{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 230,
  "global_step": 918,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010893246187363835,
      "grad_norm": 497.369384765625,
      "learning_rate": 2e-05,
      "loss": 12.1306,
      "step": 1
    },
    {
      "epoch": 0.0010893246187363835,
      "eval_loss": 3.1589934825897217,
      "eval_runtime": 4.1717,
      "eval_samples_per_second": 92.767,
      "eval_steps_per_second": 46.503,
      "step": 1
    },
    {
      "epoch": 0.002178649237472767,
      "grad_norm": 384.06939697265625,
      "learning_rate": 4e-05,
      "loss": 11.9955,
      "step": 2
    },
    {
      "epoch": 0.0032679738562091504,
      "grad_norm": 418.12237548828125,
      "learning_rate": 6e-05,
      "loss": 11.2344,
      "step": 3
    },
    {
      "epoch": 0.004357298474945534,
      "grad_norm": 319.05267333984375,
      "learning_rate": 8e-05,
      "loss": 13.0403,
      "step": 4
    },
    {
      "epoch": 0.0054466230936819175,
      "grad_norm": 275.94830322265625,
      "learning_rate": 0.0001,
      "loss": 12.7821,
      "step": 5
    },
    {
      "epoch": 0.006535947712418301,
      "grad_norm": 446.70166015625,
      "learning_rate": 0.00012,
      "loss": 12.3682,
      "step": 6
    },
    {
      "epoch": 0.007625272331154684,
      "grad_norm": 403.9425964355469,
      "learning_rate": 0.00014,
      "loss": 12.164,
      "step": 7
    },
    {
      "epoch": 0.008714596949891068,
      "grad_norm": 370.84893798828125,
      "learning_rate": 0.00016,
      "loss": 12.4891,
      "step": 8
    },
    {
      "epoch": 0.00980392156862745,
      "grad_norm": 357.8050537109375,
      "learning_rate": 0.00018,
      "loss": 12.3877,
      "step": 9
    },
    {
      "epoch": 0.010893246187363835,
      "grad_norm": 409.2673645019531,
      "learning_rate": 0.0002,
      "loss": 12.547,
      "step": 10
    },
    {
      "epoch": 0.011982570806100218,
      "grad_norm": 395.81939697265625,
      "learning_rate": 0.00019999940145388063,
      "loss": 13.2017,
      "step": 11
    },
    {
      "epoch": 0.013071895424836602,
      "grad_norm": 440.0821228027344,
      "learning_rate": 0.00019999760582268763,
      "loss": 12.3833,
      "step": 12
    },
    {
      "epoch": 0.014161220043572984,
      "grad_norm": 394.557373046875,
      "learning_rate": 0.00019999461312791638,
      "loss": 11.9402,
      "step": 13
    },
    {
      "epoch": 0.015250544662309368,
      "grad_norm": 332.89373779296875,
      "learning_rate": 0.0001999904234053922,
      "loss": 11.9442,
      "step": 14
    },
    {
      "epoch": 0.016339869281045753,
      "grad_norm": 312.7531433105469,
      "learning_rate": 0.00019998503670526994,
      "loss": 11.8069,
      "step": 15
    },
    {
      "epoch": 0.017429193899782137,
      "grad_norm": 472.9444580078125,
      "learning_rate": 0.00019997845309203334,
      "loss": 12.1533,
      "step": 16
    },
    {
      "epoch": 0.018518518518518517,
      "grad_norm": 397.2251892089844,
      "learning_rate": 0.00019997067264449433,
      "loss": 11.0714,
      "step": 17
    },
    {
      "epoch": 0.0196078431372549,
      "grad_norm": 369.2944641113281,
      "learning_rate": 0.00019996169545579207,
      "loss": 12.2346,
      "step": 18
    },
    {
      "epoch": 0.020697167755991286,
      "grad_norm": 574.8084106445312,
      "learning_rate": 0.00019995152163339178,
      "loss": 11.162,
      "step": 19
    },
    {
      "epoch": 0.02178649237472767,
      "grad_norm": 611.6285400390625,
      "learning_rate": 0.00019994015129908346,
      "loss": 12.0132,
      "step": 20
    },
    {
      "epoch": 0.02287581699346405,
      "grad_norm": 425.55072021484375,
      "learning_rate": 0.00019992758458898055,
      "loss": 12.0108,
      "step": 21
    },
    {
      "epoch": 0.023965141612200435,
      "grad_norm": 498.50439453125,
      "learning_rate": 0.00019991382165351814,
      "loss": 11.8223,
      "step": 22
    },
    {
      "epoch": 0.02505446623093682,
      "grad_norm": 454.2897644042969,
      "learning_rate": 0.00019989886265745128,
      "loss": 12.0506,
      "step": 23
    },
    {
      "epoch": 0.026143790849673203,
      "grad_norm": 557.8632202148438,
      "learning_rate": 0.00019988270777985292,
      "loss": 11.8529,
      "step": 24
    },
    {
      "epoch": 0.027233115468409588,
      "grad_norm": 412.888916015625,
      "learning_rate": 0.00019986535721411186,
      "loss": 11.4545,
      "step": 25
    },
    {
      "epoch": 0.02832244008714597,
      "grad_norm": 375.8723449707031,
      "learning_rate": 0.00019984681116793038,
      "loss": 10.8877,
      "step": 26
    },
    {
      "epoch": 0.029411764705882353,
      "grad_norm": 445.2213134765625,
      "learning_rate": 0.00019982706986332175,
      "loss": 11.8088,
      "step": 27
    },
    {
      "epoch": 0.030501089324618737,
      "grad_norm": 390.9215087890625,
      "learning_rate": 0.00019980613353660763,
      "loss": 12.8907,
      "step": 28
    },
    {
      "epoch": 0.03159041394335512,
      "grad_norm": 876.1612548828125,
      "learning_rate": 0.00019978400243841508,
      "loss": 12.3549,
      "step": 29
    },
    {
      "epoch": 0.032679738562091505,
      "grad_norm": 573.1425170898438,
      "learning_rate": 0.00019976067683367385,
      "loss": 12.5321,
      "step": 30
    },
    {
      "epoch": 0.03376906318082789,
      "grad_norm": 470.3711853027344,
      "learning_rate": 0.0001997361570016129,
      "loss": 12.0172,
      "step": 31
    },
    {
      "epoch": 0.034858387799564274,
      "grad_norm": 679.3441772460938,
      "learning_rate": 0.00019971044323575728,
      "loss": 10.8628,
      "step": 32
    },
    {
      "epoch": 0.03594771241830065,
      "grad_norm": 626.5648193359375,
      "learning_rate": 0.0001996835358439244,
      "loss": 11.3385,
      "step": 33
    },
    {
      "epoch": 0.037037037037037035,
      "grad_norm": 733.734375,
      "learning_rate": 0.00019965543514822062,
      "loss": 10.4336,
      "step": 34
    },
    {
      "epoch": 0.03812636165577342,
      "grad_norm": 587.365966796875,
      "learning_rate": 0.00019962614148503718,
      "loss": 12.0594,
      "step": 35
    },
    {
      "epoch": 0.0392156862745098,
      "grad_norm": 405.00531005859375,
      "learning_rate": 0.00019959565520504623,
      "loss": 11.8103,
      "step": 36
    },
    {
      "epoch": 0.04030501089324619,
      "grad_norm": 398.32415771484375,
      "learning_rate": 0.00019956397667319668,
      "loss": 11.0375,
      "step": 37
    },
    {
      "epoch": 0.04139433551198257,
      "grad_norm": 422.72808837890625,
      "learning_rate": 0.00019953110626870979,
      "loss": 12.0295,
      "step": 38
    },
    {
      "epoch": 0.042483660130718956,
      "grad_norm": 482.6229553222656,
      "learning_rate": 0.00019949704438507459,
      "loss": 10.7632,
      "step": 39
    },
    {
      "epoch": 0.04357298474945534,
      "grad_norm": 526.7586059570312,
      "learning_rate": 0.00019946179143004325,
      "loss": 11.2995,
      "step": 40
    },
    {
      "epoch": 0.044662309368191724,
      "grad_norm": 503.21221923828125,
      "learning_rate": 0.0001994253478256262,
      "loss": 12.1067,
      "step": 41
    },
    {
      "epoch": 0.0457516339869281,
      "grad_norm": 655.2205810546875,
      "learning_rate": 0.0001993877140080869,
      "loss": 11.5295,
      "step": 42
    },
    {
      "epoch": 0.046840958605664486,
      "grad_norm": 440.0908203125,
      "learning_rate": 0.000199348890427937,
      "loss": 11.6821,
      "step": 43
    },
    {
      "epoch": 0.04793028322440087,
      "grad_norm": 456.47406005859375,
      "learning_rate": 0.00019930887754993044,
      "loss": 11.7873,
      "step": 44
    },
    {
      "epoch": 0.049019607843137254,
      "grad_norm": 841.0929565429688,
      "learning_rate": 0.00019926767585305835,
      "loss": 11.6674,
      "step": 45
    },
    {
      "epoch": 0.05010893246187364,
      "grad_norm": 581.7565307617188,
      "learning_rate": 0.000199225285830543,
      "loss": 11.7123,
      "step": 46
    },
    {
      "epoch": 0.05119825708061002,
      "grad_norm": 547.7841796875,
      "learning_rate": 0.00019918170798983211,
      "loss": 10.9854,
      "step": 47
    },
    {
      "epoch": 0.05228758169934641,
      "grad_norm": 685.59814453125,
      "learning_rate": 0.00019913694285259256,
      "loss": 11.8064,
      "step": 48
    },
    {
      "epoch": 0.05337690631808279,
      "grad_norm": 636.0125122070312,
      "learning_rate": 0.00019909099095470444,
      "loss": 11.1934,
      "step": 49
    },
    {
      "epoch": 0.054466230936819175,
      "grad_norm": 528.2201538085938,
      "learning_rate": 0.00019904385284625424,
      "loss": 12.0213,
      "step": 50
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 490.21875,
      "learning_rate": 0.00019899552909152866,
      "loss": 10.6632,
      "step": 51
    },
    {
      "epoch": 0.05664488017429194,
      "grad_norm": 500.16021728515625,
      "learning_rate": 0.00019894602026900758,
      "loss": 11.65,
      "step": 52
    },
    {
      "epoch": 0.05773420479302832,
      "grad_norm": 477.1259765625,
      "learning_rate": 0.00019889532697135734,
      "loss": 11.6934,
      "step": 53
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 1109.7750244140625,
      "learning_rate": 0.00019884344980542338,
      "loss": 10.7624,
      "step": 54
    },
    {
      "epoch": 0.05991285403050109,
      "grad_norm": 555.9796142578125,
      "learning_rate": 0.00019879038939222329,
      "loss": 12.7329,
      "step": 55
    },
    {
      "epoch": 0.06100217864923747,
      "grad_norm": 557.53271484375,
      "learning_rate": 0.0001987361463669392,
      "loss": 11.0998,
      "step": 56
    },
    {
      "epoch": 0.06209150326797386,
      "grad_norm": 483.9259948730469,
      "learning_rate": 0.00019868072137891002,
      "loss": 11.8222,
      "step": 57
    },
    {
      "epoch": 0.06318082788671024,
      "grad_norm": 740.6929321289062,
      "learning_rate": 0.00019862411509162406,
      "loss": 10.6816,
      "step": 58
    },
    {
      "epoch": 0.06427015250544663,
      "grad_norm": 873.5588989257812,
      "learning_rate": 0.0001985663281827108,
      "loss": 10.5937,
      "step": 59
    },
    {
      "epoch": 0.06535947712418301,
      "grad_norm": 617.0968627929688,
      "learning_rate": 0.00019850736134393286,
      "loss": 11.6752,
      "step": 60
    },
    {
      "epoch": 0.0664488017429194,
      "grad_norm": 652.836669921875,
      "learning_rate": 0.00019844721528117766,
      "loss": 10.4189,
      "step": 61
    },
    {
      "epoch": 0.06753812636165578,
      "grad_norm": 681.41064453125,
      "learning_rate": 0.00019838589071444903,
      "loss": 11.4517,
      "step": 62
    },
    {
      "epoch": 0.06862745098039216,
      "grad_norm": 610.8583984375,
      "learning_rate": 0.00019832338837785863,
      "loss": 11.4514,
      "step": 63
    },
    {
      "epoch": 0.06971677559912855,
      "grad_norm": 703.3029174804688,
      "learning_rate": 0.00019825970901961705,
      "loss": 11.0479,
      "step": 64
    },
    {
      "epoch": 0.07080610021786492,
      "grad_norm": 558.1321411132812,
      "learning_rate": 0.000198194853402025,
      "loss": 10.8034,
      "step": 65
    },
    {
      "epoch": 0.0718954248366013,
      "grad_norm": 729.005126953125,
      "learning_rate": 0.00019812882230146398,
      "loss": 11.2189,
      "step": 66
    },
    {
      "epoch": 0.07298474945533769,
      "grad_norm": 563.0946044921875,
      "learning_rate": 0.00019806161650838723,
      "loss": 11.8089,
      "step": 67
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 774.91796875,
      "learning_rate": 0.00019799323682731,
      "loss": 10.8409,
      "step": 68
    },
    {
      "epoch": 0.07516339869281045,
      "grad_norm": 636.3812866210938,
      "learning_rate": 0.00019792368407680025,
      "loss": 11.4898,
      "step": 69
    },
    {
      "epoch": 0.07625272331154684,
      "grad_norm": 741.6439208984375,
      "learning_rate": 0.00019785295908946848,
      "loss": 11.1101,
      "step": 70
    },
    {
      "epoch": 0.07734204793028322,
      "grad_norm": 902.0681762695312,
      "learning_rate": 0.00019778106271195806,
      "loss": 10.9148,
      "step": 71
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 608.00634765625,
      "learning_rate": 0.00019770799580493494,
      "loss": 12.5593,
      "step": 72
    },
    {
      "epoch": 0.07952069716775599,
      "grad_norm": 645.3396606445312,
      "learning_rate": 0.00019763375924307735,
      "loss": 11.0296,
      "step": 73
    },
    {
      "epoch": 0.08061002178649238,
      "grad_norm": 532.682861328125,
      "learning_rate": 0.0001975583539150655,
      "loss": 10.9512,
      "step": 74
    },
    {
      "epoch": 0.08169934640522876,
      "grad_norm": 606.407470703125,
      "learning_rate": 0.00019748178072357065,
      "loss": 10.0461,
      "step": 75
    },
    {
      "epoch": 0.08278867102396514,
      "grad_norm": 538.4356079101562,
      "learning_rate": 0.00019740404058524457,
      "loss": 10.923,
      "step": 76
    },
    {
      "epoch": 0.08387799564270153,
      "grad_norm": 593.4065551757812,
      "learning_rate": 0.00019732513443070836,
      "loss": 11.2116,
      "step": 77
    },
    {
      "epoch": 0.08496732026143791,
      "grad_norm": 667.9852905273438,
      "learning_rate": 0.00019724506320454153,
      "loss": 10.7769,
      "step": 78
    },
    {
      "epoch": 0.0860566448801743,
      "grad_norm": 667.6882934570312,
      "learning_rate": 0.0001971638278652705,
      "loss": 11.6724,
      "step": 79
    },
    {
      "epoch": 0.08714596949891068,
      "grad_norm": 664.25244140625,
      "learning_rate": 0.0001970814293853572,
      "loss": 11.0462,
      "step": 80
    },
    {
      "epoch": 0.08823529411764706,
      "grad_norm": 713.4217529296875,
      "learning_rate": 0.00019699786875118747,
      "loss": 11.3038,
      "step": 81
    },
    {
      "epoch": 0.08932461873638345,
      "grad_norm": 636.1452026367188,
      "learning_rate": 0.00019691314696305913,
      "loss": 11.0193,
      "step": 82
    },
    {
      "epoch": 0.09041394335511982,
      "grad_norm": 536.7321166992188,
      "learning_rate": 0.00019682726503517017,
      "loss": 11.3826,
      "step": 83
    },
    {
      "epoch": 0.0915032679738562,
      "grad_norm": 599.102783203125,
      "learning_rate": 0.00019674022399560648,
      "loss": 10.964,
      "step": 84
    },
    {
      "epoch": 0.09259259259259259,
      "grad_norm": 1010.7560424804688,
      "learning_rate": 0.00019665202488632956,
      "loss": 11.0961,
      "step": 85
    },
    {
      "epoch": 0.09368191721132897,
      "grad_norm": 651.3975219726562,
      "learning_rate": 0.0001965626687631641,
      "loss": 10.7751,
      "step": 86
    },
    {
      "epoch": 0.09477124183006536,
      "grad_norm": 691.282958984375,
      "learning_rate": 0.00019647215669578536,
      "loss": 11.463,
      "step": 87
    },
    {
      "epoch": 0.09586056644880174,
      "grad_norm": 780.7763061523438,
      "learning_rate": 0.00019638048976770628,
      "loss": 11.4905,
      "step": 88
    },
    {
      "epoch": 0.09694989106753812,
      "grad_norm": 731.2135620117188,
      "learning_rate": 0.00019628766907626446,
      "loss": 11.0423,
      "step": 89
    },
    {
      "epoch": 0.09803921568627451,
      "grad_norm": 606.6620483398438,
      "learning_rate": 0.00019619369573260924,
      "loss": 10.0837,
      "step": 90
    },
    {
      "epoch": 0.09912854030501089,
      "grad_norm": 954.3007202148438,
      "learning_rate": 0.00019609857086168823,
      "loss": 10.4564,
      "step": 91
    },
    {
      "epoch": 0.10021786492374728,
      "grad_norm": 489.27496337890625,
      "learning_rate": 0.00019600229560223388,
      "loss": 11.548,
      "step": 92
    },
    {
      "epoch": 0.10130718954248366,
      "grad_norm": 563.7119140625,
      "learning_rate": 0.00019590487110674983,
      "loss": 10.8069,
      "step": 93
    },
    {
      "epoch": 0.10239651416122005,
      "grad_norm": 460.32440185546875,
      "learning_rate": 0.0001958062985414972,
      "loss": 10.7325,
      "step": 94
    },
    {
      "epoch": 0.10348583877995643,
      "grad_norm": 642.6327514648438,
      "learning_rate": 0.00019570657908648048,
      "loss": 12.1817,
      "step": 95
    },
    {
      "epoch": 0.10457516339869281,
      "grad_norm": 686.3006591796875,
      "learning_rate": 0.0001956057139354335,
      "loss": 10.7484,
      "step": 96
    },
    {
      "epoch": 0.1056644880174292,
      "grad_norm": 1625.68359375,
      "learning_rate": 0.0001955037042958052,
      "loss": 9.9405,
      "step": 97
    },
    {
      "epoch": 0.10675381263616558,
      "grad_norm": 661.0103759765625,
      "learning_rate": 0.00019540055138874505,
      "loss": 10.6417,
      "step": 98
    },
    {
      "epoch": 0.10784313725490197,
      "grad_norm": 754.760009765625,
      "learning_rate": 0.00019529625644908847,
      "loss": 10.5583,
      "step": 99
    },
    {
      "epoch": 0.10893246187363835,
      "grad_norm": 638.9000854492188,
      "learning_rate": 0.0001951908207253421,
      "loss": 11.0231,
      "step": 100
    },
    {
      "epoch": 0.11002178649237472,
      "grad_norm": 553.3854370117188,
      "learning_rate": 0.00019508424547966884,
      "loss": 10.9961,
      "step": 101
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 587.3681030273438,
      "learning_rate": 0.00019497653198787264,
      "loss": 10.701,
      "step": 102
    },
    {
      "epoch": 0.11220043572984749,
      "grad_norm": 751.8494262695312,
      "learning_rate": 0.00019486768153938338,
      "loss": 11.4205,
      "step": 103
    },
    {
      "epoch": 0.11328976034858387,
      "grad_norm": 620.5661010742188,
      "learning_rate": 0.0001947576954372413,
      "loss": 11.6781,
      "step": 104
    },
    {
      "epoch": 0.11437908496732026,
      "grad_norm": 592.9255981445312,
      "learning_rate": 0.00019464657499808152,
      "loss": 10.2305,
      "step": 105
    },
    {
      "epoch": 0.11546840958605664,
      "grad_norm": 504.8605651855469,
      "learning_rate": 0.0001945343215521182,
      "loss": 12.4109,
      "step": 106
    },
    {
      "epoch": 0.11655773420479303,
      "grad_norm": 567.1307983398438,
      "learning_rate": 0.0001944209364431286,
      "loss": 10.948,
      "step": 107
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 802.73046875,
      "learning_rate": 0.00019430642102843707,
      "loss": 11.1612,
      "step": 108
    },
    {
      "epoch": 0.1187363834422658,
      "grad_norm": 692.0497436523438,
      "learning_rate": 0.00019419077667889872,
      "loss": 10.6958,
      "step": 109
    },
    {
      "epoch": 0.11982570806100218,
      "grad_norm": 678.1286010742188,
      "learning_rate": 0.00019407400477888315,
      "loss": 11.3934,
      "step": 110
    },
    {
      "epoch": 0.12091503267973856,
      "grad_norm": 545.8530883789062,
      "learning_rate": 0.00019395610672625767,
      "loss": 11.7126,
      "step": 111
    },
    {
      "epoch": 0.12200435729847495,
      "grad_norm": 551.471435546875,
      "learning_rate": 0.00019383708393237075,
      "loss": 11.3177,
      "step": 112
    },
    {
      "epoch": 0.12309368191721133,
      "grad_norm": 747.8993530273438,
      "learning_rate": 0.00019371693782203498,
      "loss": 10.9254,
      "step": 113
    },
    {
      "epoch": 0.12418300653594772,
      "grad_norm": 814.4208374023438,
      "learning_rate": 0.00019359566983351013,
      "loss": 9.821,
      "step": 114
    },
    {
      "epoch": 0.12527233115468409,
      "grad_norm": 862.31494140625,
      "learning_rate": 0.0001934732814184859,
      "loss": 11.2622,
      "step": 115
    },
    {
      "epoch": 0.12636165577342048,
      "grad_norm": 636.8887939453125,
      "learning_rate": 0.00019334977404206443,
      "loss": 10.6286,
      "step": 116
    },
    {
      "epoch": 0.12745098039215685,
      "grad_norm": 670.830322265625,
      "learning_rate": 0.00019322514918274308,
      "loss": 11.1403,
      "step": 117
    },
    {
      "epoch": 0.12854030501089325,
      "grad_norm": 545.7217407226562,
      "learning_rate": 0.00019309940833239626,
      "loss": 11.752,
      "step": 118
    },
    {
      "epoch": 0.12962962962962962,
      "grad_norm": 587.8623046875,
      "learning_rate": 0.00019297255299625797,
      "loss": 11.7527,
      "step": 119
    },
    {
      "epoch": 0.13071895424836602,
      "grad_norm": 1006.7312622070312,
      "learning_rate": 0.00019284458469290354,
      "loss": 10.5548,
      "step": 120
    },
    {
      "epoch": 0.1318082788671024,
      "grad_norm": 590.1889038085938,
      "learning_rate": 0.00019271550495423168,
      "loss": 10.2359,
      "step": 121
    },
    {
      "epoch": 0.1328976034858388,
      "grad_norm": 605.3558959960938,
      "learning_rate": 0.00019258531532544585,
      "loss": 10.7942,
      "step": 122
    },
    {
      "epoch": 0.13398692810457516,
      "grad_norm": 605.3388061523438,
      "learning_rate": 0.00019245401736503608,
      "loss": 10.9458,
      "step": 123
    },
    {
      "epoch": 0.13507625272331156,
      "grad_norm": 672.321533203125,
      "learning_rate": 0.00019232161264475997,
      "loss": 12.0993,
      "step": 124
    },
    {
      "epoch": 0.13616557734204793,
      "grad_norm": 554.8330688476562,
      "learning_rate": 0.00019218810274962417,
      "loss": 11.2272,
      "step": 125
    },
    {
      "epoch": 0.13725490196078433,
      "grad_norm": 865.9181518554688,
      "learning_rate": 0.00019205348927786532,
      "loss": 10.1409,
      "step": 126
    },
    {
      "epoch": 0.1383442265795207,
      "grad_norm": 788.566162109375,
      "learning_rate": 0.00019191777384093081,
      "loss": 12.0527,
      "step": 127
    },
    {
      "epoch": 0.1394335511982571,
      "grad_norm": 699.4249267578125,
      "learning_rate": 0.0001917809580634596,
      "loss": 10.8955,
      "step": 128
    },
    {
      "epoch": 0.14052287581699346,
      "grad_norm": 726.8323974609375,
      "learning_rate": 0.00019164304358326275,
      "loss": 10.264,
      "step": 129
    },
    {
      "epoch": 0.14161220043572983,
      "grad_norm": 878.0653686523438,
      "learning_rate": 0.00019150403205130383,
      "loss": 11.0059,
      "step": 130
    },
    {
      "epoch": 0.14270152505446623,
      "grad_norm": 523.306884765625,
      "learning_rate": 0.00019136392513167903,
      "loss": 10.8748,
      "step": 131
    },
    {
      "epoch": 0.1437908496732026,
      "grad_norm": 962.2786254882812,
      "learning_rate": 0.00019122272450159745,
      "loss": 11.4164,
      "step": 132
    },
    {
      "epoch": 0.144880174291939,
      "grad_norm": 911.98193359375,
      "learning_rate": 0.0001910804318513609,
      "loss": 11.5345,
      "step": 133
    },
    {
      "epoch": 0.14596949891067537,
      "grad_norm": 1103.3953857421875,
      "learning_rate": 0.0001909370488843436,
      "loss": 10.3984,
      "step": 134
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 617.2359008789062,
      "learning_rate": 0.00019079257731697196,
      "loss": 11.3573,
      "step": 135
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 704.6926879882812,
      "learning_rate": 0.0001906470188787039,
      "loss": 11.3981,
      "step": 136
    },
    {
      "epoch": 0.14923747276688454,
      "grad_norm": 702.8963012695312,
      "learning_rate": 0.00019050037531200814,
      "loss": 11.738,
      "step": 137
    },
    {
      "epoch": 0.1503267973856209,
      "grad_norm": 848.1795043945312,
      "learning_rate": 0.00019035264837234347,
      "loss": 11.4592,
      "step": 138
    },
    {
      "epoch": 0.1514161220043573,
      "grad_norm": 863.1195068359375,
      "learning_rate": 0.00019020383982813765,
      "loss": 11.1281,
      "step": 139
    },
    {
      "epoch": 0.15250544662309368,
      "grad_norm": 755.1558837890625,
      "learning_rate": 0.00019005395146076616,
      "loss": 11.8929,
      "step": 140
    },
    {
      "epoch": 0.15359477124183007,
      "grad_norm": 1292.68359375,
      "learning_rate": 0.00018990298506453104,
      "loss": 11.1247,
      "step": 141
    },
    {
      "epoch": 0.15468409586056645,
      "grad_norm": 623.200927734375,
      "learning_rate": 0.0001897509424466393,
      "loss": 10.7395,
      "step": 142
    },
    {
      "epoch": 0.15577342047930284,
      "grad_norm": 628.1660766601562,
      "learning_rate": 0.00018959782542718128,
      "loss": 11.6995,
      "step": 143
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 818.7709350585938,
      "learning_rate": 0.000189443635839109,
      "loss": 11.4782,
      "step": 144
    },
    {
      "epoch": 0.1579520697167756,
      "grad_norm": 777.9385986328125,
      "learning_rate": 0.00018928837552821404,
      "loss": 9.1531,
      "step": 145
    },
    {
      "epoch": 0.15904139433551198,
      "grad_norm": 632.1226196289062,
      "learning_rate": 0.0001891320463531055,
      "loss": 10.9672,
      "step": 146
    },
    {
      "epoch": 0.16013071895424835,
      "grad_norm": 1045.4794921875,
      "learning_rate": 0.00018897465018518782,
      "loss": 12.5507,
      "step": 147
    },
    {
      "epoch": 0.16122004357298475,
      "grad_norm": 1832.7476806640625,
      "learning_rate": 0.0001888161889086383,
      "loss": 10.2606,
      "step": 148
    },
    {
      "epoch": 0.16230936819172112,
      "grad_norm": 1147.483154296875,
      "learning_rate": 0.00018865666442038456,
      "loss": 11.287,
      "step": 149
    },
    {
      "epoch": 0.16339869281045752,
      "grad_norm": 1081.8319091796875,
      "learning_rate": 0.00018849607863008193,
      "loss": 10.9169,
      "step": 150
    },
    {
      "epoch": 0.1644880174291939,
      "grad_norm": 896.3159790039062,
      "learning_rate": 0.0001883344334600904,
      "loss": 11.1309,
      "step": 151
    },
    {
      "epoch": 0.1655773420479303,
      "grad_norm": 1027.65869140625,
      "learning_rate": 0.00018817173084545176,
      "loss": 10.6067,
      "step": 152
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 792.0274047851562,
      "learning_rate": 0.0001880079727338664,
      "loss": 11.5252,
      "step": 153
    },
    {
      "epoch": 0.16775599128540306,
      "grad_norm": 1679.2010498046875,
      "learning_rate": 0.00018784316108566996,
      "loss": 12.0362,
      "step": 154
    },
    {
      "epoch": 0.16884531590413943,
      "grad_norm": 897.4620971679688,
      "learning_rate": 0.00018767729787380985,
      "loss": 11.3795,
      "step": 155
    },
    {
      "epoch": 0.16993464052287582,
      "grad_norm": 1291.0208740234375,
      "learning_rate": 0.00018751038508382176,
      "loss": 11.0511,
      "step": 156
    },
    {
      "epoch": 0.1710239651416122,
      "grad_norm": 1339.29345703125,
      "learning_rate": 0.00018734242471380572,
      "loss": 10.7022,
      "step": 157
    },
    {
      "epoch": 0.1721132897603486,
      "grad_norm": 900.9859619140625,
      "learning_rate": 0.00018717341877440226,
      "loss": 11.0906,
      "step": 158
    },
    {
      "epoch": 0.17320261437908496,
      "grad_norm": 811.28662109375,
      "learning_rate": 0.0001870033692887684,
      "loss": 10.1273,
      "step": 159
    },
    {
      "epoch": 0.17429193899782136,
      "grad_norm": 1265.2435302734375,
      "learning_rate": 0.00018683227829255334,
      "loss": 12.1877,
      "step": 160
    },
    {
      "epoch": 0.17538126361655773,
      "grad_norm": 1323.2113037109375,
      "learning_rate": 0.00018666014783387408,
      "loss": 11.6536,
      "step": 161
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 1310.9921875,
      "learning_rate": 0.000186486979973291,
      "loss": 12.1153,
      "step": 162
    },
    {
      "epoch": 0.1775599128540305,
      "grad_norm": 806.4589233398438,
      "learning_rate": 0.0001863127767837831,
      "loss": 11.7038,
      "step": 163
    },
    {
      "epoch": 0.1786492374727669,
      "grad_norm": 845.1607666015625,
      "learning_rate": 0.0001861375403507233,
      "loss": 11.3316,
      "step": 164
    },
    {
      "epoch": 0.17973856209150327,
      "grad_norm": 1319.9588623046875,
      "learning_rate": 0.00018596127277185329,
      "loss": 12.2223,
      "step": 165
    },
    {
      "epoch": 0.18082788671023964,
      "grad_norm": 1126.65234375,
      "learning_rate": 0.0001857839761572586,
      "loss": 11.2331,
      "step": 166
    },
    {
      "epoch": 0.18191721132897604,
      "grad_norm": 827.2431640625,
      "learning_rate": 0.00018560565262934318,
      "loss": 11.2991,
      "step": 167
    },
    {
      "epoch": 0.1830065359477124,
      "grad_norm": 804.45166015625,
      "learning_rate": 0.00018542630432280422,
      "loss": 11.1516,
      "step": 168
    },
    {
      "epoch": 0.1840958605664488,
      "grad_norm": 868.1654052734375,
      "learning_rate": 0.00018524593338460635,
      "loss": 12.0214,
      "step": 169
    },
    {
      "epoch": 0.18518518518518517,
      "grad_norm": 1219.4190673828125,
      "learning_rate": 0.00018506454197395606,
      "loss": 12.0915,
      "step": 170
    },
    {
      "epoch": 0.18627450980392157,
      "grad_norm": 1242.7969970703125,
      "learning_rate": 0.00018488213226227588,
      "loss": 11.2175,
      "step": 171
    },
    {
      "epoch": 0.18736383442265794,
      "grad_norm": 984.5662231445312,
      "learning_rate": 0.0001846987064331783,
      "loss": 11.5393,
      "step": 172
    },
    {
      "epoch": 0.18845315904139434,
      "grad_norm": 1010.8961181640625,
      "learning_rate": 0.00018451426668243963,
      "loss": 11.7417,
      "step": 173
    },
    {
      "epoch": 0.1895424836601307,
      "grad_norm": 784.3214721679688,
      "learning_rate": 0.0001843288152179739,
      "loss": 11.8699,
      "step": 174
    },
    {
      "epoch": 0.1906318082788671,
      "grad_norm": 750.0485229492188,
      "learning_rate": 0.00018414235425980616,
      "loss": 10.6044,
      "step": 175
    },
    {
      "epoch": 0.19172113289760348,
      "grad_norm": 1271.091064453125,
      "learning_rate": 0.00018395488604004603,
      "loss": 10.5923,
      "step": 176
    },
    {
      "epoch": 0.19281045751633988,
      "grad_norm": 778.1675415039062,
      "learning_rate": 0.00018376641280286107,
      "loss": 11.4075,
      "step": 177
    },
    {
      "epoch": 0.19389978213507625,
      "grad_norm": 999.4036254882812,
      "learning_rate": 0.00018357693680444976,
      "loss": 11.6131,
      "step": 178
    },
    {
      "epoch": 0.19498910675381265,
      "grad_norm": 905.7958374023438,
      "learning_rate": 0.00018338646031301458,
      "loss": 10.323,
      "step": 179
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 872.347412109375,
      "learning_rate": 0.00018319498560873476,
      "loss": 11.0557,
      "step": 180
    },
    {
      "epoch": 0.19716775599128541,
      "grad_norm": 756.1710815429688,
      "learning_rate": 0.00018300251498373923,
      "loss": 10.2497,
      "step": 181
    },
    {
      "epoch": 0.19825708061002179,
      "grad_norm": 845.9336547851562,
      "learning_rate": 0.00018280905074207884,
      "loss": 10.9945,
      "step": 182
    },
    {
      "epoch": 0.19934640522875818,
      "grad_norm": 694.05224609375,
      "learning_rate": 0.000182614595199699,
      "loss": 10.545,
      "step": 183
    },
    {
      "epoch": 0.20043572984749455,
      "grad_norm": 714.1306762695312,
      "learning_rate": 0.00018241915068441196,
      "loss": 11.1389,
      "step": 184
    },
    {
      "epoch": 0.20152505446623092,
      "grad_norm": 722.7552490234375,
      "learning_rate": 0.00018222271953586883,
      "loss": 11.0365,
      "step": 185
    },
    {
      "epoch": 0.20261437908496732,
      "grad_norm": 708.6851196289062,
      "learning_rate": 0.00018202530410553163,
      "loss": 11.2533,
      "step": 186
    },
    {
      "epoch": 0.2037037037037037,
      "grad_norm": 1047.9068603515625,
      "learning_rate": 0.00018182690675664514,
      "loss": 10.9609,
      "step": 187
    },
    {
      "epoch": 0.2047930283224401,
      "grad_norm": 562.74072265625,
      "learning_rate": 0.00018162752986420868,
      "loss": 11.5543,
      "step": 188
    },
    {
      "epoch": 0.20588235294117646,
      "grad_norm": 690.9712524414062,
      "learning_rate": 0.0001814271758149475,
      "loss": 10.2702,
      "step": 189
    },
    {
      "epoch": 0.20697167755991286,
      "grad_norm": 1054.808349609375,
      "learning_rate": 0.00018122584700728443,
      "loss": 11.0848,
      "step": 190
    },
    {
      "epoch": 0.20806100217864923,
      "grad_norm": 633.4119262695312,
      "learning_rate": 0.00018102354585131092,
      "loss": 11.1002,
      "step": 191
    },
    {
      "epoch": 0.20915032679738563,
      "grad_norm": 508.4339294433594,
      "learning_rate": 0.00018082027476875847,
      "loss": 10.9004,
      "step": 192
    },
    {
      "epoch": 0.210239651416122,
      "grad_norm": 616.0364379882812,
      "learning_rate": 0.00018061603619296942,
      "loss": 10.9211,
      "step": 193
    },
    {
      "epoch": 0.2113289760348584,
      "grad_norm": 553.75,
      "learning_rate": 0.0001804108325688679,
      "loss": 10.5755,
      "step": 194
    },
    {
      "epoch": 0.21241830065359477,
      "grad_norm": 554.069580078125,
      "learning_rate": 0.00018020466635293057,
      "loss": 11.7447,
      "step": 195
    },
    {
      "epoch": 0.21350762527233116,
      "grad_norm": 681.5348510742188,
      "learning_rate": 0.0001799975400131572,
      "loss": 11.0779,
      "step": 196
    },
    {
      "epoch": 0.21459694989106753,
      "grad_norm": 632.8909301757812,
      "learning_rate": 0.00017978945602904116,
      "loss": 11.4297,
      "step": 197
    },
    {
      "epoch": 0.21568627450980393,
      "grad_norm": 485.4712829589844,
      "learning_rate": 0.0001795804168915396,
      "loss": 9.903,
      "step": 198
    },
    {
      "epoch": 0.2167755991285403,
      "grad_norm": 668.1004638671875,
      "learning_rate": 0.00017937042510304392,
      "loss": 11.2272,
      "step": 199
    },
    {
      "epoch": 0.2178649237472767,
      "grad_norm": 798.0782470703125,
      "learning_rate": 0.00017915948317734942,
      "loss": 10.4403,
      "step": 200
    },
    {
      "epoch": 0.21895424836601307,
      "grad_norm": 484.9952697753906,
      "learning_rate": 0.00017894759363962554,
      "loss": 11.3431,
      "step": 201
    },
    {
      "epoch": 0.22004357298474944,
      "grad_norm": 557.5073852539062,
      "learning_rate": 0.00017873475902638553,
      "loss": 10.0637,
      "step": 202
    },
    {
      "epoch": 0.22113289760348584,
      "grad_norm": 457.8398742675781,
      "learning_rate": 0.00017852098188545602,
      "loss": 10.7078,
      "step": 203
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1345.8021240234375,
      "learning_rate": 0.00017830626477594654,
      "loss": 10.8073,
      "step": 204
    },
    {
      "epoch": 0.2233115468409586,
      "grad_norm": 508.4226379394531,
      "learning_rate": 0.00017809061026821896,
      "loss": 11.756,
      "step": 205
    },
    {
      "epoch": 0.22440087145969498,
      "grad_norm": 543.2045288085938,
      "learning_rate": 0.00017787402094385666,
      "loss": 11.3398,
      "step": 206
    },
    {
      "epoch": 0.22549019607843138,
      "grad_norm": 600.871337890625,
      "learning_rate": 0.00017765649939563365,
      "loss": 10.9323,
      "step": 207
    },
    {
      "epoch": 0.22657952069716775,
      "grad_norm": 491.7249450683594,
      "learning_rate": 0.00017743804822748345,
      "loss": 11.4056,
      "step": 208
    },
    {
      "epoch": 0.22766884531590414,
      "grad_norm": 584.7307739257812,
      "learning_rate": 0.00017721867005446806,
      "loss": 10.6625,
      "step": 209
    },
    {
      "epoch": 0.22875816993464052,
      "grad_norm": 781.3718872070312,
      "learning_rate": 0.00017699836750274662,
      "loss": 11.4948,
      "step": 210
    },
    {
      "epoch": 0.2298474945533769,
      "grad_norm": 561.0650024414062,
      "learning_rate": 0.00017677714320954378,
      "loss": 11.5926,
      "step": 211
    },
    {
      "epoch": 0.23093681917211328,
      "grad_norm": 662.3781127929688,
      "learning_rate": 0.00017655499982311847,
      "loss": 11.2635,
      "step": 212
    },
    {
      "epoch": 0.23202614379084968,
      "grad_norm": 485.6428527832031,
      "learning_rate": 0.00017633194000273188,
      "loss": 11.3912,
      "step": 213
    },
    {
      "epoch": 0.23311546840958605,
      "grad_norm": 1000.7589111328125,
      "learning_rate": 0.00017610796641861581,
      "loss": 11.0748,
      "step": 214
    },
    {
      "epoch": 0.23420479302832245,
      "grad_norm": 546.7896728515625,
      "learning_rate": 0.0001758830817519407,
      "loss": 11.1538,
      "step": 215
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 498.8612976074219,
      "learning_rate": 0.00017565728869478337,
      "loss": 10.7306,
      "step": 216
    },
    {
      "epoch": 0.23638344226579522,
      "grad_norm": 954.06787109375,
      "learning_rate": 0.00017543058995009503,
      "loss": 9.8866,
      "step": 217
    },
    {
      "epoch": 0.2374727668845316,
      "grad_norm": 474.430419921875,
      "learning_rate": 0.00017520298823166873,
      "loss": 11.1717,
      "step": 218
    },
    {
      "epoch": 0.238562091503268,
      "grad_norm": 446.2909851074219,
      "learning_rate": 0.000174974486264107,
      "loss": 11.0507,
      "step": 219
    },
    {
      "epoch": 0.23965141612200436,
      "grad_norm": 574.3330688476562,
      "learning_rate": 0.00017474508678278915,
      "loss": 10.7534,
      "step": 220
    },
    {
      "epoch": 0.24074074074074073,
      "grad_norm": 751.9092407226562,
      "learning_rate": 0.00017451479253383857,
      "loss": 11.041,
      "step": 221
    },
    {
      "epoch": 0.24183006535947713,
      "grad_norm": 969.604248046875,
      "learning_rate": 0.00017428360627408978,
      "loss": 10.7288,
      "step": 222
    },
    {
      "epoch": 0.2429193899782135,
      "grad_norm": 716.7567138671875,
      "learning_rate": 0.0001740515307710557,
      "loss": 11.062,
      "step": 223
    },
    {
      "epoch": 0.2440087145969499,
      "grad_norm": 580.2841796875,
      "learning_rate": 0.000173818568802894,
      "loss": 10.2888,
      "step": 224
    },
    {
      "epoch": 0.24509803921568626,
      "grad_norm": 488.0480041503906,
      "learning_rate": 0.00017358472315837447,
      "loss": 10.0662,
      "step": 225
    },
    {
      "epoch": 0.24618736383442266,
      "grad_norm": 805.732666015625,
      "learning_rate": 0.00017334999663684504,
      "loss": 10.4939,
      "step": 226
    },
    {
      "epoch": 0.24727668845315903,
      "grad_norm": 610.5254516601562,
      "learning_rate": 0.00017311439204819874,
      "loss": 10.3649,
      "step": 227
    },
    {
      "epoch": 0.24836601307189543,
      "grad_norm": 464.3783264160156,
      "learning_rate": 0.00017287791221283984,
      "loss": 10.8539,
      "step": 228
    },
    {
      "epoch": 0.2494553376906318,
      "grad_norm": 645.302001953125,
      "learning_rate": 0.00017264055996165007,
      "loss": 10.8646,
      "step": 229
    },
    {
      "epoch": 0.25054466230936817,
      "grad_norm": 466.3519287109375,
      "learning_rate": 0.00017240233813595478,
      "loss": 10.1154,
      "step": 230
    },
    {
      "epoch": 0.25054466230936817,
      "eval_loss": 2.678065061569214,
      "eval_runtime": 2.7886,
      "eval_samples_per_second": 138.78,
      "eval_steps_per_second": 69.569,
      "step": 230
    },
    {
      "epoch": 0.25163398692810457,
      "grad_norm": 656.3611450195312,
      "learning_rate": 0.000172163249587489,
      "loss": 10.7119,
      "step": 231
    },
    {
      "epoch": 0.25272331154684097,
      "grad_norm": 593.1845703125,
      "learning_rate": 0.00017192329717836315,
      "loss": 11.6847,
      "step": 232
    },
    {
      "epoch": 0.25381263616557737,
      "grad_norm": 629.67578125,
      "learning_rate": 0.00017168248378102892,
      "loss": 10.5297,
      "step": 233
    },
    {
      "epoch": 0.2549019607843137,
      "grad_norm": 763.1748657226562,
      "learning_rate": 0.0001714408122782448,
      "loss": 10.4083,
      "step": 234
    },
    {
      "epoch": 0.2559912854030501,
      "grad_norm": 723.7062377929688,
      "learning_rate": 0.0001711982855630416,
      "loss": 11.5429,
      "step": 235
    },
    {
      "epoch": 0.2570806100217865,
      "grad_norm": 814.066650390625,
      "learning_rate": 0.00017095490653868778,
      "loss": 9.9942,
      "step": 236
    },
    {
      "epoch": 0.2581699346405229,
      "grad_norm": 593.72216796875,
      "learning_rate": 0.00017071067811865476,
      "loss": 9.9854,
      "step": 237
    },
    {
      "epoch": 0.25925925925925924,
      "grad_norm": 915.5996704101562,
      "learning_rate": 0.000170465603226582,
      "loss": 10.0837,
      "step": 238
    },
    {
      "epoch": 0.26034858387799564,
      "grad_norm": 775.1221923828125,
      "learning_rate": 0.00017021968479624203,
      "loss": 11.7044,
      "step": 239
    },
    {
      "epoch": 0.26143790849673204,
      "grad_norm": 569.85791015625,
      "learning_rate": 0.00016997292577150528,
      "loss": 10.7066,
      "step": 240
    },
    {
      "epoch": 0.2625272331154684,
      "grad_norm": 497.0193176269531,
      "learning_rate": 0.0001697253291063049,
      "loss": 10.1781,
      "step": 241
    },
    {
      "epoch": 0.2636165577342048,
      "grad_norm": 784.2388305664062,
      "learning_rate": 0.0001694768977646013,
      "loss": 10.108,
      "step": 242
    },
    {
      "epoch": 0.2647058823529412,
      "grad_norm": 553.2965698242188,
      "learning_rate": 0.00016922763472034685,
      "loss": 10.667,
      "step": 243
    },
    {
      "epoch": 0.2657952069716776,
      "grad_norm": 534.915771484375,
      "learning_rate": 0.00016897754295745008,
      "loss": 11.2736,
      "step": 244
    },
    {
      "epoch": 0.2668845315904139,
      "grad_norm": 593.677490234375,
      "learning_rate": 0.00016872662546974008,
      "loss": 10.4391,
      "step": 245
    },
    {
      "epoch": 0.2679738562091503,
      "grad_norm": 647.2657470703125,
      "learning_rate": 0.0001684748852609306,
      "loss": 11.6138,
      "step": 246
    },
    {
      "epoch": 0.2690631808278867,
      "grad_norm": 678.768798828125,
      "learning_rate": 0.00016822232534458416,
      "loss": 11.0187,
      "step": 247
    },
    {
      "epoch": 0.2701525054466231,
      "grad_norm": 804.8411254882812,
      "learning_rate": 0.00016796894874407595,
      "loss": 11.2911,
      "step": 248
    },
    {
      "epoch": 0.27124183006535946,
      "grad_norm": 440.8306579589844,
      "learning_rate": 0.00016771475849255754,
      "loss": 11.4534,
      "step": 249
    },
    {
      "epoch": 0.27233115468409586,
      "grad_norm": 730.95166015625,
      "learning_rate": 0.0001674597576329207,
      "loss": 11.5775,
      "step": 250
    },
    {
      "epoch": 0.27342047930283225,
      "grad_norm": 570.8353881835938,
      "learning_rate": 0.00016720394921776097,
      "loss": 11.6714,
      "step": 251
    },
    {
      "epoch": 0.27450980392156865,
      "grad_norm": 532.7552490234375,
      "learning_rate": 0.000166947336309341,
      "loss": 10.7977,
      "step": 252
    },
    {
      "epoch": 0.275599128540305,
      "grad_norm": 416.3326110839844,
      "learning_rate": 0.00016668992197955398,
      "loss": 11.0194,
      "step": 253
    },
    {
      "epoch": 0.2766884531590414,
      "grad_norm": 458.32904052734375,
      "learning_rate": 0.00016643170930988698,
      "loss": 10.1537,
      "step": 254
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 4145.63330078125,
      "learning_rate": 0.00016617270139138371,
      "loss": 10.5428,
      "step": 255
    },
    {
      "epoch": 0.2788671023965142,
      "grad_norm": 435.748291015625,
      "learning_rate": 0.0001659129013246079,
      "loss": 12.5706,
      "step": 256
    },
    {
      "epoch": 0.27995642701525053,
      "grad_norm": 448.35015869140625,
      "learning_rate": 0.000165652312219606,
      "loss": 10.5882,
      "step": 257
    },
    {
      "epoch": 0.28104575163398693,
      "grad_norm": 541.2523193359375,
      "learning_rate": 0.00016539093719586994,
      "loss": 9.4986,
      "step": 258
    },
    {
      "epoch": 0.2821350762527233,
      "grad_norm": 436.6683349609375,
      "learning_rate": 0.00016512877938229986,
      "loss": 9.6175,
      "step": 259
    },
    {
      "epoch": 0.28322440087145967,
      "grad_norm": 732.4791259765625,
      "learning_rate": 0.0001648658419171666,
      "loss": 11.2249,
      "step": 260
    },
    {
      "epoch": 0.28431372549019607,
      "grad_norm": 557.4008178710938,
      "learning_rate": 0.00016460212794807414,
      "loss": 11.2695,
      "step": 261
    },
    {
      "epoch": 0.28540305010893247,
      "grad_norm": 441.1900939941406,
      "learning_rate": 0.00016433764063192194,
      "loss": 10.2486,
      "step": 262
    },
    {
      "epoch": 0.28649237472766886,
      "grad_norm": 500.40753173828125,
      "learning_rate": 0.00016407238313486712,
      "loss": 10.4287,
      "step": 263
    },
    {
      "epoch": 0.2875816993464052,
      "grad_norm": 517.90478515625,
      "learning_rate": 0.0001638063586322866,
      "loss": 10.9458,
      "step": 264
    },
    {
      "epoch": 0.2886710239651416,
      "grad_norm": 477.6897888183594,
      "learning_rate": 0.0001635395703087391,
      "loss": 10.7061,
      "step": 265
    },
    {
      "epoch": 0.289760348583878,
      "grad_norm": 577.167724609375,
      "learning_rate": 0.00016327202135792685,
      "loss": 10.5098,
      "step": 266
    },
    {
      "epoch": 0.2908496732026144,
      "grad_norm": 502.87469482421875,
      "learning_rate": 0.00016300371498265763,
      "loss": 10.3983,
      "step": 267
    },
    {
      "epoch": 0.29193899782135074,
      "grad_norm": 584.27197265625,
      "learning_rate": 0.00016273465439480618,
      "loss": 11.3498,
      "step": 268
    },
    {
      "epoch": 0.29302832244008714,
      "grad_norm": 718.4893188476562,
      "learning_rate": 0.000162464842815276,
      "loss": 11.7879,
      "step": 269
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 415.97320556640625,
      "learning_rate": 0.00016219428347396053,
      "loss": 11.5378,
      "step": 270
    },
    {
      "epoch": 0.29520697167755994,
      "grad_norm": 965.6848754882812,
      "learning_rate": 0.0001619229796097046,
      "loss": 10.5206,
      "step": 271
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 319.21905517578125,
      "learning_rate": 0.0001616509344702658,
      "loss": 9.4684,
      "step": 272
    },
    {
      "epoch": 0.2973856209150327,
      "grad_norm": 399.8673095703125,
      "learning_rate": 0.00016137815131227526,
      "loss": 10.1383,
      "step": 273
    },
    {
      "epoch": 0.2984749455337691,
      "grad_norm": 905.3511962890625,
      "learning_rate": 0.00016110463340119913,
      "loss": 10.4312,
      "step": 274
    },
    {
      "epoch": 0.2995642701525055,
      "grad_norm": 561.3602294921875,
      "learning_rate": 0.000160830384011299,
      "loss": 10.7353,
      "step": 275
    },
    {
      "epoch": 0.3006535947712418,
      "grad_norm": 664.7014770507812,
      "learning_rate": 0.00016055540642559305,
      "loss": 11.0443,
      "step": 276
    },
    {
      "epoch": 0.3017429193899782,
      "grad_norm": 605.7849731445312,
      "learning_rate": 0.00016027970393581666,
      "loss": 10.5288,
      "step": 277
    },
    {
      "epoch": 0.3028322440087146,
      "grad_norm": 745.9011840820312,
      "learning_rate": 0.00016000327984238292,
      "loss": 11.5525,
      "step": 278
    },
    {
      "epoch": 0.30392156862745096,
      "grad_norm": 453.7929382324219,
      "learning_rate": 0.00015972613745434314,
      "loss": 10.4069,
      "step": 279
    },
    {
      "epoch": 0.30501089324618735,
      "grad_norm": 679.0534057617188,
      "learning_rate": 0.0001594482800893474,
      "loss": 10.7409,
      "step": 280
    },
    {
      "epoch": 0.30610021786492375,
      "grad_norm": 601.0300903320312,
      "learning_rate": 0.00015916971107360461,
      "loss": 10.1115,
      "step": 281
    },
    {
      "epoch": 0.30718954248366015,
      "grad_norm": 559.4019165039062,
      "learning_rate": 0.00015889043374184286,
      "loss": 11.9414,
      "step": 282
    },
    {
      "epoch": 0.3082788671023965,
      "grad_norm": 552.6649169921875,
      "learning_rate": 0.00015861045143726946,
      "loss": 11.0226,
      "step": 283
    },
    {
      "epoch": 0.3093681917211329,
      "grad_norm": 488.8516540527344,
      "learning_rate": 0.00015832976751153078,
      "loss": 11.4205,
      "step": 284
    },
    {
      "epoch": 0.3104575163398693,
      "grad_norm": 666.8306274414062,
      "learning_rate": 0.0001580483853246723,
      "loss": 11.0604,
      "step": 285
    },
    {
      "epoch": 0.3115468409586057,
      "grad_norm": 710.2818603515625,
      "learning_rate": 0.0001577663082450984,
      "loss": 10.4147,
      "step": 286
    },
    {
      "epoch": 0.31263616557734203,
      "grad_norm": 615.7288818359375,
      "learning_rate": 0.00015748353964953186,
      "loss": 11.2711,
      "step": 287
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 613.173828125,
      "learning_rate": 0.00015720008292297364,
      "loss": 12.3367,
      "step": 288
    },
    {
      "epoch": 0.3148148148148148,
      "grad_norm": 516.6195678710938,
      "learning_rate": 0.00015691594145866215,
      "loss": 11.0611,
      "step": 289
    },
    {
      "epoch": 0.3159041394335512,
      "grad_norm": 572.4487915039062,
      "learning_rate": 0.00015663111865803285,
      "loss": 10.9705,
      "step": 290
    },
    {
      "epoch": 0.31699346405228757,
      "grad_norm": 2503.04736328125,
      "learning_rate": 0.00015634561793067737,
      "loss": 11.5724,
      "step": 291
    },
    {
      "epoch": 0.31808278867102396,
      "grad_norm": 945.3132934570312,
      "learning_rate": 0.00015605944269430277,
      "loss": 10.8974,
      "step": 292
    },
    {
      "epoch": 0.31917211328976036,
      "grad_norm": 566.0651245117188,
      "learning_rate": 0.00015577259637469058,
      "loss": 11.5792,
      "step": 293
    },
    {
      "epoch": 0.3202614379084967,
      "grad_norm": 736.163818359375,
      "learning_rate": 0.00015548508240565583,
      "loss": 11.0325,
      "step": 294
    },
    {
      "epoch": 0.3213507625272331,
      "grad_norm": 757.3773803710938,
      "learning_rate": 0.00015519690422900593,
      "loss": 11.0976,
      "step": 295
    },
    {
      "epoch": 0.3224400871459695,
      "grad_norm": 936.12890625,
      "learning_rate": 0.00015490806529449945,
      "loss": 11.1016,
      "step": 296
    },
    {
      "epoch": 0.3235294117647059,
      "grad_norm": 720.9487915039062,
      "learning_rate": 0.0001546185690598049,
      "loss": 11.4778,
      "step": 297
    },
    {
      "epoch": 0.32461873638344224,
      "grad_norm": 653.9827880859375,
      "learning_rate": 0.0001543284189904592,
      "loss": 12.2587,
      "step": 298
    },
    {
      "epoch": 0.32570806100217864,
      "grad_norm": 722.68115234375,
      "learning_rate": 0.00015403761855982631,
      "loss": 10.4469,
      "step": 299
    },
    {
      "epoch": 0.32679738562091504,
      "grad_norm": 1037.672607421875,
      "learning_rate": 0.00015374617124905564,
      "loss": 11.2673,
      "step": 300
    },
    {
      "epoch": 0.32788671023965144,
      "grad_norm": 721.6505126953125,
      "learning_rate": 0.0001534540805470403,
      "loss": 11.505,
      "step": 301
    },
    {
      "epoch": 0.3289760348583878,
      "grad_norm": 855.7329711914062,
      "learning_rate": 0.00015316134995037545,
      "loss": 10.711,
      "step": 302
    },
    {
      "epoch": 0.3300653594771242,
      "grad_norm": 733.3320922851562,
      "learning_rate": 0.00015286798296331632,
      "loss": 11.2952,
      "step": 303
    },
    {
      "epoch": 0.3311546840958606,
      "grad_norm": 763.12744140625,
      "learning_rate": 0.00015257398309773633,
      "loss": 11.3105,
      "step": 304
    },
    {
      "epoch": 0.332244008714597,
      "grad_norm": 462.6761169433594,
      "learning_rate": 0.00015227935387308511,
      "loss": 11.7857,
      "step": 305
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 793.5400390625,
      "learning_rate": 0.00015198409881634617,
      "loss": 10.805,
      "step": 306
    },
    {
      "epoch": 0.3344226579520697,
      "grad_norm": 587.91455078125,
      "learning_rate": 0.0001516882214619949,
      "loss": 10.8593,
      "step": 307
    },
    {
      "epoch": 0.3355119825708061,
      "grad_norm": 587.7955322265625,
      "learning_rate": 0.00015139172535195617,
      "loss": 11.7922,
      "step": 308
    },
    {
      "epoch": 0.3366013071895425,
      "grad_norm": 534.8583374023438,
      "learning_rate": 0.0001510946140355619,
      "loss": 11.4238,
      "step": 309
    },
    {
      "epoch": 0.33769063180827885,
      "grad_norm": 458.095458984375,
      "learning_rate": 0.00015079689106950854,
      "loss": 10.8279,
      "step": 310
    },
    {
      "epoch": 0.33877995642701525,
      "grad_norm": 1109.0992431640625,
      "learning_rate": 0.0001504985600178147,
      "loss": 11.2836,
      "step": 311
    },
    {
      "epoch": 0.33986928104575165,
      "grad_norm": 678.5516967773438,
      "learning_rate": 0.00015019962445177819,
      "loss": 11.7628,
      "step": 312
    },
    {
      "epoch": 0.340958605664488,
      "grad_norm": 569.3182373046875,
      "learning_rate": 0.00014990008794993345,
      "loss": 11.4081,
      "step": 313
    },
    {
      "epoch": 0.3420479302832244,
      "grad_norm": 545.4651489257812,
      "learning_rate": 0.00014959995409800873,
      "loss": 12.5765,
      "step": 314
    },
    {
      "epoch": 0.3431372549019608,
      "grad_norm": 667.7208862304688,
      "learning_rate": 0.00014929922648888308,
      "loss": 11.5474,
      "step": 315
    },
    {
      "epoch": 0.3442265795206972,
      "grad_norm": 621.5955810546875,
      "learning_rate": 0.0001489979087225434,
      "loss": 11.6176,
      "step": 316
    },
    {
      "epoch": 0.3453159041394335,
      "grad_norm": 673.23828125,
      "learning_rate": 0.00014869600440604118,
      "loss": 10.7684,
      "step": 317
    },
    {
      "epoch": 0.3464052287581699,
      "grad_norm": 874.149658203125,
      "learning_rate": 0.00014839351715344968,
      "loss": 11.6673,
      "step": 318
    },
    {
      "epoch": 0.3474945533769063,
      "grad_norm": 869.7573852539062,
      "learning_rate": 0.00014809045058582026,
      "loss": 11.8444,
      "step": 319
    },
    {
      "epoch": 0.3485838779956427,
      "grad_norm": 971.1417236328125,
      "learning_rate": 0.00014778680833113926,
      "loss": 11.0821,
      "step": 320
    },
    {
      "epoch": 0.34967320261437906,
      "grad_norm": 1166.9410400390625,
      "learning_rate": 0.00014748259402428462,
      "loss": 11.7491,
      "step": 321
    },
    {
      "epoch": 0.35076252723311546,
      "grad_norm": 1266.657470703125,
      "learning_rate": 0.00014717781130698212,
      "loss": 11.2385,
      "step": 322
    },
    {
      "epoch": 0.35185185185185186,
      "grad_norm": 972.2568969726562,
      "learning_rate": 0.00014687246382776205,
      "loss": 10.9266,
      "step": 323
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 1106.99951171875,
      "learning_rate": 0.00014656655524191537,
      "loss": 11.3394,
      "step": 324
    },
    {
      "epoch": 0.3540305010893246,
      "grad_norm": 609.1690673828125,
      "learning_rate": 0.0001462600892114501,
      "loss": 11.5497,
      "step": 325
    },
    {
      "epoch": 0.355119825708061,
      "grad_norm": 930.3221435546875,
      "learning_rate": 0.00014595306940504716,
      "loss": 10.7787,
      "step": 326
    },
    {
      "epoch": 0.3562091503267974,
      "grad_norm": 902.9607543945312,
      "learning_rate": 0.00014564549949801694,
      "loss": 11.3869,
      "step": 327
    },
    {
      "epoch": 0.3572984749455338,
      "grad_norm": 826.9057006835938,
      "learning_rate": 0.00014533738317225485,
      "loss": 11.0435,
      "step": 328
    },
    {
      "epoch": 0.35838779956427014,
      "grad_norm": 802.20751953125,
      "learning_rate": 0.00014502872411619757,
      "loss": 11.4467,
      "step": 329
    },
    {
      "epoch": 0.35947712418300654,
      "grad_norm": 1908.7032470703125,
      "learning_rate": 0.00014471952602477866,
      "loss": 11.7794,
      "step": 330
    },
    {
      "epoch": 0.36056644880174293,
      "grad_norm": 1125.180908203125,
      "learning_rate": 0.0001444097925993845,
      "loss": 11.4379,
      "step": 331
    },
    {
      "epoch": 0.3616557734204793,
      "grad_norm": 944.010986328125,
      "learning_rate": 0.0001440995275478099,
      "loss": 11.4291,
      "step": 332
    },
    {
      "epoch": 0.3627450980392157,
      "grad_norm": 979.970703125,
      "learning_rate": 0.0001437887345842137,
      "loss": 12.0289,
      "step": 333
    },
    {
      "epoch": 0.3638344226579521,
      "grad_norm": 1132.85205078125,
      "learning_rate": 0.00014347741742907433,
      "loss": 11.7372,
      "step": 334
    },
    {
      "epoch": 0.36492374727668847,
      "grad_norm": 898.1516723632812,
      "learning_rate": 0.00014316557980914528,
      "loss": 10.979,
      "step": 335
    },
    {
      "epoch": 0.3660130718954248,
      "grad_norm": 988.018310546875,
      "learning_rate": 0.00014285322545741052,
      "loss": 12.0688,
      "step": 336
    },
    {
      "epoch": 0.3671023965141612,
      "grad_norm": 870.642822265625,
      "learning_rate": 0.0001425403581130398,
      "loss": 9.5806,
      "step": 337
    },
    {
      "epoch": 0.3681917211328976,
      "grad_norm": 1094.388671875,
      "learning_rate": 0.00014222698152134374,
      "loss": 11.5287,
      "step": 338
    },
    {
      "epoch": 0.369281045751634,
      "grad_norm": 758.43994140625,
      "learning_rate": 0.0001419130994337292,
      "loss": 10.2106,
      "step": 339
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 994.099853515625,
      "learning_rate": 0.00014159871560765432,
      "loss": 13.7915,
      "step": 340
    },
    {
      "epoch": 0.37145969498910675,
      "grad_norm": 1287.93115234375,
      "learning_rate": 0.0001412838338065835,
      "loss": 11.4081,
      "step": 341
    },
    {
      "epoch": 0.37254901960784315,
      "grad_norm": 895.5205688476562,
      "learning_rate": 0.0001409684577999423,
      "loss": 10.9777,
      "step": 342
    },
    {
      "epoch": 0.37363834422657954,
      "grad_norm": 843.1907348632812,
      "learning_rate": 0.00014065259136307242,
      "loss": 11.0484,
      "step": 343
    },
    {
      "epoch": 0.3747276688453159,
      "grad_norm": 1058.655029296875,
      "learning_rate": 0.0001403362382771865,
      "loss": 10.9008,
      "step": 344
    },
    {
      "epoch": 0.3758169934640523,
      "grad_norm": 1154.1737060546875,
      "learning_rate": 0.0001400194023293228,
      "loss": 11.1368,
      "step": 345
    },
    {
      "epoch": 0.3769063180827887,
      "grad_norm": 866.00830078125,
      "learning_rate": 0.00013970208731229974,
      "loss": 11.8841,
      "step": 346
    },
    {
      "epoch": 0.3779956427015251,
      "grad_norm": 740.2269287109375,
      "learning_rate": 0.00013938429702467086,
      "loss": 11.8118,
      "step": 347
    },
    {
      "epoch": 0.3790849673202614,
      "grad_norm": 1207.28564453125,
      "learning_rate": 0.000139066035270679,
      "loss": 10.8065,
      "step": 348
    },
    {
      "epoch": 0.3801742919389978,
      "grad_norm": 846.9767456054688,
      "learning_rate": 0.00013874730586021093,
      "loss": 10.8413,
      "step": 349
    },
    {
      "epoch": 0.3812636165577342,
      "grad_norm": 835.8090209960938,
      "learning_rate": 0.00013842811260875168,
      "loss": 11.4149,
      "step": 350
    },
    {
      "epoch": 0.38235294117647056,
      "grad_norm": 854.9374389648438,
      "learning_rate": 0.0001381084593373389,
      "loss": 11.2507,
      "step": 351
    },
    {
      "epoch": 0.38344226579520696,
      "grad_norm": 1257.4901123046875,
      "learning_rate": 0.00013778834987251707,
      "loss": 10.6451,
      "step": 352
    },
    {
      "epoch": 0.38453159041394336,
      "grad_norm": 1082.4285888671875,
      "learning_rate": 0.00013746778804629177,
      "loss": 11.6068,
      "step": 353
    },
    {
      "epoch": 0.38562091503267976,
      "grad_norm": 989.6134033203125,
      "learning_rate": 0.0001371467776960837,
      "loss": 12.2527,
      "step": 354
    },
    {
      "epoch": 0.3867102396514161,
      "grad_norm": 875.22021484375,
      "learning_rate": 0.0001368253226646829,
      "loss": 11.4817,
      "step": 355
    },
    {
      "epoch": 0.3877995642701525,
      "grad_norm": 1244.9932861328125,
      "learning_rate": 0.00013650342680020258,
      "loss": 11.4714,
      "step": 356
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 1482.9927978515625,
      "learning_rate": 0.00013618109395603317,
      "loss": 12.19,
      "step": 357
    },
    {
      "epoch": 0.3899782135076253,
      "grad_norm": 1198.56298828125,
      "learning_rate": 0.0001358583279907961,
      "loss": 11.1256,
      "step": 358
    },
    {
      "epoch": 0.39106753812636164,
      "grad_norm": 1113.496826171875,
      "learning_rate": 0.0001355351327682977,
      "loss": 10.5545,
      "step": 359
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 1492.8271484375,
      "learning_rate": 0.0001352115121574829,
      "loss": 11.6428,
      "step": 360
    },
    {
      "epoch": 0.39324618736383443,
      "grad_norm": 1067.7490234375,
      "learning_rate": 0.00013488747003238892,
      "loss": 11.2813,
      "step": 361
    },
    {
      "epoch": 0.39433551198257083,
      "grad_norm": 1344.341064453125,
      "learning_rate": 0.00013456301027209882,
      "loss": 11.7236,
      "step": 362
    },
    {
      "epoch": 0.3954248366013072,
      "grad_norm": 913.9457397460938,
      "learning_rate": 0.00013423813676069534,
      "loss": 11.2616,
      "step": 363
    },
    {
      "epoch": 0.39651416122004357,
      "grad_norm": 1023.7560424804688,
      "learning_rate": 0.000133912853387214,
      "loss": 11.0282,
      "step": 364
    },
    {
      "epoch": 0.39760348583877997,
      "grad_norm": 999.524658203125,
      "learning_rate": 0.0001335871640455968,
      "loss": 11.9099,
      "step": 365
    },
    {
      "epoch": 0.39869281045751637,
      "grad_norm": 1470.6510009765625,
      "learning_rate": 0.00013326107263464558,
      "loss": 10.54,
      "step": 366
    },
    {
      "epoch": 0.3997821350762527,
      "grad_norm": 1311.093017578125,
      "learning_rate": 0.00013293458305797533,
      "loss": 11.6321,
      "step": 367
    },
    {
      "epoch": 0.4008714596949891,
      "grad_norm": 958.1239013671875,
      "learning_rate": 0.0001326076992239674,
      "loss": 12.1061,
      "step": 368
    },
    {
      "epoch": 0.4019607843137255,
      "grad_norm": 1676.4158935546875,
      "learning_rate": 0.00013228042504572285,
      "loss": 12.0777,
      "step": 369
    },
    {
      "epoch": 0.40305010893246185,
      "grad_norm": 1228.0052490234375,
      "learning_rate": 0.00013195276444101547,
      "loss": 11.9471,
      "step": 370
    },
    {
      "epoch": 0.40413943355119825,
      "grad_norm": 2634.232666015625,
      "learning_rate": 0.00013162472133224483,
      "loss": 11.8953,
      "step": 371
    },
    {
      "epoch": 0.40522875816993464,
      "grad_norm": 1066.639404296875,
      "learning_rate": 0.0001312962996463896,
      "loss": 11.6727,
      "step": 372
    },
    {
      "epoch": 0.40631808278867104,
      "grad_norm": 2369.989990234375,
      "learning_rate": 0.00013096750331496033,
      "loss": 12.0725,
      "step": 373
    },
    {
      "epoch": 0.4074074074074074,
      "grad_norm": 1213.9959716796875,
      "learning_rate": 0.0001306383362739523,
      "loss": 12.2408,
      "step": 374
    },
    {
      "epoch": 0.4084967320261438,
      "grad_norm": 1465.7821044921875,
      "learning_rate": 0.00013030880246379866,
      "loss": 11.4288,
      "step": 375
    },
    {
      "epoch": 0.4095860566448802,
      "grad_norm": 1068.891357421875,
      "learning_rate": 0.00012997890582932303,
      "loss": 11.5368,
      "step": 376
    },
    {
      "epoch": 0.4106753812636166,
      "grad_norm": 1787.318115234375,
      "learning_rate": 0.00012964865031969252,
      "loss": 11.4801,
      "step": 377
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 1149.781494140625,
      "learning_rate": 0.0001293180398883701,
      "loss": 12.9816,
      "step": 378
    },
    {
      "epoch": 0.4128540305010893,
      "grad_norm": 2007.1617431640625,
      "learning_rate": 0.00012898707849306763,
      "loss": 11.9605,
      "step": 379
    },
    {
      "epoch": 0.4139433551198257,
      "grad_norm": 2235.568603515625,
      "learning_rate": 0.00012865577009569824,
      "loss": 12.821,
      "step": 380
    },
    {
      "epoch": 0.4150326797385621,
      "grad_norm": 1477.2247314453125,
      "learning_rate": 0.0001283241186623291,
      "loss": 11.7628,
      "step": 381
    },
    {
      "epoch": 0.41612200435729846,
      "grad_norm": 1697.255615234375,
      "learning_rate": 0.00012799212816313376,
      "loss": 12.1798,
      "step": 382
    },
    {
      "epoch": 0.41721132897603486,
      "grad_norm": 2948.6123046875,
      "learning_rate": 0.00012765980257234473,
      "loss": 12.3417,
      "step": 383
    },
    {
      "epoch": 0.41830065359477125,
      "grad_norm": 1735.9176025390625,
      "learning_rate": 0.00012732714586820583,
      "loss": 12.736,
      "step": 384
    },
    {
      "epoch": 0.41938997821350765,
      "grad_norm": 1974.7271728515625,
      "learning_rate": 0.00012699416203292466,
      "loss": 11.0415,
      "step": 385
    },
    {
      "epoch": 0.420479302832244,
      "grad_norm": 1942.8695068359375,
      "learning_rate": 0.00012666085505262485,
      "loss": 11.3046,
      "step": 386
    },
    {
      "epoch": 0.4215686274509804,
      "grad_norm": 2619.687744140625,
      "learning_rate": 0.00012632722891729845,
      "loss": 13.1539,
      "step": 387
    },
    {
      "epoch": 0.4226579520697168,
      "grad_norm": 2373.228271484375,
      "learning_rate": 0.000125993287620758,
      "loss": 12.7629,
      "step": 388
    },
    {
      "epoch": 0.42374727668845313,
      "grad_norm": 2018.0404052734375,
      "learning_rate": 0.00012565903516058882,
      "loss": 12.3341,
      "step": 389
    },
    {
      "epoch": 0.42483660130718953,
      "grad_norm": 1830.7042236328125,
      "learning_rate": 0.00012532447553810126,
      "loss": 12.9322,
      "step": 390
    },
    {
      "epoch": 0.42592592592592593,
      "grad_norm": 3418.471923828125,
      "learning_rate": 0.00012498961275828247,
      "loss": 12.1159,
      "step": 391
    },
    {
      "epoch": 0.42701525054466233,
      "grad_norm": 1846.1995849609375,
      "learning_rate": 0.00012465445082974886,
      "loss": 10.8845,
      "step": 392
    },
    {
      "epoch": 0.42810457516339867,
      "grad_norm": 2432.753173828125,
      "learning_rate": 0.00012431899376469784,
      "loss": 12.9188,
      "step": 393
    },
    {
      "epoch": 0.42919389978213507,
      "grad_norm": 4167.3408203125,
      "learning_rate": 0.00012398324557885994,
      "loss": 13.8031,
      "step": 394
    },
    {
      "epoch": 0.43028322440087147,
      "grad_norm": 2242.124267578125,
      "learning_rate": 0.0001236472102914506,
      "loss": 12.6964,
      "step": 395
    },
    {
      "epoch": 0.43137254901960786,
      "grad_norm": 2378.07861328125,
      "learning_rate": 0.00012331089192512218,
      "loss": 12.1325,
      "step": 396
    },
    {
      "epoch": 0.4324618736383442,
      "grad_norm": 1914.8848876953125,
      "learning_rate": 0.00012297429450591575,
      "loss": 11.3239,
      "step": 397
    },
    {
      "epoch": 0.4335511982570806,
      "grad_norm": 5270.39697265625,
      "learning_rate": 0.00012263742206321287,
      "loss": 13.2649,
      "step": 398
    },
    {
      "epoch": 0.434640522875817,
      "grad_norm": 4266.642578125,
      "learning_rate": 0.00012230027862968743,
      "loss": 13.9944,
      "step": 399
    },
    {
      "epoch": 0.4357298474945534,
      "grad_norm": 2909.033203125,
      "learning_rate": 0.00012196286824125726,
      "loss": 12.5143,
      "step": 400
    },
    {
      "epoch": 0.43681917211328974,
      "grad_norm": 2279.06494140625,
      "learning_rate": 0.000121625194937036,
      "loss": 13.4036,
      "step": 401
    },
    {
      "epoch": 0.43790849673202614,
      "grad_norm": 3242.800048828125,
      "learning_rate": 0.0001212872627592845,
      "loss": 12.6792,
      "step": 402
    },
    {
      "epoch": 0.43899782135076254,
      "grad_norm": 2338.9423828125,
      "learning_rate": 0.00012094907575336267,
      "loss": 12.0955,
      "step": 403
    },
    {
      "epoch": 0.4400871459694989,
      "grad_norm": 2894.48046875,
      "learning_rate": 0.0001206106379676809,
      "loss": 13.7315,
      "step": 404
    },
    {
      "epoch": 0.4411764705882353,
      "grad_norm": 2612.1435546875,
      "learning_rate": 0.00012027195345365167,
      "loss": 13.4882,
      "step": 405
    },
    {
      "epoch": 0.4422657952069717,
      "grad_norm": 2377.11328125,
      "learning_rate": 0.00011993302626564102,
      "loss": 13.5782,
      "step": 406
    },
    {
      "epoch": 0.4433551198257081,
      "grad_norm": 3777.18701171875,
      "learning_rate": 0.00011959386046091998,
      "loss": 13.3125,
      "step": 407
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 2583.8447265625,
      "learning_rate": 0.00011925446009961607,
      "loss": 12.2817,
      "step": 408
    },
    {
      "epoch": 0.4455337690631808,
      "grad_norm": 1550.3592529296875,
      "learning_rate": 0.00011891482924466471,
      "loss": 12.1185,
      "step": 409
    },
    {
      "epoch": 0.4466230936819172,
      "grad_norm": 2206.906005859375,
      "learning_rate": 0.00011857497196176049,
      "loss": 13.4228,
      "step": 410
    },
    {
      "epoch": 0.4477124183006536,
      "grad_norm": 2955.05419921875,
      "learning_rate": 0.00011823489231930854,
      "loss": 14.0393,
      "step": 411
    },
    {
      "epoch": 0.44880174291938996,
      "grad_norm": 1879.666259765625,
      "learning_rate": 0.00011789459438837589,
      "loss": 12.6743,
      "step": 412
    },
    {
      "epoch": 0.44989106753812635,
      "grad_norm": 3064.472900390625,
      "learning_rate": 0.00011755408224264269,
      "loss": 13.7009,
      "step": 413
    },
    {
      "epoch": 0.45098039215686275,
      "grad_norm": 2606.11328125,
      "learning_rate": 0.00011721335995835336,
      "loss": 13.3532,
      "step": 414
    },
    {
      "epoch": 0.45206971677559915,
      "grad_norm": 2937.61083984375,
      "learning_rate": 0.00011687243161426793,
      "loss": 13.8534,
      "step": 415
    },
    {
      "epoch": 0.4531590413943355,
      "grad_norm": 2084.185791015625,
      "learning_rate": 0.00011653130129161316,
      "loss": 13.3512,
      "step": 416
    },
    {
      "epoch": 0.4542483660130719,
      "grad_norm": 2084.17578125,
      "learning_rate": 0.00011618997307403367,
      "loss": 13.8318,
      "step": 417
    },
    {
      "epoch": 0.4553376906318083,
      "grad_norm": 3513.52734375,
      "learning_rate": 0.00011584845104754304,
      "loss": 14.0658,
      "step": 418
    },
    {
      "epoch": 0.4564270152505447,
      "grad_norm": 1889.5738525390625,
      "learning_rate": 0.00011550673930047498,
      "loss": 13.4544,
      "step": 419
    },
    {
      "epoch": 0.45751633986928103,
      "grad_norm": 2762.25439453125,
      "learning_rate": 0.00011516484192343425,
      "loss": 13.4629,
      "step": 420
    },
    {
      "epoch": 0.45860566448801743,
      "grad_norm": 2621.8154296875,
      "learning_rate": 0.00011482276300924782,
      "loss": 13.2982,
      "step": 421
    },
    {
      "epoch": 0.4596949891067538,
      "grad_norm": 2016.842041015625,
      "learning_rate": 0.00011448050665291587,
      "loss": 12.8009,
      "step": 422
    },
    {
      "epoch": 0.46078431372549017,
      "grad_norm": 2019.709228515625,
      "learning_rate": 0.00011413807695156262,
      "loss": 12.7923,
      "step": 423
    },
    {
      "epoch": 0.46187363834422657,
      "grad_norm": 2043.737060546875,
      "learning_rate": 0.00011379547800438747,
      "loss": 13.6543,
      "step": 424
    },
    {
      "epoch": 0.46296296296296297,
      "grad_norm": 1839.9501953125,
      "learning_rate": 0.00011345271391261584,
      "loss": 12.6804,
      "step": 425
    },
    {
      "epoch": 0.46405228758169936,
      "grad_norm": 1405.8519287109375,
      "learning_rate": 0.00011310978877945007,
      "loss": 12.3567,
      "step": 426
    },
    {
      "epoch": 0.4651416122004357,
      "grad_norm": 2460.86572265625,
      "learning_rate": 0.00011276670671002028,
      "loss": 12.8338,
      "step": 427
    },
    {
      "epoch": 0.4662309368191721,
      "grad_norm": 1966.4986572265625,
      "learning_rate": 0.00011242347181133533,
      "loss": 13.6811,
      "step": 428
    },
    {
      "epoch": 0.4673202614379085,
      "grad_norm": 2038.7354736328125,
      "learning_rate": 0.00011208008819223354,
      "loss": 13.591,
      "step": 429
    },
    {
      "epoch": 0.4684095860566449,
      "grad_norm": 2427.309814453125,
      "learning_rate": 0.00011173655996333357,
      "loss": 13.6258,
      "step": 430
    },
    {
      "epoch": 0.46949891067538124,
      "grad_norm": 3444.146484375,
      "learning_rate": 0.00011139289123698518,
      "loss": 12.7897,
      "step": 431
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 1727.6273193359375,
      "learning_rate": 0.00011104908612722001,
      "loss": 13.4163,
      "step": 432
    },
    {
      "epoch": 0.47167755991285404,
      "grad_norm": 2945.229736328125,
      "learning_rate": 0.00011070514874970237,
      "loss": 12.7424,
      "step": 433
    },
    {
      "epoch": 0.47276688453159044,
      "grad_norm": 1918.266845703125,
      "learning_rate": 0.00011036108322167988,
      "loss": 12.9925,
      "step": 434
    },
    {
      "epoch": 0.4738562091503268,
      "grad_norm": 1834.51123046875,
      "learning_rate": 0.00011001689366193433,
      "loss": 13.6747,
      "step": 435
    },
    {
      "epoch": 0.4749455337690632,
      "grad_norm": 1388.4371337890625,
      "learning_rate": 0.00010967258419073217,
      "loss": 13.6131,
      "step": 436
    },
    {
      "epoch": 0.4760348583877996,
      "grad_norm": 2213.042236328125,
      "learning_rate": 0.00010932815892977535,
      "loss": 13.7537,
      "step": 437
    },
    {
      "epoch": 0.477124183006536,
      "grad_norm": 1625.793212890625,
      "learning_rate": 0.00010898362200215197,
      "loss": 13.7964,
      "step": 438
    },
    {
      "epoch": 0.4782135076252723,
      "grad_norm": 1119.17919921875,
      "learning_rate": 0.00010863897753228687,
      "loss": 13.6206,
      "step": 439
    },
    {
      "epoch": 0.4793028322440087,
      "grad_norm": 2075.09228515625,
      "learning_rate": 0.0001082942296458922,
      "loss": 13.5352,
      "step": 440
    },
    {
      "epoch": 0.4803921568627451,
      "grad_norm": 2514.07958984375,
      "learning_rate": 0.00010794938246991817,
      "loss": 14.4007,
      "step": 441
    },
    {
      "epoch": 0.48148148148148145,
      "grad_norm": 1654.1142578125,
      "learning_rate": 0.0001076044401325036,
      "loss": 14.1484,
      "step": 442
    },
    {
      "epoch": 0.48257080610021785,
      "grad_norm": 1555.315185546875,
      "learning_rate": 0.00010725940676292636,
      "loss": 12.6995,
      "step": 443
    },
    {
      "epoch": 0.48366013071895425,
      "grad_norm": 1430.394287109375,
      "learning_rate": 0.0001069142864915542,
      "loss": 12.6135,
      "step": 444
    },
    {
      "epoch": 0.48474945533769065,
      "grad_norm": 1317.9832763671875,
      "learning_rate": 0.00010656908344979506,
      "loss": 13.5794,
      "step": 445
    },
    {
      "epoch": 0.485838779956427,
      "grad_norm": 1727.5115966796875,
      "learning_rate": 0.0001062238017700478,
      "loss": 13.8786,
      "step": 446
    },
    {
      "epoch": 0.4869281045751634,
      "grad_norm": 1529.5556640625,
      "learning_rate": 0.00010587844558565261,
      "loss": 13.5966,
      "step": 447
    },
    {
      "epoch": 0.4880174291938998,
      "grad_norm": 1605.1805419921875,
      "learning_rate": 0.00010553301903084157,
      "loss": 14.9925,
      "step": 448
    },
    {
      "epoch": 0.4891067538126362,
      "grad_norm": 1363.312255859375,
      "learning_rate": 0.00010518752624068911,
      "loss": 13.2978,
      "step": 449
    },
    {
      "epoch": 0.49019607843137253,
      "grad_norm": 1511.412841796875,
      "learning_rate": 0.00010484197135106263,
      "loss": 15.2164,
      "step": 450
    },
    {
      "epoch": 0.4912854030501089,
      "grad_norm": 1557.5146484375,
      "learning_rate": 0.0001044963584985729,
      "loss": 13.6396,
      "step": 451
    },
    {
      "epoch": 0.4923747276688453,
      "grad_norm": 1717.6806640625,
      "learning_rate": 0.0001041506918205246,
      "loss": 13.6635,
      "step": 452
    },
    {
      "epoch": 0.4934640522875817,
      "grad_norm": 1665.452880859375,
      "learning_rate": 0.00010380497545486663,
      "loss": 14.2642,
      "step": 453
    },
    {
      "epoch": 0.49455337690631807,
      "grad_norm": 1279.9322509765625,
      "learning_rate": 0.00010345921354014279,
      "loss": 14.4201,
      "step": 454
    },
    {
      "epoch": 0.49564270152505446,
      "grad_norm": 1384.8275146484375,
      "learning_rate": 0.00010311341021544218,
      "loss": 15.3935,
      "step": 455
    },
    {
      "epoch": 0.49673202614379086,
      "grad_norm": 1815.6390380859375,
      "learning_rate": 0.0001027675696203495,
      "loss": 14.6728,
      "step": 456
    },
    {
      "epoch": 0.49782135076252726,
      "grad_norm": 1295.0675048828125,
      "learning_rate": 0.00010242169589489568,
      "loss": 12.9198,
      "step": 457
    },
    {
      "epoch": 0.4989106753812636,
      "grad_norm": 1431.1907958984375,
      "learning_rate": 0.00010207579317950827,
      "loss": 14.3134,
      "step": 458
    },
    {
      "epoch": 0.5,
      "grad_norm": 1028.9598388671875,
      "learning_rate": 0.0001017298656149618,
      "loss": 13.316,
      "step": 459
    },
    {
      "epoch": 0.5010893246187363,
      "grad_norm": 1440.857421875,
      "learning_rate": 0.00010138391734232832,
      "loss": 14.7119,
      "step": 460
    },
    {
      "epoch": 0.5010893246187363,
      "eval_loss": 3.616103410720825,
      "eval_runtime": 2.7823,
      "eval_samples_per_second": 139.095,
      "eval_steps_per_second": 69.727,
      "step": 460
    },
    {
      "epoch": 0.5021786492374728,
      "grad_norm": 1697.1995849609375,
      "learning_rate": 0.00010103795250292778,
      "loss": 15.0099,
      "step": 461
    },
    {
      "epoch": 0.5032679738562091,
      "grad_norm": 1438.0911865234375,
      "learning_rate": 0.00010069197523827833,
      "loss": 16.3135,
      "step": 462
    },
    {
      "epoch": 0.5043572984749455,
      "grad_norm": 1578.0306396484375,
      "learning_rate": 0.00010034598969004705,
      "loss": 14.3354,
      "step": 463
    },
    {
      "epoch": 0.5054466230936819,
      "grad_norm": 1145.7254638671875,
      "learning_rate": 0.0001,
      "loss": 15.1201,
      "step": 464
    },
    {
      "epoch": 0.5065359477124183,
      "grad_norm": 1281.2977294921875,
      "learning_rate": 9.965401030995301e-05,
      "loss": 15.0423,
      "step": 465
    },
    {
      "epoch": 0.5076252723311547,
      "grad_norm": 2497.565673828125,
      "learning_rate": 9.930802476172169e-05,
      "loss": 12.9232,
      "step": 466
    },
    {
      "epoch": 0.5087145969498911,
      "grad_norm": 1575.5599365234375,
      "learning_rate": 9.896204749707228e-05,
      "loss": 14.6681,
      "step": 467
    },
    {
      "epoch": 0.5098039215686274,
      "grad_norm": 941.0279541015625,
      "learning_rate": 9.861608265767167e-05,
      "loss": 14.7297,
      "step": 468
    },
    {
      "epoch": 0.5108932461873639,
      "grad_norm": 1556.48388671875,
      "learning_rate": 9.827013438503822e-05,
      "loss": 14.6033,
      "step": 469
    },
    {
      "epoch": 0.5119825708061002,
      "grad_norm": 1751.578125,
      "learning_rate": 9.792420682049174e-05,
      "loss": 15.0415,
      "step": 470
    },
    {
      "epoch": 0.5130718954248366,
      "grad_norm": 1077.553955078125,
      "learning_rate": 9.757830410510433e-05,
      "loss": 13.5249,
      "step": 471
    },
    {
      "epoch": 0.514161220043573,
      "grad_norm": 1023.6539916992188,
      "learning_rate": 9.723243037965056e-05,
      "loss": 13.8337,
      "step": 472
    },
    {
      "epoch": 0.5152505446623094,
      "grad_norm": 1367.58984375,
      "learning_rate": 9.688658978455784e-05,
      "loss": 15.3791,
      "step": 473
    },
    {
      "epoch": 0.5163398692810458,
      "grad_norm": 1138.333740234375,
      "learning_rate": 9.654078645985722e-05,
      "loss": 14.343,
      "step": 474
    },
    {
      "epoch": 0.5174291938997821,
      "grad_norm": 958.8590698242188,
      "learning_rate": 9.619502454513338e-05,
      "loss": 14.1471,
      "step": 475
    },
    {
      "epoch": 0.5185185185185185,
      "grad_norm": 1362.31689453125,
      "learning_rate": 9.584930817947544e-05,
      "loss": 14.2489,
      "step": 476
    },
    {
      "epoch": 0.5196078431372549,
      "grad_norm": 1034.6534423828125,
      "learning_rate": 9.550364150142713e-05,
      "loss": 13.4652,
      "step": 477
    },
    {
      "epoch": 0.5206971677559913,
      "grad_norm": 1429.5125732421875,
      "learning_rate": 9.515802864893739e-05,
      "loss": 14.1628,
      "step": 478
    },
    {
      "epoch": 0.5217864923747276,
      "grad_norm": 1044.7430419921875,
      "learning_rate": 9.481247375931094e-05,
      "loss": 13.1049,
      "step": 479
    },
    {
      "epoch": 0.5228758169934641,
      "grad_norm": 1164.5643310546875,
      "learning_rate": 9.446698096915847e-05,
      "loss": 13.8229,
      "step": 480
    },
    {
      "epoch": 0.5239651416122004,
      "grad_norm": 1188.510986328125,
      "learning_rate": 9.412155441434741e-05,
      "loss": 13.4517,
      "step": 481
    },
    {
      "epoch": 0.5250544662309368,
      "grad_norm": 1218.689208984375,
      "learning_rate": 9.377619822995219e-05,
      "loss": 15.032,
      "step": 482
    },
    {
      "epoch": 0.5261437908496732,
      "grad_norm": 802.5537109375,
      "learning_rate": 9.343091655020495e-05,
      "loss": 13.9797,
      "step": 483
    },
    {
      "epoch": 0.5272331154684096,
      "grad_norm": 938.1593627929688,
      "learning_rate": 9.308571350844584e-05,
      "loss": 13.8941,
      "step": 484
    },
    {
      "epoch": 0.528322440087146,
      "grad_norm": 1472.3194580078125,
      "learning_rate": 9.274059323707366e-05,
      "loss": 13.2638,
      "step": 485
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 1048.2354736328125,
      "learning_rate": 9.239555986749645e-05,
      "loss": 14.3531,
      "step": 486
    },
    {
      "epoch": 0.5305010893246187,
      "grad_norm": 1296.9527587890625,
      "learning_rate": 9.205061753008183e-05,
      "loss": 14.203,
      "step": 487
    },
    {
      "epoch": 0.5315904139433552,
      "grad_norm": 1657.743896484375,
      "learning_rate": 9.170577035410783e-05,
      "loss": 13.7112,
      "step": 488
    },
    {
      "epoch": 0.5326797385620915,
      "grad_norm": 1212.4053955078125,
      "learning_rate": 9.136102246771314e-05,
      "loss": 13.2756,
      "step": 489
    },
    {
      "epoch": 0.5337690631808278,
      "grad_norm": 1304.51220703125,
      "learning_rate": 9.101637799784804e-05,
      "loss": 14.4237,
      "step": 490
    },
    {
      "epoch": 0.5348583877995643,
      "grad_norm": 1542.832275390625,
      "learning_rate": 9.06718410702247e-05,
      "loss": 14.4378,
      "step": 491
    },
    {
      "epoch": 0.5359477124183006,
      "grad_norm": 1302.9571533203125,
      "learning_rate": 9.032741580926787e-05,
      "loss": 13.1055,
      "step": 492
    },
    {
      "epoch": 0.5370370370370371,
      "grad_norm": 1939.5728759765625,
      "learning_rate": 8.998310633806571e-05,
      "loss": 12.4175,
      "step": 493
    },
    {
      "epoch": 0.5381263616557734,
      "grad_norm": 1355.798095703125,
      "learning_rate": 8.963891677832011e-05,
      "loss": 12.9592,
      "step": 494
    },
    {
      "epoch": 0.5392156862745098,
      "grad_norm": 2074.481201171875,
      "learning_rate": 8.929485125029766e-05,
      "loss": 13.6742,
      "step": 495
    },
    {
      "epoch": 0.5403050108932462,
      "grad_norm": 1087.583740234375,
      "learning_rate": 8.895091387277999e-05,
      "loss": 14.9366,
      "step": 496
    },
    {
      "epoch": 0.5413943355119826,
      "grad_norm": 1182.4432373046875,
      "learning_rate": 8.860710876301484e-05,
      "loss": 12.0956,
      "step": 497
    },
    {
      "epoch": 0.5424836601307189,
      "grad_norm": 1062.96923828125,
      "learning_rate": 8.826344003666647e-05,
      "loss": 13.0663,
      "step": 498
    },
    {
      "epoch": 0.5435729847494554,
      "grad_norm": 1257.4434814453125,
      "learning_rate": 8.791991180776648e-05,
      "loss": 14.0272,
      "step": 499
    },
    {
      "epoch": 0.5446623093681917,
      "grad_norm": 1445.0548095703125,
      "learning_rate": 8.757652818866471e-05,
      "loss": 12.8504,
      "step": 500
    },
    {
      "epoch": 0.545751633986928,
      "grad_norm": 1665.5859375,
      "learning_rate": 8.723329328997973e-05,
      "loss": 13.8356,
      "step": 501
    },
    {
      "epoch": 0.5468409586056645,
      "grad_norm": 995.88427734375,
      "learning_rate": 8.689021122054996e-05,
      "loss": 12.8165,
      "step": 502
    },
    {
      "epoch": 0.5479302832244008,
      "grad_norm": 1658.434326171875,
      "learning_rate": 8.654728608738418e-05,
      "loss": 14.196,
      "step": 503
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 984.2662963867188,
      "learning_rate": 8.620452199561254e-05,
      "loss": 13.6334,
      "step": 504
    },
    {
      "epoch": 0.5501089324618736,
      "grad_norm": 1411.3946533203125,
      "learning_rate": 8.58619230484374e-05,
      "loss": 13.7294,
      "step": 505
    },
    {
      "epoch": 0.55119825708061,
      "grad_norm": 1083.3192138671875,
      "learning_rate": 8.551949334708415e-05,
      "loss": 12.9704,
      "step": 506
    },
    {
      "epoch": 0.5522875816993464,
      "grad_norm": 1257.832275390625,
      "learning_rate": 8.51772369907522e-05,
      "loss": 12.5718,
      "step": 507
    },
    {
      "epoch": 0.5533769063180828,
      "grad_norm": 871.9485473632812,
      "learning_rate": 8.483515807656576e-05,
      "loss": 13.1949,
      "step": 508
    },
    {
      "epoch": 0.5544662309368191,
      "grad_norm": 1094.6038818359375,
      "learning_rate": 8.449326069952506e-05,
      "loss": 14.2022,
      "step": 509
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 1254.0933837890625,
      "learning_rate": 8.415154895245697e-05,
      "loss": 13.1929,
      "step": 510
    },
    {
      "epoch": 0.5566448801742919,
      "grad_norm": 819.5022583007812,
      "learning_rate": 8.381002692596635e-05,
      "loss": 13.739,
      "step": 511
    },
    {
      "epoch": 0.5577342047930284,
      "grad_norm": 1080.34521484375,
      "learning_rate": 8.346869870838685e-05,
      "loss": 12.4172,
      "step": 512
    },
    {
      "epoch": 0.5588235294117647,
      "grad_norm": 844.1549072265625,
      "learning_rate": 8.312756838573208e-05,
      "loss": 14.1843,
      "step": 513
    },
    {
      "epoch": 0.5599128540305011,
      "grad_norm": 1664.24462890625,
      "learning_rate": 8.278664004164665e-05,
      "loss": 13.9817,
      "step": 514
    },
    {
      "epoch": 0.5610021786492375,
      "grad_norm": 1837.83544921875,
      "learning_rate": 8.244591775735732e-05,
      "loss": 14.2748,
      "step": 515
    },
    {
      "epoch": 0.5620915032679739,
      "grad_norm": 794.2677612304688,
      "learning_rate": 8.210540561162412e-05,
      "loss": 13.3806,
      "step": 516
    },
    {
      "epoch": 0.5631808278867102,
      "grad_norm": 1501.6959228515625,
      "learning_rate": 8.176510768069147e-05,
      "loss": 13.5997,
      "step": 517
    },
    {
      "epoch": 0.5642701525054467,
      "grad_norm": 1058.7418212890625,
      "learning_rate": 8.142502803823955e-05,
      "loss": 12.9171,
      "step": 518
    },
    {
      "epoch": 0.565359477124183,
      "grad_norm": 1096.179443359375,
      "learning_rate": 8.108517075533531e-05,
      "loss": 13.1182,
      "step": 519
    },
    {
      "epoch": 0.5664488017429193,
      "grad_norm": 1459.3277587890625,
      "learning_rate": 8.074553990038395e-05,
      "loss": 14.0618,
      "step": 520
    },
    {
      "epoch": 0.5675381263616558,
      "grad_norm": 1117.51318359375,
      "learning_rate": 8.040613953908005e-05,
      "loss": 12.8453,
      "step": 521
    },
    {
      "epoch": 0.5686274509803921,
      "grad_norm": 1164.1474609375,
      "learning_rate": 8.0066973734359e-05,
      "loss": 13.228,
      "step": 522
    },
    {
      "epoch": 0.5697167755991286,
      "grad_norm": 1310.04296875,
      "learning_rate": 7.972804654634834e-05,
      "loss": 13.3479,
      "step": 523
    },
    {
      "epoch": 0.5708061002178649,
      "grad_norm": 1271.54833984375,
      "learning_rate": 7.938936203231912e-05,
      "loss": 12.7188,
      "step": 524
    },
    {
      "epoch": 0.5718954248366013,
      "grad_norm": 1231.713623046875,
      "learning_rate": 7.905092424663735e-05,
      "loss": 14.596,
      "step": 525
    },
    {
      "epoch": 0.5729847494553377,
      "grad_norm": 2087.1767578125,
      "learning_rate": 7.871273724071553e-05,
      "loss": 12.5966,
      "step": 526
    },
    {
      "epoch": 0.5740740740740741,
      "grad_norm": 1349.419189453125,
      "learning_rate": 7.837480506296404e-05,
      "loss": 12.99,
      "step": 527
    },
    {
      "epoch": 0.5751633986928104,
      "grad_norm": 1333.2598876953125,
      "learning_rate": 7.803713175874275e-05,
      "loss": 12.8456,
      "step": 528
    },
    {
      "epoch": 0.5762527233115469,
      "grad_norm": 1339.005615234375,
      "learning_rate": 7.769972137031262e-05,
      "loss": 14.4523,
      "step": 529
    },
    {
      "epoch": 0.5773420479302832,
      "grad_norm": 1520.0718994140625,
      "learning_rate": 7.736257793678714e-05,
      "loss": 13.4148,
      "step": 530
    },
    {
      "epoch": 0.5784313725490197,
      "grad_norm": 1333.5904541015625,
      "learning_rate": 7.702570549408428e-05,
      "loss": 13.4515,
      "step": 531
    },
    {
      "epoch": 0.579520697167756,
      "grad_norm": 1032.6591796875,
      "learning_rate": 7.668910807487783e-05,
      "loss": 13.9835,
      "step": 532
    },
    {
      "epoch": 0.5806100217864923,
      "grad_norm": 1167.9801025390625,
      "learning_rate": 7.635278970854943e-05,
      "loss": 13.1408,
      "step": 533
    },
    {
      "epoch": 0.5816993464052288,
      "grad_norm": 1407.249267578125,
      "learning_rate": 7.601675442114009e-05,
      "loss": 13.9407,
      "step": 534
    },
    {
      "epoch": 0.5827886710239651,
      "grad_norm": 2209.746826171875,
      "learning_rate": 7.568100623530217e-05,
      "loss": 12.4755,
      "step": 535
    },
    {
      "epoch": 0.5838779956427015,
      "grad_norm": 1797.746826171875,
      "learning_rate": 7.534554917025119e-05,
      "loss": 12.8268,
      "step": 536
    },
    {
      "epoch": 0.5849673202614379,
      "grad_norm": 1302.199462890625,
      "learning_rate": 7.501038724171756e-05,
      "loss": 13.197,
      "step": 537
    },
    {
      "epoch": 0.5860566448801743,
      "grad_norm": 2067.767822265625,
      "learning_rate": 7.46755244618988e-05,
      "loss": 12.8685,
      "step": 538
    },
    {
      "epoch": 0.5871459694989106,
      "grad_norm": 1580.0186767578125,
      "learning_rate": 7.434096483941115e-05,
      "loss": 13.4972,
      "step": 539
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 894.298828125,
      "learning_rate": 7.400671237924202e-05,
      "loss": 13.1393,
      "step": 540
    },
    {
      "epoch": 0.5893246187363834,
      "grad_norm": 1165.6744384765625,
      "learning_rate": 7.367277108270156e-05,
      "loss": 13.9111,
      "step": 541
    },
    {
      "epoch": 0.5904139433551199,
      "grad_norm": 1790.1239013671875,
      "learning_rate": 7.333914494737514e-05,
      "loss": 13.351,
      "step": 542
    },
    {
      "epoch": 0.5915032679738562,
      "grad_norm": 984.1708374023438,
      "learning_rate": 7.300583796707539e-05,
      "loss": 14.5447,
      "step": 543
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 1341.8519287109375,
      "learning_rate": 7.267285413179421e-05,
      "loss": 14.1327,
      "step": 544
    },
    {
      "epoch": 0.593681917211329,
      "grad_norm": 1000.2693481445312,
      "learning_rate": 7.234019742765532e-05,
      "loss": 13.3989,
      "step": 545
    },
    {
      "epoch": 0.5947712418300654,
      "grad_norm": 1023.481689453125,
      "learning_rate": 7.200787183686625e-05,
      "loss": 13.069,
      "step": 546
    },
    {
      "epoch": 0.5958605664488017,
      "grad_norm": 1108.57470703125,
      "learning_rate": 7.167588133767091e-05,
      "loss": 12.2698,
      "step": 547
    },
    {
      "epoch": 0.5969498910675382,
      "grad_norm": 964.7139282226562,
      "learning_rate": 7.134422990430176e-05,
      "loss": 13.6585,
      "step": 548
    },
    {
      "epoch": 0.5980392156862745,
      "grad_norm": 1239.9150390625,
      "learning_rate": 7.101292150693241e-05,
      "loss": 13.0273,
      "step": 549
    },
    {
      "epoch": 0.599128540305011,
      "grad_norm": 846.2174072265625,
      "learning_rate": 7.068196011162994e-05,
      "loss": 13.0756,
      "step": 550
    },
    {
      "epoch": 0.6002178649237473,
      "grad_norm": 987.310546875,
      "learning_rate": 7.03513496803075e-05,
      "loss": 14.1704,
      "step": 551
    },
    {
      "epoch": 0.6013071895424836,
      "grad_norm": 5245.39697265625,
      "learning_rate": 7.002109417067697e-05,
      "loss": 14.4467,
      "step": 552
    },
    {
      "epoch": 0.6023965141612201,
      "grad_norm": 1111.834228515625,
      "learning_rate": 6.969119753620135e-05,
      "loss": 13.517,
      "step": 553
    },
    {
      "epoch": 0.6034858387799564,
      "grad_norm": 996.4967041015625,
      "learning_rate": 6.936166372604773e-05,
      "loss": 13.5025,
      "step": 554
    },
    {
      "epoch": 0.6045751633986928,
      "grad_norm": 867.002685546875,
      "learning_rate": 6.903249668503972e-05,
      "loss": 12.8567,
      "step": 555
    },
    {
      "epoch": 0.6056644880174292,
      "grad_norm": 1155.432373046875,
      "learning_rate": 6.87037003536104e-05,
      "loss": 13.0582,
      "step": 556
    },
    {
      "epoch": 0.6067538126361656,
      "grad_norm": 1074.625732421875,
      "learning_rate": 6.837527866775522e-05,
      "loss": 13.5709,
      "step": 557
    },
    {
      "epoch": 0.6078431372549019,
      "grad_norm": 1409.87841796875,
      "learning_rate": 6.804723555898458e-05,
      "loss": 13.7728,
      "step": 558
    },
    {
      "epoch": 0.6089324618736384,
      "grad_norm": 987.9152221679688,
      "learning_rate": 6.771957495427716e-05,
      "loss": 13.0499,
      "step": 559
    },
    {
      "epoch": 0.6100217864923747,
      "grad_norm": 1056.71240234375,
      "learning_rate": 6.739230077603259e-05,
      "loss": 13.4185,
      "step": 560
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 965.7537841796875,
      "learning_rate": 6.706541694202471e-05,
      "loss": 13.3033,
      "step": 561
    },
    {
      "epoch": 0.6122004357298475,
      "grad_norm": 950.843505859375,
      "learning_rate": 6.673892736535448e-05,
      "loss": 13.2638,
      "step": 562
    },
    {
      "epoch": 0.6132897603485838,
      "grad_norm": 888.200439453125,
      "learning_rate": 6.641283595440323e-05,
      "loss": 14.2555,
      "step": 563
    },
    {
      "epoch": 0.6143790849673203,
      "grad_norm": 1082.6995849609375,
      "learning_rate": 6.608714661278606e-05,
      "loss": 13.0727,
      "step": 564
    },
    {
      "epoch": 0.6154684095860566,
      "grad_norm": 1750.720947265625,
      "learning_rate": 6.576186323930466e-05,
      "loss": 12.9111,
      "step": 565
    },
    {
      "epoch": 0.616557734204793,
      "grad_norm": 1437.4244384765625,
      "learning_rate": 6.543698972790117e-05,
      "loss": 14.7961,
      "step": 566
    },
    {
      "epoch": 0.6176470588235294,
      "grad_norm": 1005.5140380859375,
      "learning_rate": 6.51125299676111e-05,
      "loss": 14.3845,
      "step": 567
    },
    {
      "epoch": 0.6187363834422658,
      "grad_norm": 802.7247924804688,
      "learning_rate": 6.478848784251713e-05,
      "loss": 14.1054,
      "step": 568
    },
    {
      "epoch": 0.6198257080610022,
      "grad_norm": 780.930908203125,
      "learning_rate": 6.446486723170236e-05,
      "loss": 13.4999,
      "step": 569
    },
    {
      "epoch": 0.6209150326797386,
      "grad_norm": 1568.1295166015625,
      "learning_rate": 6.414167200920391e-05,
      "loss": 14.3808,
      "step": 570
    },
    {
      "epoch": 0.6220043572984749,
      "grad_norm": 649.1087036132812,
      "learning_rate": 6.381890604396687e-05,
      "loss": 14.4266,
      "step": 571
    },
    {
      "epoch": 0.6230936819172114,
      "grad_norm": 1113.2542724609375,
      "learning_rate": 6.349657319979742e-05,
      "loss": 14.2873,
      "step": 572
    },
    {
      "epoch": 0.6241830065359477,
      "grad_norm": 735.5198974609375,
      "learning_rate": 6.317467733531712e-05,
      "loss": 14.1431,
      "step": 573
    },
    {
      "epoch": 0.6252723311546841,
      "grad_norm": 889.42822265625,
      "learning_rate": 6.28532223039163e-05,
      "loss": 14.5709,
      "step": 574
    },
    {
      "epoch": 0.6263616557734205,
      "grad_norm": 1000.8912963867188,
      "learning_rate": 6.253221195370826e-05,
      "loss": 14.9789,
      "step": 575
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 1058.2774658203125,
      "learning_rate": 6.221165012748297e-05,
      "loss": 13.2536,
      "step": 576
    },
    {
      "epoch": 0.6285403050108932,
      "grad_norm": 752.534912109375,
      "learning_rate": 6.189154066266112e-05,
      "loss": 14.1625,
      "step": 577
    },
    {
      "epoch": 0.6296296296296297,
      "grad_norm": 756.6798095703125,
      "learning_rate": 6.157188739124834e-05,
      "loss": 13.6298,
      "step": 578
    },
    {
      "epoch": 0.630718954248366,
      "grad_norm": 1065.2354736328125,
      "learning_rate": 6.125269413978907e-05,
      "loss": 13.4351,
      "step": 579
    },
    {
      "epoch": 0.6318082788671024,
      "grad_norm": 1031.8475341796875,
      "learning_rate": 6.093396472932103e-05,
      "loss": 12.9427,
      "step": 580
    },
    {
      "epoch": 0.6328976034858388,
      "grad_norm": 545.2572631835938,
      "learning_rate": 6.0615702975329194e-05,
      "loss": 13.4271,
      "step": 581
    },
    {
      "epoch": 0.6339869281045751,
      "grad_norm": 629.153076171875,
      "learning_rate": 6.029791268770029e-05,
      "loss": 13.7342,
      "step": 582
    },
    {
      "epoch": 0.6350762527233116,
      "grad_norm": 808.3855590820312,
      "learning_rate": 5.998059767067728e-05,
      "loss": 12.8523,
      "step": 583
    },
    {
      "epoch": 0.6361655773420479,
      "grad_norm": 966.9733276367188,
      "learning_rate": 5.9663761722813495e-05,
      "loss": 13.8446,
      "step": 584
    },
    {
      "epoch": 0.6372549019607843,
      "grad_norm": 832.753173828125,
      "learning_rate": 5.934740863692759e-05,
      "loss": 14.0291,
      "step": 585
    },
    {
      "epoch": 0.6383442265795207,
      "grad_norm": 870.2808837890625,
      "learning_rate": 5.903154220005771e-05,
      "loss": 14.7135,
      "step": 586
    },
    {
      "epoch": 0.6394335511982571,
      "grad_norm": 888.448974609375,
      "learning_rate": 5.871616619341653e-05,
      "loss": 13.6237,
      "step": 587
    },
    {
      "epoch": 0.6405228758169934,
      "grad_norm": 604.5468139648438,
      "learning_rate": 5.840128439234571e-05,
      "loss": 13.1112,
      "step": 588
    },
    {
      "epoch": 0.6416122004357299,
      "grad_norm": 740.3869018554688,
      "learning_rate": 5.80869005662708e-05,
      "loss": 13.2927,
      "step": 589
    },
    {
      "epoch": 0.6427015250544662,
      "grad_norm": 729.2908935546875,
      "learning_rate": 5.777301847865629e-05,
      "loss": 13.1883,
      "step": 590
    },
    {
      "epoch": 0.6437908496732027,
      "grad_norm": 679.76171875,
      "learning_rate": 5.7459641886960244e-05,
      "loss": 12.7278,
      "step": 591
    },
    {
      "epoch": 0.644880174291939,
      "grad_norm": 771.1044921875,
      "learning_rate": 5.714677454258947e-05,
      "loss": 14.2043,
      "step": 592
    },
    {
      "epoch": 0.6459694989106753,
      "grad_norm": 525.4684448242188,
      "learning_rate": 5.6834420190854745e-05,
      "loss": 13.7949,
      "step": 593
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 1367.8231201171875,
      "learning_rate": 5.652258257092569e-05,
      "loss": 13.5713,
      "step": 594
    },
    {
      "epoch": 0.6481481481481481,
      "grad_norm": 1199.23876953125,
      "learning_rate": 5.621126541578632e-05,
      "loss": 13.6268,
      "step": 595
    },
    {
      "epoch": 0.6492374727668845,
      "grad_norm": 995.0260009765625,
      "learning_rate": 5.590047245219009e-05,
      "loss": 14.3565,
      "step": 596
    },
    {
      "epoch": 0.6503267973856209,
      "grad_norm": 805.2388916015625,
      "learning_rate": 5.559020740061549e-05,
      "loss": 14.2207,
      "step": 597
    },
    {
      "epoch": 0.6514161220043573,
      "grad_norm": 779.6170654296875,
      "learning_rate": 5.528047397522133e-05,
      "loss": 13.4574,
      "step": 598
    },
    {
      "epoch": 0.6525054466230937,
      "grad_norm": 974.6607055664062,
      "learning_rate": 5.497127588380244e-05,
      "loss": 13.5537,
      "step": 599
    },
    {
      "epoch": 0.6535947712418301,
      "grad_norm": 957.2470092773438,
      "learning_rate": 5.4662616827745185e-05,
      "loss": 13.6918,
      "step": 600
    },
    {
      "epoch": 0.6546840958605664,
      "grad_norm": 564.2131958007812,
      "learning_rate": 5.4354500501983074e-05,
      "loss": 14.5732,
      "step": 601
    },
    {
      "epoch": 0.6557734204793029,
      "grad_norm": 970.4561157226562,
      "learning_rate": 5.404693059495285e-05,
      "loss": 14.259,
      "step": 602
    },
    {
      "epoch": 0.6568627450980392,
      "grad_norm": 1475.91943359375,
      "learning_rate": 5.373991078854992e-05,
      "loss": 13.7511,
      "step": 603
    },
    {
      "epoch": 0.6579520697167756,
      "grad_norm": 1206.491943359375,
      "learning_rate": 5.3433444758084604e-05,
      "loss": 13.9658,
      "step": 604
    },
    {
      "epoch": 0.659041394335512,
      "grad_norm": 729.8560791015625,
      "learning_rate": 5.312753617223794e-05,
      "loss": 12.3194,
      "step": 605
    },
    {
      "epoch": 0.6601307189542484,
      "grad_norm": 833.2518310546875,
      "learning_rate": 5.282218869301788e-05,
      "loss": 13.2262,
      "step": 606
    },
    {
      "epoch": 0.6612200435729847,
      "grad_norm": 1121.98388671875,
      "learning_rate": 5.251740597571542e-05,
      "loss": 13.1887,
      "step": 607
    },
    {
      "epoch": 0.6623093681917211,
      "grad_norm": 732.55810546875,
      "learning_rate": 5.221319166886073e-05,
      "loss": 11.9579,
      "step": 608
    },
    {
      "epoch": 0.6633986928104575,
      "grad_norm": 795.4283447265625,
      "learning_rate": 5.190954941417977e-05,
      "loss": 13.6673,
      "step": 609
    },
    {
      "epoch": 0.664488017429194,
      "grad_norm": 742.3599243164062,
      "learning_rate": 5.160648284655032e-05,
      "loss": 13.4396,
      "step": 610
    },
    {
      "epoch": 0.6655773420479303,
      "grad_norm": 872.8551635742188,
      "learning_rate": 5.1303995593958824e-05,
      "loss": 14.0764,
      "step": 611
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 865.8281860351562,
      "learning_rate": 5.100209127745661e-05,
      "loss": 13.2594,
      "step": 612
    },
    {
      "epoch": 0.6677559912854031,
      "grad_norm": 553.1524658203125,
      "learning_rate": 5.0700773511116906e-05,
      "loss": 13.4783,
      "step": 613
    },
    {
      "epoch": 0.6688453159041394,
      "grad_norm": 981.0257568359375,
      "learning_rate": 5.040004590199128e-05,
      "loss": 14.2846,
      "step": 614
    },
    {
      "epoch": 0.6699346405228758,
      "grad_norm": 817.2470703125,
      "learning_rate": 5.0099912050066556e-05,
      "loss": 12.6324,
      "step": 615
    },
    {
      "epoch": 0.6710239651416122,
      "grad_norm": 593.127197265625,
      "learning_rate": 4.9800375548221845e-05,
      "loss": 13.0678,
      "step": 616
    },
    {
      "epoch": 0.6721132897603486,
      "grad_norm": 940.2001953125,
      "learning_rate": 4.950143998218531e-05,
      "loss": 13.5529,
      "step": 617
    },
    {
      "epoch": 0.673202614379085,
      "grad_norm": 717.47021484375,
      "learning_rate": 4.920310893049146e-05,
      "loss": 12.7114,
      "step": 618
    },
    {
      "epoch": 0.6742919389978214,
      "grad_norm": 874.2285766601562,
      "learning_rate": 4.89053859644381e-05,
      "loss": 14.4422,
      "step": 619
    },
    {
      "epoch": 0.6753812636165577,
      "grad_norm": 847.6348876953125,
      "learning_rate": 4.860827464804383e-05,
      "loss": 12.8582,
      "step": 620
    },
    {
      "epoch": 0.6764705882352942,
      "grad_norm": 552.3037719726562,
      "learning_rate": 4.831177853800511e-05,
      "loss": 13.5564,
      "step": 621
    },
    {
      "epoch": 0.6775599128540305,
      "grad_norm": 704.1598510742188,
      "learning_rate": 4.801590118365383e-05,
      "loss": 14.2366,
      "step": 622
    },
    {
      "epoch": 0.6786492374727668,
      "grad_norm": 535.1793823242188,
      "learning_rate": 4.77206461269149e-05,
      "loss": 13.5502,
      "step": 623
    },
    {
      "epoch": 0.6797385620915033,
      "grad_norm": 1089.44775390625,
      "learning_rate": 4.7426016902263636e-05,
      "loss": 13.391,
      "step": 624
    },
    {
      "epoch": 0.6808278867102396,
      "grad_norm": 793.9139404296875,
      "learning_rate": 4.713201703668367e-05,
      "loss": 13.9273,
      "step": 625
    },
    {
      "epoch": 0.681917211328976,
      "grad_norm": 1263.728759765625,
      "learning_rate": 4.683865004962452e-05,
      "loss": 13.2224,
      "step": 626
    },
    {
      "epoch": 0.6830065359477124,
      "grad_norm": 583.6502685546875,
      "learning_rate": 4.654591945295969e-05,
      "loss": 13.1198,
      "step": 627
    },
    {
      "epoch": 0.6840958605664488,
      "grad_norm": 546.5349731445312,
      "learning_rate": 4.6253828750944375e-05,
      "loss": 14.2708,
      "step": 628
    },
    {
      "epoch": 0.6851851851851852,
      "grad_norm": 573.7505493164062,
      "learning_rate": 4.596238144017369e-05,
      "loss": 14.4553,
      "step": 629
    },
    {
      "epoch": 0.6862745098039216,
      "grad_norm": 485.56976318359375,
      "learning_rate": 4.567158100954083e-05,
      "loss": 13.9009,
      "step": 630
    },
    {
      "epoch": 0.6873638344226579,
      "grad_norm": 859.7692260742188,
      "learning_rate": 4.53814309401951e-05,
      "loss": 13.4996,
      "step": 631
    },
    {
      "epoch": 0.6884531590413944,
      "grad_norm": 806.7265014648438,
      "learning_rate": 4.509193470550056e-05,
      "loss": 13.8831,
      "step": 632
    },
    {
      "epoch": 0.6895424836601307,
      "grad_norm": 678.3446044921875,
      "learning_rate": 4.4803095770994106e-05,
      "loss": 12.5697,
      "step": 633
    },
    {
      "epoch": 0.690631808278867,
      "grad_norm": 760.7838745117188,
      "learning_rate": 4.4514917594344184e-05,
      "loss": 13.5064,
      "step": 634
    },
    {
      "epoch": 0.6917211328976035,
      "grad_norm": 1103.8131103515625,
      "learning_rate": 4.422740362530945e-05,
      "loss": 14.8133,
      "step": 635
    },
    {
      "epoch": 0.6928104575163399,
      "grad_norm": 737.2034912109375,
      "learning_rate": 4.3940557305697226e-05,
      "loss": 14.7038,
      "step": 636
    },
    {
      "epoch": 0.6938997821350763,
      "grad_norm": 579.3258056640625,
      "learning_rate": 4.3654382069322644e-05,
      "loss": 14.0494,
      "step": 637
    },
    {
      "epoch": 0.6949891067538126,
      "grad_norm": 680.4292602539062,
      "learning_rate": 4.3368881341967135e-05,
      "loss": 14.3115,
      "step": 638
    },
    {
      "epoch": 0.696078431372549,
      "grad_norm": 617.1260986328125,
      "learning_rate": 4.308405854133786e-05,
      "loss": 13.788,
      "step": 639
    },
    {
      "epoch": 0.6971677559912854,
      "grad_norm": 951.0760498046875,
      "learning_rate": 4.2799917077026394e-05,
      "loss": 13.1248,
      "step": 640
    },
    {
      "epoch": 0.6982570806100218,
      "grad_norm": 606.4050903320312,
      "learning_rate": 4.251646035046814e-05,
      "loss": 14.0699,
      "step": 641
    },
    {
      "epoch": 0.6993464052287581,
      "grad_norm": 596.975830078125,
      "learning_rate": 4.223369175490162e-05,
      "loss": 11.8161,
      "step": 642
    },
    {
      "epoch": 0.7004357298474946,
      "grad_norm": 867.7359619140625,
      "learning_rate": 4.195161467532769e-05,
      "loss": 13.4987,
      "step": 643
    },
    {
      "epoch": 0.7015250544662309,
      "grad_norm": 802.6876831054688,
      "learning_rate": 4.167023248846925e-05,
      "loss": 12.8087,
      "step": 644
    },
    {
      "epoch": 0.7026143790849673,
      "grad_norm": 687.2286987304688,
      "learning_rate": 4.138954856273054e-05,
      "loss": 12.1043,
      "step": 645
    },
    {
      "epoch": 0.7037037037037037,
      "grad_norm": 669.9190673828125,
      "learning_rate": 4.110956625815713e-05,
      "loss": 12.1478,
      "step": 646
    },
    {
      "epoch": 0.7047930283224401,
      "grad_norm": 841.2921752929688,
      "learning_rate": 4.083028892639541e-05,
      "loss": 12.806,
      "step": 647
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 808.080078125,
      "learning_rate": 4.055171991065262e-05,
      "loss": 13.3545,
      "step": 648
    },
    {
      "epoch": 0.7069716775599129,
      "grad_norm": 919.4234619140625,
      "learning_rate": 4.027386254565688e-05,
      "loss": 13.9735,
      "step": 649
    },
    {
      "epoch": 0.7080610021786492,
      "grad_norm": 769.6393432617188,
      "learning_rate": 3.9996720157617094e-05,
      "loss": 14.6133,
      "step": 650
    },
    {
      "epoch": 0.7091503267973857,
      "grad_norm": 880.525146484375,
      "learning_rate": 3.972029606418335e-05,
      "loss": 13.0994,
      "step": 651
    },
    {
      "epoch": 0.710239651416122,
      "grad_norm": 945.1436767578125,
      "learning_rate": 3.9444593574406915e-05,
      "loss": 14.2055,
      "step": 652
    },
    {
      "epoch": 0.7113289760348583,
      "grad_norm": 690.2129516601562,
      "learning_rate": 3.9169615988701e-05,
      "loss": 13.2737,
      "step": 653
    },
    {
      "epoch": 0.7124183006535948,
      "grad_norm": 563.0156860351562,
      "learning_rate": 3.8895366598800896e-05,
      "loss": 12.8401,
      "step": 654
    },
    {
      "epoch": 0.7135076252723311,
      "grad_norm": 814.678466796875,
      "learning_rate": 3.862184868772473e-05,
      "loss": 12.2818,
      "step": 655
    },
    {
      "epoch": 0.7145969498910676,
      "grad_norm": 712.4736938476562,
      "learning_rate": 3.834906552973424e-05,
      "loss": 12.5831,
      "step": 656
    },
    {
      "epoch": 0.7156862745098039,
      "grad_norm": 1027.3265380859375,
      "learning_rate": 3.807702039029539e-05,
      "loss": 13.827,
      "step": 657
    },
    {
      "epoch": 0.7167755991285403,
      "grad_norm": 1087.4530029296875,
      "learning_rate": 3.780571652603949e-05,
      "loss": 13.3193,
      "step": 658
    },
    {
      "epoch": 0.7178649237472767,
      "grad_norm": 926.59326171875,
      "learning_rate": 3.753515718472402e-05,
      "loss": 13.3761,
      "step": 659
    },
    {
      "epoch": 0.7189542483660131,
      "grad_norm": 758.7875366210938,
      "learning_rate": 3.726534560519381e-05,
      "loss": 12.2786,
      "step": 660
    },
    {
      "epoch": 0.7200435729847494,
      "grad_norm": 693.2098999023438,
      "learning_rate": 3.6996285017342406e-05,
      "loss": 12.9392,
      "step": 661
    },
    {
      "epoch": 0.7211328976034859,
      "grad_norm": 1132.9970703125,
      "learning_rate": 3.672797864207316e-05,
      "loss": 12.6866,
      "step": 662
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 868.7568969726562,
      "learning_rate": 3.646042969126093e-05,
      "loss": 12.7426,
      "step": 663
    },
    {
      "epoch": 0.7233115468409586,
      "grad_norm": 800.123291015625,
      "learning_rate": 3.619364136771337e-05,
      "loss": 12.4544,
      "step": 664
    },
    {
      "epoch": 0.724400871459695,
      "grad_norm": 509.1244201660156,
      "learning_rate": 3.5927616865132884e-05,
      "loss": 13.2459,
      "step": 665
    },
    {
      "epoch": 0.7254901960784313,
      "grad_norm": 568.7617797851562,
      "learning_rate": 3.566235936807808e-05,
      "loss": 13.3732,
      "step": 666
    },
    {
      "epoch": 0.7265795206971678,
      "grad_norm": 839.1197509765625,
      "learning_rate": 3.539787205192586e-05,
      "loss": 12.6018,
      "step": 667
    },
    {
      "epoch": 0.7276688453159041,
      "grad_norm": 750.1957397460938,
      "learning_rate": 3.513415808283341e-05,
      "loss": 13.1899,
      "step": 668
    },
    {
      "epoch": 0.7287581699346405,
      "grad_norm": 604.1763305664062,
      "learning_rate": 3.4871220617700126e-05,
      "loss": 13.0681,
      "step": 669
    },
    {
      "epoch": 0.7298474945533769,
      "grad_norm": 637.3112182617188,
      "learning_rate": 3.460906280413007e-05,
      "loss": 13.0931,
      "step": 670
    },
    {
      "epoch": 0.7309368191721133,
      "grad_norm": 778.7605590820312,
      "learning_rate": 3.4347687780394e-05,
      "loss": 13.7031,
      "step": 671
    },
    {
      "epoch": 0.7320261437908496,
      "grad_norm": 1044.1873779296875,
      "learning_rate": 3.4087098675392104e-05,
      "loss": 12.2163,
      "step": 672
    },
    {
      "epoch": 0.7331154684095861,
      "grad_norm": 833.3889770507812,
      "learning_rate": 3.382729860861632e-05,
      "loss": 13.2927,
      "step": 673
    },
    {
      "epoch": 0.7342047930283224,
      "grad_norm": 655.1200561523438,
      "learning_rate": 3.3568290690113034e-05,
      "loss": 12.0668,
      "step": 674
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 702.258056640625,
      "learning_rate": 3.331007802044601e-05,
      "loss": 11.1181,
      "step": 675
    },
    {
      "epoch": 0.7363834422657952,
      "grad_norm": 791.837890625,
      "learning_rate": 3.305266369065901e-05,
      "loss": 12.7031,
      "step": 676
    },
    {
      "epoch": 0.7374727668845316,
      "grad_norm": 636.5592041015625,
      "learning_rate": 3.279605078223906e-05,
      "loss": 14.1468,
      "step": 677
    },
    {
      "epoch": 0.738562091503268,
      "grad_norm": 1960.34228515625,
      "learning_rate": 3.25402423670793e-05,
      "loss": 12.7793,
      "step": 678
    },
    {
      "epoch": 0.7396514161220044,
      "grad_norm": 987.132080078125,
      "learning_rate": 3.228524150744249e-05,
      "loss": 13.8105,
      "step": 679
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 536.3349609375,
      "learning_rate": 3.2031051255924085e-05,
      "loss": 13.0451,
      "step": 680
    },
    {
      "epoch": 0.7418300653594772,
      "grad_norm": 829.6603393554688,
      "learning_rate": 3.1777674655415834e-05,
      "loss": 13.352,
      "step": 681
    },
    {
      "epoch": 0.7429193899782135,
      "grad_norm": 825.07275390625,
      "learning_rate": 3.1525114739069415e-05,
      "loss": 13.4622,
      "step": 682
    },
    {
      "epoch": 0.7440087145969498,
      "grad_norm": 1563.3502197265625,
      "learning_rate": 3.127337453025994e-05,
      "loss": 13.9679,
      "step": 683
    },
    {
      "epoch": 0.7450980392156863,
      "grad_norm": 955.0587158203125,
      "learning_rate": 3.102245704254995e-05,
      "loss": 12.3828,
      "step": 684
    },
    {
      "epoch": 0.7461873638344226,
      "grad_norm": 784.834716796875,
      "learning_rate": 3.077236527965318e-05,
      "loss": 13.1804,
      "step": 685
    },
    {
      "epoch": 0.7472766884531591,
      "grad_norm": 735.6173706054688,
      "learning_rate": 3.0523102235398714e-05,
      "loss": 13.0926,
      "step": 686
    },
    {
      "epoch": 0.7483660130718954,
      "grad_norm": 1419.035888671875,
      "learning_rate": 3.0274670893695147e-05,
      "loss": 13.365,
      "step": 687
    },
    {
      "epoch": 0.7494553376906318,
      "grad_norm": 1074.7969970703125,
      "learning_rate": 3.002707422849472e-05,
      "loss": 13.357,
      "step": 688
    },
    {
      "epoch": 0.7505446623093682,
      "grad_norm": 1391.6951904296875,
      "learning_rate": 2.978031520375798e-05,
      "loss": 13.65,
      "step": 689
    },
    {
      "epoch": 0.7516339869281046,
      "grad_norm": 789.4784545898438,
      "learning_rate": 2.9534396773417994e-05,
      "loss": 12.0752,
      "step": 690
    },
    {
      "epoch": 0.7516339869281046,
      "eval_loss": 3.2088863849639893,
      "eval_runtime": 2.6648,
      "eval_samples_per_second": 145.229,
      "eval_steps_per_second": 72.802,
      "step": 690
    },
    {
      "epoch": 0.7527233115468409,
      "grad_norm": 880.0890502929688,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 12.8765,
      "step": 691
    },
    {
      "epoch": 0.7538126361655774,
      "grad_norm": 1300.8441162109375,
      "learning_rate": 2.9045093461312258e-05,
      "loss": 12.9364,
      "step": 692
    },
    {
      "epoch": 0.7549019607843137,
      "grad_norm": 837.1845092773438,
      "learning_rate": 2.8801714436958416e-05,
      "loss": 11.9844,
      "step": 693
    },
    {
      "epoch": 0.7559912854030502,
      "grad_norm": 837.4990844726562,
      "learning_rate": 2.855918772175522e-05,
      "loss": 12.274,
      "step": 694
    },
    {
      "epoch": 0.7570806100217865,
      "grad_norm": 1010.6368408203125,
      "learning_rate": 2.8317516218971073e-05,
      "loss": 14.0263,
      "step": 695
    },
    {
      "epoch": 0.7581699346405228,
      "grad_norm": 691.3543701171875,
      "learning_rate": 2.8076702821636867e-05,
      "loss": 12.8702,
      "step": 696
    },
    {
      "epoch": 0.7592592592592593,
      "grad_norm": 884.868896484375,
      "learning_rate": 2.7836750412511016e-05,
      "loss": 13.9273,
      "step": 697
    },
    {
      "epoch": 0.7603485838779956,
      "grad_norm": 1038.74462890625,
      "learning_rate": 2.7597661864045233e-05,
      "loss": 13.7905,
      "step": 698
    },
    {
      "epoch": 0.761437908496732,
      "grad_norm": 861.4619140625,
      "learning_rate": 2.735944003834997e-05,
      "loss": 13.3323,
      "step": 699
    },
    {
      "epoch": 0.7625272331154684,
      "grad_norm": 781.8521118164062,
      "learning_rate": 2.7122087787160166e-05,
      "loss": 13.0336,
      "step": 700
    },
    {
      "epoch": 0.7636165577342048,
      "grad_norm": 1335.78076171875,
      "learning_rate": 2.688560795180126e-05,
      "loss": 12.7611,
      "step": 701
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 704.1229248046875,
      "learning_rate": 2.6650003363154963e-05,
      "loss": 13.3398,
      "step": 702
    },
    {
      "epoch": 0.7657952069716776,
      "grad_norm": 814.8448486328125,
      "learning_rate": 2.641527684162556e-05,
      "loss": 12.0726,
      "step": 703
    },
    {
      "epoch": 0.7668845315904139,
      "grad_norm": 866.6876220703125,
      "learning_rate": 2.6181431197105998e-05,
      "loss": 12.4249,
      "step": 704
    },
    {
      "epoch": 0.7679738562091504,
      "grad_norm": 820.3846435546875,
      "learning_rate": 2.5948469228944318e-05,
      "loss": 12.3404,
      "step": 705
    },
    {
      "epoch": 0.7690631808278867,
      "grad_norm": 1044.1195068359375,
      "learning_rate": 2.5716393725910215e-05,
      "loss": 12.7415,
      "step": 706
    },
    {
      "epoch": 0.7701525054466231,
      "grad_norm": 731.3047485351562,
      "learning_rate": 2.5485207466161466e-05,
      "loss": 12.301,
      "step": 707
    },
    {
      "epoch": 0.7712418300653595,
      "grad_norm": 996.6677856445312,
      "learning_rate": 2.5254913217210886e-05,
      "loss": 12.2375,
      "step": 708
    },
    {
      "epoch": 0.7723311546840959,
      "grad_norm": 1071.778076171875,
      "learning_rate": 2.5025513735893014e-05,
      "loss": 12.9786,
      "step": 709
    },
    {
      "epoch": 0.7734204793028322,
      "grad_norm": 694.3305053710938,
      "learning_rate": 2.47970117683313e-05,
      "loss": 12.5304,
      "step": 710
    },
    {
      "epoch": 0.7745098039215687,
      "grad_norm": 568.67626953125,
      "learning_rate": 2.4569410049905016e-05,
      "loss": 13.4328,
      "step": 711
    },
    {
      "epoch": 0.775599128540305,
      "grad_norm": 781.4323120117188,
      "learning_rate": 2.434271130521666e-05,
      "loss": 13.7223,
      "step": 712
    },
    {
      "epoch": 0.7766884531590414,
      "grad_norm": 707.454345703125,
      "learning_rate": 2.411691824805934e-05,
      "loss": 13.2585,
      "step": 713
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 662.977783203125,
      "learning_rate": 2.389203358138419e-05,
      "loss": 13.1885,
      "step": 714
    },
    {
      "epoch": 0.7788671023965141,
      "grad_norm": 1106.654296875,
      "learning_rate": 2.3668059997268144e-05,
      "loss": 12.8549,
      "step": 715
    },
    {
      "epoch": 0.7799564270152506,
      "grad_norm": 809.3428955078125,
      "learning_rate": 2.3445000176881537e-05,
      "loss": 13.6922,
      "step": 716
    },
    {
      "epoch": 0.7810457516339869,
      "grad_norm": 1407.9627685546875,
      "learning_rate": 2.3222856790456226e-05,
      "loss": 12.3901,
      "step": 717
    },
    {
      "epoch": 0.7821350762527233,
      "grad_norm": 1036.4595947265625,
      "learning_rate": 2.3001632497253424e-05,
      "loss": 14.0636,
      "step": 718
    },
    {
      "epoch": 0.7832244008714597,
      "grad_norm": 713.0249633789062,
      "learning_rate": 2.2781329945531936e-05,
      "loss": 13.3694,
      "step": 719
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 946.2935180664062,
      "learning_rate": 2.2561951772516587e-05,
      "loss": 12.7461,
      "step": 720
    },
    {
      "epoch": 0.7854030501089324,
      "grad_norm": 1000.2520751953125,
      "learning_rate": 2.2343500604366374e-05,
      "loss": 14.3916,
      "step": 721
    },
    {
      "epoch": 0.7864923747276689,
      "grad_norm": 1008.6942749023438,
      "learning_rate": 2.2125979056143364e-05,
      "loss": 13.3091,
      "step": 722
    },
    {
      "epoch": 0.7875816993464052,
      "grad_norm": 720.97314453125,
      "learning_rate": 2.190938973178105e-05,
      "loss": 12.963,
      "step": 723
    },
    {
      "epoch": 0.7886710239651417,
      "grad_norm": 704.6624145507812,
      "learning_rate": 2.169373522405349e-05,
      "loss": 13.3057,
      "step": 724
    },
    {
      "epoch": 0.789760348583878,
      "grad_norm": 944.1021728515625,
      "learning_rate": 2.1479018114544026e-05,
      "loss": 12.621,
      "step": 725
    },
    {
      "epoch": 0.7908496732026143,
      "grad_norm": 552.0089111328125,
      "learning_rate": 2.1265240973614486e-05,
      "loss": 12.7888,
      "step": 726
    },
    {
      "epoch": 0.7919389978213508,
      "grad_norm": 685.5350341796875,
      "learning_rate": 2.105240636037449e-05,
      "loss": 13.6543,
      "step": 727
    },
    {
      "epoch": 0.7930283224400871,
      "grad_norm": 627.4608764648438,
      "learning_rate": 2.0840516822650614e-05,
      "loss": 12.6445,
      "step": 728
    },
    {
      "epoch": 0.7941176470588235,
      "grad_norm": 645.2318725585938,
      "learning_rate": 2.0629574896956126e-05,
      "loss": 12.6309,
      "step": 729
    },
    {
      "epoch": 0.7952069716775599,
      "grad_norm": 888.5296630859375,
      "learning_rate": 2.0419583108460418e-05,
      "loss": 11.886,
      "step": 730
    },
    {
      "epoch": 0.7962962962962963,
      "grad_norm": 1192.0579833984375,
      "learning_rate": 2.0210543970958872e-05,
      "loss": 12.3713,
      "step": 731
    },
    {
      "epoch": 0.7973856209150327,
      "grad_norm": 635.3908081054688,
      "learning_rate": 2.0002459986842825e-05,
      "loss": 13.2216,
      "step": 732
    },
    {
      "epoch": 0.7984749455337691,
      "grad_norm": 657.1572265625,
      "learning_rate": 1.9795333647069448e-05,
      "loss": 12.6068,
      "step": 733
    },
    {
      "epoch": 0.7995642701525054,
      "grad_norm": 1367.3870849609375,
      "learning_rate": 1.958916743113214e-05,
      "loss": 12.8324,
      "step": 734
    },
    {
      "epoch": 0.8006535947712419,
      "grad_norm": 1032.5328369140625,
      "learning_rate": 1.93839638070306e-05,
      "loss": 12.6028,
      "step": 735
    },
    {
      "epoch": 0.8017429193899782,
      "grad_norm": 762.1034545898438,
      "learning_rate": 1.9179725231241564e-05,
      "loss": 12.5233,
      "step": 736
    },
    {
      "epoch": 0.8028322440087146,
      "grad_norm": 1431.1048583984375,
      "learning_rate": 1.8976454148689127e-05,
      "loss": 12.9953,
      "step": 737
    },
    {
      "epoch": 0.803921568627451,
      "grad_norm": 1034.8011474609375,
      "learning_rate": 1.877415299271561e-05,
      "loss": 13.2373,
      "step": 738
    },
    {
      "epoch": 0.8050108932461874,
      "grad_norm": 752.5025024414062,
      "learning_rate": 1.857282418505253e-05,
      "loss": 13.1138,
      "step": 739
    },
    {
      "epoch": 0.8061002178649237,
      "grad_norm": 911.1949462890625,
      "learning_rate": 1.8372470135791344e-05,
      "loss": 12.7149,
      "step": 740
    },
    {
      "epoch": 0.8071895424836601,
      "grad_norm": 1050.7984619140625,
      "learning_rate": 1.8173093243354878e-05,
      "loss": 12.862,
      "step": 741
    },
    {
      "epoch": 0.8082788671023965,
      "grad_norm": 943.8112182617188,
      "learning_rate": 1.7974695894468384e-05,
      "loss": 12.6691,
      "step": 742
    },
    {
      "epoch": 0.809368191721133,
      "grad_norm": 860.751708984375,
      "learning_rate": 1.7777280464131197e-05,
      "loss": 12.9838,
      "step": 743
    },
    {
      "epoch": 0.8104575163398693,
      "grad_norm": 812.5382690429688,
      "learning_rate": 1.7580849315588068e-05,
      "loss": 12.5751,
      "step": 744
    },
    {
      "epoch": 0.8115468409586056,
      "grad_norm": 684.0803833007812,
      "learning_rate": 1.7385404800301007e-05,
      "loss": 12.6821,
      "step": 745
    },
    {
      "epoch": 0.8126361655773421,
      "grad_norm": 637.1691284179688,
      "learning_rate": 1.7190949257921196e-05,
      "loss": 12.704,
      "step": 746
    },
    {
      "epoch": 0.8137254901960784,
      "grad_norm": 1279.00146484375,
      "learning_rate": 1.6997485016260793e-05,
      "loss": 13.6321,
      "step": 747
    },
    {
      "epoch": 0.8148148148148148,
      "grad_norm": 1138.228515625,
      "learning_rate": 1.680501439126525e-05,
      "loss": 12.3677,
      "step": 748
    },
    {
      "epoch": 0.8159041394335512,
      "grad_norm": 718.772705078125,
      "learning_rate": 1.6613539686985458e-05,
      "loss": 13.3372,
      "step": 749
    },
    {
      "epoch": 0.8169934640522876,
      "grad_norm": 960.2557373046875,
      "learning_rate": 1.642306319555027e-05,
      "loss": 11.9433,
      "step": 750
    },
    {
      "epoch": 0.818082788671024,
      "grad_norm": 745.1720581054688,
      "learning_rate": 1.6233587197138968e-05,
      "loss": 11.8539,
      "step": 751
    },
    {
      "epoch": 0.8191721132897604,
      "grad_norm": 837.4060668945312,
      "learning_rate": 1.6045113959953985e-05,
      "loss": 12.3409,
      "step": 752
    },
    {
      "epoch": 0.8202614379084967,
      "grad_norm": 854.7937622070312,
      "learning_rate": 1.585764574019388e-05,
      "loss": 13.3537,
      "step": 753
    },
    {
      "epoch": 0.8213507625272332,
      "grad_norm": 879.7656860351562,
      "learning_rate": 1.5671184782026106e-05,
      "loss": 12.7348,
      "step": 754
    },
    {
      "epoch": 0.8224400871459695,
      "grad_norm": 941.7662353515625,
      "learning_rate": 1.548573331756038e-05,
      "loss": 12.8492,
      "step": 755
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 926.8936767578125,
      "learning_rate": 1.530129356682175e-05,
      "loss": 12.8661,
      "step": 756
    },
    {
      "epoch": 0.8246187363834423,
      "grad_norm": 1937.478271484375,
      "learning_rate": 1.5117867737724134e-05,
      "loss": 12.9826,
      "step": 757
    },
    {
      "epoch": 0.8257080610021786,
      "grad_norm": 909.1730346679688,
      "learning_rate": 1.4935458026043959e-05,
      "loss": 11.884,
      "step": 758
    },
    {
      "epoch": 0.826797385620915,
      "grad_norm": 833.2159423828125,
      "learning_rate": 1.4754066615393668e-05,
      "loss": 13.1729,
      "step": 759
    },
    {
      "epoch": 0.8278867102396514,
      "grad_norm": 1762.9833984375,
      "learning_rate": 1.457369567719581e-05,
      "loss": 12.5088,
      "step": 760
    },
    {
      "epoch": 0.8289760348583878,
      "grad_norm": 774.0321044921875,
      "learning_rate": 1.4394347370656836e-05,
      "loss": 12.6099,
      "step": 761
    },
    {
      "epoch": 0.8300653594771242,
      "grad_norm": 833.3447875976562,
      "learning_rate": 1.4216023842741455e-05,
      "loss": 11.8071,
      "step": 762
    },
    {
      "epoch": 0.8311546840958606,
      "grad_norm": 659.9131469726562,
      "learning_rate": 1.4038727228146753e-05,
      "loss": 12.0735,
      "step": 763
    },
    {
      "epoch": 0.8322440087145969,
      "grad_norm": 888.9996948242188,
      "learning_rate": 1.3862459649276715e-05,
      "loss": 12.47,
      "step": 764
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 759.5191040039062,
      "learning_rate": 1.3687223216216904e-05,
      "loss": 12.5613,
      "step": 765
    },
    {
      "epoch": 0.8344226579520697,
      "grad_norm": 805.9093627929688,
      "learning_rate": 1.3513020026709023e-05,
      "loss": 12.0561,
      "step": 766
    },
    {
      "epoch": 0.835511982570806,
      "grad_norm": 763.7495727539062,
      "learning_rate": 1.3339852166125954e-05,
      "loss": 12.959,
      "step": 767
    },
    {
      "epoch": 0.8366013071895425,
      "grad_norm": 610.1506958007812,
      "learning_rate": 1.3167721707446678e-05,
      "loss": 12.2513,
      "step": 768
    },
    {
      "epoch": 0.8376906318082789,
      "grad_norm": 958.7525634765625,
      "learning_rate": 1.2996630711231616e-05,
      "loss": 14.1049,
      "step": 769
    },
    {
      "epoch": 0.8387799564270153,
      "grad_norm": 1156.2576904296875,
      "learning_rate": 1.2826581225597767e-05,
      "loss": 12.8887,
      "step": 770
    },
    {
      "epoch": 0.8398692810457516,
      "grad_norm": 1074.29345703125,
      "learning_rate": 1.26575752861943e-05,
      "loss": 13.7254,
      "step": 771
    },
    {
      "epoch": 0.840958605664488,
      "grad_norm": 1287.4105224609375,
      "learning_rate": 1.248961491617826e-05,
      "loss": 13.033,
      "step": 772
    },
    {
      "epoch": 0.8420479302832244,
      "grad_norm": 945.987548828125,
      "learning_rate": 1.2322702126190156e-05,
      "loss": 11.7015,
      "step": 773
    },
    {
      "epoch": 0.8431372549019608,
      "grad_norm": 763.9169921875,
      "learning_rate": 1.2156838914330072e-05,
      "loss": 12.8775,
      "step": 774
    },
    {
      "epoch": 0.8442265795206971,
      "grad_norm": 869.27880859375,
      "learning_rate": 1.1992027266133598e-05,
      "loss": 12.8051,
      "step": 775
    },
    {
      "epoch": 0.8453159041394336,
      "grad_norm": 919.6107177734375,
      "learning_rate": 1.1828269154548244e-05,
      "loss": 12.7364,
      "step": 776
    },
    {
      "epoch": 0.8464052287581699,
      "grad_norm": 840.4449462890625,
      "learning_rate": 1.1665566539909623e-05,
      "loss": 12.2462,
      "step": 777
    },
    {
      "epoch": 0.8474945533769063,
      "grad_norm": 875.924072265625,
      "learning_rate": 1.1503921369918091e-05,
      "loss": 12.9855,
      "step": 778
    },
    {
      "epoch": 0.8485838779956427,
      "grad_norm": 2267.00927734375,
      "learning_rate": 1.1343335579615467e-05,
      "loss": 12.9955,
      "step": 779
    },
    {
      "epoch": 0.8496732026143791,
      "grad_norm": 934.5510864257812,
      "learning_rate": 1.118381109136174e-05,
      "loss": 12.6251,
      "step": 780
    },
    {
      "epoch": 0.8507625272331155,
      "grad_norm": 599.7080078125,
      "learning_rate": 1.1025349814812224e-05,
      "loss": 12.7546,
      "step": 781
    },
    {
      "epoch": 0.8518518518518519,
      "grad_norm": 1287.09912109375,
      "learning_rate": 1.0867953646894525e-05,
      "loss": 12.8024,
      "step": 782
    },
    {
      "epoch": 0.8529411764705882,
      "grad_norm": 924.6251831054688,
      "learning_rate": 1.0711624471785986e-05,
      "loss": 12.9911,
      "step": 783
    },
    {
      "epoch": 0.8540305010893247,
      "grad_norm": 904.363525390625,
      "learning_rate": 1.055636416089102e-05,
      "loss": 13.8365,
      "step": 784
    },
    {
      "epoch": 0.855119825708061,
      "grad_norm": 998.8102416992188,
      "learning_rate": 1.0402174572818723e-05,
      "loss": 11.9683,
      "step": 785
    },
    {
      "epoch": 0.8562091503267973,
      "grad_norm": 890.9308471679688,
      "learning_rate": 1.0249057553360742e-05,
      "loss": 12.8672,
      "step": 786
    },
    {
      "epoch": 0.8572984749455338,
      "grad_norm": 944.108642578125,
      "learning_rate": 1.0097014935468984e-05,
      "loss": 13.2105,
      "step": 787
    },
    {
      "epoch": 0.8583877995642701,
      "grad_norm": 956.6070556640625,
      "learning_rate": 9.946048539233865e-06,
      "loss": 12.8339,
      "step": 788
    },
    {
      "epoch": 0.8594771241830066,
      "grad_norm": 783.9725341796875,
      "learning_rate": 9.796160171862367e-06,
      "loss": 12.0513,
      "step": 789
    },
    {
      "epoch": 0.8605664488017429,
      "grad_norm": 678.079833984375,
      "learning_rate": 9.647351627656543e-06,
      "loss": 12.4566,
      "step": 790
    },
    {
      "epoch": 0.8616557734204793,
      "grad_norm": 795.08154296875,
      "learning_rate": 9.499624687991871e-06,
      "loss": 13.0777,
      "step": 791
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 1114.5704345703125,
      "learning_rate": 9.352981121296134e-06,
      "loss": 12.3807,
      "step": 792
    },
    {
      "epoch": 0.8638344226579521,
      "grad_norm": 810.4409790039062,
      "learning_rate": 9.207422683028066e-06,
      "loss": 12.0099,
      "step": 793
    },
    {
      "epoch": 0.8649237472766884,
      "grad_norm": 659.8233642578125,
      "learning_rate": 9.062951115656403e-06,
      "loss": 12.5349,
      "step": 794
    },
    {
      "epoch": 0.8660130718954249,
      "grad_norm": 858.1024780273438,
      "learning_rate": 8.919568148639123e-06,
      "loss": 12.1748,
      "step": 795
    },
    {
      "epoch": 0.8671023965141612,
      "grad_norm": 888.814453125,
      "learning_rate": 8.777275498402548e-06,
      "loss": 12.9867,
      "step": 796
    },
    {
      "epoch": 0.8681917211328976,
      "grad_norm": 845.94140625,
      "learning_rate": 8.636074868320987e-06,
      "loss": 13.1836,
      "step": 797
    },
    {
      "epoch": 0.869281045751634,
      "grad_norm": 970.3507080078125,
      "learning_rate": 8.495967948696192e-06,
      "loss": 11.6568,
      "step": 798
    },
    {
      "epoch": 0.8703703703703703,
      "grad_norm": 865.5740966796875,
      "learning_rate": 8.35695641673725e-06,
      "loss": 12.0325,
      "step": 799
    },
    {
      "epoch": 0.8714596949891068,
      "grad_norm": 916.9160766601562,
      "learning_rate": 8.219041936540395e-06,
      "loss": 13.141,
      "step": 800
    },
    {
      "epoch": 0.8725490196078431,
      "grad_norm": 746.8594970703125,
      "learning_rate": 8.082226159069196e-06,
      "loss": 13.3162,
      "step": 801
    },
    {
      "epoch": 0.8736383442265795,
      "grad_norm": 847.8311157226562,
      "learning_rate": 7.946510722134692e-06,
      "loss": 12.9385,
      "step": 802
    },
    {
      "epoch": 0.8747276688453159,
      "grad_norm": 1479.700439453125,
      "learning_rate": 7.811897250375833e-06,
      "loss": 13.0996,
      "step": 803
    },
    {
      "epoch": 0.8758169934640523,
      "grad_norm": 983.4318237304688,
      "learning_rate": 7.678387355240057e-06,
      "loss": 13.556,
      "step": 804
    },
    {
      "epoch": 0.8769063180827886,
      "grad_norm": 1585.536376953125,
      "learning_rate": 7.5459826349639436e-06,
      "loss": 13.0711,
      "step": 805
    },
    {
      "epoch": 0.8779956427015251,
      "grad_norm": 818.268310546875,
      "learning_rate": 7.4146846745541506e-06,
      "loss": 12.364,
      "step": 806
    },
    {
      "epoch": 0.8790849673202614,
      "grad_norm": 732.67626953125,
      "learning_rate": 7.284495045768325e-06,
      "loss": 11.9299,
      "step": 807
    },
    {
      "epoch": 0.8801742919389978,
      "grad_norm": 785.660888671875,
      "learning_rate": 7.155415307096458e-06,
      "loss": 13.0072,
      "step": 808
    },
    {
      "epoch": 0.8812636165577342,
      "grad_norm": 1017.672607421875,
      "learning_rate": 7.027447003742071e-06,
      "loss": 11.7492,
      "step": 809
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 961.2965087890625,
      "learning_rate": 6.900591667603751e-06,
      "loss": 12.7767,
      "step": 810
    },
    {
      "epoch": 0.883442265795207,
      "grad_norm": 742.8497314453125,
      "learning_rate": 6.774850817256939e-06,
      "loss": 11.4043,
      "step": 811
    },
    {
      "epoch": 0.8845315904139434,
      "grad_norm": 624.9981079101562,
      "learning_rate": 6.650225957935552e-06,
      "loss": 12.2672,
      "step": 812
    },
    {
      "epoch": 0.8856209150326797,
      "grad_norm": 779.7870483398438,
      "learning_rate": 6.5267185815141355e-06,
      "loss": 11.9513,
      "step": 813
    },
    {
      "epoch": 0.8867102396514162,
      "grad_norm": 1622.0540771484375,
      "learning_rate": 6.40433016648988e-06,
      "loss": 13.6973,
      "step": 814
    },
    {
      "epoch": 0.8877995642701525,
      "grad_norm": 1286.1015625,
      "learning_rate": 6.283062177965038e-06,
      "loss": 11.6033,
      "step": 815
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 1373.9736328125,
      "learning_rate": 6.162916067629254e-06,
      "loss": 13.2463,
      "step": 816
    },
    {
      "epoch": 0.8899782135076253,
      "grad_norm": 1360.0540771484375,
      "learning_rate": 6.043893273742329e-06,
      "loss": 11.335,
      "step": 817
    },
    {
      "epoch": 0.8910675381263616,
      "grad_norm": 1296.9820556640625,
      "learning_rate": 5.925995221116853e-06,
      "loss": 12.4607,
      "step": 818
    },
    {
      "epoch": 0.8921568627450981,
      "grad_norm": 1495.9398193359375,
      "learning_rate": 5.809223321101276e-06,
      "loss": 11.7732,
      "step": 819
    },
    {
      "epoch": 0.8932461873638344,
      "grad_norm": 880.9378662109375,
      "learning_rate": 5.693578971562963e-06,
      "loss": 12.8642,
      "step": 820
    },
    {
      "epoch": 0.8943355119825708,
      "grad_norm": 1063.22216796875,
      "learning_rate": 5.5790635568714224e-06,
      "loss": 12.4356,
      "step": 821
    },
    {
      "epoch": 0.8954248366013072,
      "grad_norm": 657.4075927734375,
      "learning_rate": 5.465678447881828e-06,
      "loss": 12.4044,
      "step": 822
    },
    {
      "epoch": 0.8965141612200436,
      "grad_norm": 1261.831787109375,
      "learning_rate": 5.3534250019184774e-06,
      "loss": 13.5235,
      "step": 823
    },
    {
      "epoch": 0.8976034858387799,
      "grad_norm": 601.0888061523438,
      "learning_rate": 5.242304562758704e-06,
      "loss": 11.9985,
      "step": 824
    },
    {
      "epoch": 0.8986928104575164,
      "grad_norm": 781.6649780273438,
      "learning_rate": 5.132318460616625e-06,
      "loss": 11.9329,
      "step": 825
    },
    {
      "epoch": 0.8997821350762527,
      "grad_norm": 1356.27294921875,
      "learning_rate": 5.023468012127364e-06,
      "loss": 12.6829,
      "step": 826
    },
    {
      "epoch": 0.900871459694989,
      "grad_norm": 929.7411499023438,
      "learning_rate": 4.915754520331173e-06,
      "loss": 11.9567,
      "step": 827
    },
    {
      "epoch": 0.9019607843137255,
      "grad_norm": 1020.9713745117188,
      "learning_rate": 4.8091792746578935e-06,
      "loss": 11.6725,
      "step": 828
    },
    {
      "epoch": 0.9030501089324618,
      "grad_norm": 741.8558349609375,
      "learning_rate": 4.703743550911543e-06,
      "loss": 12.448,
      "step": 829
    },
    {
      "epoch": 0.9041394335511983,
      "grad_norm": 637.9666137695312,
      "learning_rate": 4.599448611254964e-06,
      "loss": 12.7454,
      "step": 830
    },
    {
      "epoch": 0.9052287581699346,
      "grad_norm": 1003.8346557617188,
      "learning_rate": 4.496295704194819e-06,
      "loss": 13.0511,
      "step": 831
    },
    {
      "epoch": 0.906318082788671,
      "grad_norm": 819.62451171875,
      "learning_rate": 4.394286064566511e-06,
      "loss": 13.2029,
      "step": 832
    },
    {
      "epoch": 0.9074074074074074,
      "grad_norm": 1023.1488647460938,
      "learning_rate": 4.293420913519541e-06,
      "loss": 12.9151,
      "step": 833
    },
    {
      "epoch": 0.9084967320261438,
      "grad_norm": 788.2391967773438,
      "learning_rate": 4.193701458502807e-06,
      "loss": 12.7021,
      "step": 834
    },
    {
      "epoch": 0.9095860566448801,
      "grad_norm": 837.283447265625,
      "learning_rate": 4.095128893250156e-06,
      "loss": 13.0428,
      "step": 835
    },
    {
      "epoch": 0.9106753812636166,
      "grad_norm": 1206.6253662109375,
      "learning_rate": 3.997704397766122e-06,
      "loss": 12.3555,
      "step": 836
    },
    {
      "epoch": 0.9117647058823529,
      "grad_norm": 856.0335083007812,
      "learning_rate": 3.901429138311763e-06,
      "loss": 12.2161,
      "step": 837
    },
    {
      "epoch": 0.9128540305010894,
      "grad_norm": 840.4730834960938,
      "learning_rate": 3.80630426739077e-06,
      "loss": 12.3642,
      "step": 838
    },
    {
      "epoch": 0.9139433551198257,
      "grad_norm": 767.2365112304688,
      "learning_rate": 3.712330923735563e-06,
      "loss": 12.338,
      "step": 839
    },
    {
      "epoch": 0.9150326797385621,
      "grad_norm": 844.8609619140625,
      "learning_rate": 3.6195102322937545e-06,
      "loss": 12.4701,
      "step": 840
    },
    {
      "epoch": 0.9161220043572985,
      "grad_norm": 808.1879272460938,
      "learning_rate": 3.5278433042146397e-06,
      "loss": 12.4012,
      "step": 841
    },
    {
      "epoch": 0.9172113289760349,
      "grad_norm": 787.5524291992188,
      "learning_rate": 3.4373312368358944e-06,
      "loss": 12.1528,
      "step": 842
    },
    {
      "epoch": 0.9183006535947712,
      "grad_norm": 861.0357666015625,
      "learning_rate": 3.347975113670454e-06,
      "loss": 13.0016,
      "step": 843
    },
    {
      "epoch": 0.9193899782135077,
      "grad_norm": 1098.5736083984375,
      "learning_rate": 3.259776004393533e-06,
      "loss": 12.3925,
      "step": 844
    },
    {
      "epoch": 0.920479302832244,
      "grad_norm": 1017.1472778320312,
      "learning_rate": 3.1727349648298267e-06,
      "loss": 13.2196,
      "step": 845
    },
    {
      "epoch": 0.9215686274509803,
      "grad_norm": 729.2058715820312,
      "learning_rate": 3.086853036940862e-06,
      "loss": 11.8512,
      "step": 846
    },
    {
      "epoch": 0.9226579520697168,
      "grad_norm": 809.0467529296875,
      "learning_rate": 3.0021312488125454e-06,
      "loss": 11.8526,
      "step": 847
    },
    {
      "epoch": 0.9237472766884531,
      "grad_norm": 948.0686645507812,
      "learning_rate": 2.9185706146428017e-06,
      "loss": 12.4416,
      "step": 848
    },
    {
      "epoch": 0.9248366013071896,
      "grad_norm": 631.7981567382812,
      "learning_rate": 2.836172134729509e-06,
      "loss": 12.836,
      "step": 849
    },
    {
      "epoch": 0.9259259259259259,
      "grad_norm": 641.8875732421875,
      "learning_rate": 2.754936795458485e-06,
      "loss": 12.6921,
      "step": 850
    },
    {
      "epoch": 0.9270152505446623,
      "grad_norm": 860.4398803710938,
      "learning_rate": 2.674865569291651e-06,
      "loss": 13.6532,
      "step": 851
    },
    {
      "epoch": 0.9281045751633987,
      "grad_norm": 760.2114868164062,
      "learning_rate": 2.5959594147554667e-06,
      "loss": 13.178,
      "step": 852
    },
    {
      "epoch": 0.9291938997821351,
      "grad_norm": 1154.3988037109375,
      "learning_rate": 2.5182192764293567e-06,
      "loss": 12.716,
      "step": 853
    },
    {
      "epoch": 0.9302832244008714,
      "grad_norm": 762.8448486328125,
      "learning_rate": 2.4416460849345123e-06,
      "loss": 12.5605,
      "step": 854
    },
    {
      "epoch": 0.9313725490196079,
      "grad_norm": 1097.0977783203125,
      "learning_rate": 2.366240756922644e-06,
      "loss": 12.0626,
      "step": 855
    },
    {
      "epoch": 0.9324618736383442,
      "grad_norm": 989.987060546875,
      "learning_rate": 2.2920041950650783e-06,
      "loss": 12.1056,
      "step": 856
    },
    {
      "epoch": 0.9335511982570807,
      "grad_norm": 761.228271484375,
      "learning_rate": 2.218937288041956e-06,
      "loss": 12.0653,
      "step": 857
    },
    {
      "epoch": 0.934640522875817,
      "grad_norm": 698.48486328125,
      "learning_rate": 2.1470409105315283e-06,
      "loss": 12.689,
      "step": 858
    },
    {
      "epoch": 0.9357298474945533,
      "grad_norm": 1059.8203125,
      "learning_rate": 2.0763159231997674e-06,
      "loss": 12.268,
      "step": 859
    },
    {
      "epoch": 0.9368191721132898,
      "grad_norm": 1506.7076416015625,
      "learning_rate": 2.0067631726899962e-06,
      "loss": 13.0918,
      "step": 860
    },
    {
      "epoch": 0.9379084967320261,
      "grad_norm": 686.2244262695312,
      "learning_rate": 1.938383491612794e-06,
      "loss": 13.2276,
      "step": 861
    },
    {
      "epoch": 0.9389978213507625,
      "grad_norm": 1398.2288818359375,
      "learning_rate": 1.8711776985360308e-06,
      "loss": 12.4386,
      "step": 862
    },
    {
      "epoch": 0.9400871459694989,
      "grad_norm": 787.7633666992188,
      "learning_rate": 1.805146597975016e-06,
      "loss": 12.0259,
      "step": 863
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 1021.8810424804688,
      "learning_rate": 1.7402909803829525e-06,
      "loss": 13.2275,
      "step": 864
    },
    {
      "epoch": 0.9422657952069716,
      "grad_norm": 724.1331176757812,
      "learning_rate": 1.6766116221413774e-06,
      "loss": 12.9674,
      "step": 865
    },
    {
      "epoch": 0.9433551198257081,
      "grad_norm": 889.06201171875,
      "learning_rate": 1.61410928555098e-06,
      "loss": 12.9815,
      "step": 866
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 694.7861938476562,
      "learning_rate": 1.5527847188223644e-06,
      "loss": 11.78,
      "step": 867
    },
    {
      "epoch": 0.9455337690631809,
      "grad_norm": 1126.9815673828125,
      "learning_rate": 1.4926386560671358e-06,
      "loss": 13.2658,
      "step": 868
    },
    {
      "epoch": 0.9466230936819172,
      "grad_norm": 1251.4708251953125,
      "learning_rate": 1.433671817289184e-06,
      "loss": 13.5727,
      "step": 869
    },
    {
      "epoch": 0.9477124183006536,
      "grad_norm": 928.8302001953125,
      "learning_rate": 1.3758849083759352e-06,
      "loss": 12.5813,
      "step": 870
    },
    {
      "epoch": 0.94880174291939,
      "grad_norm": 906.8268432617188,
      "learning_rate": 1.3192786210900033e-06,
      "loss": 12.3939,
      "step": 871
    },
    {
      "epoch": 0.9498910675381264,
      "grad_norm": 1406.4046630859375,
      "learning_rate": 1.2638536330608408e-06,
      "loss": 12.5356,
      "step": 872
    },
    {
      "epoch": 0.9509803921568627,
      "grad_norm": 734.1050415039062,
      "learning_rate": 1.2096106077767011e-06,
      "loss": 12.0204,
      "step": 873
    },
    {
      "epoch": 0.9520697167755992,
      "grad_norm": 1201.427734375,
      "learning_rate": 1.1565501945766222e-06,
      "loss": 11.8001,
      "step": 874
    },
    {
      "epoch": 0.9531590413943355,
      "grad_norm": 964.8887329101562,
      "learning_rate": 1.1046730286426775e-06,
      "loss": 13.3203,
      "step": 875
    },
    {
      "epoch": 0.954248366013072,
      "grad_norm": 1066.5933837890625,
      "learning_rate": 1.053979730992416e-06,
      "loss": 12.9455,
      "step": 876
    },
    {
      "epoch": 0.9553376906318083,
      "grad_norm": 867.9620361328125,
      "learning_rate": 1.0044709084713554e-06,
      "loss": 12.5005,
      "step": 877
    },
    {
      "epoch": 0.9564270152505446,
      "grad_norm": 1209.2740478515625,
      "learning_rate": 9.56147153745779e-07,
      "loss": 12.6775,
      "step": 878
    },
    {
      "epoch": 0.9575163398692811,
      "grad_norm": 920.9111938476562,
      "learning_rate": 9.090090452955835e-07,
      "loss": 12.656,
      "step": 879
    },
    {
      "epoch": 0.9586056644880174,
      "grad_norm": 1178.7100830078125,
      "learning_rate": 8.630571474074311e-07,
      "loss": 11.8858,
      "step": 880
    },
    {
      "epoch": 0.9596949891067538,
      "grad_norm": 827.2933349609375,
      "learning_rate": 8.182920101679092e-07,
      "loss": 12.7875,
      "step": 881
    },
    {
      "epoch": 0.9607843137254902,
      "grad_norm": 970.2561645507812,
      "learning_rate": 7.747141694570026e-07,
      "loss": 12.1683,
      "step": 882
    },
    {
      "epoch": 0.9618736383442266,
      "grad_norm": 961.2227783203125,
      "learning_rate": 7.323241469416764e-07,
      "loss": 13.4533,
      "step": 883
    },
    {
      "epoch": 0.9629629629629629,
      "grad_norm": 1200.774658203125,
      "learning_rate": 6.911224500695702e-07,
      "loss": 13.4708,
      "step": 884
    },
    {
      "epoch": 0.9640522875816994,
      "grad_norm": 961.2836303710938,
      "learning_rate": 6.511095720630244e-07,
      "loss": 12.1307,
      "step": 885
    },
    {
      "epoch": 0.9651416122004357,
      "grad_norm": 949.4239501953125,
      "learning_rate": 6.122859919130974e-07,
      "loss": 11.6507,
      "step": 886
    },
    {
      "epoch": 0.9662309368191722,
      "grad_norm": 1107.519287109375,
      "learning_rate": 5.746521743738354e-07,
      "loss": 13.5597,
      "step": 887
    },
    {
      "epoch": 0.9673202614379085,
      "grad_norm": 713.34619140625,
      "learning_rate": 5.382085699567552e-07,
      "loss": 13.0859,
      "step": 888
    },
    {
      "epoch": 0.9684095860566448,
      "grad_norm": 763.1091918945312,
      "learning_rate": 5.029556149254266e-07,
      "loss": 12.9483,
      "step": 889
    },
    {
      "epoch": 0.9694989106753813,
      "grad_norm": 913.829345703125,
      "learning_rate": 4.6889373129022085e-07,
      "loss": 11.2009,
      "step": 890
    },
    {
      "epoch": 0.9705882352941176,
      "grad_norm": 1170.449462890625,
      "learning_rate": 4.3602332680331425e-07,
      "loss": 12.748,
      "step": 891
    },
    {
      "epoch": 0.971677559912854,
      "grad_norm": 877.0611572265625,
      "learning_rate": 4.0434479495378155e-07,
      "loss": 12.1814,
      "step": 892
    },
    {
      "epoch": 0.9727668845315904,
      "grad_norm": 1181.8524169921875,
      "learning_rate": 3.7385851496284374e-07,
      "loss": 12.2686,
      "step": 893
    },
    {
      "epoch": 0.9738562091503268,
      "grad_norm": 919.8130493164062,
      "learning_rate": 3.445648517793942e-07,
      "loss": 12.535,
      "step": 894
    },
    {
      "epoch": 0.9749455337690632,
      "grad_norm": 736.0946044921875,
      "learning_rate": 3.164641560756132e-07,
      "loss": 13.5968,
      "step": 895
    },
    {
      "epoch": 0.9760348583877996,
      "grad_norm": 1050.005126953125,
      "learning_rate": 2.895567642427488e-07,
      "loss": 13.5456,
      "step": 896
    },
    {
      "epoch": 0.9771241830065359,
      "grad_norm": 1240.2838134765625,
      "learning_rate": 2.638429983870983e-07,
      "loss": 12.4089,
      "step": 897
    },
    {
      "epoch": 0.9782135076252724,
      "grad_norm": 1174.1912841796875,
      "learning_rate": 2.3932316632614416e-07,
      "loss": 12.5838,
      "step": 898
    },
    {
      "epoch": 0.9793028322440087,
      "grad_norm": 900.91845703125,
      "learning_rate": 2.15997561584913e-07,
      "loss": 12.2009,
      "step": 899
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 1263.928466796875,
      "learning_rate": 1.9386646339238924e-07,
      "loss": 13.1211,
      "step": 900
    },
    {
      "epoch": 0.9814814814814815,
      "grad_norm": 1037.0902099609375,
      "learning_rate": 1.7293013667825098e-07,
      "loss": 12.676,
      "step": 901
    },
    {
      "epoch": 0.9825708061002179,
      "grad_norm": 1237.9251708984375,
      "learning_rate": 1.5318883206962842e-07,
      "loss": 12.2276,
      "step": 902
    },
    {
      "epoch": 0.9836601307189542,
      "grad_norm": 630.1217041015625,
      "learning_rate": 1.3464278588815048e-07,
      "loss": 12.1347,
      "step": 903
    },
    {
      "epoch": 0.9847494553376906,
      "grad_norm": 859.3702392578125,
      "learning_rate": 1.1729222014709162e-07,
      "loss": 12.4704,
      "step": 904
    },
    {
      "epoch": 0.985838779956427,
      "grad_norm": 1073.7818603515625,
      "learning_rate": 1.0113734254872942e-07,
      "loss": 12.5898,
      "step": 905
    },
    {
      "epoch": 0.9869281045751634,
      "grad_norm": 1251.8004150390625,
      "learning_rate": 8.617834648185774e-08,
      "loss": 13.251,
      "step": 906
    },
    {
      "epoch": 0.9880174291938998,
      "grad_norm": 1012.1276245117188,
      "learning_rate": 7.241541101945526e-08,
      "loss": 11.1789,
      "step": 907
    },
    {
      "epoch": 0.9891067538126361,
      "grad_norm": 662.820556640625,
      "learning_rate": 5.984870091654271e-08,
      "loss": 13.627,
      "step": 908
    },
    {
      "epoch": 0.9901960784313726,
      "grad_norm": 771.7218627929688,
      "learning_rate": 4.847836660824001e-08,
      "loss": 12.1929,
      "step": 909
    },
    {
      "epoch": 0.9912854030501089,
      "grad_norm": 654.5511474609375,
      "learning_rate": 3.8304544207945495e-08,
      "loss": 11.9869,
      "step": 910
    },
    {
      "epoch": 0.9923747276688453,
      "grad_norm": 1049.4482421875,
      "learning_rate": 2.9327355505681663e-08,
      "loss": 12.3512,
      "step": 911
    },
    {
      "epoch": 0.9934640522875817,
      "grad_norm": 861.9254150390625,
      "learning_rate": 2.1546907966685236e-08,
      "loss": 12.2443,
      "step": 912
    },
    {
      "epoch": 0.9945533769063181,
      "grad_norm": 1027.2105712890625,
      "learning_rate": 1.496329473008595e-08,
      "loss": 12.2233,
      "step": 913
    },
    {
      "epoch": 0.9956427015250545,
      "grad_norm": 783.8712158203125,
      "learning_rate": 9.576594607807465e-09,
      "loss": 13.0014,
      "step": 914
    },
    {
      "epoch": 0.9967320261437909,
      "grad_norm": 1203.1295166015625,
      "learning_rate": 5.3868720836236506e-09,
      "loss": 13.0101,
      "step": 915
    },
    {
      "epoch": 0.9978213507625272,
      "grad_norm": 702.3920288085938,
      "learning_rate": 2.3941773123814516e-09,
      "loss": 13.0705,
      "step": 916
    },
    {
      "epoch": 0.9989106753812637,
      "grad_norm": 858.6944580078125,
      "learning_rate": 5.985461193791509e-10,
      "loss": 12.3027,
      "step": 917
    },
    {
      "epoch": 1.0,
      "grad_norm": 1008.3307495117188,
      "learning_rate": 0.0,
      "loss": 12.5451,
      "step": 918
    }
  ],
  "logging_steps": 1,
  "max_steps": 918,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 230,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3120952812503040.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}