "grad_norm": 0.8990418314933777,
      "learning_rate": 4.730880876664402e-06,
      "loss": 0.0053,
      "step": 289
    },
    {
      "epoch": 1.2671532846715328,
      "grad_norm": 0.21372799575328827,
      "learning_rate": 4.72807843966517e-06,
      "loss": 0.0042,
      "step": 290
    },
    {
      "epoch": 1.2715328467153284,
      "grad_norm": 1.0377510786056519,
      "learning_rate": 4.725262325645615e-06,
      "loss": 0.0083,
      "step": 291
    },
    {
      "epoch": 1.275912408759124,
      "grad_norm": 0.29527121782302856,
      "learning_rate": 4.722432551892402e-06,
      "loss": 0.0023,
      "step": 292
    },
    {
      "epoch": 1.2802919708029197,
      "grad_norm": 1.5753306150436401,
      "learning_rate": 4.719589135776048e-06,
      "loss": 0.0089,
      "step": 293
    },
    {
      "epoch": 1.2846715328467153,
      "grad_norm": 0.3794252574443817,
      "learning_rate": 4.716732094750813e-06,
      "loss": 0.003,
      "step": 294
    },
    {
      "epoch": 1.289051094890511,
      "grad_norm": 0.5407822132110596,
      "learning_rate": 4.7138614463545926e-06,
      "loss": 0.0075,
      "step": 295
    },
    {
      "epoch": 1.2934306569343066,
      "grad_norm": 0.8722830414772034,
      "learning_rate": 4.710977208208812e-06,
      "loss": 0.0051,
      "step": 296
    },
    {
      "epoch": 1.2978102189781022,
      "grad_norm": 0.6819527745246887,
      "learning_rate": 4.708079398018316e-06,
      "loss": 0.0094,
      "step": 297
    },
    {
      "epoch": 1.3021897810218979,
      "grad_norm": 0.7198041677474976,
      "learning_rate": 4.7051680335712626e-06,
      "loss": 0.0068,
      "step": 298
    },
    {
      "epoch": 1.3065693430656935,
      "grad_norm": 0.467638224363327,
      "learning_rate": 4.70224313273901e-06,
      "loss": 0.0059,
      "step": 299
    },
    {
      "epoch": 1.310948905109489,
      "grad_norm": 0.4593437612056732,
      "learning_rate": 4.699304713476009e-06,
      "loss": 0.0039,
      "step": 300
    },
    {
      "epoch": 1.3153284671532846,
      "grad_norm": 0.5790926814079285,
      "learning_rate": 4.696352793819698e-06,
      "loss": 0.0057,
      "step": 301
    },
    {
      "epoch": 1.3197080291970802,
      "grad_norm": 0.3413192331790924,
      "learning_rate": 4.693387391890382e-06,
      "loss": 0.0055,
      "step": 302
    },
    {
      "epoch": 1.3240875912408758,
      "grad_norm": 0.5049291849136353,
      "learning_rate": 4.690408525891129e-06,
      "loss": 0.0061,
      "step": 303
    },
    {
      "epoch": 1.3284671532846715,
      "grad_norm": 0.25111323595046997,
      "learning_rate": 4.687416214107655e-06,
      "loss": 0.0041,
      "step": 304
    },
    {
      "epoch": 1.332846715328467,
      "grad_norm": 0.5559152364730835,
      "learning_rate": 4.684410474908214e-06,
      "loss": 0.0093,
      "step": 305
    },
    {
      "epoch": 1.3372262773722627,
      "grad_norm": 0.3842668831348419,
      "learning_rate": 4.681391326743484e-06,
      "loss": 0.0051,
      "step": 306
    },
    {
      "epoch": 1.3416058394160584,
      "grad_norm": 1.6264209747314453,
      "learning_rate": 4.67835878814645e-06,
      "loss": 0.0063,
      "step": 307
    },
    {
      "epoch": 1.345985401459854,
      "grad_norm": 0.5829497575759888,
      "learning_rate": 4.6753128777323e-06,
      "loss": 0.0054,
      "step": 308
    },
    {
      "epoch": 1.3503649635036497,
      "grad_norm": 0.6949307322502136,
      "learning_rate": 4.6722536141982995e-06,
      "loss": 0.0055,
      "step": 309
    },
    {
      "epoch": 1.3547445255474453,
      "grad_norm": 0.6198911070823669,
      "learning_rate": 4.669181016323686e-06,
      "loss": 0.0063,
      "step": 310
    },
    {
      "epoch": 1.359124087591241,
      "grad_norm": 0.4557003080844879,
      "learning_rate": 4.666095102969545e-06,
      "loss": 0.0053,
      "step": 311
    },
    {
      "epoch": 1.3635036496350366,
      "grad_norm": 0.7198585271835327,
      "learning_rate": 4.662995893078702e-06,
      "loss": 0.0048,
      "step": 312
    },
    {
      "epoch": 1.3678832116788322,
      "grad_norm": 0.4380558431148529,
      "learning_rate": 4.659883405675604e-06,
      "loss": 0.0057,
      "step": 313
    },
    {
      "epoch": 1.3722627737226278,
      "grad_norm": 0.986754298210144,
      "learning_rate": 4.656757659866199e-06,
      "loss": 0.0091,
      "step": 314
    },
    {
      "epoch": 1.3766423357664235,
      "grad_norm": 1.1282256841659546,
      "learning_rate": 4.6536186748378236e-06,
      "loss": 0.0058,
      "step": 315
    },
    {
      "epoch": 1.3810218978102191,
      "grad_norm": 0.3973119854927063,
      "learning_rate": 4.6504664698590795e-06,
      "loss": 0.0048,
      "step": 316
    },
    {
      "epoch": 1.3854014598540145,
      "grad_norm": 0.4406156837940216,
      "learning_rate": 4.647301064279725e-06,
      "loss": 0.0039,
      "step": 317
    },
    {
      "epoch": 1.3897810218978102,
      "grad_norm": 0.8249232172966003,
      "learning_rate": 4.644122477530545e-06,
      "loss": 0.0084,
      "step": 318
    },
    {
      "epoch": 1.3941605839416058,
      "grad_norm": 1.4877322912216187,
      "learning_rate": 4.640930729123237e-06,
      "loss": 0.0054,
      "step": 319
    },
    {
      "epoch": 1.3985401459854014,
      "grad_norm": 0.4890510141849518,
      "learning_rate": 4.6377258386502956e-06,
      "loss": 0.0021,
      "step": 320
    },
    {
      "epoch": 1.402919708029197,
      "grad_norm": 0.36471042037010193,
      "learning_rate": 4.634507825784882e-06,
      "loss": 0.004,
      "step": 321
    },
    {
      "epoch": 1.4072992700729927,
      "grad_norm": 1.1714568138122559,
      "learning_rate": 4.631276710280713e-06,
      "loss": 0.0079,
      "step": 322
    },
    {
      "epoch": 1.4116788321167884,
      "grad_norm": 0.509325385093689,
      "learning_rate": 4.628032511971934e-06,
      "loss": 0.0027,
      "step": 323
    },
    {
      "epoch": 1.416058394160584,
      "grad_norm": 0.34730231761932373,
      "learning_rate": 4.624775250772999e-06,
      "loss": 0.004,
      "step": 324
    },
    {
      "epoch": 1.4204379562043796,
      "grad_norm": 0.4304009974002838,
      "learning_rate": 4.6215049466785484e-06,
      "loss": 0.0046,
      "step": 325
    },
    {
      "epoch": 1.4248175182481753,
      "grad_norm": 0.721092700958252,
      "learning_rate": 4.618221619763287e-06,
      "loss": 0.0042,
      "step": 326
    },
    {
      "epoch": 1.4291970802919707,
      "grad_norm": 0.9019221067428589,
      "learning_rate": 4.6149252901818585e-06,
      "loss": 0.008,
      "step": 327
    },
    {
      "epoch": 1.4335766423357663,
      "grad_norm": 3.142669439315796,
      "learning_rate": 4.611615978168725e-06,
      "loss": 0.0053,
      "step": 328
    },
    {
      "epoch": 1.437956204379562,
      "grad_norm": 0.8218545317649841,
      "learning_rate": 4.608293704038039e-06,
      "loss": 0.007,
      "step": 329
    },
    {
      "epoch": 1.4423357664233576,
      "grad_norm": 0.49122154712677,
      "learning_rate": 4.604958488183523e-06,
      "loss": 0.0056,
      "step": 330
    },
    {
      "epoch": 1.4467153284671532,
      "grad_norm": 0.7947913408279419,
      "learning_rate": 4.6016103510783405e-06,
      "loss": 0.0069,
      "step": 331
    },
    {
      "epoch": 1.4510948905109489,
      "grad_norm": 0.38262632489204407,
      "learning_rate": 4.598249313274972e-06,
      "loss": 0.0054,
      "step": 332
    },
    {
      "epoch": 1.4554744525547445,
      "grad_norm": 0.7605669498443604,
      "learning_rate": 4.59487539540509e-06,
      "loss": 0.0074,
      "step": 333
    },
    {
      "epoch": 1.4598540145985401,
      "grad_norm": 0.4355056583881378,
      "learning_rate": 4.591488618179428e-06,
      "loss": 0.0027,
      "step": 334
    },
    {
      "epoch": 1.4642335766423358,
      "grad_norm": 0.4696539640426636,
      "learning_rate": 4.58808900238766e-06,
      "loss": 0.0063,
      "step": 335
    },
    {
      "epoch": 1.4686131386861314,
      "grad_norm": 0.4078298807144165,
      "learning_rate": 4.584676568898267e-06,
      "loss": 0.0039,
      "step": 336
    },
    {
      "epoch": 1.472992700729927,
      "grad_norm": 0.22500784695148468,
      "learning_rate": 4.581251338658412e-06,
      "loss": 0.0027,
      "step": 337
    },
    {
      "epoch": 1.4773722627737227,
      "grad_norm": 0.28224533796310425,
      "learning_rate": 4.577813332693812e-06,
      "loss": 0.0037,
      "step": 338
    },
    {
      "epoch": 1.4817518248175183,
      "grad_norm": 0.4234824478626251,
      "learning_rate": 4.574362572108604e-06,
      "loss": 0.0057,
      "step": 339
    },
    {
      "epoch": 1.486131386861314,
      "grad_norm": 0.4610466957092285,
      "learning_rate": 4.570899078085223e-06,
      "loss": 0.0033,
      "step": 340
    },
    {
      "epoch": 1.4905109489051096,
      "grad_norm": 0.8538670539855957,
      "learning_rate": 4.567422871884265e-06,
      "loss": 0.0044,
      "step": 341
    },
    {
      "epoch": 1.4948905109489052,
      "grad_norm": 0.4335832893848419,
      "learning_rate": 4.563933974844361e-06,
      "loss": 0.0041,
      "step": 342
    },
    {
      "epoch": 1.4992700729927007,
      "grad_norm": 0.4888335168361664,
      "learning_rate": 4.560432408382045e-06,
      "loss": 0.003,
      "step": 343
    },
    {
      "epoch": 1.5036496350364965,
      "grad_norm": 0.545806884765625,
      "learning_rate": 4.5569181939916195e-06,
      "loss": 0.0062,
      "step": 344
    },
    {
      "epoch": 1.508029197080292,
      "grad_norm": 0.7364339828491211,
      "learning_rate": 4.553391353245029e-06,
      "loss": 0.0068,
      "step": 345
    },
    {
      "epoch": 1.5124087591240876,
      "grad_norm": 0.7074061036109924,
      "learning_rate": 4.549851907791722e-06,
      "loss": 0.0034,
      "step": 346
    },
    {
      "epoch": 1.5167883211678832,
      "grad_norm": 0.39756596088409424,
      "learning_rate": 4.546299879358524e-06,
      "loss": 0.0032,
      "step": 347
    },
    {
      "epoch": 1.5211678832116788,
      "grad_norm": 0.6966583728790283,
      "learning_rate": 4.542735289749498e-06,
      "loss": 0.0013,
      "step": 348
    },
    {
      "epoch": 1.5255474452554745,
      "grad_norm": 0.19892163574695587,
      "learning_rate": 4.5391581608458144e-06,
      "loss": 0.0011,
      "step": 349
    },
    {
      "epoch": 1.5299270072992701,
      "grad_norm": 0.718493640422821,
      "learning_rate": 4.535568514605617e-06,
      "loss": 0.0026,
      "step": 350
    },
    {
      "epoch": 1.5343065693430655,
      "grad_norm": 0.8941331505775452,
      "learning_rate": 4.5319663730638865e-06,
      "loss": 0.0034,
      "step": 351
    },
    {
      "epoch": 1.5386861313868612,
      "grad_norm": 0.33956244587898254,
      "learning_rate": 4.528351758332303e-06,
      "loss": 0.002,
      "step": 352
    },
    {
      "epoch": 1.5430656934306568,
      "grad_norm": 0.557651937007904,
      "learning_rate": 4.5247246925991185e-06,
      "loss": 0.0013,
      "step": 353
    },
    {
      "epoch": 1.5474452554744524,
      "grad_norm": 0.7165636420249939,
      "learning_rate": 4.5210851981290096e-06,
      "loss": 0.003,
      "step": 354
    },
    {
      "epoch": 1.551824817518248,
      "grad_norm": 0.36456218361854553,
      "learning_rate": 4.5174332972629505e-06,
      "loss": 0.0022,
      "step": 355
    },
    {
      "epoch": 1.5562043795620437,
      "grad_norm": 0.1896594613790512,
      "learning_rate": 4.5137690124180714e-06,
      "loss": 0.0056,
      "step": 356
    },
    {
      "epoch": 1.5605839416058394,
      "grad_norm": 0.6159863471984863,
      "learning_rate": 4.510092366087518e-06,
      "loss": 0.0057,
      "step": 357
    },
    {
      "epoch": 1.564963503649635,
      "grad_norm": 1.0295354127883911,
      "learning_rate": 4.506403380840321e-06,
      "loss": 0.0011,
      "step": 358
    },
    {
      "epoch": 1.5693430656934306,
      "grad_norm": 0.33694684505462646,
      "learning_rate": 4.50270207932125e-06,
      "loss": 0.0024,
      "step": 359
    },
    {
      "epoch": 1.5737226277372263,
      "grad_norm": 0.8961917757987976,
      "learning_rate": 4.498988484250681e-06,
      "loss": 0.0058,
      "step": 360
    },
    {
      "epoch": 1.578102189781022,
      "grad_norm": 1.736559510231018,
      "learning_rate": 4.4952626184244504e-06,
      "loss": 0.006,
      "step": 361
    },
    {
      "epoch": 1.5824817518248175,
      "grad_norm": 0.41748425364494324,
      "learning_rate": 4.491524504713722e-06,
      "loss": 0.0017,
      "step": 362
    },
    {
      "epoch": 1.5868613138686132,
      "grad_norm": 0.501815140247345,
      "learning_rate": 4.487774166064839e-06,
      "loss": 0.0018,
      "step": 363
    },
    {
      "epoch": 1.5912408759124088,
      "grad_norm": 0.4359874427318573,
      "learning_rate": 4.48401162549919e-06,
      "loss": 0.0044,
      "step": 364
    },
    {
      "epoch": 1.5956204379562045,
      "grad_norm": 0.3699054718017578,
      "learning_rate": 4.480236906113066e-06,
      "loss": 0.0036,
      "step": 365
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.5684164762496948,
      "learning_rate": 4.476450031077512e-06,
      "loss": 0.0023,
      "step": 366
    },
    {
      "epoch": 1.6043795620437957,
      "grad_norm": 0.6451728343963623,
      "learning_rate": 4.4726510236381956e-06,
      "loss": 0.0044,
      "step": 367
    },
    {
      "epoch": 1.6087591240875914,
      "grad_norm": 2.3887782096862793,
      "learning_rate": 4.468839907115259e-06,
      "loss": 0.0059,
      "step": 368
    },
    {
      "epoch": 1.613138686131387,
      "grad_norm": 0.6304333806037903,
      "learning_rate": 4.465016704903171e-06,
      "loss": 0.0023,
      "step": 369
    },
    {
      "epoch": 1.6175182481751826,
      "grad_norm": 0.38788676261901855,
      "learning_rate": 4.461181440470592e-06,
      "loss": 0.0027,
      "step": 370
    },
    {
      "epoch": 1.621897810218978,
      "grad_norm": 0.3805489242076874,
      "learning_rate": 4.457334137360226e-06,
      "loss": 0.0012,
      "step": 371
    },
    {
      "epoch": 1.6262773722627737,
      "grad_norm": 0.3548617660999298,
      "learning_rate": 4.453474819188676e-06,
      "loss": 0.0032,
      "step": 372
    },
    {
      "epoch": 1.6306569343065693,
      "grad_norm": 0.8332701921463013,
      "learning_rate": 4.449603509646297e-06,
      "loss": 0.0028,
      "step": 373
    },
    {
      "epoch": 1.635036496350365,
      "grad_norm": 0.7843290567398071,
      "learning_rate": 4.445720232497055e-06,
      "loss": 0.0037,
      "step": 374
    },
    {
      "epoch": 1.6394160583941606,
      "grad_norm": 0.7074784636497498,
      "learning_rate": 4.44182501157838e-06,
      "loss": 0.003,
      "step": 375
    },
    {
      "epoch": 1.6437956204379562,
      "grad_norm": 0.6076835989952087,
      "learning_rate": 4.4379178708010155e-06,
      "loss": 0.0019,
      "step": 376
    },
    {
      "epoch": 1.6481751824817519,
      "grad_norm": 0.5793138146400452,
      "learning_rate": 4.433998834148877e-06,
      "loss": 0.0036,
      "step": 377
    },
    {
      "epoch": 1.6525547445255473,
      "grad_norm": 0.7881670594215393,
      "learning_rate": 4.430067925678902e-06,
      "loss": 0.0025,
      "step": 378
    },
    {
      "epoch": 1.656934306569343,
      "grad_norm": 0.24594959616661072,
      "learning_rate": 4.426125169520903e-06,
      "loss": 0.0022,
      "step": 379
    },
    {
      "epoch": 1.6613138686131386,
      "grad_norm": 0.2806392312049866,
      "learning_rate": 4.42217058987742e-06,
      "loss": 0.0005,
      "step": 380
    },
    {
      "epoch": 1.6656934306569342,
      "grad_norm": 0.4979081153869629,
      "learning_rate": 4.418204211023569e-06,
      "loss": 0.0021,
      "step": 381
    },
    {
      "epoch": 1.6700729927007298,
      "grad_norm": 0.42502567172050476,
      "learning_rate": 4.4142260573068995e-06,
      "loss": 0.0053,
      "step": 382
    },
    {
      "epoch": 1.6744525547445255,
      "grad_norm": 1.1811860799789429,
      "learning_rate": 4.410236153147235e-06,
      "loss": 0.0026,
      "step": 383
    },
    {
      "epoch": 1.6788321167883211,
      "grad_norm": 0.4582519829273224,
      "learning_rate": 4.4062345230365345e-06,
      "loss": 0.0024,
      "step": 384
    },
    {
      "epoch": 1.6832116788321168,
      "grad_norm": 0.30464282631874084,
      "learning_rate": 4.402221191538733e-06,
      "loss": 0.0055,
      "step": 385
    },
    {
      "epoch": 1.6875912408759124,
      "grad_norm": 0.22526738047599792,
      "learning_rate": 4.3981961832895945e-06,
      "loss": 0.0003,
      "step": 386
    },
    {
      "epoch": 1.691970802919708,
      "grad_norm": 0.32826468348503113,
      "learning_rate": 4.394159522996564e-06,
      "loss": 0.0009,
      "step": 387
    },
    {
      "epoch": 1.6963503649635037,
      "grad_norm": 0.5943058133125305,
      "learning_rate": 4.390111235438606e-06,
      "loss": 0.0028,
      "step": 388
    },
    {
      "epoch": 1.7007299270072993,
      "grad_norm": 1.7098802328109741,
      "learning_rate": 4.3860513454660666e-06,
      "loss": 0.0035,
      "step": 389
    },
    {
      "epoch": 1.705109489051095,
      "grad_norm": 0.36092230677604675,
      "learning_rate": 4.381979878000506e-06,
      "loss": 0.0037,
      "step": 390
    },
    {
      "epoch": 1.7094890510948906,
      "grad_norm": 0.2771202027797699,
      "learning_rate": 4.377896858034557e-06,
      "loss": 0.0018,
      "step": 391
    },
    {
      "epoch": 1.7138686131386862,
      "grad_norm": 0.12323533743619919,
      "learning_rate": 4.373802310631765e-06,
      "loss": 0.0008,
      "step": 392
    },
    {
      "epoch": 1.7182481751824819,
      "grad_norm": 0.19630667567253113,
      "learning_rate": 4.3696962609264375e-06,
      "loss": 0.0008,
      "step": 393
    },
    {
      "epoch": 1.7226277372262775,
      "grad_norm": 0.4139691889286041,
      "learning_rate": 4.365578734123489e-06,
      "loss": 0.0031,
      "step": 394
    },
    {
      "epoch": 1.7270072992700731,
      "grad_norm": 0.6594070196151733,
      "learning_rate": 4.3614497554982845e-06,
      "loss": 0.0044,
      "step": 395
    },
    {
      "epoch": 1.7313868613138688,
      "grad_norm": 0.2723977863788605,
      "learning_rate": 4.357309350396488e-06,
      "loss": 0.0018,
      "step": 396
    },
    {
      "epoch": 1.7357664233576642,
      "grad_norm": 0.16032417118549347,
      "learning_rate": 4.3531575442339025e-06,
      "loss": 0.0005,
      "step": 397
    },
    {
      "epoch": 1.7401459854014598,
      "grad_norm": 0.3799298107624054,
      "learning_rate": 4.348994362496316e-06,
      "loss": 0.006,
      "step": 398
    },
    {
      "epoch": 1.7445255474452555,
      "grad_norm": 0.28333285450935364,
      "learning_rate": 4.344819830739349e-06,
      "loss": 0.0015,
      "step": 399
    },
    {
      "epoch": 1.748905109489051,
      "grad_norm": 0.3942627012729645,
      "learning_rate": 4.34063397458829e-06,
      "loss": 0.0018,
      "step": 400
    },
    {
      "epoch": 1.7532846715328467,
      "grad_norm": 0.8048702478408813,
      "learning_rate": 4.336436819737942e-06,
      "loss": 0.0021,
      "step": 401
    },
    {
      "epoch": 1.7576642335766424,
      "grad_norm": 0.1157551184296608,
      "learning_rate": 4.332228391952469e-06,
      "loss": 0.0009,
      "step": 402
    },
    {
      "epoch": 1.762043795620438,
      "grad_norm": 0.18697626888751984,
      "learning_rate": 4.328008717065228e-06,
      "loss": 0.0031,
      "step": 403
    },
    {
      "epoch": 1.7664233576642334,
      "grad_norm": 0.6587929129600525,
      "learning_rate": 4.323777820978622e-06,
      "loss": 0.0011,
      "step": 404
    },
    {
      "epoch": 1.770802919708029,
      "grad_norm": 0.40322232246398926,
      "learning_rate": 4.319535729663929e-06,
      "loss": 0.0013,
      "step": 405
    },
    {
      "epoch": 1.7751824817518247,
      "grad_norm": 0.33533793687820435,
      "learning_rate": 4.315282469161156e-06,
      "loss": 0.0008,
      "step": 406
    },
    {
      "epoch": 1.7795620437956203,
      "grad_norm": 0.2024499624967575,
      "learning_rate": 4.3110180655788645e-06,
      "loss": 0.0022,
      "step": 407
    },
    {
      "epoch": 1.783941605839416,
      "grad_norm": 0.5895872116088867,
      "learning_rate": 4.306742545094022e-06,
      "loss": 0.0019,
      "step": 408
    },
    {
      "epoch": 1.7883211678832116,
      "grad_norm": 0.3792962431907654,
      "learning_rate": 4.3024559339518355e-06,
      "loss": 0.0017,
      "step": 409
    },
    {
      "epoch": 1.7927007299270072,
      "grad_norm": 0.7945428490638733,
      "learning_rate": 4.298158258465593e-06,
      "loss": 0.0027,
      "step": 410
    },
    {
      "epoch": 1.7970802919708029,
      "grad_norm": 0.37964075803756714,
      "learning_rate": 4.2938495450164984e-06,
      "loss": 0.0014,
      "step": 411
    },
    {
      "epoch": 1.8014598540145985,
      "grad_norm": 0.08326616883277893,
      "learning_rate": 4.289529820053515e-06,
      "loss": 0.0005,
      "step": 412
    },
    {
      "epoch": 1.8058394160583942,
      "grad_norm": 0.14445550739765167,
      "learning_rate": 4.285199110093198e-06,
      "loss": 0.0021,
      "step": 413
    },
    {
      "epoch": 1.8102189781021898,
      "grad_norm": 0.24620558321475983,
      "learning_rate": 4.280857441719533e-06,
      "loss": 0.0007,
      "step": 414
    },
    {
      "epoch": 1.8145985401459854,
      "grad_norm": 0.2617506980895996,
      "learning_rate": 4.276504841583778e-06,
      "loss": 0.0011,
      "step": 415
    },
    {
      "epoch": 1.818978102189781,
      "grad_norm": 0.22467154264450073,
      "learning_rate": 4.27214133640429e-06,
      "loss": 0.0006,
      "step": 416
    },
    {
      "epoch": 1.8233576642335767,
      "grad_norm": 0.25831958651542664,
      "learning_rate": 4.267766952966369e-06,
      "loss": 0.0029,
      "step": 417
    },
    {
      "epoch": 1.8277372262773723,
      "grad_norm": 0.30368125438690186,
      "learning_rate": 4.263381718122092e-06,
      "loss": 0.0016,
      "step": 418
    },
    {
      "epoch": 1.832116788321168,
      "grad_norm": 0.6697282195091248,
      "learning_rate": 4.258985658790144e-06,
      "loss": 0.0044,
      "step": 419
    },
    {
      "epoch": 1.8364963503649636,
      "grad_norm": 0.4912242293357849,
      "learning_rate": 4.25457880195566e-06,
      "loss": 0.0014,
      "step": 420
    },
    {
      "epoch": 1.8408759124087593,
      "grad_norm": 0.17477519810199738,
      "learning_rate": 4.2501611746700526e-06,
      "loss": 0.0002,
      "step": 421
    },
    {
      "epoch": 1.845255474452555,
      "grad_norm": 0.09962823987007141,
      "learning_rate": 4.245732804050848e-06,
      "loss": 0.0009,
      "step": 422
    },
    {
      "epoch": 1.8496350364963505,
      "grad_norm": 0.5256549119949341,
      "learning_rate": 4.241293717281523e-06,
      "loss": 0.0005,
      "step": 423
    },
    {
      "epoch": 1.854014598540146,
      "grad_norm": 0.1596180498600006,
      "learning_rate": 4.236843941611332e-06,
      "loss": 0.001,
      "step": 424
    },
    {
      "epoch": 1.8583941605839416,
      "grad_norm": 0.3437536656856537,
      "learning_rate": 4.232383504355147e-06,
      "loss": 0.002,
      "step": 425
    },
    {
      "epoch": 1.8627737226277372,
      "grad_norm": 0.32742857933044434,
      "learning_rate": 4.227912432893282e-06,
      "loss": 0.0018,
      "step": 426
    },
    {
      "epoch": 1.8671532846715329,
      "grad_norm": 0.5527262091636658,
      "learning_rate": 4.223430754671331e-06,
      "loss": 0.0004,
      "step": 427
    },
    {
      "epoch": 1.8715328467153285,
      "grad_norm": 0.11191878467798233,
      "learning_rate": 4.218938497199996e-06,
      "loss": 0.0003,
      "step": 428
    },
    {
      "epoch": 1.8759124087591241,
      "grad_norm": 0.09846347570419312,
      "learning_rate": 4.214435688054922e-06,
      "loss": 0.0004,
      "step": 429
    },
    {
      "epoch": 1.8802919708029195,
      "grad_norm": 0.16991831362247467,
      "learning_rate": 4.209922354876523e-06,
      "loss": 0.0008,
      "step": 430
    },
    {
      "epoch": 1.8846715328467152,
      "grad_norm": 0.126469686627388,
      "learning_rate": 4.2053985253698155e-06,
      "loss": 0.0004,
      "step": 431
    },
    {
      "epoch": 1.8890510948905108,
      "grad_norm": 0.3232942521572113,
      "learning_rate": 4.200864227304247e-06,
      "loss": 0.0022,
      "step": 432
    },
    {
      "epoch": 1.8934306569343065,
      "grad_norm": 0.3737439811229706,
      "learning_rate": 4.196319488513528e-06,
      "loss": 0.0017,
      "step": 433
    },
    {
      "epoch": 1.897810218978102,
      "grad_norm": 0.14488628506660461,
      "learning_rate": 4.191764336895455e-06,
      "loss": 0.0002,
      "step": 434
    },
    {
      "epoch": 1.9021897810218977,
      "grad_norm": 0.16040323674678802,
      "learning_rate": 4.187198800411748e-06,
      "loss": 0.0005,
      "step": 435
    },
    {
      "epoch": 1.9065693430656934,
      "grad_norm": 0.19812235236167908,
      "learning_rate": 4.182622907087872e-06,
      "loss": 0.0002,
      "step": 436
    },
    {
      "epoch": 1.910948905109489,
      "grad_norm": 0.059883181005716324,
      "learning_rate": 4.178036685012869e-06,
      "loss": 0.0005,
      "step": 437
    },
    {
      "epoch": 1.9153284671532846,
      "grad_norm": 0.2905563414096832,
      "learning_rate": 4.1734401623391794e-06,
      "loss": 0.001,
      "step": 438
    },
    {
      "epoch": 1.9197080291970803,
      "grad_norm": 0.38163650035858154,
      "learning_rate": 4.168833367282479e-06,
      "loss": 0.0007,
      "step": 439
    },
    {
      "epoch": 1.924087591240876,
      "grad_norm": 0.04273957014083862,
      "learning_rate": 4.164216328121499e-06,
      "loss": 0.0001,
      "step": 440
    },
    {
      "epoch": 1.9284671532846716,
      "grad_norm": 0.02129952795803547,
      "learning_rate": 4.15958907319785e-06,
      "loss": 0.0001,
      "step": 441
    },
    {
      "epoch": 1.9328467153284672,
      "grad_norm": 0.016533153131604195,
      "learning_rate": 4.154951630915859e-06,
      "loss": 0.0001,
      "step": 442
    },
    {
      "epoch": 1.9372262773722628,
      "grad_norm": 0.11019770801067352,
      "learning_rate": 4.150304029742381e-06,
      "loss": 0.0002,
      "step": 443
    },
    {
      "epoch": 1.9416058394160585,
      "grad_norm": 0.05474651977419853,
      "learning_rate": 4.145646298206636e-06,
      "loss": 0.0002,
      "step": 444
    },
    {
      "epoch": 1.945985401459854,
      "grad_norm": 0.10142989456653595,
      "learning_rate": 4.1409784649000255e-06,
      "loss": 0.0001,
      "step": 445
    },
    {
      "epoch": 1.9503649635036497,
      "grad_norm": 0.21639519929885864,
      "learning_rate": 4.136300558475962e-06,
      "loss": 0.0019,
      "step": 446
    },
    {
      "epoch": 1.9547445255474454,
      "grad_norm": 0.45263969898223877,
      "learning_rate": 4.131612607649694e-06,
      "loss": 0.0034,
      "step": 447
    },
    {
      "epoch": 1.959124087591241,
      "grad_norm": 0.38673898577690125,
      "learning_rate": 4.126914641198123e-06,
      "loss": 0.0005,
      "step": 448
    },
    {
      "epoch": 1.9635036496350367,
      "grad_norm": 0.29815611243247986,
      "learning_rate": 4.1222066879596344e-06,
      "loss": 0.0004,
      "step": 449
    },
    {
      "epoch": 1.967883211678832,
      "grad_norm": 0.029003242030739784,
      "learning_rate": 4.1174887768339165e-06,
      "loss": 0.0002,
      "step": 450
    },
    {
      "epoch": 1.9722627737226277,
      "grad_norm": 0.21172675490379333,
      "learning_rate": 4.112760936781783e-06,
      "loss": 0.0002,
      "step": 451
    },
    {
      "epoch": 1.9766423357664233,
      "grad_norm": 0.27625802159309387,
      "learning_rate": 4.108023196824998e-06,
      "loss": 0.0038,
      "step": 452
    },
    {
      "epoch": 1.981021897810219,
      "grad_norm": 0.26207876205444336,
      "learning_rate": 4.103275586046095e-06,
      "loss": 0.0002,
      "step": 453
    },
    {
      "epoch": 1.9854014598540146,
      "grad_norm": 0.6478922367095947,
      "learning_rate": 4.098518133588198e-06,
      "loss": 0.0015,
      "step": 454
    },
    {
      "epoch": 1.9897810218978103,
      "grad_norm": 0.10226385295391083,
      "learning_rate": 4.093750868654845e-06,
      "loss": 0.0005,
      "step": 455
    },
    {
      "epoch": 1.994160583941606,
      "grad_norm": 0.18696191906929016,
      "learning_rate": 4.088973820509811e-06,
      "loss": 0.0004,
      "step": 456
    },
    {
      "epoch": 1.9985401459854013,
      "grad_norm": 0.26377999782562256,
      "learning_rate": 4.0841870184769184e-06,
      "loss": 0.0006,
      "step": 457
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.26377999782562256,
      "learning_rate": 4.079390491939868e-06,
      "loss": 0.0002,
      "step": 458
    },
    {
      "epoch": 2.0043795620437956,
      "grad_norm": 0.1395256221294403,
      "learning_rate": 4.074584270342057e-06,
      "loss": 0.0001,
      "step": 459
    },
    {
      "epoch": 2.0087591240875913,
      "grad_norm": 0.028063921257853508,
      "learning_rate": 4.069768383186388e-06,
      "loss": 0.0001,
      "step": 460
    },
    {
      "epoch": 2.013138686131387,
      "grad_norm": 0.12402255833148956,
      "learning_rate": 4.064942860035102e-06,
      "loss": 0.0008,
      "step": 461
    },
    {
      "epoch": 2.0175182481751825,
      "grad_norm": 1.1348239183425903,
      "learning_rate": 4.060107730509587e-06,
      "loss": 0.0002,
      "step": 462
    },
    {
      "epoch": 2.021897810218978,
      "grad_norm": 0.08540225028991699,
      "learning_rate": 4.055263024290201e-06,
      "loss": 0.0001,
      "step": 463
    },
    {
      "epoch": 2.026277372262774,
      "grad_norm": 0.021235302090644836,
      "learning_rate": 4.0504087711160875e-06,
      "loss": 0.0006,
      "step": 464
    },
    {
      "epoch": 2.0306569343065695,
      "grad_norm": 0.11977211385965347,
      "learning_rate": 4.045545000784995e-06,
      "loss": 0.0001,
      "step": 465
    },
    {
      "epoch": 2.035036496350365,
      "grad_norm": 0.03654933720827103,
      "learning_rate": 4.040671743153091e-06,
      "loss": 0.0,
      "step": 466
    },
    {
      "epoch": 2.0394160583941607,
      "grad_norm": 0.2099592536687851,
      "learning_rate": 4.035789028134782e-06,
      "loss": 0.0015,
      "step": 467
    },
    {
      "epoch": 2.0437956204379564,
      "grad_norm": 0.3698459267616272,
      "learning_rate": 4.03089688570253e-06,
      "loss": 0.0005,
      "step": 468
    },
    {
      "epoch": 2.048175182481752,
      "grad_norm": 0.12982334196567535,
      "learning_rate": 4.025995345886663e-06,
      "loss": 0.0005,
      "step": 469
    },
    {
      "epoch": 2.0525547445255476,
      "grad_norm": 0.2783415615558624,
      "learning_rate": 4.021084438775199e-06,
      "loss": 0.0001,
      "step": 470
    },
    {
      "epoch": 2.0569343065693433,
      "grad_norm": 0.0721098855137825,
      "learning_rate": 4.016164194513654e-06,
      "loss": 0.0001,
      "step": 471
    },
    {
      "epoch": 2.0613138686131385,
      "grad_norm": 0.013497601263225079,
      "learning_rate": 4.01123464330486e-06,
      "loss": 0.0001,
      "step": 472
    },
    {
      "epoch": 2.065693430656934,
      "grad_norm": 0.025111181661486626,
      "learning_rate": 4.006295815408781e-06,
      "loss": 0.0,
      "step": 473
    },
    {
      "epoch": 2.0700729927007298,
      "grad_norm": 0.01752329058945179,
      "learning_rate": 4.001347741142327e-06,
      "loss": 0.0016,
      "step": 474
    },
    {
      "epoch": 2.0744525547445254,
      "grad_norm": 0.5526646971702576,
      "learning_rate": 3.996390450879163e-06,
      "loss": 0.0002,
      "step": 475
    },
    {
      "epoch": 2.078832116788321,
      "grad_norm": 0.3007132112979889,
      "learning_rate": 3.9914239750495276e-06,
      "loss": 0.0003,
      "step": 476
    },
    {
      "epoch": 2.0832116788321167,
      "grad_norm": 0.42960840463638306,
      "learning_rate": 3.986448344140047e-06,
      "loss": 0.0011,
      "step": 477
    },
    {
      "epoch": 2.0875912408759123,
      "grad_norm": 0.396423876285553,
      "learning_rate": 3.9814635886935425e-06,
      "loss": 0.0001,
      "step": 478
    },
    {
      "epoch": 2.091970802919708,
      "grad_norm": 0.04294263571500778,
      "learning_rate": 3.976469739308849e-06,
      "loss": 0.0001,
      "step": 479
    },
    {
      "epoch": 2.0963503649635036,
      "grad_norm": 0.04891116917133331,
      "learning_rate": 3.971466826640623e-06,
      "loss": 0.0,
      "step": 480
    },
    {
      "epoch": 2.100729927007299,
      "grad_norm": 0.028048237785696983,
      "learning_rate": 3.966454881399155e-06,
      "loss": 0.0002,
      "step": 481
    },
    {
      "epoch": 2.105109489051095,
      "grad_norm": 0.08178327232599258,
      "learning_rate": 3.961433934350183e-06,
      "loss": 0.0001,
      "step": 482
    },
    {
      "epoch": 2.1094890510948905,
      "grad_norm": 0.1565946638584137,
      "learning_rate": 3.956404016314703e-06,
      "loss": 0.0,
      "step": 483
    },
    {
      "epoch": 2.113868613138686,
      "grad_norm": 0.017099319025874138,
      "learning_rate": 3.951365158168778e-06,
      "loss": 0.0,
      "step": 484
    },
    {
      "epoch": 2.1182481751824818,
      "grad_norm": 0.013942725956439972,
      "learning_rate": 3.9463173908433505e-06,
      "loss": 0.0001,
      "step": 485
    },
    {
      "epoch": 2.1226277372262774,
      "grad_norm": 0.031127380207180977,
      "learning_rate": 3.94126074532405e-06,
      "loss": 0.0001,
      "step": 486
    },
    {
      "epoch": 2.127007299270073,
      "grad_norm": 0.0673493817448616,
      "learning_rate": 3.936195252651008e-06,
      "loss": 0.0,
      "step": 487
    },
    {
      "epoch": 2.1313868613138687,
      "grad_norm": 0.009533442556858063,
      "learning_rate": 3.931120943918661e-06,
      "loss": 0.0001,
      "step": 488
    },
    {
      "epoch": 2.1357664233576643,
      "grad_norm": 0.2884984016418457,
      "learning_rate": 3.9260378502755644e-06,
      "loss": 0.0,
      "step": 489
    },
    {
      "epoch": 2.14014598540146,
      "grad_norm": 0.012546413578093052,
      "learning_rate": 3.9209460029242e-06,
      "loss": 0.0002,
      "step": 490
    },
    {
      "epoch": 2.1445255474452556,
      "grad_norm": 0.06351834535598755,
      "learning_rate": 3.915845433120781e-06,
      "loss": 0.0001,
      "step": 491
    },
    {
      "epoch": 2.1489051094890512,
      "grad_norm": 0.06691469997167587,
      "learning_rate": 3.910736172175066e-06,
      "loss": 0.0002,
      "step": 492
    },
    {
      "epoch": 2.153284671532847,
      "grad_norm": 0.13160394132137299,
      "learning_rate": 3.905618251450165e-06,
      "loss": 0.0002,
      "step": 493
    },
    {
      "epoch": 2.1576642335766425,
      "grad_norm": 0.13259641826152802,
      "learning_rate": 3.900491702362344e-06,
      "loss": 0.0001,
      "step": 494
    },
    {
      "epoch": 2.162043795620438,
      "grad_norm": 0.0303142461925745,
      "learning_rate": 3.895356556380833e-06,
      "loss": 0.0,
      "step": 495
    },
    {
      "epoch": 2.1664233576642338,
      "grad_norm": 0.013111322186887264,
      "learning_rate": 3.890212845027637e-06,
      "loss": 0.0001,
      "step": 496
    },
    {
      "epoch": 2.170802919708029,
      "grad_norm": 0.10938696563243866,
      "learning_rate": 3.8850605998773374e-06,
      "loss": 0.0002,
      "step": 497
    },
    {
      "epoch": 2.1751824817518246,
      "grad_norm": 0.032018087804317474,
      "learning_rate": 3.8798998525568985e-06,
      "loss": 0.0001,
      "step": 498
    },
    {
      "epoch": 2.1795620437956202,
      "grad_norm": 0.0468863919377327,
      "learning_rate": 3.87473063474548e-06,
      "loss": 0.0001,
      "step": 499
    },
    {
      "epoch": 2.183941605839416,
      "grad_norm": 0.07472304254770279,
      "learning_rate": 3.869552978174233e-06,
      "loss": 0.0013,
      "step": 500
    },
    {
      "epoch": 2.1883211678832115,
      "grad_norm": 1.0564846992492676,
      "learning_rate": 3.8643669146261105e-06,
      "loss": 0.0,
      "step": 501
    },
    {
      "epoch": 2.192700729927007,
      "grad_norm": 0.014464089646935463,
      "learning_rate": 3.859172475935674e-06,
      "loss": 0.0003,
      "step": 502
    },
    {
      "epoch": 2.197080291970803,
      "grad_norm": 0.18908506631851196,
      "learning_rate": 3.853969693988892e-06,
      "loss": 0.0,
      "step": 503
    },
    {
      "epoch": 2.2014598540145984,
      "grad_norm": 0.059129055589437485,
      "learning_rate": 3.848758600722953e-06,
      "loss": 0.0001,
      "step": 504
    },
    {
      "epoch": 2.205839416058394,
      "grad_norm": 0.008117246441543102,
      "learning_rate": 3.843539228126059e-06,
      "loss": 0.0017,
      "step": 505
    },
    {
      "epoch": 2.2102189781021897,
      "grad_norm": 0.8072256445884705,
      "learning_rate": 3.838311608237239e-06,
      "loss": 0.0036,
      "step": 506
    },
    {
      "epoch": 2.2145985401459853,
      "grad_norm": 1.5086314678192139,
      "learning_rate": 3.833075773146142e-06,
      "loss": 0.0001,
      "step": 507
    },
    {
      "epoch": 2.218978102189781,
      "grad_norm": 0.19763076305389404,
      "learning_rate": 3.827831754992854e-06,
      "loss": 0.0,
      "step": 508
    },
    {
      "epoch": 2.2233576642335766,
      "grad_norm": 0.007341600954532623,
      "learning_rate": 3.822579585967685e-06,
      "loss": 0.0,
      "step": 509
    },
    {
      "epoch": 2.2277372262773723,
      "grad_norm": 0.014308220706880093,
      "learning_rate": 3.817319298310984e-06,
      "loss": 0.0001,
      "step": 510
    },
    {
      "epoch": 2.232116788321168,
      "grad_norm": 0.11574184894561768,
      "learning_rate": 3.812050924312934e-06,
      "loss": 0.0001,
      "step": 511
    },
    {
      "epoch": 2.2364963503649635,
      "grad_norm": 0.04201903194189072,
      "learning_rate": 3.8067744963133555e-06,
      "loss": 0.0004,
      "step": 512
    },
    {
      "epoch": 2.240875912408759,
      "grad_norm": 0.2121448516845703,
      "learning_rate": 3.8014900467015093e-06,
      "loss": 0.0008,
      "step": 513
    },
    {
      "epoch": 2.245255474452555,
      "grad_norm": 0.41363075375556946,
      "learning_rate": 3.7961976079158964e-06,
      "loss": 0.0014,
      "step": 514
    },
    {
      "epoch": 2.2496350364963504,
      "grad_norm": 0.0038762835320085287,
      "learning_rate": 3.79089721244406e-06,
      "loss": 0.0001,
      "step": 515
    },
    {
      "epoch": 2.254014598540146,
      "grad_norm": 0.018642032518982887,
      "learning_rate": 3.785588892822383e-06,
      "loss": 0.0003,
      "step": 516
    },
    {
      "epoch": 2.2583941605839417,
      "grad_norm": 0.12630635499954224,
      "learning_rate": 3.780272681635894e-06,
      "loss": 0.0,
      "step": 517
    },
    {
      "epoch": 2.2627737226277373,
      "grad_norm": 0.34310439229011536,
      "learning_rate": 3.77494861151806e-06,
      "loss": 0.0013,
      "step": 518
    },
    {
      "epoch": 2.267153284671533,
      "grad_norm": 0.044213853776454926,
      "learning_rate": 3.769616715150593e-06,
      "loss": 0.0005,
      "step": 519
    },
    {
      "epoch": 2.2715328467153286,
      "grad_norm": 0.17094682157039642,
      "learning_rate": 3.7642770252632444e-06,
      "loss": 0.0001,
      "step": 520
    },
    {
      "epoch": 2.2759124087591243,
      "grad_norm": 0.0257079117000103,
      "learning_rate": 3.7589295746336074e-06,
      "loss": 0.0,
      "step": 521
    },
    {
      "epoch": 2.28029197080292,
      "grad_norm": 0.01195420790463686,
      "learning_rate": 3.753574396086913e-06,
      "loss": 0.0001,
      "step": 522
    },
    {
      "epoch": 2.2846715328467155,
      "grad_norm": 0.426551878452301,
      "learning_rate": 3.748211522495831e-06,
      "loss": 0.001,
      "step": 523
    },
    {
      "epoch": 2.289051094890511,
      "grad_norm": 0.02542410045862198,
      "learning_rate": 3.742840986780266e-06,
      "loss": 0.0002,
      "step": 524
    },
    {
      "epoch": 2.293430656934307,
      "grad_norm": 0.2616007924079895,
      "learning_rate": 3.737462821907158e-06,
      "loss": 0.002,
      "step": 525
    },
    {
      "epoch": 2.297810218978102,
      "grad_norm": 0.1279967725276947,
      "learning_rate": 3.732077060890277e-06,
      "loss": 0.0004,
      "step": 526
    },
    {
      "epoch": 2.3021897810218976,
      "grad_norm": 0.21853570640087128,
      "learning_rate": 3.7266837367900214e-06,
      "loss": 0.0011,
      "step": 527
    },
    {
      "epoch": 2.3065693430656933,
      "grad_norm": 0.21419601142406464,
      "learning_rate": 3.721282882713218e-06,
      "loss": 0.0005,
      "step": 528
    },
    {
      "epoch": 2.310948905109489,
      "grad_norm": 0.2496281862258911,
      "learning_rate": 3.7158745318129135e-06,
      "loss": 0.0003,
      "step": 529
    },
    {
      "epoch": 2.3153284671532846,
      "grad_norm": 0.10131111741065979,
      "learning_rate": 3.710458717288176e-06,
      "loss": 0.0003,
      "step": 530
    },
    {
      "epoch": 2.31970802919708,
      "grad_norm": 0.023186631500720978,
      "learning_rate": 3.7050354723838855e-06,
      "loss": 0.0005,
      "step": 531
    },
    {
      "epoch": 2.324087591240876,
      "grad_norm": 0.3965378403663635,
      "learning_rate": 3.6996048303905373e-06,
      "loss": 0.0003,
      "step": 532
    },
    {
      "epoch": 2.3284671532846715,
      "grad_norm": 0.03234691172838211,
      "learning_rate": 3.6941668246440323e-06,
      "loss": 0.0002,
      "step": 533
    },
    {
      "epoch": 2.332846715328467,
      "grad_norm": 0.14468444883823395,
      "learning_rate": 3.688721488525471e-06,
      "loss": 0.0003,
      "step": 534
    },
    {
      "epoch": 2.3372262773722627,
      "grad_norm": 0.09114562720060349,
      "learning_rate": 3.683268855460955e-06,
      "loss": 0.0,
      "step": 535
    },
    {
      "epoch": 2.3416058394160584,
      "grad_norm": 0.011452419683337212,
      "learning_rate": 3.6778089589213756e-06,
      "loss": 0.0,
      "step": 536
    },
    {
      "epoch": 2.345985401459854,
      "grad_norm": 0.026728764176368713,
      "learning_rate": 3.6723418324222126e-06,
      "loss": 0.0,
      "step": 537
    },
    {
      "epoch": 2.3503649635036497,
      "grad_norm": 0.010162679478526115,
      "learning_rate": 3.666867509523325e-06,
      "loss": 0.0003,
      "step": 538
    },
    {
      "epoch": 2.3547445255474453,
      "grad_norm": 0.2717076241970062,
      "learning_rate": 3.661386023828749e-06,
      "loss": 0.0002,
      "step": 539
    },
    {
      "epoch": 2.359124087591241,
      "grad_norm": 0.17123937606811523,
      "learning_rate": 3.6558974089864875e-06,
      "loss": 0.0,
      "step": 540
    },
    {
      "epoch": 2.3635036496350366,
      "grad_norm": 0.05525152385234833,
      "learning_rate": 3.650401698688305e-06,
      "loss": 0.0001,
      "step": 541
    },
    {
      "epoch": 2.367883211678832,
      "grad_norm": 0.0844765231013298,
      "learning_rate": 3.644898926669524e-06,
      "loss": 0.0001,
      "step": 542
    },
    {
      "epoch": 2.372262773722628,
      "grad_norm": 0.05303291603922844,
      "learning_rate": 3.6393891267088132e-06,
      "loss": 0.0,
      "step": 543
    },
    {
      "epoch": 2.3766423357664235,
      "grad_norm": 0.02489842288196087,
      "learning_rate": 3.633872332627983e-06,
      "loss": 0.0014,
      "step": 544
    },
    {
      "epoch": 2.381021897810219,
      "grad_norm": 0.4054766893386841,
      "learning_rate": 3.628348578291776e-06,
      "loss": 0.0,
      "step": 545
    },
    {
      "epoch": 2.3854014598540147,
      "grad_norm": 0.2564426064491272,
      "learning_rate": 3.6228178976076626e-06,
      "loss": 0.0005,
      "step": 546
    },
    {
      "epoch": 2.3897810218978104,
      "grad_norm": 0.4229254722595215,
      "learning_rate": 3.6172803245256283e-06,
      "loss": 0.0001,
      "step": 547
    },
    {
      "epoch": 2.394160583941606,
      "grad_norm": 0.05393270403146744,
      "learning_rate": 3.611735893037967e-06,
      "loss": 0.0,
      "step": 548
    },
    {
      "epoch": 2.398540145985401,
      "grad_norm": 0.004804346710443497,
      "learning_rate": 3.6061846371790754e-06,
      "loss": 0.0001,
      "step": 549
    },
    {
      "epoch": 2.402919708029197,
      "grad_norm": 0.2905990481376648,
      "learning_rate": 3.6006265910252393e-06,
      "loss": 0.0,
      "step": 550
    },
    {
      "epoch": 2.4072992700729925,
      "grad_norm": 0.05537007004022598,
      "learning_rate": 3.5950617886944272e-06,
      "loss": 0.0001,
      "step": 551
    },
    {
      "epoch": 2.411678832116788,
      "grad_norm": 0.006207960192114115,
      "learning_rate": 3.5894902643460807e-06,
      "loss": 0.0,
      "step": 552
    },
    {
      "epoch": 2.4160583941605838,
      "grad_norm": 0.14528508484363556,
      "learning_rate": 3.5839120521809036e-06,
      "loss": 0.0003,
      "step": 553
    },
    {
      "epoch": 2.4204379562043794,
      "grad_norm": 0.007411749102175236,
      "learning_rate": 3.578327186440654e-06,
      "loss": 0.0002,
      "step": 554
    },
    {
      "epoch": 2.424817518248175,
      "grad_norm": 0.1338014453649521,
      "learning_rate": 3.5727357014079306e-06,
      "loss": 0.0001,
      "step": 555
    },
    {
      "epoch": 2.4291970802919707,
      "grad_norm": 0.0037137684412300587,
      "learning_rate": 3.5671376314059676e-06,
      "loss": 0.0,
      "step": 556
    },
    {
      "epoch": 2.4335766423357663,
      "grad_norm": 0.26890304684638977,
      "learning_rate": 3.561533010798418e-06,
      "loss": 0.0004,
      "step": 557
    },
    {
      "epoch": 2.437956204379562,
      "grad_norm": 0.10123980045318604,
      "learning_rate": 3.555921873989148e-06,
      "loss": 0.0,
      "step": 558
    },
    {
      "epoch": 2.4423357664233576,
      "grad_norm": 0.033905990421772,
      "learning_rate": 3.5503042554220206e-06,
      "loss": 0.0001,
      "step": 559
    },
    {
      "epoch": 2.4467153284671532,
      "grad_norm": 0.2706269323825836,
      "learning_rate": 3.5446801895806904e-06,
      "loss": 0.0003,
      "step": 560
    },
    {
      "epoch": 2.451094890510949,
      "grad_norm": 0.017258645966649055,
      "learning_rate": 3.539049710988386e-06,
      "loss": 0.0,
      "step": 561
    },
    {
      "epoch": 2.4554744525547445,
      "grad_norm": 0.010611895471811295,
      "learning_rate": 3.5334128542077007e-06,
      "loss": 0.0,
      "step": 562
    },
    {
      "epoch": 2.45985401459854,
      "grad_norm": 0.0026555017102509737,
      "learning_rate": 3.527769653840381e-06,
      "loss": 0.0,
      "step": 563
    },
    {
      "epoch": 2.4642335766423358,
      "grad_norm": 0.15091587603092194,
      "learning_rate": 3.5221201445271136e-06,
      "loss": 0.0001,
      "step": 564
    },
    {
      "epoch": 2.4686131386861314,
      "grad_norm": 0.05164916068315506,
      "learning_rate": 3.5164643609473115e-06,
      "loss": 0.0004,
      "step": 565
    },
    {
      "epoch": 2.472992700729927,
      "grad_norm": 0.48653078079223633,
      "learning_rate": 3.5108023378189036e-06,
      "loss": 0.0,
      "step": 566
    },
    {
      "epoch": 2.4773722627737227,
      "grad_norm": 0.015059034340083599,
      "learning_rate": 3.5051341098981184e-06,
      "loss": 0.0,
      "step": 567
    },
    {
      "epoch": 2.4817518248175183,
      "grad_norm": 0.013922065496444702,
      "learning_rate": 3.499459711979274e-06,
      "loss": 0.0,
      "step": 568
    },
    {
      "epoch": 2.486131386861314,
      "grad_norm": 0.2496362030506134,
      "learning_rate": 3.493779178894561e-06,
      "loss": 0.0004,
      "step": 569
    },
    {
      "epoch": 2.4905109489051096,
      "grad_norm": 0.07800492644309998,
      "learning_rate": 3.488092545513833e-06,
      "loss": 0.0,
      "step": 570
    },
    {
      "epoch": 2.4948905109489052,
      "grad_norm": 0.613246738910675,
      "learning_rate": 3.4823998467443886e-06,
      "loss": 0.001,
      "step": 571
    },
    {
      "epoch": 2.499270072992701,
      "grad_norm": 0.5939133167266846,
      "learning_rate": 3.4767011175307596e-06,
      "loss": 0.0005,
      "step": 572
    },
    {
      "epoch": 2.5036496350364965,
      "grad_norm": 0.02257866971194744,
      "learning_rate": 3.4709963928544952e-06,
      "loss": 0.0001,
      "step": 573
    },
    {
      "epoch": 2.508029197080292,
      "grad_norm": 0.31688934564590454,
      "learning_rate": 3.4652857077339464e-06,
      "loss": 0.0016,
      "step": 574
    },
    {
      "epoch": 2.512408759124088,
      "grad_norm": 0.24091926217079163,
      "learning_rate": 3.459569097224054e-06,
      "loss": 0.0,
      "step": 575
    },
    {
      "epoch": 2.5167883211678834,
      "grad_norm": 0.0038259460125118494,
      "learning_rate": 3.4538465964161315e-06,
      "loss": 0.0009,
      "step": 576
    },
    {
      "epoch": 2.521167883211679,
      "grad_norm": 0.8010972738265991,
      "learning_rate": 3.448118240437649e-06,
      "loss": 0.0,
      "step": 577
    },
    {
      "epoch": 2.5255474452554747,
      "grad_norm": 0.10152903199195862,
      "learning_rate": 3.442384064452019e-06,
      "loss": 0.0,
      "step": 578
    },
    {
      "epoch": 2.5299270072992703,
      "grad_norm": 0.0037650642916560173,
      "learning_rate": 3.4366441036583803e-06,
      "loss": 0.0,
      "step": 579
    },
    {
      "epoch": 2.5343065693430655,
      "grad_norm": 0.002322736894711852,
      "learning_rate": 3.4308983932913806e-06,
      "loss": 0.0,
      "step": 580
    },
    {
      "epoch": 2.538686131386861,
      "grad_norm": 0.010998697020113468,
      "learning_rate": 3.4251469686209626e-06,
      "loss": 0.0,
      "step": 581
    },
    {
      "epoch": 2.543065693430657,
      "grad_norm": 0.009687647223472595,
      "learning_rate": 3.419389864952145e-06,
      "loss": 0.0,
      "step": 582
    },
    {
      "epoch": 2.5474452554744524,
      "grad_norm": 0.003304305486381054,
      "learning_rate": 3.413627117624808e-06,
      "loss": 0.0001,
      "step": 583
    },
    {
      "epoch": 2.551824817518248,
      "grad_norm": 0.04525240138173103,
      "learning_rate": 3.4078587620134747e-06,
      "loss": 0.0,
      "step": 584
    },
    {
      "epoch": 2.5562043795620437,
      "grad_norm": 0.14142946898937225,
      "learning_rate": 3.4020848335270946e-06,
      "loss": 0.0013,
      "step": 585
    },
    {
      "epoch": 2.5605839416058394,
      "grad_norm": 0.18268825113773346,
      "learning_rate": 3.3963053676088253e-06,
      "loss": 0.0,
      "step": 586
    },
    {
      "epoch": 2.564963503649635,
      "grad_norm": 0.07787997275590897,
      "learning_rate": 3.390520399735818e-06,
      "loss": 0.0008,
      "step": 587
    },
    {
      "epoch": 2.5693430656934306,
      "grad_norm": 0.6478791832923889,
      "learning_rate": 3.3847299654189947e-06,
      "loss": 0.0,
      "step": 588
    },
    {
      "epoch": 2.5737226277372263,
      "grad_norm": 0.029990263283252716,
      "learning_rate": 3.3789341002028364e-06,
      "loss": 0.0,
      "step": 589
    },
    {
      "epoch": 2.578102189781022,
      "grad_norm": 0.029525484889745712,
      "learning_rate": 3.3731328396651586e-06,
      "loss": 0.0,
      "step": 590
    },
    {
      "epoch": 2.5824817518248175,
      "grad_norm": 0.008475772105157375,
      "learning_rate": 3.3673262194168976e-06,
      "loss": 0.0,
      "step": 591
    },
    {
      "epoch": 2.586861313868613,
      "grad_norm": 0.013320226222276688,
      "learning_rate": 3.3615142751018893e-06,
      "loss": 0.0,
      "step": 592
    },
    {
      "epoch": 2.591240875912409,
      "grad_norm": 0.009780370630323887,
      "learning_rate": 3.3556970423966515e-06,
      "loss": 0.0,
      "step": 593
    },
    {
      "epoch": 2.5956204379562045,
      "grad_norm": 0.015174774453043938,
      "learning_rate": 3.349874557010166e-06,
      "loss": 0.0,
      "step": 594
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.011269648559391499,
      "learning_rate": 3.3440468546836564e-06,
      "loss": 0.0,
      "step": 595
    },
    {
      "epoch": 2.6043795620437957,
      "grad_norm": 0.014948060736060143,
      "learning_rate": 3.3382139711903707e-06,
      "loss": 0.0,
      "step": 596
    },
    {
      "epoch": 2.6087591240875914,
      "grad_norm": 0.005212848540395498,
      "learning_rate": 3.3323759423353618e-06,
      "loss": 0.0001,
      "step": 597
    },
    {
      "epoch": 2.613138686131387,
      "grad_norm": 0.3330608308315277,
      "learning_rate": 3.3265328039552676e-06,
      "loss": 0.0013,
      "step": 598
    },
    {
      "epoch": 2.6175182481751826,
      "grad_norm": 0.01612473465502262,
      "learning_rate": 3.320684591918089e-06,
      "loss": 0.002,
      "step": 599
    },
    {
      "epoch": 2.621897810218978,
      "grad_norm": 1.6749413013458252,
      "learning_rate": 3.3148313421229743e-06,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 2.6262773722627735,
      "grad_norm": 0.007137598004192114,
      "learning_rate": 3.308973090499994e-06,
      "loss": 0.0003,
      "step": 601
    },
    {
      "epoch": 2.630656934306569,
      "grad_norm": 0.2040940374135971,
      "learning_rate": 3.303109873009922e-06,
      "loss": 0.0,
      "step": 602
    },
    {
      "epoch": 2.6350364963503647,
      "grad_norm": 0.016712144017219543,
      "learning_rate": 3.297241725644016e-06,
      "loss": 0.0,
      "step": 603
    },
    {
      "epoch": 2.6394160583941604,
      "grad_norm": 0.05457067862153053,
      "learning_rate": 3.2913686844237963e-06,
      "loss": 0.0002,
      "step": 604
    },
    {
      "epoch": 2.643795620437956,
      "grad_norm": 0.05992881581187248,
      "learning_rate": 3.2854907854008224e-06,
      "loss": 0.0018,
      "step": 605
    },
    {
      "epoch": 2.6481751824817517,
      "grad_norm": 0.2480854094028473,
      "learning_rate": 3.2796080646564738e-06,
      "loss": 0.0,
      "step": 606
    },
    {
      "epoch": 2.6525547445255473,
      "grad_norm": 0.009658033028244972,
      "learning_rate": 3.273720558301729e-06,
      "loss": 0.0003,
      "step": 607
    },
    {
      "epoch": 2.656934306569343,
      "grad_norm": 0.06701422482728958,
      "learning_rate": 3.267828302476942e-06,
      "loss": 0.0001,
      "step": 608
    },
    {
      "epoch": 2.6613138686131386,
      "grad_norm": 0.04225367307662964,
      "learning_rate": 3.2619313333516213e-06,
      "loss": 0.0001,
      "step": 609
    },
    {
      "epoch": 2.665693430656934,
      "grad_norm": 0.01937365159392357,
      "learning_rate": 3.2560296871242085e-06,
      "loss": 0.0,
      "step": 610
    },
    {
      "epoch": 2.67007299270073,
      "grad_norm": 0.010443875566124916,
      "learning_rate": 3.2501234000218558e-06,
      "loss": 0.0001,
      "step": 611
    },
    {
      "epoch": 2.6744525547445255,
      "grad_norm": 0.039485372602939606,
      "learning_rate": 3.2442125083002014e-06,
      "loss": 0.0001,
      "step": 612
    },
    {
      "epoch": 2.678832116788321,
      "grad_norm": 0.01544400304555893,
      "learning_rate": 3.238297048243151e-06,
      "loss": 0.0,
      "step": 613
    },
    {
      "epoch": 2.6832116788321168,
      "grad_norm": 0.014269383624196053,
      "learning_rate": 3.2323770561626523e-06,
      "loss": 0.0,
      "step": 614
    },
    {
      "epoch": 2.6875912408759124,
      "grad_norm": 0.00998217985033989,
      "learning_rate": 3.2264525683984717e-06,
      "loss": 0.0001,
      "step": 615
    },
    {
      "epoch": 2.691970802919708,
      "grad_norm": 0.0631580650806427,
      "learning_rate": 3.2205236213179736e-06,
      "loss": 0.0001,
      "step": 616
    },
    {
      "epoch": 2.6963503649635037,
      "grad_norm": 0.022673817351460457,
      "learning_rate": 3.2145902513158963e-06,
      "loss": 0.0034,
      "step": 617
    },
    {
      "epoch": 2.7007299270072993,
      "grad_norm": 0.24634726345539093,
      "learning_rate": 3.2086524948141263e-06,
      "loss": 0.0,
      "step": 618
    },
    {
      "epoch": 2.705109489051095,
      "grad_norm": 0.018728157505393028,
      "learning_rate": 3.2027103882614772e-06,
      "loss": 0.0001,
      "step": 619
    },
    {
      "epoch": 2.7094890510948906,
      "grad_norm": 1.7468042373657227,
      "learning_rate": 3.1967639681334668e-06,
      "loss": 0.0002,
      "step": 620
    },
    {
      "epoch": 2.713868613138686,
      "grad_norm": 0.014051638543605804,
      "learning_rate": 3.1908132709320895e-06,
      "loss": 0.0003,
      "step": 621
    },
    {
      "epoch": 2.718248175182482,
      "grad_norm": 0.1467590481042862,
      "learning_rate": 3.1848583331855952e-06,
      "loss": 0.0,
      "step": 622
    },
    {
      "epoch": 2.7226277372262775,
      "grad_norm": 0.0301121287047863,
      "learning_rate": 3.178899191448266e-06,
      "loss": 0.0001,
      "step": 623
    },
    {
      "epoch": 2.727007299270073,
      "grad_norm": 0.011421700939536095,
      "learning_rate": 3.1729358823001873e-06,
      "loss": 0.0,
      "step": 624
    },
    {
      "epoch": 2.7313868613138688,
      "grad_norm": 0.018829964101314545,
      "learning_rate": 3.1669684423470277e-06,
      "loss": 0.0001,
      "step": 625
    },
    {
      "epoch": 2.7357664233576644,
      "grad_norm": 0.2821716070175171,
      "learning_rate": 3.1609969082198124e-06,
      "loss": 0.0001,
      "step": 626
    },
    {
      "epoch": 2.74014598540146,
      "grad_norm": 0.03946685045957565,
      "learning_rate": 3.155021316574699e-06,
      "loss": 0.0001,
      "step": 627
    },
    {
      "epoch": 2.7445255474452557,
      "grad_norm": 0.03428944945335388,
      "learning_rate": 3.1490417040927513e-06,
      "loss": 0.0001,
      "step": 628
    },
    {
      "epoch": 2.7489051094890513,
      "grad_norm": 0.0415036678314209,
      "learning_rate": 3.143058107479716e-06,
      "loss": 0.0001,
      "step": 629
    },
    {
      "epoch": 2.753284671532847,
      "grad_norm": 0.011763577349483967,
      "learning_rate": 3.1370705634657953e-06,
      "loss": 0.0005,
      "step": 630
    },
    {
      "epoch": 2.7576642335766426,
      "grad_norm": 0.2982889413833618,
      "learning_rate": 3.1310791088054225e-06,
      "loss": 0.0001,
      "step": 631
    },
    {
      "epoch": 2.7620437956204382,
      "grad_norm": 0.07167187333106995,
      "learning_rate": 3.1250837802770378e-06,
      "loss": 0.0001,
      "step": 632
    },
    {
      "epoch": 2.7664233576642334,
      "grad_norm": 0.009533251635730267,
      "learning_rate": 3.1190846146828587e-06,
      "loss": 0.0,
      "step": 633
    },
    {
      "epoch": 2.770802919708029,
      "grad_norm": 0.09512810409069061,
      "learning_rate": 3.1130816488486582e-06,
      "loss": 0.0003,
      "step": 634
    },
    {
      "epoch": 2.7751824817518247,
      "grad_norm": 0.024417169392108917,
      "learning_rate": 3.1070749196235366e-06,
      "loss": 0.0001,
      "step": 635
    },
    {
      "epoch": 2.7795620437956203,
      "grad_norm": 0.026550231501460075,
      "learning_rate": 3.1010644638796956e-06,
      "loss": 0.0002,
      "step": 636
    },
    {
      "epoch": 2.783941605839416,
      "grad_norm": 0.04400951415300369,
      "learning_rate": 3.0950503185122116e-06,
      "loss": 0.0001,
      "step": 637
    },
    {
      "epoch": 2.7883211678832116,
      "grad_norm": 0.037575673311948776,
      "learning_rate": 3.0890325204388107e-06,
      "loss": 0.0001,
      "step": 638
    },
    {
      "epoch": 2.7927007299270072,
      "grad_norm": 0.017704375088214874,
      "learning_rate": 3.083011106599641e-06,
      "loss": 0.0,
      "step": 639
    },
    {
      "epoch": 2.797080291970803,
      "grad_norm": 0.5097067356109619,
      "learning_rate": 3.0769861139570446e-06,
      "loss": 0.0002,
      "step": 640
    },
    {
      "epoch": 2.8014598540145985,
      "grad_norm": 0.03586648404598236,
      "learning_rate": 3.0709575794953333e-06,
      "loss": 0.0001,
      "step": 641
    },
    {
      "epoch": 2.805839416058394,
      "grad_norm": 0.08737017959356308,
      "learning_rate": 3.06492554022056e-06,
      "loss": 0.0001,
      "step": 642
    },
    {
      "epoch": 2.81021897810219,
      "grad_norm": 0.03558499738574028,
      "learning_rate": 3.0588900331602915e-06,
      "loss": 0.0,
      "step": 643
    },
    {
      "epoch": 2.8145985401459854,
      "grad_norm": 0.015121969394385815,
      "learning_rate": 3.0528510953633824e-06,
      "loss": 0.0,
      "step": 644
    },
    {
      "epoch": 2.818978102189781,
      "grad_norm": 0.005024883430451155,
      "learning_rate": 3.046808763899745e-06,
      "loss": 0.0,
      "step": 645
    },
    {
      "epoch": 2.8233576642335767,
      "grad_norm": 0.10004691034555435,
      "learning_rate": 3.0407630758601257e-06,
      "loss": 0.0002,
      "step": 646
    },
    {
      "epoch": 2.8277372262773723,
      "grad_norm": 0.0034653018228709698,
      "learning_rate": 3.034714068355874e-06,
      "loss": 0.0,
      "step": 647
    },
    {
      "epoch": 2.832116788321168,
      "grad_norm": 0.005470994859933853,
      "learning_rate": 3.0286617785187157e-06,
      "loss": 0.0,
      "step": 648
    },
    {
      "epoch": 2.8364963503649636,
      "grad_norm": 0.004967759363353252,
      "learning_rate": 3.022606243500526e-06,
      "loss": 0.0,
      "step": 649
    },
    {
      "epoch": 2.8408759124087593,
      "grad_norm": 0.005472187884151936,
      "learning_rate": 3.0165475004730994e-06,
      "loss": 0.0,
      "step": 650
    },
    {
      "epoch": 2.845255474452555,
      "grad_norm": 0.02719656005501747,
      "learning_rate": 3.0104855866279244e-06,
      "loss": 0.0,
      "step": 651
    },
    {
      "epoch": 2.8496350364963505,
      "grad_norm": 0.006005981471389532,
      "learning_rate": 3.0044205391759517e-06,
      "loss": 0.0001,
      "step": 652
    },
    {
      "epoch": 2.8540145985401457,
      "grad_norm": 0.07548055797815323,
      "learning_rate": 2.9983523953473697e-06,
      "loss": 0.0001,
      "step": 653
    },
    {
      "epoch": 2.8583941605839414,
      "grad_norm": 0.022410407662391663,
      "learning_rate": 2.9922811923913712e-06,
      "loss": 0.0,
      "step": 654
    },
    {
      "epoch": 2.862773722627737,
      "grad_norm": 0.0035277260467410088,
      "learning_rate": 2.9862069675759297e-06,
      "loss": 0.0,
      "step": 655
    },
    {
      "epoch": 2.8671532846715326,
      "grad_norm": 0.00670978520065546,
      "learning_rate": 2.980129758187567e-06,
      "loss": 0.0,
      "step": 656
    },
    {
      "epoch": 2.8715328467153283,
      "grad_norm": 0.0031098946928977966,
      "learning_rate": 2.974049601531126e-06,
      "loss": 0.0,
      "step": 657
    },
    {
      "epoch": 2.875912408759124,
      "grad_norm": 0.013540991581976414,
      "learning_rate": 2.9679665349295417e-06,
      "loss": 0.0,
      "step": 658
    },
    {
      "epoch": 2.8802919708029195,
      "grad_norm": 0.015702638775110245,
      "learning_rate": 2.9618805957236113e-06,
      "loss": 0.0,
      "step": 659
    },
    {
      "epoch": 2.884671532846715,
      "grad_norm": 0.03669784590601921,
      "learning_rate": 2.955791821271766e-06,
      "loss": 0.0,
      "step": 660
    },
    {
      "epoch": 2.889051094890511,
      "grad_norm": 0.011530515737831593,
      "learning_rate": 2.9497002489498394e-06,
      "loss": 0.0,
      "step": 661
    },
    {
      "epoch": 2.8934306569343065,
      "grad_norm": 0.004896043334156275,
      "learning_rate": 2.9436059161508425e-06,
      "loss": 0.0,
      "step": 662
    },
    {
      "epoch": 2.897810218978102,
      "grad_norm": 0.006100150756537914,
      "learning_rate": 2.9375088602847303e-06,
      "loss": 0.0,
      "step": 663
    },
    {
      "epoch": 2.9021897810218977,
      "grad_norm": 0.0026638389099389315,
      "learning_rate": 2.931409118778172e-06,
      "loss": 0.0,
      "step": 664
    },
    {
      "epoch": 2.9065693430656934,
      "grad_norm": 0.03918790817260742,
      "learning_rate": 2.9253067290743237e-06,
      "loss": 0.0001,
      "step": 665
    },
    {
      "epoch": 2.910948905109489,
      "grad_norm": 0.7945866584777832,
      "learning_rate": 2.9192017286325975e-06,
      "loss": 0.0,
      "step": 666
    },
    {
      "epoch": 2.9153284671532846,
      "grad_norm": 0.004381685517728329,
      "learning_rate": 2.913094154928431e-06,
      "loss": 0.0,
      "step": 667
    },
    {
      "epoch": 2.9197080291970803,
      "grad_norm": 0.006551735103130341,
      "learning_rate": 2.9069840454530583e-06,
      "loss": 0.0,
      "step": 668
    },
    {
      "epoch": 2.924087591240876,
      "grad_norm": 0.003194763557985425,
      "learning_rate": 2.900871437713279e-06,
      "loss": 0.0002,
      "step": 669
    },
    {
      "epoch": 2.9284671532846716,
      "grad_norm": 0.05487615615129471,
      "learning_rate": 2.894756369231228e-06,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 2.932846715328467,
      "grad_norm": 0.008219748735427856,
      "learning_rate": 2.888638877544146e-06,
      "loss": 0.0,
      "step": 671
    },
    {
      "epoch": 2.937226277372263,
      "grad_norm": 0.0017953907372429967,
      "learning_rate": 2.8825190002041475e-06,
      "loss": 0.0,
      "step": 672
    },
    {
      "epoch": 2.9416058394160585,
      "grad_norm": 0.00900351069867611,
      "learning_rate": 2.8763967747779926e-06,
      "loss": 0.0,
      "step": 673
    },
    {
      "epoch": 2.945985401459854,
      "grad_norm": 0.0021545395720750093,
      "learning_rate": 2.8702722388468544e-06,
      "loss": 0.0,
      "step": 674
    },
    {
      "epoch": 2.9503649635036497,
      "grad_norm": 0.01696695201098919,
      "learning_rate": 2.864145430006089e-06,
      "loss": 0.0,
      "step": 675
    },
    {
      "epoch": 2.9547445255474454,
      "grad_norm": 0.018733413890004158,
      "learning_rate": 2.858016385865004e-06,
      "loss": 0.0001,
      "step": 676
    },
    {
      "epoch": 2.959124087591241,
      "grad_norm": 0.0550026074051857,
      "learning_rate": 2.85188514404663e-06,
      "loss": 0.0,
      "step": 677
    },
    {
      "epoch": 2.9635036496350367,
      "grad_norm": 0.00857064314186573,
      "learning_rate": 2.845751742187487e-06,
      "loss": 0.0,
      "step": 678
    },
    {
      "epoch": 2.9678832116788323,
      "grad_norm": 0.008322341367602348,
      "learning_rate": 2.839616217937354e-06,
      "loss": 0.0,
      "step": 679
    },
    {
      "epoch": 2.972262773722628,
      "grad_norm": 0.001857755589298904,
      "learning_rate": 2.833478608959038e-06,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 2.9766423357664236,
      "grad_norm": 0.014696883969008923,
      "learning_rate": 2.827338952928146e-06,
      "loss": 0.0,
      "step": 681
    },
    {
      "epoch": 2.981021897810219,
      "grad_norm": 0.0031583639793097973,
      "learning_rate": 2.821197287532847e-06,
      "loss": 0.0,
      "step": 682
    },
    {
      "epoch": 2.985401459854015,
      "grad_norm": 0.0016461616614833474,
      "learning_rate": 2.8150536504736457e-06,
      "loss": 0.0,
      "step": 683
    },
    {
      "epoch": 2.9897810218978105,
      "grad_norm": 0.0036859370302408934,
      "learning_rate": 2.8089080794631514e-06,
      "loss": 0.0,
      "step": 684
    },
    {
      "epoch": 2.994160583941606,
      "grad_norm": 0.0019211465260013938,
      "learning_rate": 2.8027606122258435e-06,
      "loss": 0.0,
      "step": 685
    },
    {
      "epoch": 2.9985401459854013,
      "grad_norm": 0.001531517249532044,
      "learning_rate": 2.79661128649784e-06,
      "loss": 0.0,
      "step": 686
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.002470475621521473,
      "learning_rate": 2.7904601400266706e-06,
      "loss": 0.0,
      "step": 687
    },
    {
      "epoch": 3.0043795620437956,
      "grad_norm": 0.017151040956377983,
      "learning_rate": 2.784307210571039e-06,
      "loss": 0.0,
      "step": 688
    },
    {
      "epoch": 3.0087591240875913,
      "grad_norm": 0.0010016716551035643,
      "learning_rate": 2.7781525359005945e-06,
      "loss": 0.0,
      "step": 689
    },
    {
      "epoch": 3.013138686131387,
      "grad_norm": 0.0022327261976897717,
      "learning_rate": 2.771996153795699e-06,
      "loss": 0.0,
      "step": 690
    },
    {
      "epoch": 3.0175182481751825,
      "grad_norm": 0.001771376933902502,
      "learning_rate": 2.7658381020471965e-06,
      "loss": 0.0,
      "step": 691
    },
    {
      "epoch": 3.021897810218978,
      "grad_norm": 0.0039240955375134945,
      "learning_rate": 2.7596784184561788e-06,
      "loss": 0.0,
      "step": 692
    },
    {
      "epoch": 3.026277372262774,
      "grad_norm": 0.002869370160624385,
      "learning_rate": 2.7535171408337556e-06,
      "loss": 0.0,
      "step": 693
    },
    {
      "epoch": 3.0306569343065695,
      "grad_norm": 0.01324407197535038,
      "learning_rate": 2.7473543070008213e-06,
      "loss": 0.0,
      "step": 694
    },
    {
      "epoch": 3.035036496350365,
      "grad_norm": 0.015128228813409805,
      "learning_rate": 2.7411899547878223e-06,
      "loss": 0.0,
      "step": 695
    },
    {
      "epoch": 3.0394160583941607,
      "grad_norm": 0.0010945587418973446,
      "learning_rate": 2.7350241220345273e-06,
      "loss": 0.0,
      "step": 696
    },
    {
      "epoch": 3.0437956204379564,
      "grad_norm": 0.006534193176776171,
      "learning_rate": 2.7288568465897918e-06,
      "loss": 0.0,
      "step": 697
    },
    {
      "epoch": 3.048175182481752,
      "grad_norm": 0.0015636439202353358,
      "learning_rate": 2.722688166311328e-06,
      "loss": 0.0,
      "step": 698
    },
    {
      "epoch": 3.0525547445255476,
      "grad_norm": 0.0023577925749123096,
      "learning_rate": 2.7165181190654705e-06,
      "loss": 0.0,
      "step": 699
    },
    {
      "epoch": 3.0569343065693433,
      "grad_norm": 0.0021843581926077604,
      "learning_rate": 2.7103467427269466e-06,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 3.0613138686131385,
      "grad_norm": 0.0018377688247710466,
      "learning_rate": 2.704174075178641e-06,
      "loss": 0.0,
      "step": 701
    },
    {
      "epoch": 3.065693430656934,
      "grad_norm": 0.0018055844120681286,
      "learning_rate": 2.6980001543113653e-06,
      "loss": 0.0,
      "step": 702
    },
    {
      "epoch": 3.0700729927007298,
      "grad_norm": 0.0028426465578377247,
      "learning_rate": 2.691825018023624e-06,
      "loss": 0.0,
      "step": 703
    },
    {
      "epoch": 3.0744525547445254,
      "grad_norm": 0.006463038269430399,
      "learning_rate": 2.6856487042213825e-06,
      "loss": 0.0,
      "step": 704
    },
    {
      "epoch": 3.078832116788321,
      "grad_norm": 0.0850614532828331,
      "learning_rate": 2.6794712508178345e-06,
      "loss": 0.0001,
      "step": 705
    },
    {
      "epoch": 3.0832116788321167,
      "grad_norm": 0.0032790934201329947,
      "learning_rate": 2.673292695733169e-06,
      "loss": 0.0,
      "step": 706
    },
    {
      "epoch": 3.0875912408759123,
      "grad_norm": 0.0014041299000382423,
      "learning_rate": 2.6671130768943375e-06,
      "loss": 0.0,
      "step": 707
    },
    {
      "epoch": 3.091970802919708,
      "grad_norm": 0.0034296936355531216,
      "learning_rate": 2.660932432234823e-06,
      "loss": 0.0,
      "step": 708
    },
    {
      "epoch": 3.0963503649635036,
      "grad_norm": 0.0021895640529692173,
      "learning_rate": 2.654750799694402e-06,
      "loss": 0.0,
      "step": 709
    },
    {
      "epoch": 3.100729927007299,
      "grad_norm": 0.0013117026537656784,
      "learning_rate": 2.648568217218919e-06,
      "loss": 0.0,
      "step": 710
    },
    {
      "epoch": 3.105109489051095,
      "grad_norm": 0.0008849436999298632,
      "learning_rate": 2.6423847227600462e-06,
      "loss": 0.0,
      "step": 711
    },
    {
      "epoch": 3.1094890510948905,
      "grad_norm": 0.015677910298109055,
      "learning_rate": 2.636200354275057e-06,
      "loss": 0.0,
      "step": 712
    },
    {
      "epoch": 3.113868613138686,
      "grad_norm": 0.0022430643439292908,
      "learning_rate": 2.630015149726588e-06,
      "loss": 0.0,
      "step": 713
    },
    {
      "epoch": 3.1182481751824818,
      "grad_norm": 0.0011902175610885024,
      "learning_rate": 2.6238291470824084e-06,
      "loss": 0.0,
      "step": 714
    },
    {
      "epoch": 3.1226277372262774,
      "grad_norm": 0.002818114822730422,
      "learning_rate": 2.6176423843151866e-06,
      "loss": 0.0,
      "step": 715
    },
    {
      "epoch": 3.127007299270073,
      "grad_norm": 0.0025780019350349903,
      "learning_rate": 2.6114548994022575e-06,
      "loss": 0.0,
      "step": 716
    },
    {
      "epoch": 3.1313868613138687,
      "grad_norm": 0.008761986158788204,
      "learning_rate": 2.6052667303253886e-06,
      "loss": 0.0,
      "step": 717
    },
    {
      "epoch": 3.1357664233576643,
      "grad_norm": 0.0025233717169612646,
      "learning_rate": 2.5990779150705454e-06,
      "loss": 0.0,
      "step": 718
    },
    {
      "epoch": 3.14014598540146,
      "grad_norm": 0.0033938682172447443,
      "learning_rate": 2.5928884916276638e-06,
      "loss": 0.0,
      "step": 719
    },
    {
      "epoch": 3.1445255474452556,
      "grad_norm": 0.026820925995707512,
      "learning_rate": 2.586698497990409e-06,
      "loss": 0.0001,
      "step": 720
    },
    {
      "epoch": 3.1489051094890512,
      "grad_norm": 0.0023744499776512384,
      "learning_rate": 2.5805079721559496e-06,
      "loss": 0.0,
      "step": 721
    },
    {
      "epoch": 3.153284671532847,
      "grad_norm": 0.001529409782961011,
      "learning_rate": 2.574316952124718e-06,
      "loss": 0.0,
      "step": 722
    },
    {
      "epoch": 3.1576642335766425,
      "grad_norm": 0.005077557172626257,
      "learning_rate": 2.5681254759001828e-06,
      "loss": 0.0,
      "step": 723
    },
    {
      "epoch": 3.162043795620438,
      "grad_norm": 0.002494523301720619,
      "learning_rate": 2.561933581488612e-06,
      "loss": 0.0,
      "step": 724
    },
    {
      "epoch": 3.1664233576642338,
      "grad_norm": 0.0017919996753335,
      "learning_rate": 2.55574130689884e-06,
      "loss": 0.0,
      "step": 725
    },
    {
      "epoch": 3.170802919708029,
      "grad_norm": 0.0017449480947107077,
      "learning_rate": 2.549548690142036e-06,
      "loss": 0.0,
      "step": 726
    },
    {
      "epoch": 3.1751824817518246,
      "grad_norm": 0.0017062796978279948,
      "learning_rate": 2.5433557692314687e-06,
      "loss": 0.0,
      "step": 727
    },
    {
      "epoch": 3.1795620437956202,
      "grad_norm": 0.002210992621257901,
      "learning_rate": 2.5371625821822743e-06,
      "loss": 0.0,
      "step": 728
    },
    {
      "epoch": 3.183941605839416,
      "grad_norm": 0.0025367604102939367,
      "learning_rate": 2.530969167011222e-06,
      "loss": 0.0,
      "step": 729
    },
    {
      "epoch": 3.1883211678832115,
      "grad_norm": 0.0028825236950069666,
      "learning_rate": 2.5247755617364826e-06,
      "loss": 0.0,
      "step": 730
    },
    {
      "epoch": 3.192700729927007,
      "grad_norm": 0.0011007302673533559,
      "learning_rate": 2.5185818043773942e-06,
      "loss": 0.0,
      "step": 731
    },
    {
      "epoch": 3.197080291970803,
      "grad_norm": 0.008204109966754913,
      "learning_rate": 2.5123879329542255e-06,
      "loss": 0.0,
      "step": 732
    },
    {
      "epoch": 3.2014598540145984,
      "grad_norm": 0.001251939800567925,
      "learning_rate": 2.5061939854879485e-06,
      "loss": 0.0,
      "step": 733
    },
    {
      "epoch": 3.205839416058394,
      "grad_norm": 0.0024482880253344774,
      "learning_rate": 2.5e-06,
      "loss": 0.0,
      "step": 734
    },
    {
      "epoch": 3.2102189781021897,
      "grad_norm": 0.002128303749486804,
      "learning_rate": 2.4938060145120523e-06,
      "loss": 0.0,
      "step": 735
    },
    {
      "epoch": 3.2145985401459853,
      "grad_norm": 0.0013024493819102645,
      "learning_rate": 2.4876120670457753e-06,
      "loss": 0.0,
      "step": 736
    },
    {
      "epoch": 3.218978102189781,
      "grad_norm": 0.0012688646093010902,
      "learning_rate": 2.481418195622607e-06,
      "loss": 0.0,
      "step": 737
    },
    {
      "epoch": 3.2233576642335766,
      "grad_norm": 0.0021316749043762684,
      "learning_rate": 2.475224438263518e-06,
      "loss": 0.0,
      "step": 738
    },
    {
      "epoch": 3.2277372262773723,
      "grad_norm": 0.0019681300036609173,
      "learning_rate": 2.469030832988779e-06,
      "loss": 0.0,
      "step": 739
    },
    {
      "epoch": 3.232116788321168,
      "grad_norm": 0.0013115162728354335,
      "learning_rate": 2.4628374178177274e-06,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 3.2364963503649635,
      "grad_norm": 0.002466258592903614,
      "learning_rate": 2.4566442307685325e-06,
      "loss": 0.0,
      "step": 741
    },
    {
      "epoch": 3.240875912408759,
      "grad_norm": 0.005171903874725103,
      "learning_rate": 2.450451309857965e-06,
      "loss": 0.0,
      "step": 742
    },
    {
      "epoch": 3.245255474452555,
      "grad_norm": 0.0015806729206815362,
      "learning_rate": 2.4442586931011607e-06,
      "loss": 0.0,
      "step": 743
    },
    {
      "epoch": 3.2496350364963504,
      "grad_norm": 0.0023869157303124666,
      "learning_rate": 2.438066418511389e-06,
      "loss": 0.0,
      "step": 744
    },
    {
      "epoch": 3.254014598540146,
      "grad_norm": 0.0015038090059533715,
      "learning_rate": 2.431874524099818e-06,
      "loss": 0.0,
      "step": 745
    },
    {
      "epoch": 3.2583941605839417,
      "grad_norm": 0.0022007508669048548,
      "learning_rate": 2.4256830478752823e-06,
      "loss": 0.0,
      "step": 746
    },
    {
      "epoch": 3.2627737226277373,
      "grad_norm": 0.000799484783783555,
      "learning_rate": 2.419492027844051e-06,
      "loss": 0.0,
      "step": 747
    },
    {
      "epoch": 3.267153284671533,
      "grad_norm": 0.0014924613060429692,
      "learning_rate": 2.413301502009591e-06,
      "loss": 0.0,
      "step": 748
    },
    {
      "epoch": 3.2715328467153286,
      "grad_norm": 0.0013423648197203875,
      "learning_rate": 2.4071115083723367e-06,
      "loss": 0.0,
      "step": 749
    },
    {
      "epoch": 3.2759124087591243,
      "grad_norm": 0.0016152571188285947,
      "learning_rate": 2.4009220849294546e-06,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 3.28029197080292,
      "grad_norm": 0.002237207954749465,
      "learning_rate": 2.394733269674612e-06,
      "loss": 0.0,
      "step": 751
    },
    {
      "epoch": 3.2846715328467155,
      "grad_norm": 0.0030521079897880554,
      "learning_rate": 2.388545100597743e-06,
      "loss": 0.0,
      "step": 752
    },
    {
      "epoch": 3.289051094890511,
      "grad_norm": 0.0032652418594807386,
      "learning_rate": 2.3823576156848138e-06,
      "loss": 0.0,
      "step": 753
    },
    {
      "epoch": 3.293430656934307,
      "grad_norm": 0.0014183277962729335,
      "learning_rate": 2.3761708529175924e-06,
      "loss": 0.0,
      "step": 754
    },
    {
      "epoch": 3.297810218978102,
      "grad_norm": 0.0078318752348423,
      "learning_rate": 2.3699848502734126e-06,
      "loss": 0.0,
      "step": 755
    },
    {
      "epoch": 3.3021897810218976,
      "grad_norm": 0.003864576341584325,
      "learning_rate": 2.3637996457249433e-06,
      "loss": 0.0,
      "step": 756
    },
    {
      "epoch": 3.3065693430656933,
      "grad_norm": 0.002102078404277563,
      "learning_rate": 2.357615277239954e-06,
      "loss": 0.0,
      "step": 757
    },
    {
      "epoch": 3.310948905109489,
      "grad_norm": 0.0014949225587770343,
      "learning_rate": 2.3514317827810816e-06,
      "loss": 0.0,
      "step": 758
    },
    {
      "epoch": 3.3153284671532846,
      "grad_norm": 0.00100362254306674,
      "learning_rate": 2.3452492003055987e-06,
      "loss": 0.0,
      "step": 759
    },
    {
      "epoch": 3.31970802919708,
      "grad_norm": 0.2444867193698883,
      "learning_rate": 2.3390675677651778e-06,
      "loss": 0.0011,
      "step": 760
    },
    {
      "epoch": 3.324087591240876,
      "grad_norm": 0.0020129296462982893,
      "learning_rate": 2.332886923105663e-06,
      "loss": 0.0,
      "step": 761
    },
    {
      "epoch": 3.3284671532846715,
      "grad_norm": 0.22737674415111542,
      "learning_rate": 2.326707304266832e-06,
      "loss": 0.0006,
      "step": 762
    },
    {
      "epoch": 3.332846715328467,
      "grad_norm": 0.008498843759298325,
      "learning_rate": 2.3205287491821663e-06,
      "loss": 0.0,
      "step": 763
    },
    {
      "epoch": 3.3372262773722627,
      "grad_norm": 0.05480220168828964,
      "learning_rate": 2.3143512957786184e-06,
      "loss": 0.0001,
      "step": 764
    },
    {
      "epoch": 3.3416058394160584,
      "grad_norm": 0.0011567015899345279,
      "learning_rate": 2.308174981976377e-06,
      "loss": 0.0,
      "step": 765
    },
    {
      "epoch": 3.345985401459854,
      "grad_norm": 0.0017249404918402433,
      "learning_rate": 2.301999845688635e-06,
      "loss": 0.0,
      "step": 766
    },
    {
      "epoch": 3.3503649635036497,
      "grad_norm": 0.04111481457948685,
      "learning_rate": 2.2958259248213595e-06,
      "loss": 0.0,
      "step": 767
    },
    {
      "epoch": 3.3547445255474453,
      "grad_norm": 0.0038588643074035645,
      "learning_rate": 2.2896532572730534e-06,
      "loss": 0.0,
      "step": 768
    },
    {
      "epoch": 3.359124087591241,
      "grad_norm": 0.001527044572867453,
      "learning_rate": 2.2834818809345295e-06,
      "loss": 0.0,
      "step": 769
    },
    {
      "epoch": 3.3635036496350366,
      "grad_norm": 0.009431148879230022,
      "learning_rate": 2.2773118336886723e-06,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 3.367883211678832,
      "grad_norm": 0.0027088054921478033,
      "learning_rate": 2.271143153410208e-06,
      "loss": 0.0,
      "step": 771
    },
    {
      "epoch": 3.372262773722628,
      "grad_norm": 0.002571119461208582,
      "learning_rate": 2.264975877965473e-06,
      "loss": 0.0,
      "step": 772
    },
    {
      "epoch": 3.3766423357664235,
      "grad_norm": 0.0013004938373342156,
      "learning_rate": 2.258810045212178e-06,
      "loss": 0.0,
      "step": 773
    },
    {
      "epoch": 3.381021897810219,
      "grad_norm": 0.0035143231507390738,
      "learning_rate": 2.2526456929991795e-06,
      "loss": 0.0,
      "step": 774
    },
    {
      "epoch": 3.3854014598540147,
      "grad_norm": 0.0021215202286839485,
      "learning_rate": 2.2464828591662452e-06,
      "loss": 0.0,
      "step": 775
    },
    {
      "epoch": 3.3897810218978104,
      "grad_norm": 0.004478083923459053,
      "learning_rate": 2.240321581543822e-06,
      "loss": 0.0,
      "step": 776
    },
    {
      "epoch": 3.394160583941606,
      "grad_norm": 0.0010630427859723568,
      "learning_rate": 2.2341618979528044e-06,
      "loss": 0.0,
      "step": 777
    },
    {
      "epoch": 3.398540145985401,
      "grad_norm": 0.010118206031620502,
      "learning_rate": 2.2280038462043017e-06,
      "loss": 0.0,
      "step": 778
    },
    {
      "epoch": 3.402919708029197,
      "grad_norm": 0.005418050102889538,
      "learning_rate": 2.2218474640994064e-06,
      "loss": 0.0,
      "step": 779
    },
    {
      "epoch": 3.4072992700729925,
      "grad_norm": 0.00403578719124198,
      "learning_rate": 2.215692789428962e-06,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 3.411678832116788,
      "grad_norm": 0.004099252633750439,
      "learning_rate": 2.20953985997333e-06,
      "loss": 0.0,
      "step": 781
    },
    {
      "epoch": 3.4160583941605838,
      "grad_norm": 0.0032652115914970636,
      "learning_rate": 2.2033887135021605e-06,
      "loss": 0.0,
      "step": 782
    },
    {
      "epoch": 3.4204379562043794,
      "grad_norm": 0.002060306491330266,
      "learning_rate": 2.1972393877741578e-06,
      "loss": 0.0,
      "step": 783
    },
    {
      "epoch": 3.424817518248175,
      "grad_norm": 0.005281209945678711,
      "learning_rate": 2.191091920536849e-06,
      "loss": 0.0,
      "step": 784
    },
    {
      "epoch": 3.4291970802919707,
      "grad_norm": 0.006397861056029797,
      "learning_rate": 2.1849463495263547e-06,
      "loss": 0.0,
      "step": 785
    },
    {
      "epoch": 3.4335766423357663,
      "grad_norm": 0.008003026247024536,
      "learning_rate": 2.1788027124671542e-06,
      "loss": 0.0,
      "step": 786
    },
    {
      "epoch": 3.437956204379562,
      "grad_norm": 0.005707223899662495,
      "learning_rate": 2.1726610470718553e-06,
      "loss": 0.0,
      "step": 787
    },
    {
      "epoch": 3.4423357664233576,
      "grad_norm": 0.007127422373741865,
      "learning_rate": 2.166521391040963e-06,
      "loss": 0.0,
      "step": 788
    },
    {
      "epoch": 3.4467153284671532,
      "grad_norm": 0.0022657839581370354,
      "learning_rate": 2.1603837820626478e-06,
      "loss": 0.0,
      "step": 789
    },
    {
      "epoch": 3.451094890510949,
      "grad_norm": 0.004881497472524643,
      "learning_rate": 2.1542482578125148e-06,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 3.4554744525547445,
      "grad_norm": 0.0031092618592083454,
      "learning_rate": 2.1481148559533703e-06,
      "loss": 0.0,
      "step": 791
    },
    {
      "epoch": 3.45985401459854,
      "grad_norm": 0.010201151482760906,
      "learning_rate": 2.1419836141349964e-06,
      "loss": 0.0,
      "step": 792
    },
    {
      "epoch": 3.4642335766423358,
      "grad_norm": 0.0022016458678990602,
      "learning_rate": 2.1358545699939114e-06,
      "loss": 0.0,
      "step": 793
    },
    {
      "epoch": 3.4686131386861314,
      "grad_norm": 0.001897580805234611,
      "learning_rate": 2.129727761153146e-06,
      "loss": 0.0,
      "step": 794
    },
    {
      "epoch": 3.472992700729927,
      "grad_norm": 0.0018193572759628296,
      "learning_rate": 2.1236032252220074e-06,
      "loss": 0.0,
      "step": 795
    },
    {
      "epoch": 3.4773722627737227,
      "grad_norm": 0.0025425944477319717,
      "learning_rate": 2.117480999795853e-06,
      "loss": 0.0,
      "step": 796
    },
    {
      "epoch": 3.4817518248175183,
      "grad_norm": 0.0025497935712337494,
      "learning_rate": 2.1113611224558545e-06,
      "loss": 0.0,
      "step": 797
    },
    {
      "epoch": 3.486131386861314,
      "grad_norm": 0.002034541452303529,
      "learning_rate": 2.1052436307687725e-06,
      "loss": 0.0,
      "step": 798
    },
    {
      "epoch": 3.4905109489051096,
      "grad_norm": 0.004022255074232817,
      "learning_rate": 2.0991285622867215e-06,
      "loss": 0.0,
      "step": 799
    },
    {
      "epoch": 3.4948905109489052,
      "grad_norm": 0.003748381743207574,
      "learning_rate": 2.093015954546942e-06,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 3.499270072992701,
      "grad_norm": 0.005397303961217403,
      "learning_rate": 2.0869058450715694e-06,
      "loss": 0.0,
      "step": 801
    },
    {
      "epoch": 3.5036496350364965,
      "grad_norm": 0.0028435164131224155,
      "learning_rate": 2.0807982713674037e-06,
      "loss": 0.0,
      "step": 802
    },
    {
      "epoch": 3.508029197080292,
      "grad_norm": 0.01242138259112835,
      "learning_rate": 2.074693270925677e-06,
      "loss": 0.0,
      "step": 803
    },
    {
      "epoch": 3.512408759124088,
      "grad_norm": 0.08091861754655838,
      "learning_rate": 2.068590881221829e-06,
      "loss": 0.0002,
      "step": 804
    },
    {
      "epoch": 3.5167883211678834,
      "grad_norm": 0.007172738667577505,
      "learning_rate": 2.062491139715271e-06,
      "loss": 0.0,
      "step": 805
    },
    {
      "epoch": 3.521167883211679,
      "grad_norm": 0.0017966198502108455,
      "learning_rate": 2.056394083849158e-06,
      "loss": 0.0,
      "step": 806
    },
    {
      "epoch": 3.5255474452554747,
      "grad_norm": 0.0038727717474102974,
      "learning_rate": 2.0502997510501614e-06,
      "loss": 0.0,
      "step": 807
    },
    {
      "epoch": 3.5299270072992703,
      "grad_norm": 0.004094596020877361,
      "learning_rate": 2.0442081787282354e-06,
      "loss": 0.0,
      "step": 808
    },
    {
      "epoch": 3.5343065693430655,
      "grad_norm": 0.006543958559632301,
      "learning_rate": 2.03811940427639e-06,
      "loss": 0.0,
      "step": 809
    },
    {
      "epoch": 3.538686131386861,
      "grad_norm": 0.0020556284580379725,
      "learning_rate": 2.0320334650704595e-06,
      "loss": 0.0,
      "step": 810
    },
    {
      "epoch": 3.543065693430657,
      "grad_norm": 0.0070542385801672935,
      "learning_rate": 2.025950398468875e-06,
      "loss": 0.0,
      "step": 811
    },
    {
      "epoch": 3.5474452554744524,
      "grad_norm": 0.017733529210090637,
      "learning_rate": 2.0198702418124345e-06,
      "loss": 0.0,
      "step": 812
    },
    {
      "epoch": 3.551824817518248,
      "grad_norm": 0.009895303286612034,
      "learning_rate": 2.013793032424072e-06,
      "loss": 0.0,
      "step": 813
    },
    {
      "epoch": 3.5562043795620437,
      "grad_norm": 0.001627814257517457,
      "learning_rate": 2.007718807608629e-06,
      "loss": 0.0,
      "step": 814
    },
    {
      "epoch": 3.5605839416058394,
      "grad_norm": 0.005018120631575584,
      "learning_rate": 2.0016476046526308e-06,
      "loss": 0.0,
      "step": 815
    },
    {
      "epoch": 3.564963503649635,
      "grad_norm": 0.0010407187510281801,
      "learning_rate": 1.995579460824048e-06,
      "loss": 0.0,
      "step": 816
    },
    {
      "epoch": 3.5693430656934306,
      "grad_norm": 0.0016328482888638973,
      "learning_rate": 1.989514413372076e-06,
      "loss": 0.0,
      "step": 817
    },
    {
      "epoch": 3.5737226277372263,
      "grad_norm": 0.007556082680821419,
      "learning_rate": 1.983452499526901e-06,
      "loss": 0.0,
      "step": 818
    },
    {
      "epoch": 3.578102189781022,
      "grad_norm": 0.0021091431844979525,
      "learning_rate": 1.9773937564994747e-06,
      "loss": 0.0,
      "step": 819
    },
    {
      "epoch": 3.5824817518248175,
      "grad_norm": 0.0016330553917214274,
      "learning_rate": 1.9713382214812847e-06,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 3.586861313868613,
      "grad_norm": 0.0024340248201042414,
      "learning_rate": 1.9652859316441266e-06,
      "loss": 0.0,
      "step": 821
    },
    {
      "epoch": 3.591240875912409,
      "grad_norm": 0.001374023617245257,
      "learning_rate": 1.9592369241398747e-06,
      "loss": 0.0,
      "step": 822
    },
    {
      "epoch": 3.5956204379562045,
      "grad_norm": 0.0006934819975867867,
      "learning_rate": 1.9531912361002554e-06,
      "loss": 0.0,
      "step": 823
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.0039897337555885315,
      "learning_rate": 1.9471489046366184e-06,
      "loss": 0.0,
      "step": 824
    },
    {
      "epoch": 3.6043795620437957,
      "grad_norm": 0.002902815816923976,
      "learning_rate": 1.941109966839709e-06,
      "loss": 0.0,
      "step": 825
    },
    {
      "epoch": 3.6087591240875914,
      "grad_norm": 0.009976208209991455,
      "learning_rate": 1.9350744597794407e-06,
      "loss": 0.0,
      "step": 826
    },
    {
      "epoch": 3.613138686131387,
      "grad_norm": 0.0016394915292039514,
      "learning_rate": 1.929042420504667e-06,
      "loss": 0.0,
      "step": 827
    },
    {
      "epoch": 3.6175182481751826,
      "grad_norm": 0.011417590081691742,
      "learning_rate": 1.923013886042956e-06,
      "loss": 0.0,
      "step": 828
    },
    {
      "epoch": 3.621897810218978,
      "grad_norm": 0.006547160446643829,
      "learning_rate": 1.91698889340036e-06,
      "loss": 0.0,
      "step": 829
    },
    {
      "epoch": 3.6262773722627735,
      "grad_norm": 0.0013417567824944854,
      "learning_rate": 1.9109674795611897e-06,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 3.630656934306569,
      "grad_norm": 0.0021602141205221415,
      "learning_rate": 1.9049496814877894e-06,
      "loss": 0.0,
      "step": 831
    },
    {
      "epoch": 3.6350364963503647,
      "grad_norm": 0.004333204589784145,
      "learning_rate": 1.8989355361203057e-06,
      "loss": 0.0,
      "step": 832
    },
    {
      "epoch": 3.6394160583941604,
      "grad_norm": 0.0013988579157739878,
      "learning_rate": 1.892925080376465e-06,
      "loss": 0.0,
      "step": 833
    },
    {
      "epoch": 3.643795620437956,
      "grad_norm": 0.0024188708048313856,
      "learning_rate": 1.886918351151343e-06,
      "loss": 0.0,
      "step": 834
    },
    {
      "epoch": 3.6481751824817517,
      "grad_norm": 0.0826224833726883,
      "learning_rate": 1.8809153853171428e-06,
      "loss": 0.0,
      "step": 835
    },
    {
      "epoch": 3.6525547445255473,
      "grad_norm": 0.0022762177977710962,
      "learning_rate": 1.8749162197229626e-06,
      "loss": 0.0,
      "step": 836
    },
    {
      "epoch": 3.656934306569343,
      "grad_norm": 0.0012058281572535634,
      "learning_rate": 1.8689208911945771e-06,
      "loss": 0.0,
      "step": 837
    },
    {
      "epoch": 3.6613138686131386,
      "grad_norm": 0.0028202789835631847,
      "learning_rate": 1.8629294365342049e-06,
      "loss": 0.0,
      "step": 838
    },
    {
      "epoch": 3.665693430656934,
      "grad_norm": 0.0023832700680941343,
      "learning_rate": 1.8569418925202841e-06,
      "loss": 0.0,
      "step": 839
    },
    {
      "epoch": 3.67007299270073,
      "grad_norm": 0.0048092082142829895,
      "learning_rate": 1.8509582959072487e-06,
      "loss": 0.0,
      "step": 840
    },
    {
      "epoch": 3.6744525547445255,
      "grad_norm": 0.004336558748036623,
      "learning_rate": 1.8449786834253016e-06,
      "loss": 0.0,
      "step": 841
    },
    {
      "epoch": 3.678832116788321,
      "grad_norm": 0.008016410283744335,
      "learning_rate": 1.8390030917801883e-06,
      "loss": 0.0,
      "step": 842
    },
    {
      "epoch": 3.6832116788321168,
      "grad_norm": 0.003487163921818137,
      "learning_rate": 1.8330315576529733e-06,
      "loss": 0.0,
      "step": 843
    },
    {
      "epoch": 3.6875912408759124,
      "grad_norm": 0.005607697181403637,
      "learning_rate": 1.8270641176998138e-06,
      "loss": 0.0,
      "step": 844
    },
    {
      "epoch": 3.691970802919708,
      "grad_norm": 0.013032430782914162,
      "learning_rate": 1.8211008085517348e-06,
      "loss": 0.0,
      "step": 845
    },
    {
      "epoch": 3.6963503649635037,
      "grad_norm": 0.0017879578517749906,
      "learning_rate": 1.815141666814405e-06,
      "loss": 0.0,
      "step": 846
    },
    {
      "epoch": 3.7007299270072993,
      "grad_norm": 0.0024302792735397816,
      "learning_rate": 1.809186729067911e-06,
      "loss": 0.0,
      "step": 847
    },
    {
      "epoch": 3.705109489051095,
      "grad_norm": 0.0009780308464542031,
      "learning_rate": 1.8032360318665337e-06,
      "loss": 0.0,
      "step": 848
    },
    {
      "epoch": 3.7094890510948906,
      "grad_norm": 0.19592009484767914,
      "learning_rate": 1.797289611738523e-06,
      "loss": 0.0019,
      "step": 849
    },
    {
      "epoch": 3.713868613138686,
      "grad_norm": 0.002372670453041792,
      "learning_rate": 1.7913475051858746e-06,
      "loss": 0.0,
      "step": 850
    },
    {
      "epoch": 3.718248175182482,
      "grad_norm": 0.002017020247876644,
      "learning_rate": 1.7854097486841043e-06,
      "loss": 0.0,
      "step": 851
    },
    {
      "epoch": 3.7226277372262775,
      "grad_norm": 0.0022830108646303415,
      "learning_rate": 1.7794763786820268e-06,
      "loss": 0.0,
      "step": 852
    },
    {
      "epoch": 3.727007299270073,
      "grad_norm": 0.002295683603733778,
      "learning_rate": 1.7735474316015294e-06,
      "loss": 0.0,
      "step": 853
    },
    {
      "epoch": 3.7313868613138688,
      "grad_norm": 0.001965237082913518,
      "learning_rate": 1.767622943837349e-06,
      "loss": 0.0,
      "step": 854
    },
    {
      "epoch": 3.7357664233576644,
      "grad_norm": 0.0012728808214887977,
      "learning_rate": 1.7617029517568502e-06,
      "loss": 0.0,
      "step": 855
    },
    {
      "epoch": 3.74014598540146,
      "grad_norm": 0.0020083924755454063,
      "learning_rate": 1.7557874916997996e-06,
      "loss": 0.0,
      "step": 856
    },
    {
      "epoch": 3.7445255474452557,
      "grad_norm": 0.0020820496138185263,
      "learning_rate": 1.7498765999781455e-06,
      "loss": 0.0,
      "step": 857
    },
    {
      "epoch": 3.7489051094890513,
      "grad_norm": 0.0020251362584531307,
      "learning_rate": 1.7439703128757923e-06,
      "loss": 0.0,
      "step": 858
    },
    {
      "epoch": 3.753284671532847,
      "grad_norm": 0.0015859343111515045,
      "learning_rate": 1.7380686666483793e-06,
      "loss": 0.0,
      "step": 859
    },
    {
      "epoch": 3.7576642335766426,
      "grad_norm": 0.001591153210029006,
      "learning_rate": 1.7321716975230588e-06,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 3.7620437956204382,
      "grad_norm": 0.0018416156526654959,
      "learning_rate": 1.7262794416982717e-06,
      "loss": 0.0,
      "step": 861
    },
    {
      "epoch": 3.7664233576642334,
      "grad_norm": 0.0017293720738962293,
      "learning_rate": 1.7203919353435269e-06,
      "loss": 0.0,
      "step": 862
    },
    {
      "epoch": 3.770802919708029,
      "grad_norm": 0.0010803790064528584,
      "learning_rate": 1.7145092145991786e-06,
      "loss": 0.0,
      "step": 863
    },
    {
      "epoch": 3.7751824817518247,
      "grad_norm": 0.002193450927734375,
      "learning_rate": 1.7086313155762046e-06,
      "loss": 0.0,
      "step": 864
    },
    {
      "epoch": 3.7795620437956203,
      "grad_norm": 0.11664100736379623,
      "learning_rate": 1.7027582743559845e-06,
      "loss": 0.0008,
      "step": 865
    },
    {
      "epoch": 3.783941605839416,
      "grad_norm": 0.0010278144618496299,
      "learning_rate": 1.696890126990079e-06,
      "loss": 0.0,
      "step": 866
    },
    {
      "epoch": 3.7883211678832116,
      "grad_norm": 0.0027571087703108788,
      "learning_rate": 1.691026909500007e-06,
      "loss": 0.0,
      "step": 867
    },
    {
      "epoch": 3.7927007299270072,
      "grad_norm": 0.001730642979964614,
      "learning_rate": 1.6851686578770263e-06,
      "loss": 0.0,
      "step": 868
    },
    {
      "epoch": 3.797080291970803,
      "grad_norm": 0.0017755437875166535,
      "learning_rate": 1.6793154080819112e-06,
      "loss": 0.0,
      "step": 869
    },
    {
      "epoch": 3.8014598540145985,
      "grad_norm": 0.0018059754511341453,
      "learning_rate": 1.6734671960447335e-06,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 3.805839416058394,
      "grad_norm": 0.0010772132081910968,
      "learning_rate": 1.6676240576646389e-06,
      "loss": 0.0,
      "step": 871
    },
    {
      "epoch": 3.81021897810219,
      "grad_norm": 0.0013907999964430928,
      "learning_rate": 1.66178602880963e-06,
      "loss": 0.0,
      "step": 872
    },
    {
      "epoch": 3.8145985401459854,
      "grad_norm": 0.0017574954545125365,
      "learning_rate": 1.655953145316344e-06,
      "loss": 0.0,
      "step": 873
    },
    {
      "epoch": 3.818978102189781,
      "grad_norm": 0.0023797417525202036,
      "learning_rate": 1.6501254429898345e-06,
      "loss": 0.0,
      "step": 874
    },
    {
      "epoch": 3.8233576642335767,
      "grad_norm": 0.0010705847525969148,
      "learning_rate": 1.6443029576033493e-06,
      "loss": 0.0,
      "step": 875
    },
    {
      "epoch": 3.8277372262773723,
      "grad_norm": 0.0013967688428238034,
      "learning_rate": 1.6384857248981117e-06,
      "loss": 0.0,
      "step": 876
    },
    {
      "epoch": 3.832116788321168,
      "grad_norm": 0.0027486851904541254,
      "learning_rate": 1.6326737805831039e-06,
      "loss": 0.0,
      "step": 877
    },
    {
      "epoch": 3.8364963503649636,
      "grad_norm": 0.0009487051866017282,
      "learning_rate": 1.6268671603348428e-06,
      "loss": 0.0,
      "step": 878
    },
    {
      "epoch": 3.8408759124087593,
      "grad_norm": 0.0026726871728897095,
      "learning_rate": 1.621065899797165e-06,
      "loss": 0.0,
      "step": 879
    },
    {
      "epoch": 3.845255474452555,
      "grad_norm": 0.0028086488600820303,
      "learning_rate": 1.6152700345810063e-06,
      "loss": 0.0,
      "step": 880
    },
    {
      "epoch": 3.8496350364963505,
      "grad_norm": 0.005458397325128317,
      "learning_rate": 1.6094796002641836e-06,
      "loss": 0.0,
      "step": 881
    },
    {
      "epoch": 3.8540145985401457,
      "grad_norm": 0.0014935546787455678,
      "learning_rate": 1.6036946323911753e-06,
      "loss": 0.0,
      "step": 882
    },
    {
      "epoch": 3.8583941605839414,
      "grad_norm": 0.002109678229317069,
      "learning_rate": 1.5979151664729063e-06,
      "loss": 0.0,
      "step": 883
    },
    {
      "epoch": 3.862773722627737,
      "grad_norm": 0.0017332076095044613,
      "learning_rate": 1.5921412379865259e-06,
      "loss": 0.0,
      "step": 884
    },
    {
      "epoch": 3.8671532846715326,
      "grad_norm": 0.009105571545660496,
      "learning_rate": 1.5863728823751922e-06,
      "loss": 0.0,
      "step": 885
    },
    {
      "epoch": 3.8715328467153283,
      "grad_norm": 0.0020738227758556604,
      "learning_rate": 1.5806101350478552e-06,
      "loss": 0.0,
      "step": 886
    },
    {
      "epoch": 3.875912408759124,
      "grad_norm": 0.0010305409086868167,
      "learning_rate": 1.5748530313790379e-06,
      "loss": 0.0,
      "step": 887
    },
    {
      "epoch": 3.8802919708029195,
      "grad_norm": 0.002297050319612026,
      "learning_rate": 1.5691016067086198e-06,
      "loss": 0.0,
      "step": 888
    },
    {
      "epoch": 3.884671532846715,
      "grad_norm": 0.002647539833560586,
      "learning_rate": 1.5633558963416203e-06,
      "loss": 0.0,
      "step": 889
    },
    {
      "epoch": 3.889051094890511,
      "grad_norm": 0.0012281707022339106,
      "learning_rate": 1.5576159355479814e-06,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 3.8934306569343065,
      "grad_norm": 0.0018704022513702512,
      "learning_rate": 1.5518817595623514e-06,
      "loss": 0.0,
      "step": 891
    },
    {
      "epoch": 3.897810218978102,
      "grad_norm": 0.001662683323957026,
      "learning_rate": 1.546153403583869e-06,
      "loss": 0.0,
      "step": 892
    },
    {
      "epoch": 3.9021897810218977,
      "grad_norm": 0.0023420508950948715,
      "learning_rate": 1.540430902775946e-06,
      "loss": 0.0,
      "step": 893
    },
    {
      "epoch": 3.9065693430656934,
      "grad_norm": 0.0019655104260891676,
      "learning_rate": 1.534714292266054e-06,
      "loss": 0.0,
      "step": 894
    },
    {
      "epoch": 3.910948905109489,
      "grad_norm": 0.0009503703331574798,
      "learning_rate": 1.5290036071455056e-06,
      "loss": 0.0,
      "step": 895
    },
    {
      "epoch": 3.9153284671532846,
      "grad_norm": 0.04980412498116493,
      "learning_rate": 1.5232988824692406e-06,
      "loss": 0.0,
      "step": 896
    },
    {
      "epoch": 3.9197080291970803,
      "grad_norm": 0.0007013222202658653,
      "learning_rate": 1.5176001532556118e-06,
      "loss": 0.0,
      "step": 897
    },
    {
      "epoch": 3.924087591240876,
      "grad_norm": 0.0010505859972909093,
      "learning_rate": 1.511907454486168e-06,
      "loss": 0.0,
      "step": 898
    },
    {
      "epoch": 3.9284671532846716,
      "grad_norm": 0.0009482012246735394,
      "learning_rate": 1.5062208211054398e-06,
      "loss": 0.0,
      "step": 899
    },
    {
      "epoch": 3.932846715328467,
      "grad_norm": 0.004321451764553785,
      "learning_rate": 1.5005402880207272e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 3.937226277372263,
      "grad_norm": 0.00210862560197711,
      "learning_rate": 1.4948658901018826e-06,
      "loss": 0.0,
      "step": 901
    },
    {
      "epoch": 3.9416058394160585,
      "grad_norm": 0.00250151171348989,
      "learning_rate": 1.4891976621810972e-06,
      "loss": 0.0,
      "step": 902
    },
    {
      "epoch": 3.945985401459854,
      "grad_norm": 0.0026350142434239388,
      "learning_rate": 1.483535639052689e-06,
      "loss": 0.0,
      "step": 903
    },
    {
      "epoch": 3.9503649635036497,
      "grad_norm": 0.004716258030384779,
      "learning_rate": 1.4778798554728866e-06,
      "loss": 0.0,
      "step": 904
    },
    {
      "epoch": 3.9547445255474454,
      "grad_norm": 0.0018248335691168904,
      "learning_rate": 1.4722303461596192e-06,
      "loss": 0.0,
      "step": 905
    },
    {
      "epoch": 3.959124087591241,
      "grad_norm": 0.013006918132305145,
      "learning_rate": 1.4665871457922997e-06,
      "loss": 0.0,
      "step": 906
    },
    {
      "epoch": 3.9635036496350367,
      "grad_norm": 0.0023916151840239763,
      "learning_rate": 1.4609502890116146e-06,
      "loss": 0.0,
      "step": 907
    },
    {
      "epoch": 3.9678832116788323,
      "grad_norm": 0.001576004782691598,
      "learning_rate": 1.4553198104193094e-06,
      "loss": 0.0,
      "step": 908
    },
    {
      "epoch": 3.972262773722628,
      "grad_norm": 0.0026557703968137503,
      "learning_rate": 1.4496957445779792e-06,
      "loss": 0.0,
      "step": 909
    },
    {
      "epoch": 3.9766423357664236,
      "grad_norm": 0.002321658656001091,
      "learning_rate": 1.4440781260108521e-06,
      "loss": 0.0,
      "step": 910
    },
    {
      "epoch": 3.981021897810219,
      "grad_norm": 0.0010188892483711243,
      "learning_rate": 1.438466989201583e-06,
      "loss": 0.0,
      "step": 911
    },
    {
      "epoch": 3.985401459854015,
      "grad_norm": 0.0009046494378708303,
      "learning_rate": 1.4328623685940335e-06,
      "loss": 0.0,
      "step": 912
    }
  ],
  "logging_steps": 1,
  "max_steps": 1368,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 228,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.2978513818196378e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}